Not really what works on social media and as a fact, it cost me ~300 subscribers since the start of the year, who decided that my content was not for them as I did not want to participate anymore in this confirmation bias echo chamber X and Substack are slowly becoming. And it served me right - and you who chose to stay, as my stock picking delivered ~40% alpha YTD - which is pretty massive and you can follow it all for free here.

This article is just one more of those going against the crowd as trying to justify the current SaaS sell-off in an environment where most are saying those stocks are lifetime opportunities won’t make me more friends.

But I’m here to make money, not friends - although I’d love to make both. So let’s dissect what is happening on the SaaS sector, why, and how we will profit immensely from it, when the time is right - not before.

SaaS What?

I’m not sure everyone knows what a SaaS is so let’s start here as it is key to understand why they are dying. Software-as-a-Service.

Say, you have two ways to eat nowadays. The “normal” way - at least for me, is to go do my own shopping with my small legs and wallet, come home, turn on the gas, cut my vegetables or season my beef, cook it all and serve hot. The “occasional” way is to take the same legs but go sit at someone else’s table who’s going to ask me what I want to it, prepare it for me and serve me. And if I want salad instead of fries, or to remove tomatoes from my burger then sure, the cook will comply.

This is a SaaS; I do nothing but pay and consume. Someone else works at making it good enough so it is worth my money, and I just have to ask for what I want within the circle of competence of the provider - you won’t ask for caviar at McDonald’s.

Adobe is a SaaS, they allow me to download and use a software with built in tools, a certain degree of personalization but no control over the core code. I can use it myself and that’s about it, just like I can eat a restaurant’s food but can’t go in the kitchen to cook my own steak.

SaaS companies give you access to a controlled environment with pre-defined access, liberty and customization.

Most SaaS are meant to resolve pain points - which is key to why they are being sold off nowadays as AI is just better and cheaper at doing so, with the bonus of even understanding context. Some smart guys found out there was a strong need for a service which was complex or costly to set-up and built a generic system to resolve this particular one, sold it, improved it, expanded, etc…

Take DocuSign. The founders understood early the need to sign documents remotely, that moving our small legs with a pen on our pocket would become a pain point in the years to come; they built a service to sell trust online in the form of a document signing platform, integrated different document formats to heal their users’ pain points - and generated $3.22B of revenues FY25 by doing so.

Not bad, eh?

The entire SaaS sector relies on a handful ground rules.

1. Resolve a pain point many will accept to pay for.

2. Grow user base, revenues and margins.

3. Optimize and improve the service for retention.

4. Drink margaritas.

The SaaS sector is unforgiving before reaching the margaritas as everything depends on how fast it scales, retain users and prices its service. SaaS rely on massive volume demand to resolve a pain point and the value clients agree to pay for it to be resolved externally.

The SaaSpocalypse

This is the name given to the so called “silly” sell-off of the SaaS sector - which isn’t silly at all as I am about to explain.

The SaaS ETF, IGV, is down an astonishing 25% since January. We are talking about an ETF here, not an individual name; this should give you an idea of some of the losers’ performance… We’re talking about casual -50% YTD on Duolingo or AppLovin’, and more… Some portfolios will take years to climb back this performance.

A literal bloodbath cause by AI, with reasons.

Let’s start with a few words to remember what the market is about: safe and growing cash generation, right? You’ve heard me say this so many times now - details on this write-up.

This is how the market works. Understanding this helps avoid value traps; companies that look cheap but have no path to margin expansion, revenue acceleration, or gain the market’s trust. Without a catalyst, the market simply walks away. A value trap can become a winner, but it needs one of those triggers first.

We don’t buy those. We look for the names the market is about to fall in love with; companies that can accelerate growth or increase cash generation, creating a double appreciation: rising cash and expanding multiples.

That’s what stock picking is about.

The SaaS sector has been rewarded by high premiums for two decades because they were the only economically viable solution to structural pain points. If a company needed to sign 10,000 documents per day, it was cheaper to contract DocuSign than to develop and maintain an in-house software, just like it’s easier to pay a bit more to go to the restaurant than to cook sometimes.

This created an entire sector where winners had a strong, safe and recurrent cash generation. As long as the pain point existed, companies would rely on the company which solves it and as by design, successful companies grow, their usage of pain point resolution increase and SaaS revenue grow. With stable margins also comes increased cash generation, etc… A virtuous wheel which pushed the market to increase those companies’ premium while fundamentals were improving.

Take Adobe as an example.

The stock’s multiples followed free cash flow margins to perfection through the years, up and down alike, until… 2025. This is where we’ll need some critical thinking. Most stop at “what was will be again” and buy those stocks. This is the problem of fundamental investors: they look at the past only.

But the world moves forwards, thankfully, and what was could not be anymore. You do not ride a horse to go to work nor wash your clothes in the lake. What was eventually stopped to be when new technologies disrupted old methods. And that was for the greater good.

This is what the market is pricing now.

What is AI?

We’ve already answered that question many times from a technological point of view, but never from a usage point of view. Artificial Intelligence is an automated and intelligent tool which can be used for any kind of tasks, including… resolving pain points.

A versatile technology trained for everything whose progress has been baffling, especially these last few weeks/months. I mean, I am a nerd and most of you also are - that’s a compliment coming from me. We are aware of the new technologies and developments; heck, most of you read a 15min write up about photonics. Only nerds do that, most people I spend time with daily have no idea that the word “photonics” even exists.

I use AI every day now and see it improving each passing week, I see answers, services or corrections I was frustrated not to have six months ago finally happen. We all see Anthropic’s new Claude releases, monthly, shipping new functionalities like if they were socks, and how transformative they have been - notably Claude Code with most CEOs telling us than a significant proportion of their code is now written by AI.

I’d say maybe 20% to 30% of the code that is inside of our repos today... are probably all written by software.

Satya Nadella, Microsoft CEO

Over the past 6 months, I built 3 websites and ~100 scripts for automation tools for my investing, writing and research. I could never have done so much before. This made Anthropic the fastest growing company in history, reaching a $30B run rate in less than 3 years of activity while ChatGPT continues to scale and without including Gemini or other AI functionalities and compute demand.

AI is baffling and its progress is even more baffling, to the point that this week Anthropic CEO said they developed a security focused AI model called Mythos, which was so performant they couldn’t release it to the public for fear it would be too easy to use it to hack into current system and decided to team up with security specialists to avoid this situation and find a way to integrate this model for defense first.

This is AI. This is a massive structural shift improving at the speed of light. And you guys want to tell me that what was true a decade ago will still be true in a year, while Palo Alto, which developed the most powerful and comprehensive security service is now being pushed to its limit by a new AI model?

Cmon…

The Disruption Risk

I will steal the wise words of a reader to go straight to the point.

The market correctly sees the risks (regardless whether the risk is justified or not which is a matter of personal opinion), and, as such, is no longer willing to give them premiums as safe, stable compounders. Hence the stock price crashed despite the numbers of most earning reports are still looking great…

The sentence in bold is the most important sentence on this entire write up. As always in the market, personal opinions simply don’t matter, and they don’t make money - not by themselves. You can take 1 year or 10 years to realize this, but it will still be true even with AI pace of progress because the markets are about future’s expectations based on individuals’ sentiments. Unless AI can predict the future, the market’s structural behavior won’t change.

Back to our the point and why the current SaaS disruption is justified: because AI is a structural change. What was a large pain point might just become a 10 days of work using Claude Code and what was 10M daily users might become 5M as AI agents replace workers - or hire fewer workers/consume fewer licenses, reducing retention, recurrent revenues and therefore cash generation, even if margins stay stable. SaaS mostly resolve tech pain points and AI will reduce the friction to solve pain points.

As for the argument “Don’t be stupid AI cannot build Adobe’s software”, it completely misses the point of what a disruption and stocks are. We are not talking about replacing those companies or software, we are talking about reducing the need for them which means reducing their cash generation.

Once more: the market only cares about future and safe cash generation.

Obviously, SaaS won’t die. Today, the market treats them all equally because it has no capacity to distinguish the future winners from the future losers. In doubt, sell everything, reassess in a few months.

Even structural AI winners like Palantir are being sold off while AIP is exactly what the AI revolution strives to create: an AI centric software.

And we’d expect non AI-centric names to thrive? No. The market has no view on what the future looks like for those names - neither you nor I do. We all have bias, opinions, but there is a possibility for 90% of nowadays’ SaaS to be entirely disrupted and rendered useless by AI in the next 5 years. You might judge it small, maybe even use the word “ridiculous”. But it exists.

I would never have imagined AI become so performant, so fast. Who says the next years will not again bring exponential improvements? What then?

The SaaS Future

Today, the market doesn’t see a future for those companies. It knows they won’t disappear, everyone knows that, but it has no idea what to do with them and so it sells. Liquidity goes away as uncertainty is real, whether we like it or not.

Obviously, SaaS will survive. Some, at least, maybe with another form, business model or value added. I cannot imagine Palo Alto or CrowdStrike be disrupted by AI as their cybersecurity service is more than just an AI model, and their need is exponentially growing with the democratization of AI ironically. These companies will adapt.

First, they will need to become AI native. I’ll always remember Brian Chesky’s words during an Airbnb earning call in 2024 about how this wasn’t a choice, but a necessity, while it would take a few years to get there as AI applications are structurally different from normal software.

The founder-led companies and the companies that are prepared to change and transform are the companies that are going to benefit from AI, because AI means everyone changes. And if you don't change, you're going to be disrupted... If you don't disrupt yourself, someone else will.

This is the first complexity, while model are by design AI-centric, SaaS have to pivot from classic software to AI-centric model, which makes them late by design; it’s always more efficient to build a house from scratch than to remodel one while being careful to keep the foundations and some furniture.

Second, resolving a simple pain point won’t be enough anymore. SaaS will need to create value which goes beyond facilitating workflows as this will be done cheaper and faster in-house with AI now. The cost and time of developing and maintaining a DocuSign equivalent got divided by 50x and will be even more in a few years; companies can’t compete on pricing with AI agents. The standards are different now.

Third, the economics might need to change as AI is structurally deflationary. A business model based on low prices - which can now be competed with, and high volume - which will structurally reduce… Well, you can do the math. Again, not all SaaS are impacted but most, while the market has no clue on which.

The SaaS sector, or what it will become, needs to prove itself. It needs to prove the market its business can adapt to AI, leverage the technology to its benefit and accelerate user acquisition or monetization. If we take Adobe as an example, again… We’re not there. Some are doing well, no doubts there, but again: the market treats them all the same because it has no certainty what AI will be in five years and considering the recent pace of improvements, it assumes it will continue and could disrupt more verticals - and might be right.

The market cares about predictable and safe future cash generation and stocks trade on liquidity first, sentiment second and fundamentals third.

Liquidity is tight, sentiment is in the gutter as SaaS future is everything but predictable and their fundamentals - hence cash generation, is threatened. In the meantime, many sectors have large tailwinds and are worth betting on while the SaaS downtrend has no end in sight. It will end when most companies show that their AI centric service has a large potential, not before, and this might come in years for our legacy players, or in a few quarters for new ones. Cheap is a bullshit word in the face of a technological revolution, and being early is not a great position to be when investing, contrary to what most preach on social media - and their portfolio shows this today.

I will without any doubt be one of the biggest SaaS stock buyers in the years to come, and I’ll very probably make a killing on those names once the market sees the potential and corrects its mistakes. It will happen, but today is way too early to anticipate tomorrow’s winners - even if some look very tempting like the cybersecurity vertical. You don’t want to hear this, but most will become value traps. Let time filter those and let the market give signs of who will win the AI enhanced software era.

What was won’t necessarily be again.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

Would you rather have weekly or bi-monthly articles like this one, focusing on the names I select? Or would you prefer weekly access to the raw data (~10 names per week) so you can dig into the names I don’t highlight (which included big winners like Marvell or Iridium these last weeks on which I simply didn’t have time to write), but that still meet my uptrend criteria? I could share this in a section of my website or simply drop it every weekend in the group chat - or something else entirely.

Everything is on the table, I just want to make sure it actually brings you value, you might also be satisfied with the way things are right now. Any feedback is welcome!

Intuitive Machines

Ticker:

IM is about space logistics and lunar infrastructure. Their core business is to design, manufacture and operate lunar landers (Nova-C and Nova-D) to transport scientific and commercial payloads to the surface of the Moon. They basically manufacture the luggage that contains what “travelers” want to bring to the moon and goes inside the rocket - except in this case the luggage can move by itself and deliver your panties right at your hotel door autonomously.

This is the core but not the only competence they have, they also manufacture rovers, Earth re-entry vehicles and other pieces of hardware (propulsion systems, avoidance systems, autonomy landing software, lunar rovers…), communication solutions on or close to the moon - basically setting up internet up there, software necessary for any companies travelling in the region, and a few other smaller things.

If I were to summarize, IM is one of the key player on hardware/software requirements to explore the moon but also later on, exploit it.

Their business model today is to fund R&D and capital intensive missions with governmental contracts; they are the primary NASA partner for lots of hardware and software, with a five years $4.82B contract to set up communication and navigation services for missions in the region - the famous moon’s internet, and late March a new $180M contract for IM’s fifth landing mission for NASA’s Commercial Lunar Payload Services (“CLPS”) initiative. NASA financed the four missions for a total ~$390M.

IM-1 was the first commercial payload to successfully land on the moon ever in February 2024, and transmitted 350 megabytes of information during ~140h, a pretty large success even if the NOVA tipped on landing, preventing perfect conditions and only transmitted for a few days. Can’t succeed on the first try right?

IM-2 confirmed the technology’s capacity to land but did not land where it was supposed - ~250m away, and could not set its solar panel right ending in a power outage. Initial tests were successful, with more complex tests being cancelled.

IM-3 is supposed to launch H2-26 with no defined dates yet, with autonomous robots, radiation sensors, a lunar plant experiment, Intuitive Machines’ first lunar data relay satellite - to start building a commercial data relay network in lunar orbit, and more… And IM-4 is scheduled in 2027, launched with SpaceX Falcon for always more hardware on the moon.

I think the case is pretty clear already. The technology still requires fine tuning which can only happen with tests and IM is NASA’s preferred partner for those, with Firefly Aerospace as its main competitor as it is a wish for NASA to diversify - meaning contracts will continue to go towards both, and others. IM is the undisputable leader today and proved its technology already so as NASA intends to accelerate the frequency of launches and tests, IM will naturally benefit from more contracts.

This is our generation’s space conquest and it is a very serious one which the U.S. intend to win. If you wonder why, the answers are simply: the moon is filled of key resources we’d love to extract, its gravity and conditions make it an interesting geography for any kind of manufacturing (pharmaceuticals and semiconductors are already being manufactured in space with many advantages), plus the need to make humanity an interplanetary species.

These kind of missions and the global interest around the moon will only increase in the years to come. Becoming the go-to hardware manufacturer for moon exploration and later on exploitation is something, just like becoming the preferred launcher for commercial or touristic space travels for SpaceX or Rocket Lab is.

There is an entire economy and supply chain on the space exploration sector which is set to explode in the years to come - I could write on that if interested.

And as said so many time, with the markets being about liquidity first, sentiment second and fundamentals third, the space sector is a large focus for #1 and #2 while #3 are still speculative but also very real, with potential. With SpaceX IPO coming this year, the excitement around the entire sector is palpable and should only increase - at least until D-day, boosting #1 and #2 further. It is already happening to be honest with SpaceX expected to IPO ~$2T which is massive, Rocket Lab up 135%, PlanetLabs up 724% or ASTMobile up 323% since 2025. Space is hot.

Without any surprises, IM is pushing higher and meeting my criteria - growing volume on rising stock with breakouts, and Firefly Aerospace which is also a publicly traded company (IPOd in 2025 so chart is in daily timeframe) has the exact same profile. Same sector, same tailwinds, same momentum.

Both are on a pretty tough resistance right now ~$23 and ~$35, breaking those levels on volume would trigger a clear uptrend for months - hence giving a great entry, while retesting lower breakout ~$20 and ~$28 respectively would give a great risk reward accumulation point… Two really strong setups either way.

This is the bottom line of the trades, speculative but based on real fundamentals and potential, carried by a heavy momentum with SpaceX and the Artemis II mission, the push governments are giving to space exploration, etc…

Two positions I could personally take with a few months timeframe as I’d also expect SpaceX IPO to print a top in optimism and anticipation - too early to say for now but probable.

Sigma Lithium Corp

Ticker:

Much more classic within a well-known sector, a small capitalization still at ~$1.6B but with “less” execution risks and clearer demand for its products as we are talking about lithium - as advertised in the name. Sigma is a pure play on high quality lithium extraction, operating in Brazil with a few twists, transforming it from just another lithium player to a potential great opportunity.

First, the company certifies a “quintuple green extraction” (zero coal power, zero tailings dams, zero utilization of potable water, zero use of hazardous chemicals and zero accidents), which you might not care about but many companies do as they have environmental restrictions – the famous ESG regulations. This became important as most of the resource comes from China or Chinese operated mines and doesn’t answer these regulations.

As the west is working on independence from extraction to refining and focuses on ESG-approved commodities, Sigma will receive more demand - as my second point will confirm.

Second, the company signed a $100M collateralized bank guarantee fully collateralized by its own clients through a blend of corporate guarantees, letters of credit and export receivables to be mutually agreed. This is the equivalent of Meta and Microsoft pre-paying for Nebius’ compute before it is installed, highlighting the demand for its products and allowing them to invest healthily in their second Greentech plant to double production from 270,000t to 520,000t annually.

Healthy balance sheet and the financing to double their capacity – hence revenues, in the medium term. This comes with execution risks obviously but this is excellent news for the financial health of the company and its growth profile.

Third, as a pure play lithium extractor, Sigma is optimized with ~40% operating cash margins – pretty high for a commodity miner, and extracts the commodity for cheaper than competition - hence very competitive on the market.

Sigma is guiding to a $592/t extraction and shipping of their lithium, which puts them on the lowest extraction price on the sector - green position on the chart, while the red, blue and green lines are their profitability margins depending on lithium’s price, which as we’ll see has been rising and sits ~$2,000/t today.

Fourth, the geopolitical situation helps western companies with Trump’s tariffs forcing local production - Brazil isn’t local but it is close from U.S. factories which are being built to refine the raw material locally, while battery productions are already within the U.S. for some companies. As the continent works on building a supply chain locally - meaning without China, Sigma will become a logical go-to partner.

Natural resources are also on the rise for many reason - geopolitics and fears of supply chain shocks, raising oil being, refocus of liquidity… Raising resources are never great for an economy but we can benefit from it.

Lastly, the dollar falling is a big tailwind for Latin America comapnies as most of their loans – as confirmed with Sigma’s own loan, are denominated in dollars while their expenses to pay local workers and resources are in local currencies. I’ve wrote about this currency mix and how it impacts LATAM on MercadoLibre’s thesis.

In summary, Sigma is an optimized pure play with a very healthy balance sheet, an accelerating growth profile, healthy cash generation with an advantage with its ESG product compliance and cost competitiveness, while the global macro is a tailwind – less competition from China combined with cheaper operations in Brazil.

To put some perspectives, Albemarle Corp – its main competitor, has already ran ~170% over the last year without explosive news and while having a less attractive financial profile – but being a much bigger diversified company.

The second tailwind for the industry is the global electrification happening around the world now while lithium is the most important component for batteries – which are globally getting cheaper and more efficient for households and businesses alike. If we look at Tesla’s success with Megapacks and the rebound for solar installations for households, on which I am overly bullish and is pushed by the Iran war, then demand for lithium should continue to increase, and if the commodity is also more expensive, profits for producers simply accelerates.

And both Albemarle and Sigma charts show this kind of anticipation by the market. As usual for the latter, the same chart patterns as always.

Now sitting below a strong resistance which creates the same condition than both IM and Firefly Aerospace: a breakout on volume triggers a buy on the retest as the trend is starting, while a rebound on the W50 triggers accumulation anticipating for a potential breakout.

If we start looking at valuation, the stock looks more expensive than its peers but it also has higher cash generation, healthier balance sheet, a premium for the potential revenue growth, its ESG compliance, accelerating demand for local supply chain and cost competitivity, with the risk being execution risk for its expansions.

Another name I would gladly buy on the right price action conditions.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

We will start by the need and the technology itself, what it is and why is their a need to upgrade our datacenters, and then go over the entire supply chain to have a global overview of the companies involved in this new technologies, their role in the supply chain and how they interact with each other.

The sector has received lots of liquidity and stocks already returned hundreds of percent, but as this technology could be the next foundation for scaling inference, the opportunity wouldn’t stop after those returns and we should be ready for when (not if) the breathers happen.

The opportunity is clear, the technology is backed by serious capital and people, the speculative part of investing, betting on if the technology will be adopted, is behind us and what lies ahead is scaling - with order books and revenue growth. Analysts have no idea on what to expect and this is when the risk reward will be the best - on the breathers happening after fundamental confirmations but before the first financial ones.

This is what makes the opportunity so interesting today; while it is just starting, while analysts have no idea of what is coming, while most of these companies are still pretty small and ignored by the biggest players.

Late enough to avoid all the risks. Early enough to reap more than enough returns.

Physics & Bottlenecks

Before going over photonics themselves, we need to review some concepts we’ve already seen - like datacenters or GPUs, which I’ll go over again briefly.

Datacenters are the interconnexion of thousands of hardware, each with its own specialty; a hardware is the interconnexion of hundreds of semiconductors. Data transits through them via electrical or optical “cables” between them in the datacenter or conductive metals - copper as the cheapest conductor to scale, within the hardware.

Artificial Intelligence is this famous new technology, born from data and compute. A model learning from billions of iterations to guess the next token - which could be anything from a letter to a pixel passing by an action, each mistake forcing the model to update its parameters during training by adjusting weights in how the next token is guessed, until reaching satisfaction. AI learns just like you do, by repeating a task and adapting as your mistakes teach you how you should do things, until you get it right - until you are satisfied at least.

For decades, the AI bottleneck was compute, the question being how do we treat so much data, how do we guess the next token fast enough so a simple task like writing a one-line text doesn’t take weeks?

This was resolved by GPUs, semiconductors designed to do parallel computing and behave like a hive. Thousands of GPUs would be allocated to one task, each only working on a fraction of the problem, be fed the required data to solve it and share the result with the hive. This new computing method, GPUs parallel computing instead of CPUs sequential computing, allowed model training with human timeframes.

GPUs are made from specific semiconductors, optimized and small. AI democratized, its potential became very clear and demand for compute rose exponentially, reaching a point where physics were limiting its deployment. This created an entire ecosystem, some would call it a circular economy, with many large winners in the stock market, as you all already know.

Then came inference, the usage of those models. Musicians learn music theory early on, but eventually they need to pick up their instruments. Imagine you have trained all your life and your final exam will be someone - let’s say me, giving you a song. Any song, with the required music sheet, and you’ll need to play it well enough so I’m satisfied. This is inference, the model knows, now it has to be used for what it was trained - whatever it is from driving to generating images.

The issue with inference is, it is limited. The model itself and its capacities aren’t, it could answer perfectly to your needs - assuming it is trained perfectly, just like you will take your guitar and play me Sweet Home Alabama if I give you the music sheet. But imagine if 10M people gave you a different music sheet at the same time and asked you to play them all, simultaneously. You couldn’t.

When a model is going live, if usage grows exponentially, the resources allocated to that model have to grow as well. Just like you’d need to hire more workers if your workload was 1,254h/day. Except, more workers wouldn’t be your only need in term of logistics. No, you’d also need roads large enough for your materials to reach your workers, a parking lot large enough to unload those and food for everyone.

AI isn’t just about stacking GPUs together, it is about data computing. You’ll need the workers (GPUs), you’ll need the data (materials), you’ll need the parking (memory) and you’ll need those wide large roads for your materials to pass (bandwidth). You’ll also need everyone to be there on time so they can work (latency). If one is missing, well… Your task will be delayed, maybe some of your workers will just spend their day playing cards not just because they’re lazy, but because they simply cannot work. And you don’t want to pay your workers doing nothing, right?

All this introduction brings us to today’s actual bottleneck with AI. Inference is reaching exponential real world demand growth, and we realize that more workers isn’t the solution anymore – we do need more but that won’t solve our problem. No, our biggest problem today are the parking, the punctuality and the highways.

Today, we’ll only talk about those damn highways.

What Are Photonics?

Photonics are the science behind light generation and manipulation. On this write up, tech focused, we will consider them as optical semiconductors, chips designed for optical connectivity within the hardware, supposed to replace copper which is now physically reaching its limit for our needs. Without going into details, copper highways simply aren’t big enough for our materials to circulate on them and be on time, combined with a much lager energy consumption that light. There’s no solution with those highways, we just need to change them.

Enter optics, light, photons. Now those would answer our need in term of bandwidth and latency, simply because light travels faster than anything else while generating less heat than conductive copper - or the amount of hardware needed to push copper to its limit and deliver the bandwidth and latency we need today. It isn’t a matter of preference but of physics.

But using photons isn’t easy, nor cheap, as it comes with many more requirements and complications than copper, and creates two very large verticals to work properly, for those photons to be generated and travel where we want them, how we want them. Two verticals, many more opportunities.

1. A photon generator - a laser. To guide light, well… we need light.

2. A semiconductor to guide light, called SiPho - for Silicon Photonics.

On their final forms - in a few quarters time, hundreds of both will be combined together into a final usable product to transmit data through light within hardware: Co-Packaged Optics.

Today, we will go over this new hardware which should be responsible for the next phase of scaling AI datacenter by improving our current deficient highways; what it is, how it works, its supply chain, manufacturing process and key players in each vertical, so we understand why it matters and why the opportunity is real - and massive.

Here’s is an overview of where we’re going.

SiPho: Silicon extraction → SOI wafers (Soitec) → EDA simulation (Synopsys) → SiPho engineering (Broadcom, Marvell) → Foundry etching (GlobalFoundries, Tower Semi, TSM) → Testing and validation (Aehr Test Systems, Teradyne, FormFactor).

Lasers: Indium/Phosphorus extraction → InP substrate (Axti, Sumitono) → Epitaxy (IQE, Coherent, Lumentum) → Laser Chip Fabrication (Coherent, Lumentum, Sivers) → Testing and validation (Aehr Test Systems).

CPO: Final assembling (Fabrinet).

The Opportunity.

We understood the global physics and the bottleneck, so before diving into the specifics let’s talk about the opportunity and why I am writing all this here after sending you an article few days ago about a potential bear market.

But as I always say: there is always a bull run somewhere.

Even during weak markets, companies continue their operations. Cash flow is still generated and spent. AI hyperscalers and compute providers in general have commitments for hundreds of GW of new datacenters over the next ~5 years. Buildouts could slow and some could even be cancelled but the net increase in total usable GW from today to 2030 would still be massive, and photonics are expected to be ready within hardware by next year and play a large role in those new datacenters, scale AI further, which means it should reflect in photonics companies’ revenues.

This isn’t just a narrative or a maybe. Key players in the industry are investing in this including Nvidia. Jensen Huang not only confirmed that CPOs were the only viable solution to scale AI further but also invested directly - through Nvidia, with $2B investments/partnerships with Coherent and Lumentum to accelerate R&D and manufacturing on InP lasers - we’ll detail what they are later, and $2B more with Marvell to strengthen the NVLink networking system with photonic solutions.

Sure, $6B isn’t much for Nvidia and its ~$30B quarterly FCF but it shows the direction and that management is putting money where its mouth is, while Nvidia is the company with the clearest view on AI scaling needs so pushing $6B on a technology is signal, not noise. They did the same a few years ago with networking companies and hardware manufacturers to push GPU capacities further for training - and they were right to do so.

Here are Jensen’s words at the latest Nvidia GTC.

The AI factory is the new unit of computing. To connect a million GPUs, we can no longer rely on the old physics of copper. Silicon photonics is now central to the next generation of AI infrastructure.

By integrating silicon photonics directly into our switches, we are shattering the old limitations of hyperscale networks. We are moving from a world of pluggable parts to a world of integrated light.

We have squeezed every last drop out of copper. At 224G per lane, the physics simply demand light. The transition is no longer a choice; it is a thermal necessity.

The technology is ready, but the supply chain is not yet at the ‘Industrial Scale’ required for zero-defect AI factories.

Nvidia GTC keynote

The bottleneck is real and specialists of the sector believe the solution is photonics. As hyperscalers expect a combined ~$0.5T spending in AI buildouts next year, it is fair to assume that part of it will go to the first usable photonics technologies, or at least their engineering while spending shouldn’t slow down by 2027 - even if it doesn’t increase, while spending on photonics should increase as the next technology to scale AI further.

I mean, big spenders with only one new technology to move their capacities forward points to a pretty obvious investment path, but who knows? I could be wrong.

And if we look at the sizes and revenues of these companies, we see that they wouldn’t need much spending to be transformed from small caps to key AI players.

We are looking at a sector whose combined market cap is lower than the yearly CapEx expected from hyperscalers this year, and total cumulated revenues are lower than $70B, while some of those have a core business outside of photonics which means the photonics related revenues of this sector are even lower.

Don’t read too much into this, this is just to illustrate that the sector is still relatively small compared to the opportunity once CapEx flows down onto photonics - although obviously the entirety of the $0.5T won’t flow to optics, but even 0.5% of this total spending would be material for this sector.

There is no knowing exactly where the industry is heading. There’s lots of optimism and anticipation in the stocks today, with most of those names up hundreds of percent over the last few quarters, but there is a reason for such a performance and no saying where it stops. Some of these companies are still small caps, ignored by most of the large players because it is too early in the cycle to pour hundreds of millions on them, and analysts don’t even know half of them - even less their potential.

If you were to look at analysts’ expectations on Nvidia, they were wrong almost every quarter - still are today. Why? Because it was impossible to understand the sector and anticipate demand for their hardware and the spending frenzy which was unlocked over the last years. Some things cannot be modelled.

They aren’t stupid, they are simply playing a game which cannot be played as Nvidia - and AI, is an irregularity in the markets, a company and sector you only see a handful times in a lifetime, but it isn’t the only one and even if photonics names ran 500% over the last few quarters, nothing says they have reached their potential. It took Nvidia 1 year and a half to return 500% after being known as a key AI player and delivering its first GPUs to OpenAI in 2016. The stock is up 3,000% since.

I wouldn’t expect photonics to become Nvidia. But if photonics were indeed the next foundation for AI scaling, analysts won’t be able to anticipate growth, profitability and cash generation just like they couldn’t with Nvidia, because demand will be faster than they’d expect it.

Massive past returns do not mean future returns are unlikely. It means we missed the discovery, the riskiest part of investing where the potential is here but unconfirmed. It now is confirmed by Nvidia’s endorsement, while the potential is still unknown. If it were to be as transformative as expected by the market and Jensen himself, then the safest part of the gains are in front of us, with fundamental confirmations and stocks which should be lifted up by revenues, growth and cash generation.

Everything the market likes.

The SiPho Supply Chain

SiPho: Silicon extraction → SOI wafers (Soitec) → EDA Simulation (Synopsys) → SiPho design (Broadcom, Marvell) → Foundry etching (GlobalFoundries, Tower Semi, TSM) → Testing and validation (Aehr Test Systems, Teradyne, FormFactor).

Silicon Extraction & SOI Wafers

Key player: Soitec.

Wafers are the foundation of all semiconductors, the platform on which elements can be etched or deposited. You can think about them as your plate during dinner, on which you will serve your dinner - except your food isn’t microscopic hopefully.

I have a detailed write up on Soitec, its business and its importance in the supply chain. They manufacture SOI wafers - a very specific wafer architecture optimized for photons, which are the industry standard for SiPhos, and not only do they manufacture them but they also own the patent for the industry standard technology to do so.

They are the first bottleneck of the «chain: every SiPho requires an SOI wafer which will either be produced by Soitec or by its partners who pay a fee to use their proprietary technology - except if buyers accept to pay more for another provider not using Soitec’s technology but that would be less scalable.

Investing wise.

As noted in my detailed review, Soitec is in the sweet spot to start an investment in my opinion - I own it, after a clear W50 breakout on massive volume after years of downtrend as its core business declined.

The market is ignoring its key position on the supply chain and focuses on the company as a whole, not on this SiPho position and revenue growth potential, which could eclipse the revenue decline in other verticals during the next quarters.

EDA Simulation

Key players: Synopsys.

The software providers. They do not design the chips, they build the complex digital physics simulators (Electronic Design Automation) that the architects will use to design the chips. Every single chip is first designed and tested within a software environment before sent to manufacturing, to theoretically prove the viability of the design.

Investing wise.

Synopsys has been part of the entire AI revolution as its software isn’t exclusively optic related which means its core business and revenue growth have already been rewarded in the past years. It remains a key player in the industry but I wouldn’t focus on it.

SiPho Design

Key players: Marvell, Broadcom.

The architects. Companies with the engineering capacities to design how light should interact within the SiPho to reach its maximum potential. They receive requirements from final users and work with them to design the most efficient SiPho for their needs, just like a house architect would design the best house based on your demands.

For the sake of clarity, while Marvell is only an architect - and specializes in optimized connectivity which is why it is such a key player in SiPho, Broadcom also has manufacturing capacities. Just to be thorough.

Investing wise.

Just like Synopsys, both have been part of the entire AI revolution as their design aren’t exclusively optic-related, and both have been rewarded although Marvell was left behind as its core focus is networking - despite other verticals, which might give the company a push now as they will receive orders to fit optics within hardware.

The stock saw a healthy breakout on large volume post Nvidia’s investment and any retest ~$100 seems like a great entry.

This would be my focus if I wanted a stable giant with upside, ignoring the small-cap risks; Broadcom is retesting its W50 this week which is always a great entry during uptrends, but at a much higher valuation due to its partnership with Google for TPU designs. But a W50 is a W50, all great companies deserve to be bought at this level and any optics-focused partnerships - especially for Google’s TPUs, would push the stock even higher.

Foundry Etching

Key Players: GlobalFoundries, Tower Semi, TSM.

Just like Soitec, I have a detailed write-up on foundries.

Those foundries will etch the required components from the SiPho designs - waveguides, photodetectors, modulators, etc… on SOI wafers. These companies own fabs around the world and manufacture the finished product based on specifications.

Investing wise.

Tower Semi is up 290% since 2025, boosted in 2026 by Nvidia’s partnership, and is extended now. The opportunity is real but I wouldn’t touch such a chart, especially while they are being sued by GlobalFoundries for using patented technologies.

Both companies are key for optics and one is much cheaper than the other while having comparable potential. Just like Soitec, GlobalFoundries is dragged down by its core slowing business as it isn’t a SiPho pure-play, and is a much bigger company than Tower.

To me, GFS is the right play today with the perfect chart signaling accumulation at low valuation with a perfect retest of the W50 given this week ~$42, the price I highlighted in my review.

Tower would be a great name to own after a breather and TSM, which is obviously a great name, is also a major player for every semiconductor and a great opportunity but not a company I would focus on personally as I’d rather look at higher potential and volatility. As with Broadcom, a great focus to limit risks and volatility.

Testing and validation

Key players: Aehr Test Systems, Teradyne, FormFactor.

Once a semiconductor is finished, it must be tested as it won’t be shipped/paid for without confirmation of its performance. These companies sell the go-to systems or/and the components to test SiPhos components - and lasers as we’ll see later. They do not necessarily perform the tests themselves but usually sell the hardware to do so.

Investing wise.

AEHR will be the best play fundamentally as the most used hardware to test optics and SiPhos (with wafer-level burn-in), and the market rewarded it. As with any uptrend, buying the W50 is key and AEHR gave many retests over the last months, waiting for another one is the logical next step to me.

Teradyne and FormFactor are both extremely extended at the moment, patience will be key on those to avoid taking the elevator down and seeing a position destroyed because of FOMO. But those three names would be great opportunities at the right price be exposed to the SiPho testing sector which should also see demand grow as production increases.

The Laser Supply Chain

Lasers: Indium/Phosphorus extraction → InP substrate production (Axti, Sumitono) → Epitaxy (IQE, Coherent, Lumentum) → Laser Chip Fabrication (Coherent, Lumentum) → Testing and validation (Aehr Test Systems).

I will not go over the raw material extraction as most of the production comes from China and the refinement of the material is then done during substrate manufacturing, so we’ll start there right away.

InP Substrate

Key players: Axti, Sumitomo

This is the name given to a wafer made from a different chemical (Indium Phosphide). Concretely, it remains a wafer, optimized for laser usage.

We will focus on Axti, which is comparable to Soitec. InP substrates are the industry standard for semiconductor lasers - yes we are talking about semiconductors which produce light (photons), Axti owns patent for specific process within the InP industry standard manufacturing technology called Vertical Gradient Freeze (VGF) which allow them to yield low-defect wafers and therefore cost competitive. And while this is their most important advantage, it isn’t the only one.

They also own the raw material extraction vertical through subsidiaries in China which gives them privileged access at a preferential price. Their operation in China aren’t just material extraction, they also manufacture InP substrate there which reduce costs massively and increase its competitiveness and their manufacturing scale.

Axti is the only InP substrate vertically integrated pure play and the very best in the entire market, meaning most of lasers demand for SiPho optical connectivity will pass through them at one point due to their prices.

Investing wise.

Without much surprise, it comes at a premium but this is where my speech about Nvidia and analysts’ expectation comes into play. The stock is up 2,222% since 2025 to a ~$3B market cap for ~$90M of revenues. It is extended, very much, but it has everything we are looking for: massive volume and rapid expansion on strong potential and fundamentals.

I would not buy here, but it is and will be a leader in photonics, one of the key players and most important company to scale optics within hardware, without any doubts, while having massive exposure to… China. Volatility will happen and as usual: the W50 is key in uptrends.

Massive past returns do not mean the run is over, it means we should be patient and wait for those retests which always happen.

Epitaxy

Key players: IQE, Visual Photonics, Applied Optoelectronics, Coherent, Lumentum.

We have a blank InP substrate, a perfect, flat crystal which cannot generate photons by itself. We need something else, a phenomenon called electroluminescence; epitaxy is the technological process to achieve it, by depositing microscopically thin layers of crystals on top of our InP base, called “Quantum Wells”. The finished product will be called an epi-wafer.

That’s where physics is amazing and our understanding of it is mind-blowing.

Here’s how those wells work. Electricity will be conducted within the hardware and electrical impulses will enter the quantum wells, which are designed to change the energy state of electrons and force the release of a photon. This is how a “simple” well generates light from an electrical signal.

IQE and Visual Photonics are the epitaxy leaders, I added Applied Optoelectronics, Lumentum and Coherent because all three also do epitaxy - but not only as we’ll see below. When it comes to pure epitaxy needs, a fab to transform InP substrates into epi-wafers, IQE is the western leader.

Investing wise.

It won’t surprise you if I tell you that IQE is up 382% year-to-date I guess. And yet, compared to the rest of the sector this isn’t much as the stock is only up 122% since 2025, while being the go-to company for epitaxy. The stock has been on a downtrend since 2017 as the need for epi-wafers was very niche and never met expectations, until this year, with demand for lasers growing.

If the thesis of this write-up is right - that photonics are the future, then epitaxy is a mandatory step to scale it. The question is: will it pass through dedicated fabs or vertically integrated ones? In the first case, IQE has a long way to go, although the company has other structural issues that any investor should look at before investing - this write up is focused on the photonics supply chain not individual financial profiles.

Laser Chip Fabrication

Key players: Coherent, Lumentum, Applied Optoelectronics, Sivers Semi.

The last step of the chain, which consists of many different actions including etching waveguides, cleaving the wafer into individual laser chips, adding conductive metals, etc… This is the equivalent of the SiPho’s foundry stage, once the wafer (epi wafer in this case), is ready and the final touches are missing to manufacture the finished product meeting clients’ specifications.

We can find names we already know; those companies are vertically integrated, they do their own epitaxy from the InP substrate, they work on the design and the final touches on the wafer - they do not need to pass through IQE or other fabs, they do it themselves.

Once treated by one of these companies, the InP substrate leaves the factories as a finished and functional semiconductor laser which can then be packaged with SiPhos to offer a complete optical connectivity within hardware.

Investing wise.

As you can imagine, those are loved by the market as vertically integrated on one of the most important sectors to develop and scale photonics. They are extremely extended today and could remain like this for long, but they will be key in the development of photonics and any retest on the W50 would be a buying opportunity as long as it happens on healthy fundamentals - not the rejection of their technology.

Testing and validation

Key players: Aehr Test Systems, Teradyne, FormFactor.

We already went through them as they propose complete testing systems - which include both the lasers and the optic channels.

CPO Packaging

Key players: Fabrinet, Coherent, Applied Optoelectronics, Lumentum.

We’ve seen all the key companies required to manufacture both SiPhos and lasers, the only step missing is to combine both technologies into a packaged solution capable of delivering a customized optical connectivity within hardware.

The industry will rely on pluggable optical transceivers for the next quarters until the final product, Co-Packaged Optics (CPO), is ready to scale but both will use the same components so the supply chain is identical.

Packaging is probably the most important step of them all as electrons don't care if a copper wire is slightly crooked, but photons will miss the target if a laser is misaligned with a silicon waveguide by even a fraction of a micrometer - and the device will be useless.

We find our three leaders once again - vertically integrated laser semiconductor manufacturers as advertised. The only step they do not perform is the creation of the InP substrate, which they will always need to purchase. But we also find another company: Fabrinet.

The known giant and the only technology-agnostic one as the company receives the required products/specifications and assembles them as required by their final client. This technology-agnostic giant has the advantage of flexibility.

Investing wise.

The stock has been rewarded for its packaging services globally, up 142% since 2025 as the company is much larger and exposed to more optical verticals than just these new technologies. Another large and stable company to invest in for those who want to avoid small caps.

Conclusion

The photonics sector is just starting. Many of these names, which are still ignored by most market participants, will accelerate once real revenues and continued orders flow really kicks in. We’ll need to buy them at a correct price but the thesis and the signals are strong, execution is the only missing link to have successful investments.

This write-up was designed to give a complete overview of the entire AI optical supply chain, from raw physics to hardware, from blank substrates to pluggable transceivers and CPOs. The complete supply chain to replace our congested copper highways with light/photons.

Keep in mind those companies are part of the same sector but all have their particularities and financial profiles, their own risks. Depending on whether you want to invest in them long-term or simply trade them by buying key support levels and moving averages you will need to do your own research and follow your own process. I am here to share the ideas and the logic behind the narratives only.

Personally, I intend to wait for key supports on daily/weekly charts. I intend to take small positions with a trading mindset - clear entry and exit points, while waiting for deeper breathers. Hopefully, we’ll soon see pullbacks on their W50 where, if the fundamentals stay as strong as today, I’d build heavy positions in those names and ride the entire photonics wave - while I already have large positions in GlobalFoundries and Soitec.

I hope this write-up gave you a bit more clarity on the supply chain, the key actors and the fundamentals of this massive new AI vertical, and that you will be able to leverage this information into your own systems.

There’s always a bull run somewhere.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

I didn’t expect 2026 to be a good year for most of the market and shared it extensively, especially for the stocks most of social medias were looking at, and the last months have proven me right with most of them strongly sold or really destroyed in some cases - losing 50% plus of their value in some cases.

And finally, this month, the S&P500 broke its weekly 50 on growing volume after months of weakness.

I’ll speak bluntly and as I see it: this might be the start of a bear market. I struggle to see why the market would push higher and will assume that any coming bounce is nothing but a bounce, an exit liquidity for traders. I will not take part on most of the market activity until I see strength again.

This is the main subject of this month investment plan. Why do I believe this and why am I being even more cautious now.

Liquidity

As shared on my February review, the single most important factor to a bull run is liquidity. This is what most investors who focus on fundamentals and intrinsic value only misunderstand. It isn’t because you sell the best bananas on the world that you’ll be billionaire. If no one has the money to buy your bananas… Then you don’t make money.

If no one wants to buy your stock… It won’t go up.

Now fundamental investors will ask “Why wouldn’t I want to buy Meta at 20x earnings? It is so cheap!”. No, it isn’t. Valuation is relative, 20x earnings is cheap for someone who has money to buy it, but it isn’t necessarily everyone. As long as most think that, the stock will go up but the opposite will also be true. There is a factor of possibility first - some cannot buy, and a factor of wish - some don’t want to buy. The first is mathematical, the second is sentimental.

Mathematically, global liquidity is shrinking with constant - not increasing, fiscal policies, stable interest rates for longer around the world and no real stimuli, while the current geopolitics are shrinking liquidity even more.

The situation is Iran isn’t improving at all and I believe lots are losing faith in Trump and his institution as the messages are changing everyday from total annihilation to let’s make peace tomorrow, the latest - at time of writing, being “we’ll get out and let Hormuz close” while also sending more soldier in the region.

Inconsistent, growing uncertainty, crushing will to invest - but that’s about sentiment and we’ll dig this after.

On liquidity, the situation impacts oil’s supply chain which reduces capacities to invest because of global inflation. If prices increase, everything we pay is more expensive leaving us with less money for other activities, purchases or investments, and what is true for individuals is also true for companies.

I said last month I wouldn’t be worried until we see weeks of disruption without short term potential resolution.

Still, it is impossible to know what will happen. You know what you lose, never what you get. The removal of Khamenei doesn’t necessarily means peace and prosperity from Monday. It remains to be seen who will take the lead and what their objectives will be; if its army commander or someone with less restrain takes command, it could lead to a long fight.

We still have uncertainty. Maybe less worries than if Khamenei held the reins. Nothing is certain. As I write this - Sunday lunch in Europe, missiles are still flying, commands of the Iranian regimes just went to the next one and the Strait of Hormuz is blocked. The question is can this be sustained and for how long without a stable decisional structure in Iran.

There’s no point trying to predict Monday’s open, it will depend on what happens in the next hours and we could have another very eventful week end into a non eventful week. Or an eventful week. I talked often about weakness in the market lately and that could be another excuse for sellers to continue their biding.

We are at a month of disruption now, effects on the economy are real and won’t be fixed that easily even if we were to reopen the canal today, and no countries will be immune to the consequences. Even America who has no/very limited exposition to Middle East oil will see inflation due to globalization. Any product manufacturing process goes through clear steps.

Material extraction → Transformation → Manufacturing → Commercialization.

If any of these step get more expensive, the final product will be more expensive and as everything - most things, in the U.S. are imported then the life cycle of their products manufacturing will also be impacted by oil spikes.

Global oil disruptions affect the entire world.

Some places will be worse than others – notably countries getting their oil directly from the Middle East as everything will get more expensive there, but no place will be safe because of that globalization effect, while oil will also get more expensive even from other geographies as the total supply shrinks which increase demand on a shrinking supply.

If China bids for Canadian oil above what the U.S. pay for it, Canada will sell to China at a higher price or the U.S. will have to compete for it. All those dynamics on a mandatory resource for every country will generate inflation, already has and it will only get worse each passing day now.

I see it in Portugal and in Europe already in my daily life, gasoline prices are up ~20% compared to a month ago - this is real life not data set. Imagine the impact on a family who drives 40km every day. And imagine at the size of a business which uses oil daily for producing a toy, food, clothes or electrical services.

If every household/companies expenses in Portugal rise ~20%… Money spent cannot be spent again, which means lower consumption, less leisure and spending, less investing, even if stocks are “cheap”. Can’t buy stocks if they have to buy gasoline. that phenomenon is also true for companies - who could buy back stocks or invest in their operations; and that will happen worldwide.

Inflation is a vicious cycle. Less money, consumption, investments lead to decrease in revenues for companies, earnings, investments, less liquidity which in turns reinforce the circle.

You’ve heard me say often that the market works on liquidity first. Sentiment second. Fundamentals third. We’ve covered liquidity, why it dried and why it’ll continue to dry.

Sentiment

Our bigger problem lately is the wish to buy.

Individual investors and institutions need a reason to buy stocks and if you believe “Meta is at 20x earnings!” is enough, well… You’re wrong. Institutions and big players don’t invest that way, they have frameworks/conditions for the bulk of their liquidity, they are risk averse.

Leaving in our social media bubbles it looks like everyone is looking at earning multiples compared it to historical; but that’s not investing, no serious players work that way. That’s wishful thinking that the future will be identical to the past, and it never is.

The market needs optimistic catalysts to move higher, buyers need to see a positive future to spend their cash - which is needed for other things today, on stocks, they need to believe that tomorrow’s prices will be higher than yesterday. Why would you invest if you expected tomorrow to be worst?

Stability and certainty is what the market needs.

For quarters, the market had things to look up for. CapEx spending, slowdown of inflation, return of consumption, AI and its benefits in costs/growth, plus everything this new revolution brought, rate cuts, tariffs optimism, etc… There was always a reason for the situation to get better, clearer. Most of those are behind and I struggle to see what could push optimism forward in today’s situation while data are going to show inflation spikes - this isn’t a maybe anymore.

A gloomy situation with falling stocks increase negativism, triggers profit taking, making the situation more gloomy, with real financial consequences on companies, triggering more sales, pessimism, etc… Momentum is a self reinforcing mechanism which starts and ends in extremities, and we might have seen the bullish extremity.

There’s nothing to look up to, except maybe the retreat from Iran without victory and the reopening of the Hormuz straight but even then, liquidity is hurt and inflation will happen. The only question is how much - more each passing day?

Why take risks in such uncertain times without clear positive catalysts?

Liquidity first. Sentiment second.
Both are in the gutter.

So What?

Look, if you read me since some time, you know I am not a doomer. I am an optimistic and I love to invest, but I also know how to call a cat a cat. We can call it a dog if we want but it still won’t bark.

The situation is shit and even an end to the war in Iran wouldn’t fix it. Damages are done, now let’s see how much damages will be done and how the market will react. Of course good news would trigger optimism but as I said: from now on, every bounce should be considered as nothing but a bounce, an exit liquidity for traders.

The silver lining is that we have systems.

Good systems, which work in any kind of environment and protect our capital while forcing our focus on strength. My view of the market has been pretty spot on for months now and my stock picking reflects this with a large overperformance YTD despites a tough environment where many got destroyed - X is full of -50% portfolio. This is because of my system, helping me to cut losers and focus on winers.

Contrary to what social medias preach, DCA in bear market isn’t the solution. Patience and focus on execution is.

If there is one write up that you have to read on my entire Substrack, it would be this one. I believe this methodology is what will set winners and losers in term of returns over the next months - it already has those last months. You don’t want to be early during a downtrend

Strength will come back, and as our favorite stocks are falling, opportunities also will. We don’t know when nor why, but seeing stocks like Meta, Palantir or AsteraLabs fall means I’ll get to buy them later on.

So I’ll have the same conclusion than last month, while being more pessimistic myself.

I have no conclusion to give during uncertain times and as much as I like to share the situation as I see it, the only thing we can do as investors is follow our system, here to protect us from this kind of situation. If we bought great names at great prices and in great setups, we will be alright. If they break their setups, those names were maybe not as strong as we thought and that is fine. Decisions can be taken then.

Follow the system. Don’t play Pokémon. Preserve capital.

This is what I will be doing for the months to come. Sit tight on my positions and the few sectors which behave well, cut if they stop behaving, observe my watchlist until green lights turns on, with uptrend starting on large volume, so I can go all in as the market finds a new wave of optimism and liquidity.

Portfolio & My Focus

Even during a global bear market, there’s a bull run somewhere, opportunities to catch, sectors to dig and things to learn. There’s always an asset going up, even if that assets is the dollar.

We’ve seen in over the last months with all the opportunities shared on this Substrack and their performances. If you follow me since ~6 months, most of the thesis you received were outside of tech and performed wonders (Nutrien, Darlings, Smith & Wesson, Halliburton, Schlumberger, SolarEdge, Huntsman…).

Not all of them did, of course but discipline with a clear system cut losers fast and let winners run, resulting in a 34.5% YTD alpha on the market, all shared publicly and for free on my website.

This isn’t my personal portfolio, it is an equal weighted portfolio of all the thesis shared on here following my investing strategy - buying the W50 breakout retest with high volume on strong fundamentals and closing the position if the stock breaks it on volume for a handful days.

This website is the visualization of my stock picking performance.

And I’ll simply continue to follow my strategy, boringly, with a clear focus on strength which still exists today, in photonics first, supported by continuous CapEx even during uncertain economics, the energy vertical is also obviously very strong while many other defensives are also doing great - materials namely.

The portfolio is only holding on 9 positions, I personally am holding 3 after closing some losers this last days, without any pleasure…

The hardest one to sell was Transmedics ~$110 as the stock broke its W50 on high volume despites strong fundamentals. There isn’t much to do in those situation but respect the system and let go, despites personal convictions, despites what looks like an amazing opportunity. If the stock were to have such a strong potential, the market will reverse and give me another opportunity, on high volume and strong price action. Until then, I’ll continue to follow the company, but from the sidelines.

• This trade returned 48.75% vs. 23.88% for the S&P 500 on the same period, a 24.87% alpha over 11 months.

Alibaba was a different kind of deception as the market had some reasons to sell it: its liquidity is being spread on two different verticals while it should be focused on AI. Weather a great management decision or forced by competition, this will slow down AI expansions and the market won’t buy a company focused on e-com, it proved it with Amazon, MercadoLibre and Sea Limited.

• This trade returned 53.01% vs. 7.26% for the S&P 500, a 45.75% alpha over a year and three months.

I also had to close UPS which fell below its W50 on weak price action following the war in Iran; logical as the transportation business is heavily impacted and a restructuration in the middle of an energy crisis won’t be easy to succeed…

• This trade returned -4.47% loss vs. -2.09% for the S&P 500, a -2.38% loss over two months.

Lastly, Lululemon as the sector did not convince with its latest earnings and the stock lost its local bottom. We do not hold weakness.

• This trade returned -13.83% loss vs. -1.26% for the S&P 500, a -12.57% loss over three months.

And that leaves us today only holding winners and strength, four of them linked to the energy/materials sector performing really well, one holding on AI compute, one cyclical hardware forgotten by the market and three new names on a sector in very high demand: photonics.

Nine stocks, two categories.

The Runners

All these names are great holds but have been identified months ago and are simply running now. They’re just great names to hold with strong narrative and improving fundamentals, maybe a trim here or there if one needs to rotate liquidity, but nothing more.

Halliburton and Schlumberger are in this situation. Identified months ago, running very well and pretty high but with much more room to go especially as the conflict in Iran will require their service either to find oil or repair damaged infrastructures. We found the trend, studied valuation and fundamentals, and are now riding it.

I could put Nebius here as well. On the paper, we are back around the W50 and this is where we accumulate. Fundamentally, everything is strong but sentiment and liquidity are weak. And I am curious on how the market will react on those names with rising electricity costs which should increase Opex and hurt margins as their business is about electricity consumption. Nevertheless, the system says this name is a hold or a buy as long as we hold this W50 even if its sector is weakening.

They aren’t in the screenshot as the retests weren’t given, but I have to add Nutrien (up 8.4%), Darlings (up 46%) and Smith & Wesson (up 35.4%) to this list, which are in the same situation: perfect uptrend to ride quietly. Those weren’t counted in my YTD performance, which would be even better including them.

The Buyable

My photonics names remain the most interesting buys in the market to my opinion. The sector is one of the only strong in tech with very bullish narrative and fundamentals funded by continuous hyperscalers CapEx which won’t disappear because of some oil disruption.

Sure, they could be slightly reduced but that would hurt Nvidia and the likes, who depend on acceleration, not photonics who are a very niche business without much revenues and would benefit massively from a few billions spent their way - which is nothing compared to the $500B plus planned in CapEx FY26.

Soitec owns the technology to manufacture photonics wafer, the socle of everything, GlobalFoundries is the #1 worldwide SiPho fab and Silicom is the only temporary alternative to photonics, until the technology scales. The three company share one characteristic: they have a beaten down stock due to weak core business cycles over the last years.

Cheap stocks at the heart of the next technological wave for AI datacenter scaling, with perfect price action signaling strong accumulation at reasonable prices.

Nordic Semiconductor follows a comparable pattern, although it will be impacted by final consumer demand for AI wearables which could be lowered if inflation were to come back and push its cycle to a bit later if hardware companies were to slow production as consumers weaken. But the stock remains at a reasonable price for a large potential over the next years.

SolarEdge is without a doubt on par with photonics in term of attractivity, one of my favorite name in the current market. Fundamentals are excellent, narrative is perfect with the war pushing for more energy independence, and price action is as good as the rest with another breakout being retested ~$45.

The breakout volume wasn’t as high as I’d like it but I believe this name is a strong candidate for the next quarters/years at today’s valuation.

Huntsman is a bit more messy. Higher oil prices mean higher commodity price to produce its material, but it also means less competition as the material sector was hurt by Chinese competitivity for years, and as an energy crisis would be heavier on Asia than America, it is possible that it hurts Chinese competitivity, while material and chemicals prices are rising, giving a second hedge to the company. I believe the entire chemical sector is really worth looking at and I’ll try to come with more names around this narrative - I have found a few but need a bit more research.

Looking Ahead

The month to come should be quiet. I have two sectors to study and will write about them if I find interesting assets: materials (a beaten down sector providing key chemicals or goods for recession proof industries) and optics (photonics are only one part of the vertical as to transport light, we also need lasers and other mechanisms which could see a strong narrative and revenue growth).

I’ll also continue to run my screeners regularly to find new opportunities the market like and share them if necessary. This is business as usual: follow the system.

We will also start earning seasons from mid-April, with too many interesting companies to list but count on me to share the most interesting data with you, related to the verticals we are invested or looking at closely.

As to conclude.

I am not very bullish for the months to come. We might bounce, and it seems like today is off a good start as Trump said the war wouldn’t last much longer and the Strait will open by itself when the U.S. retreat. That would be the best for everyone but first, who knows? Second, when? And third, what about the damages already down?

A bounce during a downtrend is nothing but a bounce. It can be violent and give hope to investors, especially now after 5 straight red weekly candles on the S&P 500, we could expect it! But as long as we’re below key moving average, it’s still just a bounce… It doesn’t bark.

I talked about that rotation for long, the logical next step is for the sectors liquidity left to fall, until they find a bottom - when bad news don’t punish the stock anymore, stabilize, and bounce again. This is what I expect for most of the market, each sector reaching the next step at its own pace. I don’t see why the S&P 500 would shot up to new highs without liquidity growth nor positive catalysts. We can always hope for some drastic measures from Trump to push the stock higher as its performance is tightly linked to its personality, but it’s hard to see happen… And if it were? Then my portfolio would shoot up with it as I’m only involved in strength.

In the meantime, there is always a bull run somewhere and it’s my job to find it. I’ll continue doing it, share strong sectors and soon enough, you’ll see me talk about Meta, Nvidia, Palantir, and all those names that deserve to be bought and rode.

Their time will come again, and we’ll be there to buy them.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

We’ll need to go back to some basics to have a proper understanding of our subject: semiconductors. We’ve talked about all this already on my ASML/TSM review so go for it, it’ll be a great introduction, even though we’ll go over the basics again today.

What are they? Simply put, they are etched circuits on wafers. The complexity comes when we start talking about the circuits – which are meant to conduct electricity between components, and the wafers – which are made of different elements to optimize conductivity or not.

The final goal of any semiconductor is to control current between elements being cautious on signal clarity, speed and heat, which results in different circuit pattern, elements, components, materials, etc… Hence their complexity and constant improvements, as they are also very, very small - we’re talking ~1,000 time thinner than a hair.

This explains why we have many different components and how critical they are. The objective of a semiconductor within a smartphone isn’t the same as within a 5G antenna which again isn’t the same within a GPU. They’re all different in size, components, materials, etc…

Which brings us to our subject today, a new kind of semiconductor we talked about a few days ago in my GlobalFoundries and my Silicom Ltd thesis: photonics (optics semiconductors called SiPho for Silicon Photonics) and their packaged version CPOs (Co-Packaged Optics) which are the aggregation of SiPho within a board to make it usable. Those are the new innovation supposed to push AI datacenter capacities further and solve the actual networking bottleneck.

And as for any semiconductors, it all starts with the wafer.

SOI (Silicon on Insulator)

The wafer is the raw platform made from specific materials on which circuits will be etched; SOI is one kind of wafer, optimized for optical connectivity, hence bandwidth/latency. A “normal” semiconductor is a simple layer of silicon. The problem is that silicon is transparent to some wavelength, not usable to transport light by itself as it would pass through it left, right, top and bottom, losing photons to the void hence signals, data. Not very optimized…

No, for light to transit, we need it contained within a defined and reflective space so no photons are ever lost as they bounce until they reach their destination.

This is how it works in optic fiber for example, which can look weird as it is nothing but a twisted cable plugged between two pieces of hardware. In reality, the light bounces on the refractive side walls of the cable until finding the receiver at the other end, without losing a single photon in the process – nothing affecting data transmission at least, and with no heat.

If the cable was made of silicon only, the light would pass through…

When it comes to optics semiconductors, we use SOI wafers, a three-layer cake optimized to trap light where we want it: within the silicon only. SOI are manufactured with a technology called “Smart Cut” which is a patented method from the company we’re talking about today.

Here’s how they work.

• Layer 1 – Single-crystal Silicon. Classic semiconductors, except instead of etching conductive patterns for electrical impulse, we will etch “waveguides” for lights to refract and follow a direction, just like in an optic cable.

• Layer 2 – Buried Oxide (BOX). An oxide layer more reflective than silicon which prevents light from escaping below the silicon layer.

• Layer 3 – More silicon. A much thicker meant as a handler. A mechanical layer to fix the wafer and give it stability; we are talking about a semiconductor optimized to let light pass through, precision is necessary.

When connected to a light source, photons will transit through the semiconductor, trapped between “waveguides” etched on the first silicon layer and the reflective BOX around, between point A and B with minimal loss and heat.

SiPho are optimized for bandwidth and latency and when packed within a CPO, they give the possibility for data to transit through photons within a GPU, a rack, or any other hardware, pushing AI compute to a new scale.

Now again, what we call SOI is only the wafer which will be sold to a foundry to be etched according to specifications into a SiPho, then shipped or directly assembled within a CPO and installed within a GPU, rack or any kind of hardware which requires optical networking.

The wafer is step 1 of a semiconductor’s life (excluding material extraction and refinement), GlobalFoundries and Tower Semiconductor for example are buyers of wafer to etch them, build CPOs and resell those finished products. You know I am bullish on SiPho, CPOs and on both foundries so very logically, I am bullish on demand for wafer.

SOI Usage

To do a bit of history SOI isn’t new, has been used for years and is very critical in sectors which are already focused on optical connectivity.

• RF and mixed-signal devices (4G/5G, smartphones, satellite communications…)

• Low-power and high-performance CMOS (AI datacenters, wearables…)

• MEMS devices, including sensors and actuators (cars, medtech…)

• Power management ICs (EVs, IoT…)

The AI datacenter need is new as no one foresaw a decade ago that our network would be a limitation for global compute withing hardware, creating a bottleneck. The tech industry realized this maybe a few quarters ago and needed time to engineer the solution, which is coming to life nowadays in the form of massive and growing demand for SOI wafer and SiPho.

Soitec Dominance

Now that all introductions are done and we understand SOI and their applications, we can talk about Soitec, which is the leader in SOI wafer. I’ll be very straightforward, not only are they the leader, they own the technology on how to manufacture SOI wafers which gives them a quasi-monopoly on who can and cannot do so.

Two words: Smart Cut, the technology used to create the BOX layer within a silicon.

Yes, Sebastian, on the overall SOI market, of course, there is. We always say that there is competition, starting with our only licensee today, with SEH. But we don’t see big changes in the segment shares we are observing, and particularly for us, regarding the 200 and the 300 millimeters. Of course, players are claiming to have SOI solution, and some SOI solutions are not using our technology like Smart Cut. You can have also other ways to do SOI, a silicon insulator, namely BSOI, that is quite well adapted for power application in the cars. But even the domains, competition, and our positioning, it’s quite strong.

Then today, we don’t see segment shares, significant evolution, and the competition is far away from what we are doing.

Pierre Barnabé, Soitec Q4-25 Earning Call

As you could read, it isn’t the only method to manufacture SOI wafers, another main one being SIMOX (Separation by IMplanted OXygen). The objective and results are the same but as the method is more complex and more expensive due to the hardware used, and as no one wants to pay equal quality more… well, the cheapest and most scalable wins.

And that lays my investment thesis on Soitec; they are the company owning the patent to manufacture the cheapest SOI wafers at scale, which usage is supposed to blow up as SiPho and CPOs demand is also blowing up… What’s left to say?

This doesn’t mean they are the only producers of SOI wafers, others also do but have to pay a fee for using their technology so Soitec effectively makes money on 100% of SOI wafers manufactured with Smart Cut, plus their own production which is ramping up, with accelerating orders, demand and contacts from the tech industry.

Edge and Cloud AI revenue amounted to EUR 54 million, showing 27% organic growth YoY and improving from Q2-26. Activity continues to be supported by AI-related demand across edge and cloud environments, with contributions from both Photonics-SOI and FD-SOI. Specifically, Photonics-SOI remains very dynamic.

And today, we have more than five customers today engaged in our silicon photonic solutions, and more and more are coming.... And each of these customer asking for more and more volumes. I’m confident that this dynamic should last, because the need for AI are very important. And of course, we are part of this components of this industry, because the silicon photonics is today the only solution, viable solutions, that is allowing for next generation AI data centers, transfer control, to transform, electron into photons for high bandwidth. Particularly if you move up to 1.6 giga. Not only for the data centers, but tomorrow to interconnect the GPU, and particularly with the co-packaged optic solutions.

And on top of it, we are enhancing our photonics layers with the lithium niobate, let’s say, add-on, that is helping the industry to move beyond the 1.6 tera. That means that we have today a plan, a clear plan, and the industry have a clear plan to fuel the silicon photonics growth, for the coming 4-5 years, without any doubt. And our market shares, positionings, relationships with our customers, is making us confident.

Pierre Barnabé, Soitec Q4-25 Earning Call

Besides these confirmations from Soitec, we have confirmations on SiPho and CPOs demand with companies like GlobalFoundries and Tower Semiconductors talking about a ~50% CAGR in SiPho and expansions for their SiPho fabs – a $900M expansion planned this year to 5x their production for Tower while GlobalFoundries is planning to increase production after their acquisition of Advanced Micro Foundry to a $1B+ run rate by end of year 2027 - hence roughly 4x as well.

Very basic math, if Tower and GlobalFoundies both intend to 4x/5x their production in the next two years or so, it also means they will need to 4x/5x their SOI wafer purchases and as we’re looking at industrial scale, most should be produced with Smart Cut which means directly sold by Soitec or by companies paying fees to use Soitec’s technology.

I think you start to see the potential already.

The Opportunity

I won’t be able to put a price or a target/valuation on Soitec because we are talking about the company potentially responsible for the next AI datacenter innovation, the first wheel of a car which should be very fast. It’s impossible to anticipate demand, revenue growth, margins nor the potential premium which could come from their position, so I won’t it. It’s also hard to anticipate how the stock would react if CPOs weren’t the solution and demand for SOI wafers were to fall.

What I can say though is that today, Soitec is in the same state than GlobalFoundries: completely ignored by the market because of its slowing core business, looking like a weak company with declining growth and margins, unprofitable and certainly not worth any premium.

Why? Because they are the leader in SOI wafers and those are used in other sectors, AI datacenters are only the new shinny source of demand, while the other sector… Well, they’ve been extremely weak and plagued by an inventory building during/post Covid, as the period boosted demand for hardware which slowed down massively the years after.

In mobile communication, customer engagement remains active around advanced RF solution and next-generation platforms, despite constrained volumes with RF-SOI demand still impacted by customer inventory correction, mobile communication generated revenue at EUR 90 million in the third quarter, down 36% year-over-year organically.

In automotive and industrial, while the environment remains challenging, our technologies are well-positioned to support future applications once demand recovers. Automotive and industrial revenue reached EUR 16 million, improving sequentially versus the second quarter, but remaining down 32% organically year-on-year. Market conditions are still weak.

5G activity was slightly down year-over-year, while showing an improvement from Q2.

On a year-on-year basis, revenue remains down, reflecting a 22% decline at constant scope and foreign exchange, and a negative currency impact of 7%. Sequentially, revenue increased 18% organically compared with the second quarter, coming in above our guidance.

Pierre Barnabé, Soitec Q4-25 Earning Call

If that remind you another story – and opportunity, this is the exact same situation for Nordic Semiconductor.

Soitec is going through a rough time for the biggest volume of its business which is mostly for RF technologies, cars and smartphones – with the last two being in a clear slowdown/stabilization since Covid hence inventory issues and slow demand recovery, which in turn triggered slow growth over the last years.

Management was a bit more confident this last quarter might be close to demand bottom, but this wouldn’t mean return of growth on a their core vertical this year and would still hurt the profile of the company.

As for inventory reduction, which is another priority, we’re aligning production with actual demand. Of course, the downside of this is that it brings a temporary hit on margins and profitability.

Pierre Barnabé, Soitec Q4-25 Earning Call

In the middle of this situation, we have an SOI organic growth for photonics up 18% sequentially and 27% YoY combined to very positive management comments on what is yet to come and their capacity to deliver in volume with strong supply chain for materials – which will be my next area of research as if demand for SOI wafers grows, demand for materials also will and might create other opportunities, combined with lasers as transporting photons happens only if photons are generated (to give you a heads up on the potential next narratives).

We have no specific limitation, but to be sure, we organize properly our factories, because of course, we are sharing the same factories for the 200 and particularly the 300 millimeters. Then it’s a matter of anticipation. But we have access to the right level of bulks. As you know, we diversified also our sources of silicon bulk in terms of the quality, and particularly the one able to correspond to Photonics’ requirements and demands. We have today more and more customers asking for Photonics and more and more in volumes, 200 and 300 millimeters. But today, there is no specific limits to take the ride and to accompany a market that is growing by around 25%-30% per year.

Pierre Barnabé, Soitec Q4-25 Earning Call

This is the global thesis on Soitec today. A company trading at low valuation because of low growth and declining margins due to a slowing core business, with the potential to be revigorated by a massive and accelerating demand for their wafers for photonics, on which they have a quasi-monopoly as they own the patent to produce them.

This looks like a pretty strong potential here.

Investment Plan

You should be used by it now, the chart looks like every charts I look for. A clean base – although this one was pretty rapid, built in ~4 months, with a lower low on massive volume followed by stable price action on growing volume, which indicates more buyers at actual price, meaning the majority of the market finds value at this price – otherwise price would drift lower. And as usual, the final confirmation comes from a breakout and reclaim of the W50 on volume spike, meaning most of the market accepts, in sync, to pay more for this stock.

There are always reasons for liquidity to go towards a name: value and potential. It doesn’t mean every stock with this pattern will succeed, but every success starts with this pattern because a large base of investors accept to pay higher prices for the stock and bet on its success. And that’s what we want to see - and buy.

We now know why those investors bet on Soitec, our next step is to know if we want to be involved with it and at what price. I want to and bought on Tuesday as the French market had a bad day and Soitec followed – yes Soitec is a baguette company. That retest went close to the W50 and I built a pretty large position, as a concentrated investor buying a great price target.

I shared this on my Silicom Ltd write up because of the volatility of this name and was right to as it went up double digits since. I couldn’t write fast enough in time…

Once again, it is very hard to value the company and set targets - to buy or sell, as we have no idea of the wafer volume coming their way nor their capacity to execute. Soitec has never received a premium comparable to their situation today, so we can’t really model anything. This is a narrative trade with a massive potential, which could materialize - or not.

My opinion is that a EUR 1.81B company trading ~2x sales with the potential to become a critical infrastructure company for the future of AI datacenter has upside, both in term of growth and premium valuation - hence pretty explosive upside. And is actually trading at a pretty sweet spot even considering its core business – which is supposed to stabilize, compared to historical multiples.

The real risk lies in the success on SiPho and the continuous dominance of Soitec in the SOI wafer vertical. I wouldn’t touch this stock otherwise. If CPOs weren’t working or weren’t the answer to the actual networking bottleneck, or if a competitor were to find a better technology than Smart Cut, this investment could hurt, badly. There are no answers on the CPO efficiency yet – although the entire industry points to it but things could change, and when it comes to dominance, Soitec is ready to defend its position and even consolidate it with more patents.

We are filing more and more patents every year, more and more. This year we're gonna even beat what we did last year. That was quite exceptional. Then we continue to run ahead.

Pierre Barnabé, Soitec Q4-25 Earning Call

But the risks exist.

As I said I am already in and will gladly buy more if we see another retest ~W50, I only bought commons; I do not do options on such names. Low valuation and expected volatility are the perfect recipe to blow up option trades. No reasons to add risk on an already risky trade. Only shares with quarters long timeframe - hopefully years.

If SiPho and CPOs are indeed the future of AI datacenters as many – Jenssen included, believe, then we’re probably in for years of growing demand, and if AI hardware were to indeed need to include CPOs in each of them… There is a case to say that SiPho and CPOs will be the next biggest vertical of AI hardware, and we know hyperscalers aren’t done spending.

If that were really the case, then the need for SOI wafers would only be starting and demand for it would only accelerate.

Then the co-packaged optics is, is really a new architecture, for which we see, some, pull forward, because, it was expecting to come in 2028, then 2027s, and today we, we have heard that, some, first, let’s say, design gonna be, gonna be commercially available by end of, of this year. Then there is a clear accelerations.

Pierre Barnabé, Soitec Q4-25 Earning Call

If so, then the next innovation cycle relies on SOI wafers.
And then, mostly on Soitec.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

I have been working a lot on the current AI bottleneck lately - networking and bandwidth, not just because social medias are full of companies in the sector but because many names are showing up on my screeners, which means they are being accumulated and breaking out key levels - which is what we want to buy.

I already talked about GlobalFoundries a few days ago - the SiPho fab leader. We won’t talk about SiPhos today - although we will still talk about networking bottlenecks, but I want to share that I plan to write about Soitec this week, and as the stock is close to a pretty sweet spot and my write up isn’t ready yet, I wanted to share it with you before.

I wanted to share this first so you can start looking around before the full report hits your inbox in the next few days - hopefully.

The AI Networking Bottleneck

We went through the basics in the GlobalFoundries thesis, detailing how optics are the only viable solution to increase bandwidth and decrease latency within AI data centers, currently the #1 focus for compute providers.

As we move from training to inference, the needs change. Training was about sheer data compute, having as much of it as possible to train behaviors. Inference is a different beast; it requires compute but also speed. Data must transition rapidly between network points before being processed. Today, we have the power to process this data; we have so much power that GPUs end up staying idle, waiting for data to reach them.

This is today’s inference issue: the speed at which data transits between compute points, mostly due to the limitations of copper. To use my previous analogy.

Imagine a highway limiting the number of people passing through. Say 2M cars can drive simultaneously. But in the middle, there’s a section where only 1M can pass (bandwidth), controlled by tolls that require travelers to change their car engines (latency and power). That creates waiting time, slowing down the entire system.

GlobalFoundries Thesis

The market is excited about photonics because it is the only way to scale compute further. Adding more GPUs today is like adding workers on a construction site waiting for supplies. It won’t build that house faster, what we’d need is to move the supplies faster. But as noted by Jensen and Astera Labs’ management, we are still years away from industrial scale silicon photonics.

The AI factory is the new unit of computing. To connect a million GPUs, we can no longer rely on the old physics of copper. Silicon photonics is now central to the next generation of AI infrastructure.

We have squeezed every last drop out of copper. At 224G per lane, the physics simply demand light. The transition is no longer a choice; it is a thermal necessity.

The technology is ready, but the supply chain is not yet at the ‘Industrial Scale’ required for zero-defect AI factories.

Nvidia GTC keynote

What’s up until then? There might be another way to bridge the gap.

FPGA & SmartNIC

To be clear: these technologies won’t replace the long-term need for photonics, but they are usable today, faster to deploy, would remain useful even after the transition and could help with it, temporarily and in some cases.

Field-Programmable Gate Array (FPGA)

A FPGA isn’t a GPU nor a CPU. Think of it as a Ditto (the Pokémon). By now, we probably all know the difference between those two.

• CPUs are the brains, excellent at complex, vertical, serial tasks.

• GPUs are the muscles, excellent at horizontal, parallel computing.

Before going further, we need to understand the physical optimizations of GPUs and CPUs. Chips are made of silicon, a semi-conductive “sand” on which electrical paths are etched, physically, to optimize electrical flows between components depending on the final need of the semiconductor. Exactly like we would manufacture a jet to fly as fast as possible for example; you’ll understand easily it isn’t meant to drive. One path, one function, no detour, no waste of energy.

We thought GPUs/CPUs would be enough; they are compute wise but aren’t self-sufficient and hyperscalers are now looking to optimize them, and that requires other hardware; one of them being FPGAs.

FGPAs also are semiconductors, silicon. The difference between GPUs/CPUs and FGPAs is physical. While the first have finite and optimized etched circuits, the second has configurable etched circuits. It has more than it needs but those can be turned on or off, depending on the needs.

Imagine driving in a city. There are plenty of roads, right? A GPUs/CPUs will have only one road, the most efficient one and everyone would follow it. A FGPA on the contrary will have all the roads configured, but traffic lights will force drivers on this or that one, depending on where they are and want to go.

GPUs and CPUs can be optimized like with Nvidia’s CUDA, making the road smaller or bigger, slicker, changing the wheels or the cars to accelerate traffic. But the road itself will always be the same, from one point to another. It cannot be changed because it physically is built that way and you aren’t going to destroy buildings to change the road because Michel wants to pass here and not there. No one cares Michel, follow through and be optimized.

With FGPA, the road Michel wants to take already exists. It is closed but it exists and if the need were to favorize this road, then it could be opened. This kind of reconfiguration isn’t simple nor extremely rapid, but once the configuration is set - which is the hardest part, a simple reboot can transform the same hardware with one optimization, for one function to another one, “just like that”.

Your question now should be: then why do we have GPUs and CPUs?

Well, material optimization remains the best form of optimization. Imagine a car with pals which can turn into an helicopter. A normal car will be smaller, faster, consume less. Sure, you won’t fly with it, but sometimes you just need to go see grandma, not to fly. GPUs and CPUs are much better at what they do.

So the next logical question is… Why would we use those, then?

First, they are already very used in specific sectors nowadays as the ultimate focus of those semiconductors are latency, critical in cybersecurity, defense systems or algorithmic trading for example. But they are starting to become critical for a much bigger market who’s actual bottleneck is both bandwidth and latency: inference.

AI infrastructure investments are shifting from training models to querying the models at scale known as inference. Inference is continuous, distributed, and extremely latency-sensitive.

While training primarily happens via network GPU cards at the core of the data center, inference happens everywhere, continuously, at the edge, in telcos, and in enterprise data centers. This creates massive networking and interconnect bottlenecks, and that’s exactly the problem that Silicom excels in solving.

Liron Eizenman, Silicom Q4-25 Earning Call

Because they are configurable and reprogrammable, they can be used improve network congestions, acting as a GPU/CPU when needed, or a hybrid. They have two key characteristics which make them so valuable.

1. Modularity: As they are modular, they can treat information differently. A GPU needs a large amount of data before starting its compute and will stay idle until it has enough data - you don’t want million-dollar hardware to stay idle. A FPGA can be configured to compute while data flows. It can be configured to only treat one type of data while being energy optimized. Or it can be overly specialized for a temporary task companies won’t spend thousands to engineer a GPU for.

2. Interconnectivity: They can be directly interconnected within the hardware - to the GPUs/CPUs or any other chips, which largely improves latency and removes bandwidth bottlenecks as data flows to them directly at electrical speed. Which will brings us to our second subject today.

FPGA can be seen as Dittos used for niche usage, when a GPU or a CPU won’t be as efficient of isn’t worth being developed for, in very specific use case where nothing else can optimize workflows or for very specific needs.

SmartNICs

The FPGA is the silicon, it isn’t complete by itself; it needs to be included within a motherboard with potentially other components (RAM or/and DPUs) and connectors to fit within the system. The final product is called SmartNIC, a Ditto board focused on networking, optimizing packet flow between the network and components with the lowest latency possible.

FPGAs and SmartNICs are usually used in modular environments where latency has to be the lowest possible: Cybersecurity, 5G networks, virtualization, defense systems, algorithmic trading… And now, potentially inference, connected to GPUs, CPUs, or any hardware within the racks with limited copper in between.

A Ditto which can be connected to critical infrastructure, optimized for low latency and maximum bandwidth.

Silicom LTD.

This is the company we’ll look at today, a pure SmartNIC player. Silicom doesn’t do FPGAs; they buy them and manufacture SmartNICs for their clients. Which also means they do not own tech IPs and are just a component assembler at the end of the day, not the highest value in the chain but still selling a very demanded finished product.

The company and sector have been a bit distant from the AI revolution; there were always FPGAs in quantity in AI datacenters but the need for customized ones wasn’t there then, and it might be now. Might be.

Let’s talk about the environment first.

Nvidia, AMD & Co.

All the big guys already have FPGAs; as I said, it isn’t new. AMD bought a company called Xilinx in 2022, was known as the go-to for FPGAs and today’s leader in the sector. Intel acquired Altera in 2015 but didn’t do great with it and partially sold it this year, while Nvidia is focused on SmartNICs with ASICs, not FPGAs, via BlueField.

Those are for the big integrated guys, but there also exist some independent names manufacturing FPGAs: Lattice and Microchip - two stocks with a pretty crystal-clear growth trend at the moment.

And there is a crystal clear source for actual demand: AI.

And this companion role becomes foundational as AI workloads push system-level complexity higher and higher and drive to faster time to market, as AI servers disaggregate into processor boards, networking cards, security cards, power and cooling modules, storage, and other specialized blades. FPGAs show up everywhere in the data center. We are also seeing the same trend accelerating in physical AI. Intelligence is moving closer to the sensors where data is created. A single robot can have multiple vision sensors, including image, LIDAR, radar, and infrared, requiring fusion, aggregation, and preprocessing.

In addition, humanoid robots could have dozens of motors, requiring high precision and low latency control. Companion chip FPGAs sit, besides those sensors and actuators, to synchronize high-bandwidth vision streams with real-time motion and deliver deterministic responses.

Lattice Semiconductor Q4-25 Earning Call

Two stocks which could be considered, but their risk profiles aren’t comparable to Silicom, with 8x and 24x sales and large competition from Intel and AMD. No, my focus wouldn’t be on FPGAs because those are only one part of the story, the critical part, but only one part.

The key for today’s networking bottleneck is flexibility. Hyperscalers have very specific needs which require independence, something that today’s leaders do not have.

AMD and Nvidia both have closed ecosystem and will not propose customized hardware with manufacturers they do not partner with. They propose a finished product which is enough in 95% of cases but the last years showed that hyperscalers needed independence and further customization, not for the bulk of their compute but for the optimized niche.

And this is what Silicom proposes. They are a middle man, constructor agnostic, and can meet any kind of optimized hardware demand for any kind of niche.

The Thesis

To recap the thesis clearly: FPGAs are already used in AI datacenter and should only increase in usage as the need for flexibility and network optimization grows. The question is how will those be integrated and why would Silicom benefit?

The first answer is that demand will continue in their core vertical - very low latency services which now also leverage AI like cybersecurity, algorithmic trading and the others we already talked about. Demand on those sector is accelerating as confirmed by management.

Just to give an example, a few weeks ago, we announced that the global networking and security as a service leader significantly expanded its deployment of Silicom Edge devices into multiple additional use cases, increasing our expected annual revenues from this customer from $3 million-$4 million to between $8 million and $10 million, more than doubled, with some of those incremental revenues expected in the coming months.

Q4 revenues grew 17% year-over-year to $16.9 million, well ahead of our guidance range between $15 million and $16 million. It confirms that the demand for our core product is high, resilient, and strengthening.

Liron Eizenman,

This is step one, the base case. Step two is about the bull case: customized SmartNICs to optimize network bandwidth and latency within inference. Not for special sectors, but for inference itself.

Management was contacted by some hyperscalers not for a definite order, but for a POC - Proof of Concept. There is interest in the technology for a larger usage, enough to see if their product can answer the demand to resolve those bottlenecks.

We already have initial orders for our inference-optimized FPGA-based solution to be utilized by our customer at a POC with a hyperscaler end user, and we are developing a dedicated AI NIC based on a leading high-performance networking chip for another AI inference leader. We have initial orders in hand and follow-on POCs underway. We are also engaging with multiple customers, and we are in advanced discussions with additional AI inference chip vendors.

Liron Eizenman, Silicom Q4-25 Earning Call

This isn’t a thesis coming from my imaginary mind; it is real demand coming from hyperscalers to try out SmartNICs and see if they can help resolve actual bottlenecks, combined with real demand for their product in their known verticals.

In 2025, we achieved eight major new design wins across edge systems, SmartNIC, and FPGA solutions, with both new customers and existing Tier 1 customers expanding their engagements with us. Those design wins give us strong visibility into 2026 and beyond, supporting our expectations for double-digit revenue growth for the year ahead.

We are again targeting between 7 and 9 design wins in the current year, spanning across all our product lines.

Liron Eizenman, Silicom Q4-25 Earning Call

Now, keeping feet on the ground, those are what they are: POCs. They might not be conclusive and won’t yield massive revenues this year.

And even for 2026 as a whole, I think we are not expecting them to be still huge. There is an opportunity for that, but we definitely are expecting our core business to be very, very strong in 2026 and keep growing. And each of those opportunities, it can boom at any point in time, but right now, we’re still in the early stages. But we feel that we are very strong in the early stages and that we feel very strong traction on each of those.

Liron Eizenman, Silicom Q4-25 Earning Call

But the bull case doesn’t rely on massive revenue tomorrow, it relies on the potential for SmartNIC to be valuable within inference, for hyperscalers, to help with the actual bottlenecks - releasing pressure on the network and improving GPU usage, reducing idle time, etc...

Still feet on the ground, we aren’t talking about redefining inference neither, simply helping on key pressure points. If the opportunity was as huge as the new GPUs, hyperscalers wouldn’t do a POC with such a small company, they’d either develop it internally or buy the company. Working with an assembler makes sense if the need is large enough to justify the spending, but small enough not to do it internally.

The potential is real. The interest is real. The POC is real. But the opportunity is niche. That being said, “niche” at the scale of hyperscalers spending hundreds of billions per year is the Everest from the point of view of a ~$100M company.

So the inference happens everywhere. And when we say everywhere, it could be at the edge of the network, it could be a local data center, it could be a telco data center, it could be even in the enterprise. And there are so many different installation types and deployment types and types of different networking they need to support, then different cards, different companies developing different inference chips, not all of them able to provide all the different layers that they need to cope with. So they will only focus on the inference chip, but they need someone to complement it on the networking side. So if you want to do, as you said, scale out to multiple servers, multiple ports, how do you do it efficiently? How are you making sure that the network is not a bottleneck and that you actually get the most that you can out of the inference chip? That, I think, is the key. And there’s no like it’s very fragmented and very different from deployment type to another. That’s where the opportunity is created.

Liron Eizenman, Silicom Q4-25 Earning Call

If those POCs were conclusive, we’d pass from niche hardware for low-latency verticals returning to double digit growth to an hyperscaler partner with a feet in the AI inference game optimization sector. This certainly would trigger market’s rerating.

Financial Profile & Risk

To state this clearly: we are looking at a micro cap. A ~$100M capitalization, very small, very volatile and very risky - although I’ll mitigate that risk right away.

Silicom is a ~$100M cap with ~$50M of cash and equivalents in the bank for no debt. This is what we call a fortress and leaves management with many opportunities, from acquisition to investment into manufacturing capacities or buybacks - which are the preferred use for now. And management confirmed the actual demand for their products do not require much capacity expansions or R&D.

So in some of those projects, as we said, we’re leveraging existing IP and we’re leveraging existing know-how. So it’s not like we’re starting from scratch. So for some of those opportunities, it’s actually taking some of our existing products, making some changes on them so we can react very, very fast, and it can actually be a quick road to revenue here and quick road to design wins. On some of the others, we do need to do some development, but we already started with that. So we are deep into the development of some of those. So we think overall, we are expecting it to be faster than what we’ve seen in the past.

Liron Eizenman, Silicom Q4-25 Earning Call

That said, they have the cash needed if need be to accelerate production, expand, or anything else. We are talking about a company trading at 1x EV/sales with no real need for its cash - yet, but buybacks and safety cushion.

Although this safety cushion is necessary as the company remains unprofitable. We’re talking about ~30% gross margins for -20% operating margins and a loss of ~$3M per quarter. As shared earlier, Silicom is an assembler and do not own IPs or patents, which means lower margins and at the moment, no profits. So cash is important for the company.

Profitability isn’t the bull case either way here, we are with a small cap growth name and returns will be tied to demand, growth and hyperscalers POCs. But at today’s valuation, with double digit growth guidance and a very large cash cushion, risk always exists but it seems much lower than many other micro cap.

Investment Plan

As for investing, micro cap always come with their risk and volatility so always keep this in mind. This is a risky investment. Based on a clear thesis but risky nonetheless.

A company trading at 1.7x sales, less than 1x EV/S with ~50% of its capitalization in cash, a very healthy guidance of double-digit growth for its core business without including a potential tailwind from the actual largest AI bottleneck, which means growth could be much stronger if the POCs were to be conclusive.

The last time the company traded at double-digit growth, it was trading at ~3x sales and this growth is today’s guidance only based on core business.

If we were to combine a 15% FY26 growth with a 3x sales multiple, which would be the base case, without a hyperscaler POC success, we’d be looking at a ~$200M cap, ~100% higher than today. An hyperscaler POC success would justify a higher growth rate and multiple, which I wouldn’t try to model but would push the name higher.

The way I see this name is a good risk-reward with excellent potential return in the base case, and exponential returns in the bull case - public acknowledgment that their SmartNIC has value in hyperscaler AI inference systems. The market would anticipate this and won’t wait for orders to pour in to reward the stock; any positive announcement in favor of their hardware will trigger demand directly, in anticipation of future growth. Narrative will be key on this name.

I already have a position as price action confirmed demand for this stock, which probably means others share my thesis.

The base and the bottom are clear and were fixed before the hyperscaler and AI potential, which is a good sign as buyers did not have today’s expectations. The breakout happened last quarter when these announcements were made, meaning the market accumulated the return of core growth and got excited by the AI potential.

Clear base, clear volume increase, clear breakout, clear retest.

This is growth investing 101; this is the very beginning of this stock’s potential, while the downside risk is still minimum. Any negative news on the hyperscaler POC would certainly trigger selling, but at less than 1x EV/S and with ongoing buybacks, I wouldn’t expect a -50% overnight - although once again this is a micro cap.

This is a high risk, high reward name with the risk being diluted by a strong core business, massive cash and low valuation. But we are still talking about a $100M capitalization within a very volatile environment and sector. So it is important to be cautious with those speculative names.

The potential is here, demand for the stock as well. Now let’s follow the core business and the hyperscalers POCs.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

As for investing, Alibaba is officially no longer a stock I can hold, as it has been trading below its W50 for several weeks and this quarter didn’t show what I wanted and was not received positively by the market.

I have a system, and I follow it: I do not hold weakness.

We rode this name from $80 in December 2024 following a clean W50 retest after a high-volume breakout, delivering 53.01% returns vs. 7.26% for the S&P 500, an alpha of 45.75% over a year and three months.

Vertically AI Integrated

This is a major part of the bull case as Alibaba owns the entire AI stack, from infrastructure to software, including models and end users. This is what could drive above-average growth in the coming years and what the company has to leverage - and they are doing so.

Looking ahead, we are well-positioned to drive growth on both enterprise AI and consumer AI fronts, powered by our fullstack AI capabilities spanning foundation models, cloud infrastructure, and proprietary chips, alongside deep integration with our broader ecosystem

It showed this quarter with continuous acceleration - slower but still here.

Management confirmed that their AI services continue to capture a large share of the Chinese tech market. This is key to the bull case, as the country is accelerating its AI transformation and demand is - and will remain, massive.

Alibaba Cloud continues to lead the market, attracting more customers to onboard our comprehensive AI+cloud products and services, including high-performance networking, distributed storage, cloud operating system, and services for model training and inference.

They also confirmed growing adoption of their AI services by end users, beyond internal optimization and compute usage, which is another important signal as this is how they leverage the final consumers into revenues.

Since launching its Chinese New Year campaign on February 6, Qwen app has seen tremendous user engagement. By the end of February, approximately 140 million users have had their first AI-driven shopping experience through Qwen app’s agentic features from ordering food and groceries to ticketing and travel bookings. In February, consumer-facing Qwen has surpassed 300 million monthly active users across all platforms.

There isn’t much to criticize on their AI/cloud execution. This is what I want to see and the company is moving forward really well, even if growth acceleration slowed a bit this quarter; the market may be frustrated but this could simply be a temporary pause before a second ramp-up.

Local Consumption

The first issue is exactly where expected: Chinese consumption. I remain bullish long term but the Chinese government only deploys targeted and gradual stimulus, unlike the liquidity expansions in the U.S. where it is accepted to just throw money into people’s pockets. Not even talking about the cultural trend of saving...

And so what looked like a promising acceleration was hurt this quarter with a big sequential growth decline.

No magic here, deceleration comes from lower consumption and a normalization of their fee programs which was modified a year ago. Not what one wants to see for an e-commerce platform but this isn’t the source of growth nor the bull case either way, it was only meant to be the source of cash generation to fund their AI expansion.

And this is where the real problem lies this quarter.

Growth Without Cash Flow

To expand AI infrastructure, you need cash. And this quarter, Alibaba saw a 2% revenue growth - which should have been a 9% without the negative drag of Intime and SunArt, for a -74% decline in operating income… What happened?

Management decided to scale meaningfully its investments, but not only in AI and clouds, also in their e-commerce business and quick commerce infrastructures, yielding a 43% decline YoY in cash generation from the stable source, from the vertical which was supposed to generate cash to expand AI.

Can’t help but see Obiwan Kenobi screaming “You were the chosen one! It was said you were to destroy the Sith. Not join them!” Well apparently management thought it would be great to spread spending between two verticals instead of focusing only on one and do it well.

On another hand, the small foot print taught us that some of the AI CapEx spending was reattributed to “All Others”, so even if the cloud vertical seems healthy the real EBITDA and margins of this vertical alone is probably not as shinny as displayed.

I believe the market is fine with their cloud situation and even this accountability trick. We’ve seen it in the west: you need to spend to grow and Alibaba has a large opportunity ahead plus very stable finances which allow spending. The issue is about spending on an e-commerce platform without much ROI as their quick commerce, which “only” is growing 20%+ is not generating much cash…

Management is going all in on two capital-intensive verticals at the same time without comparable returns. Why? The funds required for tech are already so large, why split them? Why now?

I know my bull case relies not only on AI but on AI application and Chinese consumers, so this is a step management has to go through, they have to invest in their e-commerce platform, but there is a time for everything and trying to go through two capital intensive transformations at the same time while cutting massively your cash generation doesn’t seem to be the right approach…

It can work. But the risks are huge and vision/execution must be flawless.

Investment Strategy

So as I shared earlier, Alibaba does not meet my criteria anymore.

The thesis isn’t broken per se but it isn’t true anymore, not in the form I was expected and as management decided to play the game in hard mode, the market is not going to reward them... It already struggle to accept the risks to fund AI datacenters in the west so it certainly won’t reward a Chinese company with comparable challenges plus spendings on another vertical with lower ROI.

What if they run out of funds? What if they execute wrongly and have to come back on some of their decisions? What if they misunderstand the market or the technology’s potential? What if one vertical slows down the other? SO many more what ifs.

They have/had the chance of being the #1 AI service in China with a complete dominance while keeping their status of #1 e-commerce platform but decided to gamble both with massive spending instead of having a clear focus.

It doesn’t meant they won’t succeed. It means they are making it harder than it should and the market doesn’t like harder than it should. One mistake could set them years back and slow everything down.

The market obviously doesn’t like it. The stock lost its W50, lost its higher low and fell harshly post earnings with high volume, meaning lots of investors unloaded their shares accepting a lower price for it.

I am very bullish on China and on Alibaba, but the gamble they are taking isn’t worth it right now, the market doesn’t want it: so I wouldn’t want either. I did not have a position personally and was waiting for the earnings to take a decision, which is that I’ll stay out. And as the stock doesn’t meet the criteria anymore, I also closed the position on my stock picking follow up website.

This is the end of Alibaba for me, at least in term of positions. I will continue to follow the company as a massive potential and the #1 Chinese cloud, but investment wise, I will need more to start another position.

As shared at the beginning, I am closing a position with great returns and am very happy with the trade. Hopefully, Alibaba will give us more. Until then…

Farewell.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

I found GlobalFoundries pretty randomly, not even looking at the trendy photonics narrative, but because my screener found it which is surprising as most photonics names are already up hundreds points over the last months. And after digging for a few hours, I followed Druckenmiller’s method:

Buy first, study later.

And as I studied more I bought more, so I already own a position today and I will detail everything with you, from what photonics are to why GlobalFoundries was left behind, and what we can expect from the stock.

Copper, Optics, Photonics, Scale In/Out.

Lots of concepts, but we’ll go step by step without going too deep into the details; we aren’t physicists. I already talked about data centers and how they are built in my Arista thesis, I’ll go over some vocabulary and concepts so we all start from the same base.

• Internet is the name given to the physical interconnection between data centers, usually done by cables, sometimes submarine, sometimes dug below the ground.

• A data center is the name given to the physical interconnection of hardware at the same location, under the same roof.

• A rack is the name given to the interconnections of hardware within one vertical… well, rack.

Companies buy hardware, interconnect them together within a rack, interconnect racks together to “create” a data center, and then interconnect those data centers, creating “the internet”. Brief and oversimplified, but effective.

Now we can start talking about interconnexions themselves.

Copper & Optics.

Hardware communicate with each other via electrical signals passing through cables. Originally - and still today, many of those cables are made of copper for two main reasons: its conductivity and its price - it’s cheap.

As technology evolved, copper showed its limitations in term of bandwidth and efficiency. Signals are lost or deteriorated as density increases and the metal heats up past a certain speed of signal exchange, which creates constraints.

New technologies emerged to answer those two issues and help signal transit faster: optics. More expensive but much more efficient.

With this, data centers evolved, using optics in some cases and copper in others, the latter mostly within hardware, racks or compute clusters, while optics were used between racks or nearby data centers - what we now call scale-up and scale-out respectively in the AI era.

Let me introduce one term that matters a lot for AI data centers: east-west traffic. At a high level, east-west traffic simply means data moving between machines inside the data center rather than in and out of the data center. Traditional applications were dominated by north-south traffic, users talking to servers. AI flips that model. Most of the data now moves laterally between compute resources and the switch fabrics that interconnect them.

In AI systems, GPUs and XPUs must constantly exchange data to train and run large models. That lateral machine-to-machine communication is what now dominates network load.

This shift to what is commonly referred to as scale out and scale up networking exposes multiple system bottlenecks and forces a reset towards maximizing system level efficiency on multiple levels. At the same time, power consumption is approaching 1,000 TWh annually. When power becomes the constraint, the industry no longer optimizes individual components. It optimizes entire systems. Interconnect efficiency becomes critical.

GlobalFoundries Q4-25 Earning Call

Upgrading systems means upgrading every parts, focusing on bandwidth and latency, or in English words the speed and quantity of information which can be transferred per unit of time. The more, the better.

Compute Intensity

What we call AI is the result of compute clusters iterating together until they find the answer to your problem, like hundreds of humans brainstorming until they find the final answer.

We’ve already talked about GPUs; the brains, the compute capacity, how they work and how Nvidia revolutionized the industry. But as shared above, this isn’t enough anymore. As training/inference developed, models complexified and only increasing the brain capacity wasn’t enough, we needed those brain to communicate with each other faster; the nervous system was weak, incapable of matching the “thinking” pace.

Imagine if you find the answer to your part of the equation, but you have to wait ten second before communicating it. Imagine if hundreds of brains had that limitation. This is what I mean when I say the nervous system is weak. And this is what management meant when they talked about the industry’s need to optimize the entire system.

So the entire chain was upgraded with optimizers like AsteraLabs’, improved connectivity like Arista Network’s or Nvidia’s NVLink, and more. But we are still limited as demand and innovation isn’t slowing, on the contrary, which brings us to another barrier, not yet crossed: optics, the nervous system.

This is where physics takes over. Using copper becomes fundamentally challenged beyond 200G per lane. It cannot economically support the reach, bandwidth density, or power efficiency required for sustained GPU to GPU communication at scale.

GlobalFoundries Q4-25 Earning Call

As I said, copper and optics still coexist. In traditional data centers, each piece of hardware has a purpose and the flow of data is sequential, each fulfilling a need, which means data transit isn’t the same between all hardware and copper has its place.

With AI datacenters, data transits constantly between GPUs and all GPU have the same function: to compute and communicate the output with the “hive”, which collectively produces the final result. This means communication between hardware within the rack and across the data center is constant and must be as fast as possible; maximum bandwidth and minimum latency.

Any slowdown in the chain means a slowdown in total compute, a waste of energy and GPU uptime - and therefore money.

This is what photonics are looking to solve: transferring copper connections into optical ones further along the chain, removing the conversion delay between electrical signals/light and accelerating the transit globally, increasing data transmission speeds manyfold and therefore the compute capacity of an entire data center.

The challenge is massive. It’s not just unplugging copper and plugging in optics, it is rethinking entire hardware connectivity within and between hardware. To help you visualize, here what connectivity looks like between hardware today.

An optic fiber (yellow cable) plugged within what we call SFPs which serves as a bridge between light and electrical signals - passing through copper, within the hardware. The data transit from a hardware through copper to a converter which transit the signal in light to another converter doing the contrary and then sending the signal through copper again. Sometimes, the signal simply passes from copper within the hardware to copper cables, towards copper within the next hardware.

Imagine a highway limiting the number of people passing through. Say 2M cars can drive simultaneously. But in the middle, there’s a section where only 1M can pass (bandwidth), controlled by tolls that require travelers to change their car engines (latency and power). That creates waiting time, slowing down the entire system.

That’s what a single copper connector does within an optical system. The system is only as strong as its weakest link. This is the state of connectivity today.

SiPho and CPOs

The only solution is to replace copper and all those connectors. Enter SiPho - silicon photonics. Engineered semiconductor components designed to transmit light, and CPOs, the finished product filled with SiPho to interconnect hardware between each other or components within a hardware with photons only.

This semiconductor is meant to replace the system above entirely and you can imagine why this isn’t just a plug and play difference, this is a complete rethinking of hardware architecture - which won’t happen overnight, replacing copper within and between racks.

This doesn’t mean copper disappears; it will still be used where cost efficiency matters. But high-performance compute will increasingly rely on CPO. Jenssen confirmed this need and tendency a few days earlier at Nvidia GTC.

The AI factory is the new unit of computing. To connect a million GPUs, we can no longer rely on the old physics of copper. Silicon photonics is now central to the next generation of AI infrastructure.

By integrating silicon photonics directly into our switches, we are shattering the old limitations of hyperscale networks. We are moving from a world of pluggable parts to a world of integrated light.

We have squeezed every last drop out of copper. At 224G per lane, the physics simply demand light. The transition is no longer a choice; it is a thermal necessity.

The technology is ready, but the supply chain is not yet at the 'Industrial Scale' required for zero-defect AI factories.

Nvidia GTC keynote

As we reach physical limits in current data center architectures and demand for compute continues to accelerate, the only path forward, instead of waiting years to build more capacity, is to innovate and improve compute quality: more performance per unit of space and energy, as I’ve said many times.

This sets the stage for the company we’ll discuss today.

GlobalFoundries & Tower Semiconductor

So, again: SiPho are semiconductors, which means they are manufactured in fabs, just like GPUs and ASICs. And yes, TSMC is one of the main fab working on this as they are the preferred manufacturing partner for GPUs and ASICs for companies like Nvidia.

But they aren’t the only ones, and they’re not even considered the leader in this space. Their advantage lies in integration: they can embed CPO directly into GPUs they manufacture, making the entire workflow cheaper and faster.

GlobalFroundries is one of the key player of the sector. As I said earlier, I found it pretty randomly through my screeners and it surprised me as they are build to find potential/starting uptrend: higher lows/highs, high volume and breaking out their W50. This is partly why I believe there is something here.

Not only is the narrative hot - it’s extremely hot, but fundamentals are strong and price action is clean. Nothing happened with the stock since its IPO in 2022 but in recent months, we’ve seen rising volume and flat price action, which is typical of accumulation before a breakout. That breakout came in January and we are not going through an healthy retest.

This is demand, and I quickly understood why when I discovered that its main competitor - outside of TSMC is Tower Semiconductor, one of the best-performing stocks of 2025, up 36% year-to-date after partnering with Nvidia this February for silicon photonics.

So my first question was: why is Tower trading at such a premium compared to GlobalFoundries if both are considered key players in photonics? And I found a few reasons - despites Nvidia’s partnership. But none that justifies such a gap.

Specialization. This has become increasingly important in AI, as specialized hardware means performance. Tower focuses on low-volume but highly customized SiPho and related technologies. They don’t operate at scale but instead prioritize deep customization. They also consider themselves the leader in the sector.

They know how precious, especially from Tower, SiPho demand is, as we are truly, by far, the leader in silicon photonics.

Tower Semiconductor, Q4-25 Earning Call

GlobalFoundries is a much larger company with more verticals, more revenue streams, less specialization. They have a higher volumes and more capacity. But while Tower claims leadership, the numbers are more balanced.

Silicon germanium and silicon photonics revenues represented 27% of our corporate revenues, or $421 million, up from $241 million, or 17% in 2024.

SiPho revenues alone were $228 million in 2025, up from $106 million in 2024. Specific to the Q4, RF infrastructure revenues were 32% of corporate revenue, with SiPho having achieved $95 million, or a $380 million annual run rate.

Tower Semiconductor, Q4-25 Earning Call

GlobalFoundries does not break out SiPho revenue as explicitly, but they stated that their optical networking business generated ~$1B in revenue, with SiPho expected to reach a $1B run rate by 2028 - or more.

Clearly, CPO in the scale-up sense is gonna be an accelerant to the market. But honestly, you know, if CPO takes a little longer, you know, we still stick by that number. We’ll just ship more pluggables as CPO, you know, catches up and honestly, could go the other way as well, where CPO could be bigger sooner, bigger faster, and that scale-up networking could actually drive that percentage of that billion-dollar plus run rate, you know, up above 1/3. That’s why, you know, to your point, we’ve pulled in the expectation of, instead of a billion-dollar plus revenue in 2030, it’s really a billion-dollar plus run rate before the end of 2028.

GlobalFoundries Q4-25 Earning Call

This implies ~50% CAGR over the next three years. Tower is guiding toward comparable numbers with currently ~$380M SiPho run rate, which is significant.

Extrapolating from GlobalFoundries’ numbers - 50% CAGR to a $1B run rate by FY28, we can conclude that they generated ~$250M with SiPho in FY25, implying a run rate comparable to Tower for FY26 with expectations of nearly 100% growth for the year.

Tower might be the leader, but we are not talking about a 90/10 split, more like 55/45, especially considering both companies have comparable products and roadmaps. This contrasts with Tower’s management claim that they are the only company selling 1.6T connectivity hardware.

For long range and coherent applications, we are shipping into 400G and 800G ZR+ solutions, as well as 1.6 terabit coherent interface modules. In scale-out data center networks, we’re enabling high bandwidth, low loss links with DR4 and DR8 architectures, scaling to 1.6 T, 3.2 T, and beyond.

GlobalFoundries Q4-25 Earning Call

So yes, Tower is specialized in lower volumes and higher customization. That works well in the early stages of a new industry but as the market scales, large-volume production of standardized designs will become necessary and that requires industrial capacity and execution. Volume.

I am not saying Tower will slow down, they certainly won’t, but there is a clear competition between both companies, each with their own advantage and capacities while the market is shared, not dominated.

Size. There is a significant difference in scale and diversification between both companies.

This has a major impact on market’s perception. Tower looks like a high-growth story while GlobalFoundries looks sluggish even though the SiPho opportunity is similar with similar growth. This is basic math, it’s harder to grow from $6B than from $1.5B. At current revenues, SiPho represent ~22% of Tower’s revenue and only ~5% of GlobalFroundries’

And it showed already as Tower’s stock appreciated the last months on accelerating quarterly growth while GlobalFoundries’ acceleration is also real in SiPho but not showing at the company’s level. GlobalFoundries’ legacy business is a drag and that is probably not going to change so the alternative is to have a strong growth on SiPho which shows in the company’s profile - which can happen.

Yet, both companies have comparable operating margins which should increase with higher-value products.

So growth might look a bit slower for GlobalFoundries but its cash generation is comparable while operating verticals with lower margins and owning the full cost of maintaining/expanding its infrastructure. Not bad.

Capital efficiency. Tower is not spending as much as GlobalFoundries; they focus on using capacity rather than building it. They install their equipment within third-party infrastructure and rent space, meaning their fabs are effectively leased capacity which reduces CapEx. GlobalFoundries owns its fabs and must fund expansion/maintenance.

This is a structural difference; both models have advantages and inconvenient - lower costs and CapEx for Tower vs control and scalability for GlobalFoundries. Tower is facing issues with Intel for example which is kicking them out of its New Mexico fabs, forcing Tower to relocate production to Japan, causing delays.

Intel has expressed its intention not to perform under the September 2023 Fab 11X agreement. We are presently in a mediation process.

Tower Q4-25 Earning Call

Not owning infrastructure can be financially efficient but creates operational risks, especially in a sector as strategic and capacity-constrained as AI hardware.

On the contrary, GlobalFoundries owns and operates its fabs in strategic geographies, including New York.

To solve this challenge for our customers, GF launched plans for our silicon photonics and advanced packaging facility in 2025 in Malta, New York, to establish high-volume manufacturing capability to support the CPO ramp. The result of this investment is expected to deliver a fully integrated silicon photonics production flow from silicon substrate to known good optical modules. This capability will dramatically shorten the manufacturing cycle time and also provide rapid turnaround from wafer fabrication to full module yield all under one roof.

This could be an advantage especially given the importance of U.S. manufacturing in Trump’s administration - tariffs, regulations, capacity constraints, subsidies, taxes...

These are the main differences between the companies. They are not fundamental but mostly in term of by perception - to my view.

Analysts’ question: One of your competitors talks about having over 80% of the silicon photonics market today, but you have very similar revenues to them. It seems like you have at least a similar trajectory into 2026, I mean, to 2027. Can maybe try to explain the discrepancy in their view versus how GlobalFoundries sees things and kind of more broadly, how might your platform approach to silicon photonics versus Tower’s more customized solution set play into this perceived difference? Thanks.

Mike Hogan: Let me just put, like, some framing and some context around how we think about where we are in Silicon Photonics. You know, you can talk to the Tower about what they think, but I’ll give you our perspective. You know, I think just simply put, we think we’re the largest player in Silicon Photonics, full stop. You know, we’ve built that position based on a decade of sustained investment and benefited from the deep customer and partner learning cycles that go along with that. Just generally have, you know, benefited from a lot of time over target to feel like we have the largest market share of all the pure play Silicon photonic foundries.

GlobalFoundries Q4-25 Earning Call

GlobalFoundry is not behind, but it looks like it. But it doesn’t have less volume or less potential.

Advanced Micro Foundry & MIPS

Important to talk about those acquisitions late 2025 which changed GlobalFoundries’ profile and competitive positioning.

AMF is a Singapore-based pure-play photonics company focused on SiPho, directly competing with Tower. This is why GlobalFoundries remains the #1 SiPho foundry and the source of its potential, now integrated within the company, its scale and financial resources to expand operations.

MIPS is an autonomous entity within GlobalFoundries, focused on physical AI processing - low-latency and real-time compute for robotics, automotive, and edge applications. A key sector for future growth verticals.

The Thesis

It seems to me that Tower’s stock has been better marketed than GlobalFoundries’, but that does not mean it has a fundamentally superior business, especially in a market that will require industrial-scale production.

We had referred to as the $380 million run rate that we had in Q4, take off of that some small amount of NRE, which we don’t specify, the silicon wafers that we shipped for the Q4, that exact amount of silicon wafers by capacity, we plan to have five times more than that in the Q4 of 2026.

Over 70% of the total SiPho capacity is either presently reserved or in the process of being reserved through 2028, firmly backed with customer prepayment.

Tower Semiconductors, Q4-25 Earning Call

If I were to point to the true leader, it’d be TSMC, not Tower. Not because it has the best photonics tech but because of its integration within GPU and ASIC manufacturing workflows and its customer base, which will be happy to integrate CPOs directly into chips by the same manufacturer.

That said, demand for photonics is real, and TSMC alone won’t have the capacity to meet it. GlobalFoundries and Tower will capture part of the market and I struggle to imagine one being excluded as volumes ramp over the next few years.

Sure, GlobalFoundries doesn’t have a partnership with Nvidia but it works with major players like AMD and Broadcom, both significant in the ecosystem with AMD supplying GPUs to hyperscalers and Broadcom designing Google’s TPUs.

GlobalFoundries is a strong development partner and a key supplier for our next generation of optical networking products.

Near Margalit, VP and GM of Optical Systems Division at Broadcom

And if Nvidia is the one making the market moves, we know they can partner with many companies within the same industry as it is in their interest to have the most diversified supply chain, not just please one or another. So new partnerships could happen as volume demand scales.

Bottom line, GlobalFoundries is and will be a key player in both photonics and CPOs, it might look a bit sluggish compared to Tower Semiconductor but treat comparable volume at comparable margins. The opportunity seems be here fundamentally, while the market is ignoring it as growth looks (and is) slower due to legacy verticals…

From a risk perspective, we are looking at a key player trading at average/low multiples while the industry it represents if reaching new highs and demand will accelerate. There is always a downside to any stock, but the risk reward in this specific case looks massive if photonics were to scale and volume were to go towards GlobalFoundries - and management says it will.

We also need to keep in mind the clear stock accumulation. If the market wasn’t positive, we wouldn’t see growing volume and breakouts, holding the W50.

The $1M question is: is it anticipation, speculation? Or because something is coming?

I already bought shares at ~$42 while the W50 sits ~$40, I bought before completing my research and again yesterday while writing this article as my research confirmed my suspicions. The main question is how fast demand can grow and how will the market react to a slower global growth, while the narrative and the SiPho growth should accelerate.

Will it push the premium as high as higher growth names because of the company’s strategic importance and volume capacity, despites a slower global growth, or not? I’d either way assume that even a slower growth in a key sector is enough to justify higher multiples than today. Really higher, while any partnership or positive catalyst - and many could happen, will trigger strong reactions/momentum.

Once again, it’s all about risk/reward, and this one seems solid.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

I will detail the five best opportunities I see in the market today and for the year, following my system’s framework: Fundamentals, Narratives, Entry Price, Target, and Timeframe.

A clear why and how with five high R:R (Risk/Reward) opportunities, effective today.

Transmedics

I’ve been saying this for months but comes a time when expansion is inevitable after months of compression as fundamentals remain damn strong.

Fundamentals & Narrative.

TransMedics is the most significant innovator in the transplant industry, offering an end-to-end service with "warm organ" conservation hardware. The company enabled the market to expand, improving organ collection and transplant quality. It now is undergoing FDA clinical studies for its new lung and heart hardware, while also developing kidney hardware and expanding into Europe.

There isn't a complex narrative here but the stock has some advantages compared to many others; A defensive sector, not concerned by AI, spending, tariffs, geopolitics, consumer spending, recession… Great diversification.

Investment Plan.

My assumption is that TransMedics is a $150+ stock, probably closer to ~$180 before year-end and above $200 by end 2027, assuming solid - not even perfect, execution. My model suggests even higher in a bull case, but I prefer to stay conservative with my "optimistic base case."

From today’s $125, we are looking at ~50% returns in my base case and over 100% in my bull case within ~1.5 years.

Price action-wise, the stock has a strong support on its weekly 50 holding for more than a year by the market and the CEO himself who bought millions of dollars worth of shares at ~$116.

The risk/reward on this stock is unequalled - although our patience is tested… Even assuming a 10% drawdown as a stop-loss, you are looking at a 4:1 ratio based on conservative optimistic assumptions, and up to 10:1 in a bull case over two years.

Add to this its defensive characteristic; I’m not sure you can find much better in the entire market.

SolarEdge

Once a market darling, this sector is now completely ignored.

Fundamentals & Narrative.

A cyclical sector hammered by high interest rates. SolarEdge enables residential solar installations by selling hardware systems. Demand for energy autonomy is a logical long-term trend and current geopolitical conditions will just push the narrative even more in the next months.

Plus some structural improvements within the company with better products quality, upcoming lower interest rates and constant demand, the narrative should rapidly materialize in the financial results.

Investment Plan.

The stock is down 90% from its ATH. That isn't to say it must return there, but to highlight how violent the market was - from ridiculously high to ridiculously low.

The stock is now holding its W50 and pushing higher on improving results and returning demand - and soon a noisy narrative. My personal target is ~$100 with a two-year timeframe and a stop-loss below ~$28.

There is a solid R:R here, but it carries execution and external risks (tariffs and rates). It offers a higher R:R than TransMedics based on expected returns but with a much higher risk profile, hence why it is only second.

Nordic Semi

This is one of the few AI hardware narratives which didn’t start yet: AI Wearables.

Fundamentals & Narrative.

AI wearables require extreme optimization for both compute and battery life, a complex task even for non-AI devices, so when you add intense AI compute… This was highlighted by the Meta Smart Glasses which only have a few hours of life in usage. Not cool for glasses.

Wearables are the future. AI is also the future. Meta’s glasses are selling out, and production is expected to triple by FY26 while other tech giants enter the game, and Nordic has the best-in-class products for all of them.

Investment Plan.

The 2022 cycle ends and we are starting a new one. The stock has been making clear higher highs for months and held up well recently while other tech stocks struggled.

If the cycle ramps up as expected, this stock has ~100% upside over the next 18 months, easily. A close below the current higher low (~15% below today’s price) would invalidate the trade. This offers a ~6:1 R:R at a healthy on a not "cheap" valuation but within a strong narrative of the tech hardware market.

Solid, although again, it comes with execution risks plus a constant demand for those AI wearables, even growth.

Nebius

I’ve been wrong. I haven’t followed the system I preach. I am aware of the irony: “Do as I say, not as I do.” But I am back now.

Fundamentals & Narrative.

I am implementing more safeguards to avoid repeating past mistakes. That said, Nebius is an amazing opportunity, even more after Nvidia’s $2B investment.

Compute demand is massive, and providers are supply-constrained. Nebius is emerging as a top-tier provider due to its focus on full-stack quality. While the previous narrative focused on "unhealthy" spending and indebted customers, Oracle’s latest quarter combined with OpenAI new funding started to shift market’s perception of the sector.

If that happens and the market starts another leg up on the sector following constant growth and demand for compute, Nebius is a prime candidate for returns with analysts’ average target pointing to ~$160, as high as $290 for the most bullish.

Investment Plan.

The reason Nebius isn’t higher on my list is the R:R volatility. I’ve decided to buy again due to the massive volume and breakout following Nvidia’s investment, rewarded by a 17% jump in three days.

Breakout and high volume is a signal that buyers are accepting higher prices.
Price action matters, yeah.

This also means the conditions to close the position based on price action are more blurry without clear bottom, except for one ~25% lower. Vertiginous, but the opportunity remains massive here for a bull who sees a path to $250+, which I believe possible with a combined great execution - factual, and the return of a bullish narrative on compute demand and financial health - potential.

That being said, I would never have been in this situation if I had followed my system as I should, and held/bought more on the weekly 50 ~$90.

Edit: I wrote this Sunday before the announcement of Meta’s deal up to $27B which of course pushed the stock even higher. A great example of price action’s importance. I bought because of a large breakout on large volume. Price action anticipates.

Alibaba

This isn’t another compute play; it is a diversification and deep value play.

Fundamentals & Narrative.

Alibaba is to China what Google is to the West: a vertically integrated AI/tech giant. They own the hardware, software, logistics, and the customer base. The price is currently suppressed for two reasons:

1. The “China Discount” (Geopolitical risk).

2. Unfair growth comparisons.

Assuming cloud growth continues, Alibaba should be a 15%+ revenue growth company trading at the multiple of a dying e-com despite its accelerating AI verticals. A Western company with these metric would trade at $300. The China discount will always exist but a $200 is very reachable this year.

Investment Plan.

My concern is that it lost its W50 for two weeks now, mostly on geopolitical tensions on Iran while "propaganda" about Taiwan aren't helping.

China is supposed to be Iran’s ally; the reality is that they are opportunistic; they won’t support one if the cost outweighs the benefit.

Narratives and price action have little to do with Alibaba. But the market trades on narrative and sentiment so we cannot ignore them, and I I do not buy or hold a stock below its W50.

I will look to buy again if the price pushes back above that level in the coming days, which could happen with earnings this Thursday. For now, it’s a wait and see situation, with a high potential if the world were to quiet.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

So today we’ll review the importance of not letting sentiment take over, as well as my personal system and the safeguards I use to avoid emotional mistakes, and Tomorrow I’ll share the top five assets that currently fit my system.

Why Systems Matters

The answer is simple: investing solely based on opinion doesn’t work. It might work for some times but it fails over long decades. All super investors have a system and it’d be fullish not to have ours.

Druckenmiller sizes up aggressively on high risk-reward after a deep review of the trade and is famous for his “buy first, study later”. Buffett and Munger are known for buying value below their assumption of intrinsic value & popularized “moats” - which is misused 99% of the time nowadays. And Lynch is known for focusing on growth, categorizing stocks and ignoring volatility.

These are systems. Different from one another, but all successful because they are adapted to the investor’s personality and objectives. Looking deeper, we’d find checklists and methods to follow those systems in their stock picks.

If you don’t have a system… You’re very probably doing it wrong, or doing it for not very long, and that should change. Otherwise, you will make mistakes. I’ve shared mine this year with Nebius; a big, costly mistake that happened only because I ignored my system. We’ll talk more about that tomorrow. Bottom line being a system shields you from yourself.

Take the last year as an example. In roughly 360 days, we had

• A tariff war and escalations with China.

• A stealth attack on Iran by the U.S.

• The removal of Venezuela’s president.

• A never-ending war in Ukraine.

• More conflicts in Iran

• Now, disruptions of oil supply.

This comes after covid, other attacks on Israel, interest rate hikes, and countless other events. If you told me that none of this affected you emotionally or made you question your portfolio, I’d say you’re lying. And if I asked how many of you ended up rushing a pick - to buy or sell, I’m sure the “guilty” rate would be high.

Systems prevent that. If you follow one, you can go through your checklist and tell yourself.

No. This shouldn’t be done, so I won’t do it.

My System & Stock Picking

The situation in Iran hasn’t changed. It has now been a week since oil transport was disrupted in the region and the disruption now is real. It is being managed as most countries are using their strategic reserves, but a week of delays means ships that were navigating have reached their destinations and depleted their stocks while new ones remained stuck. And shut down or damaged production sites will take weeks or more to start again.

The question now isn’t will there be consequences, but how big. They still could be brief and reasonable but we’d need the situation to be resolved soon for it to be the case.

In such situations, you know things could happen, but you have no idea if they will. That’s where systems matter. Your system should drive your actions, not your personal perception of geopolitics which is likely as flawed as mine - nothing wrong with that the subjects are so complex.

If your system says to buy, buy. To sell? Sell. Here’s mine to illustrate

My System.

I know this sparks debates in the investing community but I do rely heavily on price action because it tells me what the market thinks and what it fears. This is one more selection process which help me filter stocks worth buying.

My process is straightforward.

I have a clear selection process.

• Strong fundamentals and narrative.

• Within an uptrend (above weekly 50 with higher highs/lows).

• At a reasonable valuation based on my assumptions.

A clear buying process.

• Identified supports.

• Breakouts with volume.

• Weekly 50 retests.

And a clear selling/trimming process.

• Trimming on extensions (RSI heating up + extension from key averages).

• Selling under three conditions only.

  • My original thesis is broken.

  • The stock falls below identified support/weekly 50.

  • I identify another stock with a better expected R:R.

That’s me, it shouldn’t be you or doesn’t have to be. What matters is having clear conditions for every action you take.

This is what this Substack and stock picking in general is all about: finding high potential stocks and concentrating liquidity on them. I track all my picks on my website for follow up purposes, but I do not own all of them. I only focus on those I consider the best opportunities available at any given time.

Liquidity Attribution

Concentration is key. I usually hold fewer than five positions at a time. I can do so because I have a crystal-clear plan for every pick. I know why a stock should appreciate, the price I want to pay, and the target price I expect it to reach.

I’m not always right, I certainly never have everything playing as expected, but I don’t need to. A roadmap helps me make emotionless decisions, leave a stock behind if necessary (as I did with Duolingo, Palantir, and Hims) or go heavy on a mispriced opportunity (as I am with Transmedics today).

It helps me make decisions based on my assumptions, not my emotions, with a clear plan, clear conditions.

Everyone calls this game “stock picking,” but its real name is position sizing. I like to say that we aren’t here to play Pokémon and holding 10 or 20 positions is most of the time a waste. Our goal is the find the best ones, not the most.

Imagine a $100k portfolio with ten positions of $10k each. If one returns 100% in a year, your total portfolio is only up 10%. You’d be better off putting 20% into a “safer” name with a 50% expectation to generate that 10%.

If you don’t think in total portfolio returns, you aren’t playing the game right. So many believe a small moonshot could make them a fortune but it is rarely the case and most would be better off with much bigger positions on safer stocks. The game isn’t to find a 10 bagger, but to be heavy in that 10 bagger.

If I expect a stock to return 100%, I go big. If I’m scared to go big, my conviction isn’t strong enough so maybe I just shouldn’t own it and should grow another position, maybe with a lower upside but with a bigger position.

You won’t always pick the #1, but you’ll still be better off with a 30% position in the #3 performer than you’d be buying 5% position of the top 10.

The game is about total portfolio returns.
Not stock pick returns.

Diversification & Dilution

We’ve discussed systems and liquidity attribution. Let’s talk about diversification and portfolio composition, both vital when managing a concentrated portfolio.

Many investors believe diversification means owning different stocks. That is plain wrong. Diversification means owning assets that are not driven by the same narratives.

If you own both IREN and Nebius, or ASML and TSMC, you aren’t diversified; you are diluted. This is poor capital allocation because these pairs are highly correlated. You’d be better off putting your capital on the highest potential, and then seek for true diversification elsewhere.

Diversification means buying assets that are uncorrelated to balance swings. At the moment I own Transmedics (Healthcare), Nebius (AI Compute), SolarEdge (Energy), and Nordic Semi (Tech Hardware). While all markets move on liquidity and will share some baseline correlation, these names won’t be as correlated as the examples above because they follow different narratives.

With my system, holding a handful of positions is easy as there aren’t that many great opportunities in every sector. We usually look at many stocks within the same sector hence the wish to hold many names but as they are correlated, this is like only collecting fire type Pokémons.

Go get that water badge.

No, when you get to the league, your team usually has six Pokémons, different types, only the best one of each, because that’s what makes a team efficient. Only the best on each of their category, with a clear plan on how to manage them during the fights to come, which to use when, to sacrifice against which, everything planned in order to win.

Clear selection. Clear strategy. Clear focus.

In Brief

We have covered the importance of having a system & the discipline to follow it. Systems are personal, they must be adapted to your personality, objectives, and strategy. There is no wrong system as long as it works for you & performs, but not having one is a guaranteed path to mistakes and underperformance.

Second, liquidity attribution is key. Small positions will not have a significant impact on your wealth. They might make you feel good, but if they don’t make you real money, what is the point? I now always ask myself this before pressing a button.

Will this matter?

This covers both the upside and the downside. Is this investment enough to matter? Is it safe enough not to jeopardize my account? Is this the best asset I can buy today?

We are all playing Pokémon, but not the catch ‘em all version. Do not mistake diversification for dilution. We are playing the final fight against the League Elite. And for that, we need the best stocks, versatile and optimized.

Not six Pikachus.

No. To win, we need to have a clear selection process, a clear method on which actions to do and when to do them, a clear focus on which names we want to optimize and use more than others, and a clear objective.

I’ll share my Top 5 tomorrow.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

How fast perception can change with lower valuation and positive comments.
Oracle was plagued by two issues.

1. Its cash generation was impacted by CapEx without any clear view on when that would slow down or end.

2. Its biggest client - OpenAI, wasn’t known to be the biggest cash machine on the private market for the same reasons: massive CapEx without any clear view on when that would slow down or end.

Both concerns were addressed during this quarter and, combined with a -53% stock drop, triggered a pretty positive reaction. Expectations are much lower at today’s price, and this is key to remember in the markets.

It’s all about perception.

Oracle is still cash flow negative as it spends more in CapEx than it makes in OpCF, but the market now has visibility on when that will change and on the potential as Oracle reported a continuous revenue acceleration made possible only due to its AI services, which continue to attract massive demand and growth.

This is step 1: demand for AI compute is through the roof; we always knew this, but it’s always great to have confirmation and show a clear ROI.

Multi-cloud database revenue grew 531% YoY. AI infrastructure revenue grew 243% YoY. Both also have demand that exceeds supply and a clear execution plan from Oracle that will rapidly turn that demand into profitable recurring revenue.

We created our multi-cloud partnerships with first Microsoft, then Google, and finally Amazon to bring the best database platform to all clouds. Those partnerships unlock an enormous backlog of demand. Our database customers who want to use our database in other clouds. This quarter, we achieved an important milestone.

Step 2 is cash generation, which comes in two verticals. First, slowing spending and second, margin capacity.

Slowing spending is not happening yet, but management did not announce any more of it which is already great news while their last round of financing - which the market judged unreasonable, was oversubscribed meaning real demand for it, despites risks.

In February, we announced our intent to raise up to $50B dollars in debt and equity financing, along with the statement that we do not expect to issue any additional bonds beyond this amount in calendar year 2026.

Within days of the announcement, Oracle raised $30 billion through a combination of investment grade bonds and mandatory convertible preferred stock, with a record order book that was substantially oversubscribed. We have not yet initiated the at-the-market equity portion of the financing program.

Add to this that, just like for Nebius, many customers are now paying upfront for their future compute, helping with the debt weight for future buildouts.

This is the first step towards slowing down debt risks: not taking even more of it. They probably will next year, but until then the market won’t punish the stock more.

The second step is to repay actual debt, and while demand is not an issue as we’ve seen, Oracle still needs to make money from revenues and that is also going in the right direction.

I know Buffett and Munger did not like to use EBITDA, but they also never invested in tech and sometimes we have to use it as this is the only method to judge compute providers’ cash generation capacity post CapEx spending. It isn’t only EBITDA margins which are growing, also AI compute gross margins ~32%, largely above the Street’s expectations.

Last but not least, the worries about OpenAI’s capacity to pay their bills are also slowly being reduced as the company successfully raised more money; a new $110B round completed at the end of February with Amazon, Nvidia and SoftBank. That will allow them to pay those famous bills, if only for some more quarters.

Bottom line, this was a good quarter driven by strong demand, less spending, growing revenues with healthy margins and cash generation. A quarter showing the Street that Oracle could actually pull it off, that its debt and the risks were maybe not as huge as they originally looked in the face of the opportunity ahead.

But that reaction is only due to the difference of valuation between now and then, the risk is certainly not the same at $300 or $150. It’s all about perception.

UiPath

I closed my position in UiPath when the stock lost its weekly 50 because that is what my system wants me to do as a concentrated investor: not hold weakness. But in the SaaS stocks’ current condition, losing this trendline doesn’t mean the thesis broke, it just means the market doesn’t want to buy those names.

The fundamental thesis was that UiPath would continue to accelerate as demand for AI enhanced automation would only grow, not quite like Palantir but following the same trend.

This quarter seems to confirm it with the second quarter in a row of ARR acceleration, and with a continuous acceleration in RPO, up 14% YoY for the quarter.

Yet, guidance points to ~$2.053B FY26 ARR or a 10.8% YoY growth compared to ~11% YoY FY25, meaning management isn’t expecting this acceleration to continue through the entire year, and this isn’t exactly what we want to see.

We are talking about a structural shift in how companies are supposed to use AI and we’ve seen this shift in the numbers with Palantir. Growth is accelerating because the tool is supposed to be so transformative companies’ cannot not use it, it would put them at a massive disadvantage compared to the competition.

Not seeing a comparable trend in UiPath doesn’t mean it won’t happen, but it means the need for their services is not as strong as I expected, yet. I did not expect Palantir like numbers today but I wanted to see acceleration, even a mild one, and it isn’t what this guidance points towards.

It might just be a bit early. I’ll continue to follow the company as the potential could be massive. But unless guidance is a bit sandbagged, I don’t expect to hold it this year. There is no point in putting money on it yet, in my opinion.

Adobe

Another name not worth buying yet to my opinion. And I don’t say it to be negative as I really like Adobe and their products, but loving their products doesn’t mean I have to love the stock.

Many believe that Adobe’s stock is down because the market is scared of disruption. I don’t entirely agree. I believe Adobe is down because the market cares about leaders and Adobe isn’t proving itself to be one in the AI era, yet.

Firefly is starting to show real results with a $250M ARR at the end of the year, up ~3x YoY and ~75% QoQ. So there is traction but this is kinda small at Adobe’s scale.

Another positive indicator is the ~80M creative MAUs, up 50% YoY, which translates to a clear interest in Adobe’s services and therefore in their AI services. The question will be about conversion and retention on premium plans, as using AI tools for fun as a freemium is one thing but holding paying customers is another.

So there is some traction, but like everywhere and for every tool with AI, and a $250M ARR doesn’t change much yet for Adobe and its $24.5B in revenues. The truth for now is that there is no acceleration, clear/rapidly growing demand, or margin expansion.

Adobe is a wonderful company but the market doesn’t reward those. It rewards future cash generation and Adobe is in a situation where they are very stable, very healthy, but without clear potential for acceleration.

And while everyone will tell you that Adobe trades at its lowest valuation ever - which is factually true, no one tells you that it also trades at its lowest growth ever and has never been priced in for such growth over the last decades.

Adobe was always a fast-growing company with expanding cash generation, and the market is only repricing the stock for what Adobe could become: a very healthy value stock without many reasons for cash generation acceleration or margin expansion.

I personally am bullish on the fundamentals and believe that Adobe will thrive in the AI era and will deserve higher premiums. But this is an opinion, a bias, not an investment strategy.

Until this is proven by the numbers and validated by the market, I’ll stay out.

Sportswear & Retail

I don’t own any of those names, but they are a great proxy for my Lululemon position as they provide a good overview of Western/worldwide demand for brands and clothing in general.

And despite stable earnings, most of them ended up selling off - although the market has been very touchy lately for other reasons so it’s hard to be sure that this sell-off won’t be bought back in the next few days.

American Brands

Four retail giants reported earnings this week (American Eagle, Abercrombie & Fitch, Victoria’s Secret and GAP). I won’t go over each one individually but will talk about the overall assessment of their earnings.

All of them show a comparable trend: quarter & year ended with better results than most - management included, expected at the beginning of the year, as 2025 started poorly.

FY25 ended well. They succeeded in pivoting and found ways to fuel revenue and comparable sales growth; clients went to their stores and bought their products more than the year before.

Guidance is very uncertain. They talked about healthy comparable sales and revenue growth in the single digits, but the key data point is that most growth comes from non-premium verticals or international demand. The Western consumer is present but picky, focused on cheap products, while real growth and demand comes from overseas.

Margin profiles should be stable; most guided to flat or expanding margins as they optimize their operations. But there is a very big “if,” as FY26 will be the first year with nonstop tariffs and those will put pressure. The real question is whether management will be able to compensate for them with price hikes or optimizations - cutting costs, AI usage, supply-chain optimizations, etc, and if the consumer will react positively to those.

If not, their guidance will miss, and with such uncertainties around consumers and external factors, the market isn’t very optimistic.

Overall, the quarters were correct, as was guidance. But these stocks had run up as management shared their plans to navigate a complex environment and succeeded during 2025 while valuations were very low. They are now back at healthier valuations with a picky local consumer and margin pressures.

Stocks were punished, not because of bad earnings or guidance, but because the market had already rewarded them and is now pricing in struggles for longer. Lululemon hasn’t been rewarded for anything yet as it hasn’t shown its strategy was yielding results, but will face comparable headwinds. It remains possible that the market rewards the stock if execution if comparable to these even with potential pressure during FY26.

On Running

I’ll treat On independently as a company I have followed closely for years which doesn’t compare with others, proved once again with strong earnings; double beat on both revenues and EPS.

The two important metrics for a brand are revenue growth and DTC proportion. As long as customers go directly to the brand to buy products, it means they want this specific brand and not just shoes, and that trend is now stable for On after years of growth.

Stable isn’t bad. When half of your products are sold directly through your own channels, upside is more limited. Still very healthy.

On faces another challenge: currency fluctuation, as the company is Swiss and reports in CHF. A 22.6% YoY growth would be 30.6% on a constant-currency basis, with America growing 23.4%, and margins staying high excluded of those currency fluctuations.

On does not face the same struggles as its peers, none. Growth is worldwide, guidance is healthy, margins are stable. The only headwind comes from being a Swiss company reporting in CHF, a currency even stronger than the U.S. dollar, which is terrible for an exporter.

I’ve already detailed how currencies impact export/import dynamics, so I’ll let you look at that, but this is what the market disliked about On’s quarter, or misunderstood. As we have more and more algorithms or investors looking only at numbers without trying to understand the situation, what they see is a 15% decline in net income, lower net margins, and lower cash generation.

But it isn’t a fundamental issue or a demand issue. It is a currency issue. That doesn’t change the fact that the stock is falling because of it but we know that this situation will change once currencies dynamics shift; On will be rewarded twice as much then.

Technology

My original thesis that hardware companies should do well and be rewarded by the market while the rest of the AI trade would fade or do nothing has been accurate enough since I shared it in December. Since then, hardware stocks have pushed higher while most of the market has stayed flat or fallen.

And hardware companies continue to do extremely well.

Broadcom & Credo

These two are excellent proxies for Astera Labs; the first designs computing units (notably Google’s TPUs), and the second is a competitor in scale-out optimizations.

Both printed revenue growth and expanding operating margins, sustained by very large/growing demand for their products as datacenter buildouts simply have no other option than buying better hardware until infrastructures are ready for physical expansions.

Broadcom notably sees massive growth in demand for custom ASICs, and revenues reflect that.

The thesis was on point. Customization is indeed the next step for AI buildouts, even if the market doesn’t reward stocks with as much optimism as it used to. Broadcom is still largely below its ATH while a quarter like this should have pushed it higher - although compared to the market this week, being green’s already something.

Most importantly, management confirmed one important detail during their call: copper is here to stay for a long time.

The best way to do that is use Direct Attach Copper. That's the lowest latency, lowest power, and lowest cost. You want to keep doing that, especially in scale-up, as long as possible. In scaling out, we're past that. We use optical. That's fine.

We can do it with copper, we can push the envelope from 100G to 200G to even to 400G. We have SerDes now running 400G that can drive distance on a rack to run copper.

All I'm trying to say is you don't need to go run into some bright, shiny objects called CPO, even as we are the lead in CPOs. CPOs will come in its time, not this year, maybe not next year, but in its time.

This is what I was already saying months ago in my AsteraLabs thesis.

Finally, a quick word on optics. This has been a big narrative before Oracle’s quarter & the debacle that followed, with many green candles printed on companies working on optical connectivity. For now, this is a fantasy, we do not have optic connectors within the hardware - they are only between racks. Copper is king and will be for some time as it will take years of R&D to get optic connectivity in such small environments. The gap in data transaction speed will be massive when the technology is ready, but we are still far from it.

It apparently doesn’t change anything in the market’s reaction as Astera Labs continues to be sold, to my biggest disappointment. I know this stock will rocket in a few months, quarters, or maybe years. This is the company that will allow datacenters to scale compute faster until infrastructures are sufficient.

But the market is worried about a few margin points instead of focusing on the bigger picture, and this dissonance between stock and fundamentals is also happening with Credo, which is more focused on scale-out but is facing comparable backlash: a gross margin guided to ~65% from ~69% today to focus on scaling.

Growth comes with spending as manufacturing faster cannot be done for free. This, plus their revenue guidance of ~50% growth, makes it look weak after 200%+ quarterly growth - volume naturally slows percentage growth. It is easier to grow from small numbers, while after more than a year of ridiculous growth, scaling becomes harder.

The stock was punished ~15% post-earnings. It has caught up most of it since, but is still stuck below key moving averages with no support below.

Both Astera Labs and Credo were punished for being hyper-growth stocks, and their stellar quarters or future R&D couldn’t change anything about that… They will continue to do wonderfully in term of business but it is hard to know when the market will reward them again.

Both are outside of my system, and their fundamentals can be as wonderful as they are; as long as a stock doesn’t fit my criteria, I won’t be buying it.

AmbiqMicro

I am looking at Ambiq as a competitor to Nordic Semiconductor. Both focus on System on a Chip (SoC) - a single chip that contains the necessary components to power the hardware. Ambiq has a specialty in compute, while Nordic has a specialty in low-energy connectivity. But now both compete to win the same market: low energy AI wearables, but do so from different verticals - we’ll see that a bit later.

And Ambiq’s quarter confirmed a very large and growing demand.

While the company was focused on low-margin power-efficient chips, mostly sold to Chinese manufacturing hubs for all kinds of wearables, this quarter showed a clear shift in demand, geographically and in terms of usage.

We took deliberate steps to prioritize customers who view our ultra-low-power technology as a critical enabler of Edge AI while reducing exposure to efficiency-focused, feature-neutral customers, primarily in mainland China.

This is the primary reason for the company’s double beat in revenues and EPS, as the AI chips sold to Western clients come at a higher price and margins. This is happening because AI is starting to be deeply integrated into wearables and requires more efficient systems.

2025 was a milestone year... we estimate that more than 80% of the units we shipped were running AI algorithms.

So demand in the sector is real, which confirms my thesis for Nordic & its pivot to integrate NPUs within its connectivity chips.

Ambiq management is ambitious for FY26, they talk about a potential $100M in revenue due to this new demand vertical with Q1-26 guidance sits at ~$21M, which suggests a clear acceleration through the year.

From my perspective, the market should still favor Nordic because its nRF54 is built to deliver the most efficient compute while Ambiq designed its chips to be energy efficient only. Nordic’s chips are additive to a core compute chip from Intel or Qualcomm, which are the real stars of the show. They only are here to help them improve compute performances, not replace them.

Integrating high-end compute into a radio chip is easier than integrating radio into a compute chip, especially when AI wearable demand is about efficiency. Sheer energy efficiency isn’t what manufacturers need. They need optimized compute and conectivities, and that is what Nordic sells.

That being said, Ambiq confirmed the rapidly growing demand for this kind of chip and as I believe Nordic has the better product; we can expect demand for their chips to be even higher.

Only good news here.

Consumers

I will not go into deep detail here, but it’s interesting to note some market behaviors, notably around Target’s earnings.

Target.

I’ve had the company on my watchlist for some time as the price action fit my criteria: a long downtrend that stabilized with growing volume and broke above its weekly 50. I did not share it because the fundamentals were shaky; yet the stock is up 21% YTD, more than 15% since it triggered my entry signal.

Why is the market rewarding such a struggling business? Restructuring.

The market stabilized the stock once it heard that word and started to reward it in anticipation of the first results, which finally became visible a year or so later. This quarter is the first one with positive comparable sales growth, expanding margins due to cost cuts, and EPS growth.

Nothing shiny for FY26, but guidance shows a business that could go back to healthy stability, and that would be enough to justify a repricing as Target was trading at its lowest multiple of the last decade only a few weeks ago.

This confirms that lots of sectors and stocks will continue to be repriced as the market turns away from risk and toward low-valuation healthy companies, even if they do not deliver triple-digit growth. If such results can yield ~20% returns in two months, I’ll continue to focus on comparable names as the risk/reward is incomparable to AI/growth names at the moment.

Smith & Wesson

A second rapid earnings review for the firearms manufacturer, which is going through a new cycle of growth, for sad reasons as it probably happens due to a feeling of insecurity, but the market is what it is.

The stock is up ~30% since I presented the thesis a bit more than a month ago, while the option play is up ~65%. The stock pushed higher on great earnings, as described a month ago: inventories cleaned up, new products in demand, price increases accepted by final clients, which also led to margin expansion and higher cash generation.

All of this is supposed to continue as guidance is also very healthy, with double-digit growth expected for the next quarter, which was my bull case.

I wouldn’t make my base case a 1.5x sales as growth is hard to predict and a double-digit growth year still seems far-fetched today, but if the trend continues we’d be looking at a higher multiple than today’s and easily a 20% - 30% push from my $10 target.

This has been an excellent trade and we’re now back at ~1.3× sales.

As this comes with double-digit growth, expanding margins, guidance pointing to more of the same, we could reach that famous 1.5× sales. As I am not in the trade I won’t need to manage it, but if some of you took it, congrats.

A great thesis yielding great returns.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

We are only a few days past their excellent Q4-25 but there are some information which weren’t shared then, so I thought this write up could interest some of you shareholders.

The Kidney Opportunity

This isn’t news: the kidney market is very large and has flaws that can be addressed by the OCS and its warm preservation and perfusion systems. Just like other transplanted organs, kidneys suffer from high organ loss and post-transplant complications, which have a financial impact on healthcare systems.

Today, kidney transplant post-transplant complication rate that requires the patient to go back on dialysis, that costs CMS up to $50,000-$60,000 per occurrence, called delayed graft function, is at all-time high at an incident of approximately 55%-60%.

Another nuance in kidney that’s very important to mention is CMS is financially responsible for all end-stage renal disease conditions in the United States.

The second is another very, very alarming rate, which is kidney discards or utilization of kidney donors. In the United States today, kidney utilization is at all-time low of approximately 60%.

This is the classic value added by the OCS across all organs: increasing both volume & outcomes, and it has already done wonders in livers, with concrete data proving it.

We are talking about a ~$3B market FY25 expecting to grow ~6% CAGR, maybe even more if TransMedics were to replicate its success with liver expanding the market and its market shares, which was ~36% FY25. This is napkin math but a 35% market share on a $3B market is ~$1B revenue.

Again, napkins math as we don’t know pricings for the service, etc… But it’s enough to illustrate the opportunity ahead.

Margin Profile & Cash Generation

TransMedics is healthy in terms of cash generation but I believe the market wasn’t happy with the potential flat or shrinking margins FY26 as the company invests in European expansion, kidney OCS, Gen-3 OCS while growing expenses to run NOP services - which are a large source of growth.

As I said in my quarterly review, these short-term expenses will increase margins and cash generation in the years to come. TransMedics’ CFO explained how margins will expand in the case of the kidney Gen-3 OCS and all Gen-3 OCS platforms after that one is approved and commercialized.

There are two areas that we are really looking for to capture additional margin.

One is the device itself. The device itself will be smaller. We have less part counts, which basically will allow us to reduce the cost of assembling.

Two, right now, a lot of processes are manual. We’re expecting to do a lot of that process with robots. Today we need a clinical specialist that needs to be babysitting the organ from the moment that is placed into the device to the moment that is removed from the device for transplantation to assess the organ. Well, imagine a scenario in which with that next generation, we can do that testing, that assessment and treatment remotely via a clinical command center in our headquarters. That’s actually the vision

The kidney OCS is expected to be commercialized - at least in trials, by H1-27, not only in the U.S. but internationally.

We already presented the design to the clinical community at the American Society of Transplant Surgeons winter meeting in January. We have another milestone in June here in Boston at the American Transplant Congress, where we are aiming to have functioning devices by then. Then it’s a race through getting all the testing required to present the dossier to FDA, and hopefully by first half of next year, we’re in the clinic.

We are seriously considering having a parallel path, both releasing the product internationally and at the U.S. at the same time for clinical programs.

TransMedics is moving very fast, and the expenses of FY26 could start yielding results as soon as 2027 with volume growth from Europe and margin stabilization with the Gen-3 OCS. We’re talking about a year of spending before concrete rewards in cash generation - if only stabilization.

Nothing says everything will work perfectly. But the potential is here & is very bullish. The market will have answers quickly: one or two years is a very rapid timeframe.

Governmental Push

I shared in a previous write-up that a one bull case would be a governmental push. Today, the transplant “rulebook” in the U.S. focuses on the initial quality of the organ. An organ which doesn’t have 100% chance of yielding a perfect transplant is thrown. This creates a situation where viable organs are wasted because Organ Procurement Organizations (OPOs) would rather play it safe than risk lowering their performance scores - and therefore waste viable organs.

But the U.S. system is trying to shift toward a model that prioritizes both volume and quality, pressuring OPOs to use all viable organs and maintain high clinical standards. It is not a “transplant everything” but a move to maximize organ usage.

This would force OPOs to do everything to ensure transplants, hence adopting the best hardware and service to preserve & monitor organs from donor to recipient.

All these initiatives are mandated on the entire community of organ transplant by bipartisan effort from U.S. government, HRSA, CMS, and others. That is ongoing, and there is new language coming out of CMS, new initiative, new language, talking about potentially opening up the opportunities for new entities to become Organ Procurement Organization to maximize competitiveness and maximize transparency.

We’re the only company that has the only ischemia-free technology for organ preservation. We’re the only company that invested in vertically integrated logistics for organ transplant, 100% dedicated to organ transplant. We’re the only company that developed a very sophisticated digital ecosystem to maximize transparency, and visibility of every organ that we carry.

We will express our interest to try to play a bigger role in collaboration with existing structures in that new Modernization Act of organ transplantation in the United States.

Europe is different. Hardware is selected and purchased by national health services, governmental agencies based on governmental budgets/clinical mandates, not on a scoring system or private economics. TransMedics’ expansion in Europe depends on these governmental decisions and will grow volume by proving superior outcomes compared to traditional methods; results already demonstrated in the U.S.

They are already operating in Europe, Italy mainly, and have a healthy growing volume while management is planning to build its NOP H1-26. They aren’t spending without visibility on demand for their OCS, this expansion is the sign that European countries want OCS/NOP, that demand is here.

They know they’ll be well received in the region, they wouldn’t move otherwise.

Hearts, Lungs & Guidance

Regarding the trials, nothing much. Management shared that the onboarding rhythm will be the key indicator of how well the trials are progressing until official results are available. This is what we will need to monitor: growth in heart and lung procedures. Any acceleration means that their new hardware is a net improvement.

Now it’s showtime. Now we need to transition that momentum into enrollment, into completion of the trial, because that’s really, that’s when we know that, you know, people vote with their commitment in enrollment and adoption. We can’t wait to see that in action in Q3 and Q4.

They also communicated that even though competitors did not want to take part in the comparison arm of the trials, they’d find a way either to force them or make the comparison themselves - to my understanding. They plan to reveal in April how they will proceed if competitors continue to refuse participation.

I think the competitor spoke loudly when they were unsure of their performance against the new OCS. I think that reaction alone is very telling. We’re not gonna allow that behavior to derail or delay the trial.

We have a plan. We’re gonna execute it, and we’re gonna announce it publicly at the upcoming ISHLT in April.

Lastly, regarding those trials, management confirmed that their FY26 guidance does not include the entire trial volume. It is effectively sandbagged to give themselves room for enrollment speed.

It doesn’t include any incremental impact. We have 300 patients in the case of lung. The first 100, which more or less what we do today, that’s included. On the heart side, part A, 175, that’s current. We’re assuming that that is part of our current baseline. However, in part B, 320 hearts, we don’t have that.

We don’t know how fast we’re gonna be able to enroll because of what we have mentioned before. That one, it can be an upside, absolutely.

Not all future volume will be part of the trials. But management confirmed they intend to complete them within maximum 18 months - by end of H1-27, meaning ~800 OCS cases both organs combined, and only guided for ~275 FY26.

Question’s how fast enrollment will happen and that is a fair question. But assuming the trials proceed without issues like clear signs that the OCS is harmful, which is hard to imagine but could happen, it would be surprising if TransMedics were to only do ~275 OCS case on the ~800 expected during 75% of the trials timeframe.

A fair expectation would be to complete at least half of them - hence ~250 hearts and ~150 lungs, if not more. For context, TransMedics performed 854 heart and 88 lung cases FY25. Lungs is a drag but as this new OCS is the only technology interesting to try for the organ, many OPOs could like to take part of the trial which is better than nothing - that is speculation though.

Based on the latest quarter, ~250 heart OCS would be worth ~$36M & 100 lung OCS ~$17M, which means a ~$11M additive revenue on top of today’s guidance. Any net increase in volume translates into net increase in FY26 revenue, and that is excluding additional sandbagging related to seasonality and European expansion.

More napkins math here as the trials aren’t billed at 100% and the comparison branch of the trial will add costs to TransMedics but this remains interesting to be aware of.

I know I may look very bullish. But this is what the data suggests, and as long as those trials continue to enroll, it means they are also producing results. I see no reason not to be enthusiastic about the future as long as the data is bullish. And it is.

I think it’s very important to note that my position is not just a hypothesis or the thesis or a dream. My position is based on the fact that we are sitting here with nearly 10,000 liver transplants in the U.S. We know what the data says. The data is unequivocal in liver transplantation. The OCS is superior to cold storage. The OCS is superior to competitive technologies. The OCS is superior to back-to-base model. The OCS is superior to other perfusion technologies that try to compete with the OCS. Period. Full stop. That’s what the market is showing. That’s how the adoption curves are being, you know, printed every quarter. It’s because of that fact. Because of that, I have a high degree of confidence that, one, our superb technology, supported by a world-class clinical leaders in the field, will continue to drive adoption in liver, and liver will continue to be an engine of growth for TransMedics.

We have a long way to go. We have a long way to go because when you’re sitting with superior technology at every metric, at every level, you can evaluate it.

It will be malpractice not to use OCS.

European Expansion

A few words on expansion, which is still ongoing with more to come.

We hope that Italy will be fully functional and you can feel the breadth of Italy by end of this year. Really, it’s a huge catalyst for 2027. We hope that by end of this year, we’ll announce who’s next in Europe, and we are working with several countries in Europe, Netherlands, Belgium, France, UK, and Switzerland are the kinda the near term opportunity in addition to others, of course.

Conclusion.

I was already bullish, and I am even more so now. Not because my bias has grown, but because the data continues to support the thesis I wrote more than a year ago.

There are many execution risks for the company: expansion, the Gen-3 OCS, the trials and the kidney hardware. Many things could go wrong - no doubt about that, but at today’s price, I’d remain bullish even with average execution. I believe valuation offers a healthy risk-reward still today. Unless we see delays, cancellations, or disappointing results in one of these areas, I expect strong performances.

I will continue to monitor heart and lung trial enrollment - the only key metric for the moment, remain attentive to European expansion and follow announcements related to kidney and the Gen-3 OCS.

Regarding the stock price, I expect Q1 to break records based on recent flight data, but the key metric to watch for stock appreciation will be margins. Growth is a given.

TransMedics is my largest position and my favorite risk-reward in today’s market. I am consistently satisfied with what I hear and see, and I hope that continues for a long time. Hopefully with the stock following the trend of the fundamentals.

That’s what we’ll talk about today; not necessarily about a defensive asset, but about a cyclical sector trading at very low valuations with potential tailwinds coming up in the next months that could push valuations higher on healthier fundamentals.

Materials & Chemicals

Both sectors have been doing really well over the last months, but let’s first talk about what they are. If we take a classic product manufacturing chain, it would look like this (very simplified).

Commodity → Transformation into materials → Material assembly/transformation to manufacture a product → Shipped to consumer

Most materials and chemical companies operate in steps 2 and 3, depending on what they sell. Some use molecules extracted from oil for example and transform them into chemicals; others buy those chemicals and mix them to form new products, etc...

The materials sector includes a wide range of companies - from selling raw chemicals extracted from commodity to selling finished products such as construction materials. The chemical branch within the sector is also very diversified in term of products and usage. Here are a few examples to illustrate.

We’ve already talked about Nutrien for example, which has largely outperformed the market since and is part of the chemical sector for agriculture.

As you can guess, these are very cyclical businesses; companies usually focus on one or two verticals serving specific sectors, demand for their products varies depending on demand for the final products. This creates boom-and-bust cycles and if we aren’t yet confirmed in a boom cycle, we’re certainly near the end of the current bust cycle.

Which is why most of these names have been very interesting in recent months, and why they could be interesting over the coming years. They’re not tech, sure. But hey, returns in tech or in chemicals are returns.

The Bear & Bull Case

Besides inherent cyclicality in their business models, chemical companies have faced other headwinds over the last few years.

Today’s Bearish Situation

1. China. Probably their biggest issue as over the years, China has specialized in exporting everything the West needs, including chemicals. And over time, the country has flooded the market with cheap ones. The infographic doesn’t detail per chemical but it shows the market shares change over the last two decades.

   

2. Demand has also slowed. Higher interest rates and weaker consumer purchasing power across the West reduced demand for end products, which in turn reduced chemical demand. Coupled with high volume from China and inventories, this did not help U.S. producers.

3. Profitability suffered, as you could expect. Lower demand, rising competition and oversupply; the worst combination for any capital-intensive sector. Growth slowed, margins shrunk and cash generation fell off the cliff, some even turning unprofitable.

This is why most chemical companies were crushed over the past 5 years.

Tomorrow’s Potential Bull Case

But dynamics may be shifting with a handful reasons for it.

1. Tariffs & local production. Trump focus on the U.S. will put pressure on imports - already is, combined with potential subsidies or tax reductions when producing locally like for solar systems, this could reduce competition as they become price competitive and increase profitability.

2. Inventory destocking. Companies ran through inventories instead of new orders after pilling up cheap Chinese products for years. As supply chains normalize and inventories are now low, any rebound in demand will lead to restocking while the global profile of demand will globally be healthier.

   As we have reported in the past few quarters, we've seen volumes improve as de-inventorying has ceased and demand has rapidly returned.

   Management characterized current inventories in the supply chain as 'quite low,' flagging potential for sudden product shortages should demand rebound rapidly.

   From Huntsman Feb-25 call

3. Interest rates. Lower rates benefit capital-intensive businesses (refinancing, new loans) and indirectly by supporting demand for financed goods like autos and housing, which are very demanding sectors in chemicals.

4. Restructuring. Many companies implemented layoffs, debt renegotiations and plant closures as they reached the bottom of the cycle and need to improve efficiency and cash generation for the next cycle.

5. Valuations. The entire sector got smashed for half a decade and many now trade at historically low multiples - for valid reasons as fundamentals also declined. If tailwinds materialize, repricing could follow.

6. Liquidity rotation. Capital has been shifting away from tech and growth toward cyclicals and defensives lately, we’ve talked about this at length already. And as the market cares about liquidity first, rotations are important to look at.

An entire sector with a few tailwinds after years of headwinds, dragging valuations to historical lows. There are execution and external risks to those tailwinds materializing, but valuation makes the risk-reward interesting here while the market is enthusiastic about those names.

Concrete Examples

As I shared in my March Investment Plan, I bought a position in Huntsman Corp., and shared Nutrien a few weeks earlier. I’ll focus on the first later but I will share others as all those names have one characteristic I like above all else: the market loves them.

This investment is a stock picked within a strong sector with high demand more than an investment on strong fundamentals of a single name. So it isn’t the only name worth looking at, the entire sector is.

I found ~10 chemical names on my screener the past 2/3 weeks as they all answer my conditions: clear higher low, rising volume, higher high breaking above the weekly 50. I won’t go over all the charts, just Huntsman to illustrate; know that many, many look like this (LYB DOW EMN CE ALB OLN).

This is the kind of chart I look for; they signal a clear demand for the stock.

Large volume while price action stabilizes means lots of shares exchanged without enough sellers to push the price lower. And as the price goes higher and volume continues to rise, the amount of shares exchanged continues to rise with demand at higher prices.

More investor accept to buy higher.
And there’s a reason for this.

Huntsman Corporation

Let’s look at Huntsman itself. The company is focused on downstream chemistries.

• Polyurethanes (60% of Revenue): They are a top-5 global producer of MDI, the key ingredient for Rigid Foam (energy-efficient housing insulation) and Flexible Foam (mattresses, furniture and car seats).

• Performance Products: They create amines and carbonates used in detergents, energy coatings, and agricultural chemicals. If a liquid needs to "stick," "foam," or "clean," Huntsman likely made the active ingredient.

• Advanced Materials: Epoxy resins and adhesives used in Aerospace (gluing carbon-fiber wings for Boeing/Airbus) and Automotive light weighting.

Products for different industries but most of their revenues come from housing and auto. As I said, tightly correlated with interest rates which hopefully should go down.

Huntsman went/is going through restructuring with the objective of $100M of saved expenses, impacting margins and cash generation.

• Headcount Reductions: They have cut or relocated approximately 500 to 600 positions, representing nearly 10% of their global workforce.

• Asset Rationalization: They closed seven manufacturing sites, targeting high-cost loss-making plants in Europe for example.

• Dividend Reset: They slashed the quarterly dividend by 65% back in 2025 to preserve cash.

• Free Cash Flow Focus: Aggressively clearing out inventory late 2025 and entering 2026 with a leaner balance sheet.

Valuation was back to all-time low a few weeks earlier, with reasons. Lower growth, declining margins and cash generation with little perspectives of amelioration won’t get a premium.

So yes, we are talking about a struggling company fundamentally. But many potential tailwinds from fundamentals, from the economy and from the market.

As usual, nothing is given in the market. But this setup looks like a pretty good setup to me - not the perfect one but an actionable one, which could hurt short-term like most of the market with the geopolitical conditions, but has potential for the years to come.

Execution Plan

The market is a machine to anticipate; struggling companies can have a rising stock as the market anticipates a reversal or is optimistic about tailwinds. This is why I look so much at price action; it tells you what the market expects. Looking only at earnings & fundamentals tells you what happened, not what can happen. What is done is done, the market cares about what’s next.

Stocks trade on expectations, not past fundamental performance.

If the market’s expectations were to materialize, we’d be talking growth instead of decline and improving margins, which might not go back to pre-COVID but could reach 4%-5% operating margins and EBITDA profitability. If the actual optimism is confirmed by expanding margins during earnings, we will see a repricing. Maybe still below historical average but even then, we’d be talking about a ~40%+ upside.

I personally bought options, details on the strategy here.

The risk/reward is very healthy today and the stock is grinding higher with demand, as volume shows. I do not want to tie too much liquidity to this position, it remains an opportunistic trade based on market’s optimism more than a strong fundamental investment like TransMedics. It allows me to diversify and participates in a sector strength without committing too much.

Reversal structure, as usual:

• Jan15’27 $10 sold puts, below the last weekly 50 and breakout retest, which should be a strong demand region for buyers and hold - assuming the thesis holds.

• Jan15’27 $18 bought call, or ~50% gain. I wouldn’t expect it to be reached by January but I don’t think I will hold than name until then - I could. I am giving myself room to capitalize on optimism & liquidity rotation, which should both accelerate H2-26, or so I expect.

This trade would pay you $0.67 per contract, meaning you would make money as long as the stock trades above $9.33 by January, which is pretty close to yearly low and largely below all breakout/averages.

Huntsman cut its dividend as shared earlier, meaning commons are less interesting but as usual, a thesis is only a thesis, execution depends on each of us, our profile & objectives.

Huntsman is a great diversification play in a cyclical industry which is demanded by the market and with a few interesting tailwinds, while most of the headwinds have been priced in after four years of bear market.

It isn’t the only interesting asset in its sector, I chose it because I believe rate cuts and return of demand for financed goods will generate optimism for final demand plus its diversification with other chemicals for key industries in the future; but other chemical names could follow the same pattern.

It is a stock pick within a strong sector.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

Starting with a reminder that the website is live and accessible for free. It shares my investment thesis, positions and performances for all investments shared.

https://thefewbetsthatmatter.com/

It is not a representation of my personal portfolio, but a representation of my stock picking and investing system. It is a centralized place for strategies and follow-ups, more organized than a Substrack page.

Content Review

Going back to my February investment plan, I’d say many of the risks materialized this month. The summary was liquidity issues and narrative shifts. Since then, tech has been flat while other defensive names I shared rallied, and names I wanted to avoid fell sharply.

I think my write-up is clear: I do not intend to buy risk assets. It doesn’t mean I won’t hold any, it means I do not plan to accumulate or start new position in the sector, not without crystal clear sign of demand at least. I need the market to show love to hold my positions and be more aggressive.

Crypto kept crumbling, which confirms the importance of cutting selling names that don’t follow the system. Holding them would have been costly, with Strategy down ~15% in a single month for example. These assets will go higher long term, but it is better to be out during uncertain times.

We also saw the famous social media darlings I mentioned underperform, some getting completely demolished. Personal conviction is not an investment thesis, even if many believe it is. Holding those names likely destroyed portfolios in just a few weeks: Adobe -12%, MercadoLibre -19%, Duolingo -26%, Hims -45%, PayPal -13%, Uber -9%, Meta -9%. All had clear signs of weakness or broken thesis.

The only one that performed well was Netflix after announcing it wouldn’t buy Warner Bros, on which I shared my thoughts here.

I share this to highlight the importance of following a system. Opinion alone will never be enough to make money in the market because you will be wrong more often than right. Systems protect us from our biases and ego. I dislike selling companies I believe in, but I believed in most the names above and my portfolio would be broken today if I had not sold them.

On the contrary, names I don’t particularly like but shared in January performed well: Darling +16%, Nutrien +5%, Smith +9%, UPS +10%. Earlier names like Schlumberger and Halliburton continued rising. Only Novo performed terribly, and I shared that the bull case was shattered after terrible guidance.

Opinions don’t matter. Systems do.

I also shared a piece explaining how markets have changed since Buffett and why the admiration of his methods may be a mistake, in my opinion. Feedback was positive and very negative, but I believe it is a piece worth reading.

With that said, let’s review market conditions and then conclude with my portfolio and outlook for next month.

Macro, Economy & Geopolitics

I started writing this article Thursday & had finished this section before the weekend, just before Trump launched his operation in Iran. So let’s write it all again.

Edit: I finished this portion Saturday morning and had to rework it Sunday morning as the U.S. are apparently capable of overthrowing a regime in less than 24h.

Kinda impressive.

So the situation changed since I wrote it first & I wouldn’t expect the market to react too violently now that Ali Khamenei and much of the Iranian decision-making sphere are confirmed dead. I’ll still leave most of the Saturday written context and will only change the investing portion.

A(nother) War in the Middle East

AS USUAL: this is a tense geopolitical situation and everyone has an opinion on what is right or wrong. I have mine, you have yours, but they don’t matter here, alright?

We’re here to talk about what is happening and how it affects investing, not to fight over personal views on Iran or Israel.

Don’t twist my words. Focus on investing consequences, not opinions.

The Context - Written Saturday

Trump said for months he’d launch another attack if they didn’t reach an agreement with Iran’s regime, mostly about renouncing their nuclear program. Tensions were already prettwe’ve had tensions for years, even more since the attack on Israel back in October 2023 which started escalations and even an American stealth strike on Iran’s nuclear facilities.

Trump said for months he would act if no agreement was reached with Iran’s regime, mostly about renouncing their nuclear program, and we’ve had tensions for years - even more since the attack on Israel in October 2023 which triggered escalations & even an American stealth strike on Iran’s nuclear facilities.

But this is different. This isn’t a stealth attack, this is Trump telling Iranian people they will get the help they have been asking for years to overthrow the regime in place and bring changes.

This is a declaration of war against that regime with a very clear objective for the west: disarm the country and stop their nuclear progress at worst, completely overthrow the regime and replace it with a more peaceful one at best.

We are going to destroy their missiles and put their missile industry to the ground […] we are going to annihilate their navy, we are going to make sure their terrorist proxies can no longer destabilize the region and the world and we are going to make sure Iran never obtain nuclear weapons.

The life of American heroes may be lost and we may have American casualties, that often happens in war, but we’re doing this for the future.

To the members of the Islamic guard, the arm forces and all of the police, I say tonight that you must lay down the weapon and have complete immunity, or face certain death.

To the great proud people of Iran, the hour of your freedom is at hand. Stay sheltered, don’t leave your home, it’s dangerous outside; bombs will be dropping everywhere. When we are finished take over your government, it will be your to take. This will be probably your only chance for generation.

For many years you have asked for America’s help. You never got it. No President was willing to do what I am willing to do tonight. Now you have a President who is giving you want you want, so let’s see how you respond. Now is the time to seize control of your destiny.

This is the moment for action, do not let it pass.

I purposely left some harsh words because his speech was that harsh. Crystal clear on his intentions.

That being said, we know Trump’s method. Every impactful decision was taken on a Friday with a resolution before market’s open or rapidly after. It is a priority for the U.S. government to keep markets high as they are what make or break the economy.

Sunday edit: they indeed “resolved” the biggest part of their problem before market open.

His language doesn’t make it look like it will be swift, although they also said that they wouldn’t commit to a long war. At time of writing, Saturday morning in Europe, lots of strikes have been launched with American assets participating; they aren’t just helping Israel with ammunition and information. They are part of the strikes and are targeted by Iran’s strikes.

Iran is also targeting other countries as we’ve seen in Dubai, with drones apparently targeting civilian regions. This is about putting pressure on the U.S. by attacking “allied” countries with high political/financial capital. Great military strategy but not really great relationship strategy; hurting your neighbors rarely makes them like you more.

The situation in the Middle East is not what it used to be a decade or two ago, many actors are now focused on economic developments and peace, not fighting the west. Iran isn’t isolated but we shouldn’t expect the Middle East to rise to aid them against the U.S. as they could have done years ago. Even less while facing Iranian attacks on their population…

Russia and China are another story entirely.

China has always been an opportunistic country and will use the situation as it. They have commercial relations with Iran for “defensive” weapons & will certainly continue. Tensions could start if their relations hurts the U.S. actions or troops’ safety - selling offensive weapons for example. Until then, China should continue to play its role of a quiet giant; involved, but never too much and only for its own interests.

I also can already hear the war drums about Taiwan and how China will without any doubt use the actual tensions to invade the country. And I will as usual ignore that noise for now.

Russia is another story; they have been Iran’s support for longer in a proxy war to weaken the U.S., just like the U.S. used Ukraine - do not twist my words, and Iran has been a support of Russia during the Ukrainian war. Their relation is more important, and harder to manage as Russia has also been negotiating peace in Ukraine, without much success yet but talks exist, which is something. Pretty uncomfortable situation. It’ll be interesting to see what their actions say, but I wouldn’t expect much.

Investing Wise - Written Sunday

Most of the uncertainty is gone with Khamenei and his close sphere killed. The market could have worried about a war without swift success and collateral damages, impacts on the Strait of Hormuz where most of the oil traffic passes, pushing oil prices higher - which the market wouldn’t have liked.

This situation would have been dramatic for the markets. This scenario is much more unlikely without a strong governing entity in Iran.

Still, it is impossible to know what will happen. You know what you lose, never what you get. The removal of Khamenei doesn’t necessarily means peace and prosperity from Monday. It remains to be seen who will take the lead and what their objectives will be; if its army commander or someone with less restrain takes command, it could lead to a long fight still.

We still have uncertainty. Maybe less worries than if Khamenei held the reins. Nothing is certain. As I write this - Sunday lunch in Europe, missiles are still flying, commands of the Iranian regimes just went to the next one and the Strait of Hormuz is blocked. The question is can this be sustained and for how long without a stable decisional structure in Iran.

There’s no point trying to predict Monday’s open, it will depend on what happens in the next hours and we could have another very eventful week end into a non eventful week. Or an eventful week. I talked often about weakness in the market lately and that could be another excuse for sellers to continue their biding.

I have no conclusion to give during uncertain times and as much as I like to share the situation as I see it, the only thing we can do as investors is follow our system, here to protect us from this kind of situation. If we bought great names at great prices and in great setups, we will be alright. If they break their setups, those names were maybe not as strong as we thought and that is fine. Decisions can be taken then.

In the meantime, we cannot anticipate the next days and even if I believe that we will see a “swift” resolution - and the last months showed it as Trump’s government has been lightning fast with its operations, we should remain cautious.

No point piling on risks.

Another “advice”: In this kind of situation, be mindful of what you read and where, as panic can spreads very fast from tabloids who sell by clicks. The situation is rarely as terrible as media makes it look like and we should stay away from all the panic and noise.

This is the kind of situation where mistakes are made.
Trust your system. I'll trust mine.

Macro & Other Subjects

Besides this situation, the month has globally been pretty calm. Most of the important data came from earnings and market’s reactions, and some specific sectors/names.

Macro

Macro wise, nothing changed compared to last month and I continue to believe that liquidity is slowing while opportunities emerge outside of the market, which hasn’t been the case for years with climbing interest rates. The conditions simply are not what they used to be so we shouldn’t expect the market to be.

The news of the month was the tariff mess which the market ignored. The Supreme Court judged tariffs illegal but concluded that the next steps will be messy and are out of their range of power...

“This is bad but this is done and we don’t know what to do from here.”

To which Trump decided that because his own court judged his politics illegal, he should add 10% tariffs on the world. Pretty special, but isn’t everything lately?

As I said in a note, there’s no saying if this ruling is positive or negative or if things will even change. It could trigger a massive liquidity inflow with billions of refunds to companies and individuals, or many problems as it also means all relocalization is useless - without tariffs everything can remain delocalized in Asia and such.

On macro, the Production Price Index came higher than expected meaning materials come at a higher price hence finished products will be sold at a higher price. This means inflation isn’t beaten yet and could slow the rate cuts cycle.

Market & Earnings

The S&P 500 is still close to all-time highs while tech and some growth names have been destroyed. Some portfolios are really bleeding, in the order of -50% or more over the last three months, while the S&P is down -1.7% from all-time highs...

This is what high beta and growth investing looks like. You don’t need to be a great stock picker when everything goes up due to constant and massive liquidity inflows. Things are very different when the music stops.

Nvidia is a great example. We don’t see such companies often; three years of beat and raise earnings, leaving the market in awe with constant growth & margins expansion. This stock, if the market was only about fundamentals, would have never stopped running and its latest quarter should have rocketed it to new highs.

It isn’t normal for such a company to still grow 70%+ and accelerate while expanding margins. This is unseen before.

But the market is about liquidity first, sentiment second and fundamentals third. If no one has money or the wish to buy that stock - and that can be because they struggle to see the upside or have better to do with their money, then the stock won’t go up.

There are no fundamental issues in AI, hardware and software. There are liquidity and sentiment issues. As I said in my previous write-up:

Pickiness is not a bull market sign, it is a low liquidity sign and we see it more.

Palantir is showing the same dynamic, with insane earnings and yet... A stock down -34% from ATH and -12% post-beat and raise earnings.

I won’t go over each name individually, but we have many examples; Arista Networks, Astera, AMD, SMCI, Google and Amazon who all reported very strong earnings and were not rewarded by the market for it, even being harshly punished in some cases, especially AsteraLabs which is down -35% after excellent print, only because they invest in their future while their actual products are in high demand.

Even names like Crowdstrike, Palo Alto and other key players for the AI security layer have seen their valuation reduced by 30% or more simply reacting to some silly ads from Claude about their new security-focused AI, as if this could replace complex interconnected network/AI security systems. Market participants know that, but it doesn’t matter; as momentum and sentiment dry, liquidity withdraws and goes where returns are. If the potential returns slow down, why stay?

Fundamentals are third, after liquidity and sentiment in short/medium term returns.

And so the tech sector continues to be “meh,” either flat or down but not up during the last months while fundamentals continue to be rock solid, with CapEx accelerating and even finally the great news that OpenAI completed its $100B funding round with key partners.

This would have rocketed the stock market in September as a renewed confidence in OpenAI’s capacity to pay their bills and commitments. But it wasn’t well received this year. Sentiment clearly shifted.

On the contrary, there were many names which were rewarded by the market even if their earnings weren’t that spectacular - nothing comparable to Palantir and Nvidia. Names like Cava (restaurant), Celsius (soft drinks), Crocs (footwear), Darling & Nutrien (materials), Columbia (sportswear), Moderna (healthcare), , Enphase and SolarEdge (energy). I have a longer list; no tech in it, nor in my screeners.

The market treated those names better than any excellent tech earnings. This speaks volume on where liquidity is going. The best illustration of this is gold, like for the last six months; it fell a large -21% in three days to climb it all back up the next 26 days. The largest swing ever seen on this asset.

I don’t expect this trend to stop if tensions with Iran aren’t resolved rapidly, although to be fair I don’t expect this trend to stop either way the demand for gold is structural, not just speculative - although we also see clear speculative behavior.

So if I were to listen to the market... I would certainly not be buying tech outside of a few exceptions still trading above their weekly 50 despite the late bloodbath and still printing healthy and strong earnings. The issue with those few names is that even if they look strong, their sector will drag them down on others’ weakness while the risk-reward is much better on other sectors.

My conclusion of this last months is that I see no point taking risks on tech, which is sold off despites great earnings, while defensive names or other sectors pump on correct earnings - sometimes even “meh” ones. I’ll let the market drag them down, I know enthusiasm will come back.

What matters is to be there when that happens.

Investing Plan & Watchlist

I had a few key changes since February which I have shared here and on the website. Some names had to be closed due to price action, others due to fundamentals. I sold entirely UiPath, Nebius, Alibaba and AsteraLabs, and ignored Novo Nordisk after the terrible earnings and guidance they shared.

I now own four stocks: Transmedics, Nordic Semiconductor, Lululemon, Huntsman Corporation.

Let’s detail everything.

Closed positions

Some will be very easy to explain, others will require a bit more words.

AsteraLabs and UiPath will be the easiest as both remain great companies with great potential. But that isn’t enough for me to hold a stock as you know. Many called me a trader because I consider price action, but both stocks are down 9% and 14% since I closed the positions, and as liquidity rotated 100% to TransMedics up ~10% since, I’m pretty happy with the outcome.

Holding a falling stock based on personal bias is simply not something I do. If that works for you and you can outperform that way, please do so. To each their system. I’ve said in the last monthly that I would close on wrong price action, and I did.

I won’t hold these names if they’re sold after good/stable earnings. If strong results aren’t rewarded, my money is better spent on defensives, it means appetite is down. In those situations, bounces happen, but medium-term potential is limited. It’s not about the companies; it’s about risk appetite.

I am 99% certain both names will see new highs in the next years and I will continue to monitor them and hope to be part of the run when that happens. Both are great companies with great potential. But the market doesn’t like their stocks, so I don’t either. Concentration requires making choices.

Both sold positions are about price action, losing their weekly 50 for a long period of time, which is my selling signal.

UiPath will report earning this month and I will follow and write about it. Let’s see the data and market's reaction.

Alibaba is about opportunity. I sold the name because I needed liquidity for another opportunity: Nordic Semiconductor. I did not have enough to make both positions significant so I chose the best risk-reward according to my opinion. I didn’t share that change because nothing changed for Alibaba, it is still to my opinion a great stock to hold and even buy today.

I’ve been seriously considering re-opening it now that the stock is back on the perfect spot to be bought for bulls: its weekly 50.

The weekly 50 is my lighthouse and I’ve been waiting long on that stock. I was looking at liquidity and how to build a serious position on Friday and didn’t press the button as I had to do something else.

With the situation in Iran and the potential implication of China - it doesn’t matter if it’s real or not, what matters is market’s perception, I could expect the stock to fall a bit more at open assuming the situation isn’t resolved by then. I cannot say what I’ll do today as it’ll depend on the situation in Iran, but if the situation seems stabilized and Alibaba is still on its weekly 50, I will certainly be a buyer now that I have enough liquidity.

Novo Nordisk was detailed in my earnings review. Terrible earnings, shattering the bull case. Nothing to do but sell. I’ll monitor it but I start to consider this name a value trap. Hope data will change my mind in the next quarters.

Nebius... The famous Nebius. I’ll be very straightforward as I’m not here to play violin. Selling was an emotional mistake. Nothing else. I broke my system. The previous weeks were complicated, I got emotional and pressed a button I shouldn’t have.

Systems are here to protect us from ourselves.
I didn’t respect mine. And it cost me.

I thought my personal bias that Nebius wouldn’t be rewarded for increased CapEx if others weren’t, would be right; that I could anticipate. It was not and the market kept treating Nebius differently because of its financing source. Ironically, Coreweave was punished post earnings exactly for what I was concerned about for Nebius.

This doesn’t mean Nebius won’t fall in a few days and break its weekly 50, but I either way shouldn’t have broken my system. If Nebius were to come back to its weekly 50 I would certainly be a buyer now that I have liquidity for one more position. It is more important to correct or wrongs than miss an opportunity by ego - as long as we don’t end up chasing.

If Nebius answers to my system, I have no reasons to say no.

Positions & Watchlist

TransMedics remains my #1 position, ~60% of my portfolio today in both shares and options as I opened a reversal trade Thursday, post-earnings, as the stock fell to $130 for what seemed to be an algorithm stupidity.

As shared in the review, the EPS were wrong on most aggregated data websites and it seems that algorithms believed TransMedics had a sharp decline in net margins, which triggered sells while it was not the case.

I bought the dip aggressively, selling shares to buy my option trade with a January ‘28 expiration. I went big on what seems a great opportunity lately: a defensive healthcare asset with strong growth, massive advantage over competition & many growth levers for the next two/three years.

Still my favorite name in today’s market conditions.

Nordic Semiconductor is the only AI/tech name I own today and I do so because the stock trades at cheap multiple and fundamentals are strong. A company focused on what could be the next wave of hardware demand: AI wearables

We aren’t talking about AsteraLabs’ 25x sales but a modest 4x. The risk profile is not comparable to 99% of the tech market while the potential is, especially after hearing Meta’s confirmations of massive demand for their smart glasses.

A position I intend to hold as long as I can, as long as my thesis holds.

Lululemon’s still in the portfolio. I’ve been surprised by the market this month & how it reacted to consumer-focused companies with restructuring narratives. Columbia or Crocs were rewarded for flat growth guidance and names like Cava/Celsius had great earnings rewarded by the market - we’re talking 20%+ pumps.

We also had tough news like a see-through scandal for their new leggings, the new tariffs which should impact margins and the management transition... Nothing was enough to push the stock lower. We’re holding the bottom with bad news while the sector and consumer names are healthy and pumping.

This is an indicator that sellers are exhausted and that we’ll need more bad news for holders to sell. This smells bottom, although it doesn’t mean we will go higher in a straight line. But my bear case is lateralization assuming no new scandals impacting the brand and its future growth.

Huntsman Corporation, lastly. A name you never heard about. I bought an option reversal trade on Thursday around $12 after it retested its weekly 50. I haven’t had time to write about it but I hope to do so this week. Rapidly:

A cyclical material name which was oversold as fundamentals struggled with Chinese competition, which could stabilize or push higher now that inventories are emptying and with the tariffs situation - reducing Chinese competition. It focuses on chemicals used in homebuilding and automotive, which demand could pick up with lower rates.

The risk here is very acceptable and the market would anticipate recovery if we were to see signs of it which, coupled with low valuation could yield returns. The sector is pushing higher the last weeks which also helped my decision. The market has reasons to push those names higher.

On SolarEdge, which I shared a few days ago, the stock fell on Friday as the market sold off. A riskier thesis which could make sense with the rate cut cycle and expected stable demand for solar systems. We are now back on the weekly 50, my lighthouse. I had a small position closed to buy TransMedics, nothing significant.

It could be a contender to Nebius and Alibaba for the next position but as for the two others, I’ll keep the cash I have in cash short term, until we have a bit more clarity on the Iran situation and impacts on the market.

Looking Ahead

Here’s what I’ll be looking at this month. I won’t have a detailed review on each, only UiPath and Lululemon are assured to have one. I’ll write on the rest if interesting, for the companies themselves or as proxies.

• Week 1 - On Running, Sea Limited, America Eagle

• Week 2 - UiPath, Adobe, Rubrik, Oracle

• Week 4 - Lululemon

February was the month of the pivot at the end. And I struggle to see how the trend shifts in March, maybe even in 2026. The bloodbath in software and downtrends on many others are a great thing, they’ll create opportunities. But those’ll need time to shape and transform into an actionable setup. Most are still falling knives today.

I have so many names I’d be excited to own and I believe I will own most before 2027. Just have to be patient; as a concentrated investor I want to have my chips on the best potential, not diversify onto 20 names. Their time will come.

Investing isn’t cheerleading.

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

This quarter was excellent. The company delivered everything I expected and a bit more. I’m baffled by the market’s reaction even though it is a bit too early to draw conclusions only a few hours later - and it has its reasons.

We are not seeing competitive dynamic impacting our ability to execute in 2026 and beyond, and I will leave it at that.

Fundamentals.

The business is stronger than ever in every vertical and the bull case is not only alive, but strengthening each passing quarter. The company closed FY25 with 5,139 OCS case used compared to 3,735 in FY24, meaning approximately 38% YoY increase in transplant volume using TransMedics, growing their market share in liver & stable in heart - which should change with the new OCS.

For liver, in 2025, OCS liver transplant represented 4,197 transplants, or 36% of the overall liver transplant volume in the United States. That is up from 26% in the same period in 2024. For heart, OCS transplant represented 854 cases, or approximately 18% of the overall heart transplant volume, and modestly up from the 17% seen in 2024. For lung, the numbers are small. OCS lung transplants represented only 88 cases, or approximately 2%.

This represented approximately 26% of the total 19,833 U.S. transplants for heart, lung and liver combined in FY25, up from 20% YoY.

Importantly, for the third consecutive year, we saw growth in overall U.S. liver and heart and lung transplant volumes... We strongly believe that the OCS NOP once again played a key role in driving overall liver and heart market growth due to the increased use of DCD and DBD donors in the U.S.

This data is enough for us to conclude that TransMedics is everything they claim to be. A safe, trusted and efficient hardware for DBD and DCD which practitioners are using and trusting more and more.

Besides the OCS itself which has proven and continues to prove itself, their service is also healthy in terms of growth with stable demand for transportation and clinical services; a 30% YoY growth for the second quarter in a row, confirming not only the demand for OCS cases but for the end-to-end service.

TransMedics didn’t need a proof of concept but seeing that kind of growth for a new and innovative end-to-end transplant service is very positive for the future and for its expansion in Europe as the need for transplant systems is the same around the world.

Speaking of Europe, it is interesting to see that since management said they would focus on the region for their hardware and plan their expansion, revenues rocketed, notably for their core product: OCS liver.

This isn’t anecdotal, it shows the clear strategy of expanding its service overseas and this should only accelerate while the potential is really large the old continent.

This program is actively launching in Italy, and few other European countries have expressed strong interest in exploring the NOP model in their local geography.

Expanding our commercial activities in Europe has the potential to nearly double our transplant total addressable market for TransMedics.

We’ll talk numbers later but fundamentally, my bull case written more than a year ago is perfectly materializing, maybe even better than what I expected back then.

Future Growth Levers

We did not have much news about the next growth levers, except to recall them to everyone’s memory:

1. European expansion ongoing & expected to be meaningful in terms of revenues by H2-26.

2. Heart and Lungs OCS trials ongoing in the U.S. unrestricted with great feedback so far, but it remains early to be overly optimistic.

3. OCS Kidney in preparation for FY26 to require FDA approval early 2027, unlocking a massive market as TransMedics continues to earn the trust of practitioners with other products and will logically use OCS Kidney when available considering the success of other OCS.

   To our knowledge, the OCS Kidney will be the first and only warm perfusion oxygenated kidney platform for kidney transplantation used from the donor to the recipient.

4. OCS Gen 3 which is in parallel to the OCS Kidney although the OCS Kidney would already be Gen 3. Improved hardware, smaller and more efficient at controlling organ quality with more automated systems.

Those four points will require investments and that’s something we will talk about in the financial part of this write-up, below, but they should all be in full swing by 2028, earlier for heart, lungs and Europe, and contribute meaningfully to the company’s revenues.

Some interesting quotes though about the trials as management is facing some issues to move forward with the phase B of the trial.

Part B is designed to allow OCS to gain a potential new clinical indication in DBD heart transplant segment that are sub 4H preservation by demonstrating superiority of outcomes in a head-to-head comparison to current cold static storage modalities.

Progress in Part B has been slightly impacted by a competitive dynamic as it relates to cold storage arm of the trial. Specifically, there is a hesitation amongst competition to a head-to-head comparison between OCS and their static cold storage modality.

This is a CEO talking so I can understand why some wouldn’t take those words at face value. But I also can understand why many would refuse to compare their systems to the OCS as when we see the growth and demand for the product, we can understand there is something practitioners like about it.

What’s coming is really the unequivocal, drop the mic, statistical superiority in the most important outcomes after liver transplantation, which would justify and support all the evidence that’s been built in having more than 14 or 15 publications now already in print, out there. Those publications that are coming, they’re aggregated of thousands of cases, they’re coming out of our registry, and many of them are already under review.

Waleed said that publications were coming H1-26 for liver OCS while the rest would take longer. Hopefully those will be out soon.

Financials.

Now, we saw that fundamentally, everything was going good to great. But as usual, and I hope you know it by now, the market cares only about… Cash generation.

And even there, TransMedics has been nothing but excellent.

Revenues beat the Street expectation at $160.8M, up 32% YoY and at $605.5M for the year, up 37% YoY. As a reminder, the FY25 first guidance came in at $575M and was raised three times through the year to $600M. Management raised their guidance three times in three quarters and still beat it.

On the “less good,” which I personally do not consider less good because of context but needs to be said, margins are slightly declining.

This is normal. As their service revenue continues to grow, their expenses also do but this is the counterpart of growing market share and a better service. Without those, TransMedics wouldn’t have comparable revenue growth nor market share.

Add to this rising expenses for trials, next gen OCS and European expansion. As I said earlier, this costs money but it will yield constant long-term growth in return & signs give me reasons to be optimistic; when market share rises in a sector like healthcare, your future products have a large head start as trust is already built.

We also need to understand that those expenses are one-off expenses. Once spent, the products/services built from it will be here and generate cash for decades. But it does require spending and this spending will continue 2026 as it will be the year of innovation, with all projects running during the period.

Based on the current revenue guidance for 2026, we expect operating margins to be approximately 250 basis points below 2025 full-year levels, primarily reflecting the timing and scale of these investments.

As investment levels normalize and the business continues to scale, we will expect operating margins to resume expansion. We continue to expect operating margins to approach 30% by 2028

Once we complete those elements, that’s what gives me the confidence that spend should normalize, and then we should be able to start capture and operating leverage as we continue to grow.

So yes, cash generation wasn’t the perfect case and won’t be in FY26. That being said, we are talking about a company with $107M of income, up 203% YoY - excluding the tax benefit, with approximately $440M of net cash which will also help on CapEx. The progress has been very impressive and the results even more.

We are talking about a company which wasn’t fully profitable a year ago and now runs at a run rate of 30% revenue growth with ~13% net margins, earned the trust of many practitioners in a very complex sector and continues to push to expand.

Guidance.

On the main meal, which once again is not falling short of expectations.

Keep in mind FY25 early guidance, read those comments from management.

In our guidance, we factored in all of the above, and issued what we believe is a realistic guidance that would enable us to execute and let the execution and performance dictate what do we do, or if we need to revisit the guidance.

We feel confident on the guidance that we are putting forth here, as it bakes in all the uncertainties, all the opportunities and uncertainties in front of us.

You know, again, we will go and execute, and we let the execution and the results and the performance dictate, if we need to revisit the guidance as we move forward throughout the year.

And look at the flight profile of the first two months of 2026, which according to my methods - which has always found the revenue floor of every quarter for more than a year now, should put TransMedics at approximately $110M of revenues with 37% of the quarter left to go.

Management is conservative by design; they learned their lesson in 2023. Seasonality is complicated to anticipate and it is better to overdeliver than overpromise.

Will they? I don’t know. No one does. But I won’t be surprised if they raise guidance through the year, even if the market would have loved a strong guidance today.

So… Why is The Stock Red?

There are no valid reasons in my opinion and we’ve seen that the market bought the stock today, passing from -7.5% to -0.9% at time of writing. There’s demand at today’s price and that’s what I’ll conclude later. But it’s important to remind ourselves that our opinion just doesn’t matter.

And we have to remain factual and keep our feet on the ground. The market, even if it sees the positives, focuses on the negatives. And there are two.

1. The margin pressure is real and will persist through 2026 as detailed above.

2. The company has execution risks: the trial results are still pending, even if the first feedbacks are positive, OCS Kidney is in development, European expansion will require expenses before yielding results - even if it takes only one semester.

TransMedics has execution risks which comes with shrinking margins in the meantime. The market is not punishing the stock, it is simply acknowledging the risks, the “what if it doesn’t work?”.

That being said, we’re also cognizant of operational challenges that could influence the pace and timing of these initiatives.

First, as stated, we are still building out our logistics infrastructure in Europe, which could moderate the initial pace of our EU NOP launch as we ensure we have the right foundation in place.

Second, timing of the full De Novo trial accrual will depend on how long and how the lung transplant market adopts machine perfusion and NOP, which remains to be proven.

Third, timing of Enhanced Part B completion will be influenced by some of the inertia created by competitive dynamics for the cold storage arm in the marketplace.

Fourth, the very common and now, I hope, well-understood annual phenomena of potential Q3 seasonality in U.S. transplant market that temporarily slows down transplant activities.

Finally, ramping our infrastructure and clinical staffing to meet the growing demand for OCS NOP will be critical to achieve our full growth potential in 2026.

The market is always worried about spending, even if we are bullish and if the team has proven to be great at executing. The advantage with TransMedics is that their plans have deadlines. Europe should give results this year, like the trials, and the OCS kidney should be approved before end of year. This isn’t a story of “we don’t know when,” this is a story of “give us two or three quarters.”

Personally, I continue to believe TransMedics is insane value at today’s price and it is the kind of stock I want to own: fair price, perfect execution, massive potential. We also had talks about integrating official government workflow which I talked about already and would honestly make sense considering the product.

Specifically, we are exploring if TransMedics can be a more integrated contributor to the national transplant ecosystem in the U.S. The goal is to maximize donor organ utilization for transplantation and continue to save more American lives and save significant healthcare dollars.

This is a long shot but one more example of the opportunities such company has.

As for my valuation model of TransMedics, which expects ~$1B revenues FY28, I stand by it today, after what I saw and the FY26 guide. My bull case with great execution pushes up to ~$1.3B of revenues by then and could happen if Europe picks up rapidly on a clearly superior technology and yields its expected growth.

As for the market, I continue to believe 7x sales is below what TransMedics deserves even with execution risks & potential margin pressure. But the market isn’t rewarding anything below perfection anymore, so we’ll maybe have to be a bit more patient on this one, until perfection happens - the European expansion yielding results would already be a great sign.

In the short term, I’ll wait to see what the next days offer. It is disappointing so far but again, there is clear buying pressure today at this price and it would be surprising to go lower on such a print for a defensive name, but who knows... I also know most data platforms are displaying a wrong EPS print due to miscalculations, which could impact algo-trading and explain the weird +10% to -7% swing post-earnings. Who knows? I wouldn’t enter into that kind of speculation, though. Let’s wait and see the next days.

The long-term goal remains Intuitive Surgical, a company growing ~20% CAGR with a very comparable profile in healthcare and 30% operating margins... Trading at 17x sales and 62x earnings. This is what a deserved healthcare premium is.

If TransMedics were to execute on its promises by FY28, this is what it could reach.

——————————————————

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

Not many, at least; and I’ll be part of the few.

Enphase and SolarEdge were two darlings of the 2022 bull run and are down 85%+ since for many reasons, mostly valid ones. We’ll go over each of them and why the situation is changing and why these two could yield great returns over the next year(s), and how to capitalize on it.

Solar Systems & Hardware

I’ll start with a rapid overview on how solar systems work to share some basics and understand what Enphase and SolarEdge sell, and how they compete.

Nothing very complicated - at least on the surface.

• Solar panels are set on the roof to catch luminosity and transform it into DC current - a certain type of current.

• They are connected to one or many inverters depending on the system, which are meant to transform the electricity generated by the solar panel from one state to another - only AC current can be used by households’ hardware. They also are the brain of the system, managing production and distribution while transforming the current if necessary.

• Inverters are connected to the grid, the household electrical system, and batteries - potentially, to manage electricity depending on the need and its most efficient usage. Sometimes reselling production is more interesting than consuming it.

Enphase and SolarEdge compete on two hardware: inverters and batteries, and on the system they propose to households which have fundamentally different architectures, both with their advantages and flaws.

Enphase’s architecture transforms current at the panel level, with inverters attached to each panel transiting AC current to an electrical panel. A system with 20 panels comes with 20 inverters; a more efficient current conversion, less impactful dysfunction and a globally more efficient system altogether, which also comes at higher expenses. The Rolls-Royce of solar systems; more efficient, more expensive.

SolarEdge’s architecture has a centralized system to transform the current to which all solar panels are connected. The representation above is SolarEdge’s system. It comes with fewer inverters and hardware, hence less expensive but with a single point of failure and a globally lower efficiency than Enphase. The Toyota of solar systems.

Both also sell batteries which are comparable in quality as of today and can simply be added to the architecture. They are the next narrative for solar systems as technology improved drastically over the last years and makes them very efficient - and worth the spending.

Both companies share the entire inverter market in the west and only have Tesla as serious competition when it comes to batteries with its powerwall system.

When it comes to quality and efficiency, Enphase is strictly better. But quality comes at a price, which gives SolarEdge a clear advantage while the latter has been improving in terms of quality and could become a more premium service than it used to be. Add to that that SolarEdge is also starting to sell to small companies who understand the value of solar systems, opening a new market entirely.

Both have qualities and flaws which will appeal to different clients. Both can thrive, as long as the sector thrives.

The Solar Industry & Pre-2023 Bull Run

The solar industry is born from a simple need: autonomy and self-reliance.

Solar is the cheapest source of energy and the only, or one of the only sources which can be used by households for self-reliance. I have a detailed write-up about energy and about how an electrical grid works for cities, which can be a great introduction, although not a necessary one.

When it comes to households, it is becoming cheaper to have an efficient solar system to power your house than to rely on the grid and electricity price fluctuation which can be due to different reasons depending where you live - electricity price rocketed when the war between Russia and Ukraine started in Europe for example, and never went back down since, while taxes... Well, in general taxes never go down.

This is a constant and mandatory expense for households who have no control on its source, price or quality. The only control households have on their electricity costs is using electricity or not. They can accept to lower their heat or air conditioning but if they want to live at peak comfort, they’ll need to pay the price. Whatever it is.

Solar changes this as it allows them to have control and potentially be self-sufficient for their electrical production. The reason for that need is different for each household around the world but the advantages are the same for everyone; they do not rely on anyone but their system, and have a clear view on prices beforehand.

The early 2020s came with many circumstances which yielded a massive demand for solar systems across the west - U.S. and Europe will be our focus in this write-up, due to the advantages of solar systems but mostly external factors.

1. The economics were interesting during the COVID period with very low interest rates, allowing very low borrowing costs; an entire solar system can range from $10k to $30k and is usually financed by loans, hence the importance of low rates as a single percentage increase can yield thousands more to pay.

2. Regulations were pro-renewable in the west with favorable financing conditions, tax reductions, state subsidies, and more... They still are favorable in Europe, but the landscape changed a bit in the U.S.

3. Electrification was in full speed with Tesla being the most sold car in the west and requiring electrical installations, which was often paired with solar installation and drove demand.

Another important factor of the last cycle was California which was a specific case for the solar industry due to a specific regulatory framework.

In a few words, households were allowed to sell their energy production to the grid without pricing restrictions while the state had the most expensive electricity prices - $0.30-0.40/kWh vs $0.14/kWh elsewhere. That allowed homeowners installing a solar system to refund it in ~5 years compared to ~15 under normal conditions elsewhere. Everything post-refund of the system was pure profit. Very favorable conditions to finally make the step and finance your installation.

The situation changed in 2023 as households wouldn’t be able to sell electricity at grid rate anymore, slowing demand as the economics changed.

The combination of those conditions triggered the perfect conditions in the west for a gold rush on solar systems, and this is what happened. And as the conditions changed and demand stabilized, the classic inventory building cycle occurred. Installers were anticipating continuous demand and bought more hardware from manufacturers. As demand did not materialize, installers ended up with quarters of inventory.

Demand flattened, inventories were full, orders stopped flowing, revenues flattened, margins shrank, cash generation slowed and stocks crashed.

We’ve seen the dynamic with Nordic Semiconductors last week, and with hardware providers globally through the years.

As you can see in the revenues, the situation stabilized since with revenues flat since end of 2024 while comments have been positive on the installers’ inventory situation, which are emptied by now.

We are back to a healthy situation for those companies.
We might even be starting to see a rebound.

The Post 2025 Situation

The situation is much different today. Not only did the regulatory framework change in California, but it also changed everywhere in the U.S. since Trump took office - not much changed in Europe in terms of regulatory incentives as with the current energy “crisis,” independence is even more important for both countries and households.

On the U.S. situation.

• The Inflation Reduction Act forced manufacturers to relocalize production within the U.S. which added costs - to build and equip the factories plus opex. This was rewarded by tax credits but the costs remain higher globally than delocalized factories, while the upfront costs had to be spent.

• The Investment Tax Credit - a 30% tax credit to install solar systems, was cancelled so installation now costs a flat 30% more for households. It still exists for business in a different form.

• Tariffs have and continue to create uncertainties and inflate prices.

On the positives, the most important factor were the last months’ rate cuts and the probable continuation of rate cuts at least in the U.S., assuming Trump were to have a puppet at the helm of the Fed, as otherwise we should expect stable rates at best, while Europe is back with pretty low rates already, healthy enough to fuel financing for solar systems.

Sector Recovery & Growth

Today, the need for solar energy and energy independence hasn’t changed. I am living in Portugal and went through the April 2025 blackout and the 2026 storms, which haven’t impacted me personally but highlighted the importance of independent sources of electricity for households.

Those blackouts and situations aren’t frequent, but we’ve had lots of them these last years in the west, the latest one in America in New York with record colds which put some tension on the grid with record electrical consumption.

Europe’s electricity prices haven’t gone down and shouldn’t go down any time soon as even with a renegotiation with Russia to access cheap energy, the pipes have been broken and will take years to rebuild - after years of negotiation, so this seems utopic. Solar is the only source of self-reliance for households and the cheapest source of electricity. That was true then, is still true now and will get truer in the future.

If given the opportunity, everyone would love to have solar systems on their house or apartments. The only variables for demand are external to the fundamentals. They are about prices, financing, purchasing power, installation times, etc...

California also adapted itself to its new conditions by adding batteries, which allow a more efficient management of their generation, storing when necessary and reselling on peak demand.

Under NEM 3.0, homeowners with battery storage can earn approximately $200 per week by storing solar electricity generated during the day and exporting it to the grid during high-value windows.

Solar.com analysis

Battery attachment rates have increased following those changes, passing from ~10% in 2022 to 70%+ last year. This trend will continue as batteries are now very efficient for an affordable price.

We see that demand’s starting to pick up, certainly due to the inventory from installers finally being cleaned up while the need for hardware continues, illustrated here with SolarEdge hardware sales which are clearly picking up.

This is the bull case from here.

Inventories are cleared, fundamentals are strong, need for solar is growing, financing conditions are easing. The questions from here are:

1. Will the financing conditions easing refuel demand and yield growth?

2. If so, which player will be the best performer in this scenario?

Enphase vs Solaredge

The market focuses on cash generation - you’ve probably heard me say that already, it’s probably the most repeated sentence on this Substack. This means Enphase will logically have a premium over SolarEdge due to its premium product, higher margins hence higher cash generation.

A company which generates more cash deserves more recognition.

Even ignoring that data issue Q3-24, you can see the difference in margins. Enphase will be a better stock and deserve a bigger premium, at least based on today’s data.

But things are changing.

SolarEdge has been transforming its business. They drastically improved the quality of their inverters after CEO transition, reducing their failure rate from ~7% to <2% compared to ~1% for Enphase, and have improved their batteries to be comparable to competition in terms of quality.

SolarEdge is a turnaround story that comes with execution risks but also potential. If they were to catch up in terms of quality and efficiency while still being cheaper than Enphase, demand would shift towards its product accelerating growth, stabilizing margins and growing cash generation. This is the vision of new management and their objective - assumed many times over the last quarters.

That said, it wouldn’t change the architecture advantages of Enphase - notably having independent panels which are globally more efficient and impactful in case of failure while also more complex. But it would make a serious competitor not only in pricing, but also in quality.

The battle for market share has been ferocious between both with Enphase at the top during the peak as regulations & financing conditions allowed most to buy premium, but SolarEdge is back at #1 as many focus on affordability again, while quality has increased making its value per dollar hard to compete with.

You can see this in the revenue profile of both companies as SolarEdge is clearly back to growth while Enphase is still a bit lagging.

One has growth & a potential transformation ongoing which would make it a serious competitor in terms of quality, market share and cash generation. The other is the established Rolls-Royce which could keep its throne and will be rewarded by the market with stable and higher cash generation.

Both require solar demand to pick up.
And both can win together.

Investment Strategy

SolarEdge’s bottom was ~0.7x sales with revenues declining ~60% at that time. This is what we can call a bottom and as this write-up showed, a situation we should not revisit now that inventories are clean and demand picks up. We’d need a comparable situation to get back to those multiples, and it shouldn’t happen short/medium term.

This ~2x sales was also the bottom before the run-up during the solar bubble with revenues were growing 100%+ and we reached the top around ~35x sales. I wouldn’t expect anything comparable but I could make the point that 1.5x-2x sales is a healthy multiple for SolarEdge - a great risk/reward, assuming healthy sector and operations.

To my opinion, normal operation is the least we could expect moving forward. We still could wish for more data about demand but we’ve seen it stable through quarters and guidance is positive - even with Q1 being the weakest due to seasonality. SolarEdge is an optimistic investment on management executing its transformation. If that were to happen, ~2x sales wouldn’t be normal, it would be cheap. Enphase did not reach 2x sales at the bottom of the inventory crash, it bottomed at ~2.3x sales.

Outside of black swans or recessions, I could see SolarEdge having a bottom around 1.5x sales while its potential could easily surpass 4x sales, probably higher, in the bull case of a successful transformation, growing margins and market share.

Price action-wise, we have what I look for.

A -97% through 3 years followed by a clear bottom late 2024, growing volume while holding that bottom which means accumulation, followed by different breakouts and a reclaim of the weekly 50 in August 2025, which have been held since except for some divergences slightly below.

The last quarter wasn’t as great as many would expect and yet was rewarded by a nice retest of the weekly 50 - which I personally bought, and a +15% right after, signaling that the potential of the turnaround was more important than the actual results, and that the market was optimistic about it.

Anything around $35 seems like a fair buy as long as the stock holds above its weekly 50 and local bottom ~$28. This is a great risk-reward I can take today with a very large potential if SolarEdge were to succeed in its transformation, even more if demand were to pick up and revenues/cash generation also accelerate.

On Enphase, we are a bit early to my taste. Growth is lower and demand for premium hardware might be slower to catch up but if SolarEdge does grow, Enphase logically should follow. The question is will they be able to retake market share while batteries are the most demanded products and their quality is on par with Tesla and SolarEdge while being more expensive.

Management said Q1-26 should be the bottom - although they have said that before as well. Personally, I believe Enphase is simply lagging SolarEdge & will follow, maybe a few more weeks to confirm the return of demand for premium hardware but that demand should return.

Price action is a mimic of SolarEdge a few quarters ago after a comparable drawdown followed by growing volume and a reclaim of its weekly 50. The stock is trading at ~4x sales which is correct as its local bottom was ~2.3x sales.

I wouldn’t buy there though. We don’t have confirmation of growth, margins, return of demand nor market share, we do not have a clear double bottom in terms of price action and 4x sales isn’t the best risk-reward for this name.

Enphase has great potential but I’d wait for a bit more development before chipping in, especially as if SolarEdge were to keep its market share or even increase it, the company - and stock, would suffer from it.

Both names have potential as long as solar demand continues, and I believe it will as long as financing is healthy - which should be the case moving forward. Enphase just needs a bit more confirmation while optimism is already here for SolarEdge. We’ll need a bit of time for those names to prove themselves, probably quarters, but there is potential for both for an optimistic and patient investor who can hold those for ~2 years or so, as long as the stocks behave properly and the news confirms the return of demand for solar systems.

Social media darlings are always ignored after they gave back all their gains.
But that’s usually when they pick up again.

I apologize for taking so long to share this report. SolarEdge ran since I started my position and finished my research; I was taken by other subjects & didn’t have the time to write…

Price action always gives second chances to the patient ones.

——————————————————

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

I’m a nerd. I read and write financial articles; I have no clue what a “Garmin” is; so they all looked at me like I was an alien when I asked, before taking time to explain that it’s their watch. The best watch in the world apparently, for runners & athletes in general - or so they say.

There’s a reason for that introduction, don’t worry.

I’m still a nerd who runs but I usually do so with my feet, a short & ear buds, nothing else, or a t-shirt maybe. But here they were, waving their Garmin under my nose and explaining what it can do for you - and it doesn’t run for you. They kept comparing them until my uncle complained about his for one reason: his battery isn’t that great.

Something I could relate to with my Bluetooth earbuds. I don’t run for three or four hours, but a long flight or work session and battery’s dead. Charges really fast but it leaves an uncomfortable gap for me sometimes. His Garmin needed a bit longer to die but he still found it too rapid, a few hours usually, less than a day while using it with GPS trackers etc…

This is a clear pain point. Not just for my uncle & me, not just for watches & earbuds, but for millions, maybe billions, and for all wearables. Batteries don’t last long enough and the more functionalities they have, the shorter they hold. We want wearables that hold at least a full day so we can go about our day and charge them only at night, like we do with our smartphones.

Guess what we’ll be talking about today?

You probably guessed batteries. Wrong; half wrong. We’re going to talk about the nRF54L, which won’t give you much of a clue either, but by the time we finish this write-up you’ll understand everything.

It does have to do with batteries, though. And I believe there is a strong bull case here for many reasons, the first being the situation I just described. Would you pay a bit more for better battery life? For your Garmin to last twice as long, or even just 50% longer? For your earbuds to last 7-8 hours instead of 4h?

I would. I think most of you would too, and frankly, the last few decades have shown that battery life is one of the most important concerns for end users.

As AI evolves and new generation of wearables are being commercialized, we will be facing battery problems as AI drains it multiple times faster than traditionally. The new Meta Smart Glasses, for example, last ~5h with constant but basic use.

So you’d put your glasses on your nose for a zoo session with your kids and you’re done by lunch, with no place to plug it, unless you carry an external battery to use during lunch break. These products are already impressive but there is still massive need for improved battery life.

This is the situation as of today. Wearables are going to improve rapidly in the coming years thanks to AI, but their battery life won’t convince any user to buy. Everyone wants AI but no one wants to spend $500 on ~4h effective battery life.

Enter Nordic Semiconductor’s nRF54L: not a battery, a chip designed not to expand batteries themselves, but to optimize usage to a point where your wearable’s battery size feels like it tripled - or more.

Nordic Semi - The 2019-2022 Boom & Burst

Before talking about AI and the nRF54L, we need to talk about cyclicality. It’s no secret that hardware - chips in particular, are a cyclical business that follows product cycles. As technology evolves, companies innovate with products requiring more advanced chips, triggering demand that creates a virtuous cycle; until demand normalizes as penetration reaches its limit and we wait for the next wave to generate momentum.

This is a well-known process in the hardware industry.

Nordic’s first major cycle happened during COVID, which raised a key question for investors: were their products only demanded because of the unique circumstances COVID created, or is Nordic a real cyclical hardware player that will see demand with new products generation?

My answer is it will. The market is still doubting.

As to what Nordic sells, it is pretty simple: chips. Semiconductors. Tiny, ultra-efficient wireless chips that let smart wearables and IoT devices run for days or weeks on small batteries by optimizing compute usage. That’s it. That’s what they do.

And their hardware became a scarce resource when COVID hit, for two reasons.

1. Stuck-at-home consumers spent more on IoT devices and Bluetooth wearables like earbuds, driven by genuine demand: wireless mouse, keyboards for remote work, audio accessories. This wasn’t purely a COVID story, some of these were already becoming mainstream. Take the AirPods, first commercialized in 2019. A flagship product that democratized Bluetooth earbuds and which everyone still uses today.

2. Supply chain shocks from the demand surge for all kinds of tech hardware made it impossible to match orders. Nordic is a fabless company, meaning they design chips and outsource manufacturing to TSMC and others. As foundries were filled of GPU demand, Nordic’s products faced delays too.

The product demand was organic; boosted by COVID, yes, but these products are still in use today, still in high demand. Take wireless earbuds as an example. Demand kept growing post-COVID notably in 2024 with 12% YoY volume growth. That’s a sustained organic demand, not a one-time spike.

Every major smartphone company now sells wireless earbuds & most of them still use Nordic’s chips. The company had ~45% market share in this segment while Apple has a ~35% market share using in-house hardware. Meaning Nordic’s extremely dominant excluding Apple.

Earbuds are just one example. Nordic’s chips are found in wireless keyboards, mouses, gaming peripherals, fitness accessories, smartwatches and countless other wearables including healthcare - I’ll list some later.

Back to cyclicality. As demand grew, hardware manufacturers built inventories to meet it while the supply chain was overwhelmed. Delays stretched, orders kept piling up & were eventually filled. To a point where final demand normalized and inventories were self-sufficient to meet it. So orders for Nordic chips stopped.

Nordic’s order book looked like demand had vanished. The truth was end demand was still there; normalized, but real. Their customers simply had enough inventory to meet it.

This is what cyclical revenues look like.

Bluetooth-connected & wireless devices have never been more in demand than today and they’ll be even more tomorrow when AI-powered. This was a story of inventory build-up and the delay to work through it, not one of a product in demand due to a one time foreign even which changed the world for a fixed period of time.

And it’s starting to show in the numbers.

The demand bottom was reached in Q1-24, and there has been a clear recovery since. A recovery that confirmed Nordic’s hardware was still very much in demand.

Q3 came in slightly above our guide, in the higher end of our guiding range at $179 million, and we are enjoying a gradual recovery continuing through 2025, which we are enjoying with strong competitiveness from relatively old products.

CEO Vegard Wollan, Q3-25

Growth from relatively old products. Key words. We might finally be entering a new cycle, driven by a new product category.

The AI Wearables

I’m not going to blow your mind by telling you that AI changed the game. But AI until recently has been what? ChatGPT, Claude, autonomous vehicles, a handful of notable software… Nothing transformative in mainstream hardware.

Innovation takes time, but as the technology is now two to three years old - outside of pure research, we’re beginning to see commercialization of serious products. One example stands out above all others, released in early 2024.

Of course, not many wanted them at first. It was a gadget, not a very useful one as AI was early and the glasses didn’t add meaningful value to most people’s lives. But as the technology improved, the hardware’s capabilities improved with it.

Last year, more than 7M glasses were sold. Meta sold ~2M of them in 2023 and 2024 combined, a 350% increase in 2025 compared to the two prior years combined. This product was the only responsible for EssilorLuxottica’s FY25 growth.

There is no denying the demand for smart glasses anymore. They now allow you to do what you’d do with a phone, except better in certain respects as AI wearables have a few key advantages over our phones.

• They don’t need a screen. Voice and gesture are enough to interact wit them.

• They can see and hear what you see and hear making them context-driven.

• They can be integrated into socially accepted hardware like glasses, rings or watches. It’s better than using your phone in many situations.

And one key disadvantage, very important for Nordic.

• They need to be constantly aware of their surroundings to answer commands.

Meta’s management kept production low because they weren’t sure consumers were ready. They were right at the start but numbers prove they are now and management asked for production be boosted next year to ~20M units to meet demand.

I’ll let you guess two things. Who manufactures the chip responsible for optimizing power consumption in those glasses? And what is the most commonly reported complaint from users?

Nordic. And battery life.

Which should bring us to the core of this article: the nRF54. But before getting there, a brief word on the broader ecosystem; smart glasses aren’t the only product Nordic powers, even if they are the most iconic in the AI era.

• Rings (Oura Ring and Samsung Galaxy Ring), used for fitness and healthcare to monitor heart rate, temperature, sleep and more. Discreet and data-rich, with ~6M units sold in FY25, and a ~50% increase expected FY26.

• Watches like Garmin, Whoop, Fitbit… market leaders across use cases from high-performance sport to casual, feeding users with personal data as everyone is now comparing their REM sleep cycle for some reason.

• Healthcare devices from Dexcom and Abbott; glucose monitors, cardiac patches, all requiring long battery life and reliable real-time response. Nordic’s industrial and healthcare segment grew approximately 60% YoY in 2025.

With many more to come with categories and wearables we don’t yet know about as all tech companies are working on their own product incorporating AI services.

Which finally brings us to the core product.

The nRF54

To recap, the core problem of wearables is battery & as we add capabilities, batteries drain faster. AI makes this exponentially worse: AI inference consumes 3x to 5x more energy than conventional processing, and the Meta glasses’ ~5h battery life under active use is already a bottleneck.

This happens because wearables face a unique constraint compared to smartphones: they need to be continuously aware of their surroundings. You want your glasses to respond when you speak, but only when you speak, not when someone nearby does. You want it to see what you see but only when you need it, not constantly. The device needs to know everything and to answer when asked while consuming almost nothing to identify your asking. It needs to see with eyes closed.

To illustrate Nordic’s product, consider the human brain: it is lazy, on purpose. When you’re sleeping, your brain doesn’t activate with every minor stimulus. Imagine if you woke up every time your wife moved a finger, you’d never sleep. Your brain would wake up your body, muscle, organ functions, and consume all your needed energy. No more REM sleep for you. That’s not what your brain does, it has habits and wakes you up only if it notices something unusual, to keep that precious energy for later. It isn’t only when you sleep, but constantly. It knows what information matters.

Nordic’s chips work on the same principle. They keep hardware in a low-power “sleep” state until something worth acting on happens, and avoid the constant wake-wake-wake cycle that drains wearable’s battery in hours. But they are pushing things even further with their latest product as they engineered it to incorporate a Neural Processing Unit capable of decision making.

AI factories train intelligence, but Nordic deploys it on device, at the edge, where the world happens. Edge AI is no longer optional, it's the only way to deliver safety, privacy, and sustainability at scale. Nordic's edge AI solution enables millisecond decisions without round-trip latency to the cloud and delivers radically improved battery life for billions of connected devices. This is the new standard for ultra-low-power edge AI, and Nordic is defining it.

Vegard Wollan at CES 2026, announcing the nRF54LM20B

Detecting gesture, filtering ambient noise or looking for voice trigger will not be done by the main processor anymore, but by the nRF54LM20B - which is the “AI” version of the nRF54. All those overly battery consuming tasks will now be done by an optimized chip while costing almost no battery. It only escalates to the processor when a real computing task demands it.

Result? the CPU sleeps 95% of the time, instead of waking thousands of times per day. This is not a small improvement; we’re talking about a leap forward.

Tests suggest wearable battery life could improve 3x to 7x from today; smart glasses approaching a full 12h of use, rings lasting a full week, medical patches lasting weeks. I started this piece by saying the minimum acceptable bar is a full day, allowing us to charge them at night like our phones - outside of healthcare. This nRF54 version is what makes that bar achievable for AI-enabled wearables.

The nRF52 was released in 2015 and remains the industry standard for Low-Energy applications today, with ~30% of the global Bluetooth LE market shares - including Apple and its in-house tech once again. The nRF54 dramatically improves processing efficiency, is 50% smaller and has better memory capacities, enabling devices to run multi processes simultaneously.

More process. More capacity. Less space. Less battery consumption.

Competitive Advantages & Competition

Nordic is a one-trick pony. They have one product, different generations. What they do, they do excellently. Companies like Meta, Samsung, and even Apple in specific cases (Find Me service) use their products. There are several reasons for this.

1. They are the best on the market due to their specialization. It is always easier to do something exceptionally well when you have only one thing to do.

2. Their SDK enables deep customization; increasingly important in the AI era, where we already see demand for custom compute systems with GPUs and specialized accelerators. The same dynamic will apply to hardware chips.

3. Switching costs are massive. We’re talking about a tiny chip present in millions of devices designed to millimeter-level precision. Changing them mean redesigning the entire hardware to optimize space, weight, form…

4. The nRF54 is the first 22 nm Bluetooth Low Energy chip with an NPU to reach the market. As of today, there is simply no direct competition for this specific product. Nordic is setting the benchmark.

By Q4-25, the nRF54 accounted for ~15% of new Bluetooth LE certifications, up from less than 10% in Q3. A pretty clear momentum.

We have leadership in hardware with our 22nm and in software with Nordic Connect SDK. Our position as a connectivity hub makes us a unique life cycle management partner for customers.

Competition is limited. The most credible competitor was Silicon Labs which was acquired by Texas Instruments last year. They still are serious competition with much more resources now, but TI is already challenged to meet demand for general-purpose chips and it is unclear how they will use this new resource.

Designing a new competitive 22nm chip with NPU would require at least a year or two of engineering alone, time during which Nordic can deepen its partnerships and take place in generations of AI wearable designs with high switching costs.

There are Chinese alternatives, the European STMicroelectronics and others but none come close to Nordic’s level of power optimization and customization - yet. And for any premium wearable manufacturer, the geopolitical and tariff risks of partnering with a Chinese supplier is seriously taken into consideration nowadays.

The Opportunity

I try to avoid projections - they are nothing but optimistic maybes. But I’ll share the known partners and products using Nordic’s nRF52 that are likely to transition to nRF54. It won’t be revenue projection but it certainly gives an idea.

• Meta Smart Glasses, projected to push production to ~30M by FY27. Compared to the 7M units this year this means a 3x net revenue increase only from Meta which should translate to ~$150M added revenue, ~25% of the company’s FY25 revenue. From one product.

• Samsung Galaxy Ring, doubled production between FY24 and FY25 with assumptions pointing to ~3M units sold, another double, in FY27.

• Oura Ring (all generations), projected to grow ~100% YoY.

• Whoop, growing slowly compared to others at ~12%, but a consistent volume.

• Garmin, low double-digit growth with increasing Nordic chip content as the growth shifts toward AI-enhanced models.

• Google Pixel Buds, Logitech wireless peripherals, Dexcom and Abbott CGM patches… large-volume, steady-growth products already running Nordic chips, whose next-generation versions will require nRF54’s efficiency for AI features or just massive improvement in battery life with the non NPU version.

The growth over the next two years looks strong even ignoring any new AI wearable, while many should be commercialized by then given that Google, Amazon & Apple are all reported to be working on smart glasses and OpenAI is working on a secret hardware; plus countless existing wearables which will upgrade their product with AI capabilities and smaller companies creating new ones.

It doesn’t mean Nordic will have a 100% of market shares, but it sounds like they will either increase or stabilize it while volume will grow. A stable 30% share of an industry growing 30% is a pretty strong growth & that’s what we start to see in the numbers.

Some customers are concerned about potential component shortages, which may lead to restocking. We are fairly optimistic that we should be getting the capacity. Excellent collaborations with TSMC and GlobalFoundries as our two main foundries.

Wollan on the Q4-25 call.

Do I hear inventory stocking? Possibly. And if so, this could mirror the early stages of 2020-2021, with demand driven by real AI wearable and without a supply chain shock this time, which will not allow backlog to reach covid-period’s order books but still yield a pretty strong enthusiasm.

The 2022 demand was “phantom” due to inventory building. Those are now empty and Nordic is back to a clear growth trajectory with FY25 up 31% YoY and the latest quarters stabilizing at a healthy double digit growth.

With the nRF54 chip supposed to ramp up volume from 2026 and the actual trends in AI wearables, this could end up in growing demand, anticipation, inventory building and trigger a covid-like reaction in the stock, fueled by optimism.

The market already reacted after Q1-26 guidance that came in more than 15% above consensus at a 19% YoY revenue growth; analysts raised their expectations from ~12% to ~19% FY25 while the IoT sector is expected to grow in the high single digits.

Something is happening at Nordic, and it has a name: the nRF54.

We believe we are on track for our long-term 20% annual growth target, supported by our product portfolio renewal.

This is where I remind you that besides all the great stories, the only thing the market cares about is projected and safe cash generation, which can also play in our favor with Nordic.

Gross margins are what you’d expect from a fabless semiconductor company: 50%+ and rising. That rise is driven by three factors.

• nRF54 is a higher-margin product, even if price is comparable to the nRF52.

• Software revenue, introduced in 2023, is now bundled with chips to help clients customize and optimize usage, and they will use it even more with AI needs; a new standard in the semiconductor industry.

• Revenue mix is improving with smaller specialized customers, notably healthcare who pays a premium with lower volume.

• Global efficiency after inventory destocking and larger orders to TSM and other fabs - more volume means cheaper units.

These tailwinds are structural, not temporary. Management has guided toward 55% gross margins, long term, hence stabilization.

Net margins are weak though for a semi company at ~2-3%, but these are set to expand as revenue scales as expenses shouldn’t increase as fast as revenues and there’ll be a point where any net growth will expand margins & cash generation.

That double acceleration is the core financial argument for the stock; it’s what we need to have a stock repricing.

Investment Strategy

This one is straightforward to me: common shares. I see no value in using options on a company of this size, at this valuation, with this level of timing uncertainty. I usually use options to boost returns on high-conviction, low-volatility names with clear near-term catalysts and low valuation. Nordic doesn’t fit.

We are not talking about a multi-billion-dollar blue chip at low valuation, but a $2.6B Norwegian company at fair multiples. This is a small-cap and it comes with risks and volatility of a small cap.

Nordic is not cheap today. At ~NOK 140 or ~$14.50, it would be if the AI wearable thesis plays out - if we are indeed entering a new inventory cycle driven by nRF54 demand. But until that is confirmed by more than one strong guidance, we are trading at or slightly above fair value for the current business.

We’re all used to 20x sales being called cheap by now but we shouldn’t.

I don’t expect Nordic to deserve a re-rating outside of my bull case. That said, a 20% growth rate at a stable multiple would still yield strong return, and that seems to be where we’re headed on the base case. But expanding margins, growth acceleration and inventory building are the bull case and cause for re-rating and a deserved ~8x sales multiple. That’s where we have real fun.

What makes this play very interesting to me is the limited downside, or reasons for a downside. Every fundamentals point to positive outcomes and the stock is trading at a fair valuation excluding this positivity, as if we were going for years of low double digit growth and stable margins. The bear case would be Meta shifting to in-house chips or another provider but as said, while possible, it is unlikely.

A 3x - 4x sales is fair for a company in Nordic’s actual state and comparable to peers. Anything lower would require an external event and we saw even in the worst of the inventory situation, Nordic fell at 2.2x sales. We shouldn’t get any close to this kind of situation in the next year or two.

What I would call a great opportunity at acceptable risks.

Price action wise, you can clearly see the post-COVID inventory crash and we’ve been in a gradual uptrend since, without fireworks but with an established bottom in March at peak order book weakness followed by healthy higher lows.

The stock reached a higher high ~8 months ago and has been consolidating since. The market is waiting for nRF54 to actually ramp in the numbers and even if Q1-26 guidance confirmed this, the market isn’t getting too optimistic yet.

It’s clear that the nRF54 Series is contributing very small in 2025. It is a year where our workhorse, the nRF52 Series, is predominantly our revenue - products that are relatively old. The Nordic team has done a great job defending that position up until having the nRF54 Series. What’s really energizing for us is that we see the strong competitiveness of our new 22nm platforms.

We are currently just below the prior higher high, above the weekly 50, on the second-highest volume spike of the decade following earnings. Good setup to my opinion.

Another issue with the stock is visibility. We’re talking about a Norwegian small cap & price action mostly moved on earning days. Boredom is a bear case and we could have to wait weeks with no price action until volatility kicks, down or up, during earnings. Under the radar opportunities need to shine to be bought.

I personally bought a starter position of ~40% final allocation. I’ll continue adding the next months. Only shares. Small-cap, uncertain timing, not worth boosting with option risk on top. I believe this could become a nice 2026 play as AI democratize outside of ChatGPT and demand for wearables is very healthy and growing. It seems like a great set up from all angles, assuming one can be patient enough to hold until the catalysts materialize.

This closes my review of Nordic Semiconductor. Hope you enjoyed it; if so, feel free to share your thoughts or give it a like. I’ll see you for the next one, have some things up my sleeves for the coming weeks!

——————————————————

Disclaimer: I am not a licensed financial advisor, analyst, or broker. This content reflects my personal opinions and investment decisions for informational and educational purposes only. I hold positions in securities discussed and may buy or sell without notice. Nothing here constitutes a recommendation to buy, sell, or hold any security. Past performance does not guarantee future results.

Always conduct your own research and consult a qualified professional before making investment decisions. I accept no responsibility for any financial losses.

Pretty big update today. I’ve been working for a few weeks on a better way to share my strategy, performances, position updates & clear “educational” content on how I invest in the market.

Substack is an amazing service, but it has limits, especially for writers in the financial space. I have ~300 write-ups as of today. If you wanted to review my positions and strategies, you’d need to search through all of them to first understand my methods, second find the stock you’re interested in, and read them all to understand process and positions over the months.

Not really practical. So I worked on making it by centralizing all key information in a website focused on my strategy, stock picking and performance.

You’ll find different sections meant to help you find the information you need when you need it, instead of going through hundreds of write-ups looking for it.

The Home section is just a home section with a rapid overview of the achievement & access to content.

My Strategy details my framework. Not only how I invest but my perception of the market, how they work, why price action matters, why I do not buy downtrends nor DCA, options, margin, positioning strategies, etc... Nothing more than what was shared on Substack over the last year, but centralized and easy to access.

Thesis & Updates regroups my investing-focused write-ups shared on Substack. No difference in content, but since everything is centralized, you’ll be able to find the key write-ups per asset and have a clear, rapid overview of my positioning - once again, easier than looking through hundreds of write-ups.

Performances covers all stock picks following my investing method. Note that this isn’t a representation of my portfolio but one of my stock picking and investment strategy; from starting a position to closing it in gain or loss.

Every asset has a card with information related to my own investing framework and links to the latest content written about the asset. Centralized and efficient.

Note that I only added assets shared recently or in position. I didn’t include positions I’m no longer involved with from January 2026; winners - Hims or Palantir closed after 200%+ and 700%+, and losers - Duolingo & PayPal closed after -48% and -12%. My objective is to move forward, not review a year and a half.

I hope you will find the website useful! This is only a beta version. I decided to share it with you now because it’s functional, but I still have work to do to improve it visually, add some functionalities and ergonomics. Nothing better than putting products out there to get feedback, even if they aren’t perfected yet.

As usual, any feedback is welcome, positive & negative. I hope you’ll find it a great addition to my Substack. Nothing will change here though, this is purely additive. Here you go:

https://thefewbetsthatmatter.com

Don’t be fooled by the earnings overview, this quarter wasn’t a miss.

I’ll start by commenting on the quarter from a fundamental point of view only – and there aren’t many bear points to comment on; I’ll talk about the market’s perspective and why I am not overly bullish on the stock itself after.

Because yes, one can be bullish on a company and not on its stock.

The Perfect Bull Case

There is nothing wrong with this quarter in my opinion; some of the bear cases are even falling, their myths being deconstructed. Nebius bull case remains to compete with hyperscalers with high compute quality, availability and pricing.

And they’re great on everything while the market continues to be supply constrained with very high demand, but we knew that.

We sold out of capacity in Q3 and Q4 last year, and we’re already now in Q1 of 2026, also sold out.

Even before we bring capacity online, it’s often, it’s often sold out. As a result, the average contract duration of new cloud customers grew by 50%, and the prices of GPUs didn’t fall, even on previous generations of GPUs, as the industry may have expected. Customers are demanding more compute and increasingly, increasingly sophisticated solutions to run their AI workloads.

And not only is demand and quality expectation high now, but it is high for the future as commitments are increasing and companies are fighting to secure their shares of compute, even if it means at a slightly higher price.

In Q4, we saw nearly twice as many transactions completed for over 12 months in duration over what we succeeded with in Q3, while average selling prices increased by more than 50%. As an example, we are sold out of hoppers, and those that are coming up for renewal, often off of short reserve agreements, are getting renewed at 12 months or longer, while we’re actually seeing pricing nudging up.

Which means management is optimistic. Logically. Especially while also being on time to deliver their (very large) contracts with Microsoft & Meta and having a very high demand even on holder GPU generation - which was one of the first bear case.

We are reiterating our annualized run rate revenue of $7B - $9B by the end of 2026. For the full year, we expect to achieve between $3B & $3.4B in revenue.

Over the last few months, our conviction in this range has become stronger. Why? Because, because we exceeded the high end of our 2025 ARR guidance and showed more than $1.2B ARR. Because we already contracted more than 2 GW of capacity and are on track to exceed 3 GW this year. Because we have already delivered all of our capacity for the Meta contract, because we are on track to deliver the capacity for Microsoft through the course of 2026, exactly as planned. And lastly, the demand, demand for our AI cloud continues to be strong. Pricing is strong.

Everything seems perfect and who wouldn’t be bullish up to here?

The challenge is the same for Nebius as for any hyperscaler and compute provider: to grow supply, which means more infrastructure, GPUs, energy consumption; and that takes two things: time & money. Two things Nebius has, which is why I’ve loved this stock more than others.

And they are showing it by expanding and accelerating their buildouts.

They exceeded their own FY25 ARR guidance, were sold out on capacity, reiterated their FY26 ARR guidance based on today’s very positive market dynamics, and increased both their FY26 contracted power guidance and the number of datacenters they intend to have running from 7 to 16.

Massive demand requires a massive expansion plan.

And compared to many companies – like Oracle, Nebius still has very strong shoulders to expand, so while the market worries about CapEx and debt, Nebius management has this kind of comment.

In order to capture the large and growing opportunity that we see for the future, we plan to invest in CapEx in the range of $16B - $20B in 2026. We already have about 60% of the capital needed for this range from our balance sheet, existing operations, and commitments.

We evaluate several funding options available to us on a consistent basis, and we’ll deploy two guardrails when we look at capital alternatives. First, we will focus on raising debt relative to our business needs and will be prudent with respect to the cost of capital. Second, we will be mindful about the shareholder dilution if we choose to issue equity. Given our balance sheet and minimal debt, we are fortunate to have many additional options to finance our growing business.

For example, we are currently exploring adding corporate debt and asset-backed financing to our balance sheet.

But most importantly, I would say, we have significant amount of cash that we received, and we will continue to receive in 2026 from the favorable terms of our long-term contracts.

In English: they have enough money and will generate enough more to finance ~60% of their expansion, while the rest could be financed by asset-backed debt, leveraging their stake in ClickHouse, Avride, Toloka or else… Minimum debt & dilution.

Everything to please the market. And yet…

Nothing is Never Enough

As I’ve said for weeks, maybe months now, nothing I shared above matters anymore.

The market is not looking at potential, it is looking at facts and pricing risks. Today, it only sees a company which cannot accelerate growth because its limit is not reach or engagement, but physical constraints with slow buildouts while spending accelerates to capitalize on an opportunity which could shift in a year’s time.

Who says demand for compute will be equal in a year?
Who says new hardware won't be enough to meet it?

I personally do not believe this, I believe buildouts are necessary & Nebius will thrive. But the last time I tried to give my opinion to the market and told it my stocks should go up… Well, it didn’t care – rightfully so.

What matters is liquidity, risk appetite, sentiment.

Today, Nebius is a $23B company telling the market its growth & cash generation are constrained by physical ceilings which cannot be avoided and is seen in this quarter’s revenue miss, while it decides to spend a small 100% of its capitalization on buildouts whose usefulness is a question mark.

That's what the market sees. And just like you, I find it stupid.
But again: My opinion doesn't really matter.

And if we could build faster, yeah, and even more than we have today, we would do it. So we're building because it's clear growth.

And if they could, the market wouldn’t punish them, because growth wouldn’t be linear; it would be exponential.

Investment Execution

A company and its stock are two very different things.

So I will stick to my opinion that Nebius is an excellent company the market will not reward because of uncertainty, execution risk, and growth ceiling. And I’d love to be wrong. If you believe the next 5 years are nothing but rainbows and unicorns and you feel capable of holding a potential growth stock in downtrend – which isn’t the case yet but could become it, Nebius is certainly a great name to accumulate at today’s price.

I’d personally hold as long as we are above the weekly 50 but I would be cautious and aware of the risks and market’s reactions. As long as Google, Microsoft & co are not rewarded for expanding and insatiable demand, I don’t see why Nebius would be, especially at today’s valuation.

Although yes, growth is incomparable as Nebius is guiding ~500% revenue growth which would push its P/S down to ~10 by end of year 2026, matching others. To that we need to add execution risk and demand concerns by then, making Nebius the highest valuation and risk of them - except maybe for CoreWeave.

I remain very bullish on the company itself and hopefully I’ll buy back in a few months – maybe years. Who knows? This is an excellent company and I understand why many would like to buy and hold it. It’s just not how I personally do things, but to each their methods. Nothing wrong in that.

In the meantime, I’ll be patient and adapt myself to the complicated year 2026 is shaping out to be.