Most investing content is built on looking so certain that readers assume you know better. No one knows better. Every investor is constantly learning, and that’s what I’ve decided to build my Substack around: constant improvement, out loud.

Building a framework that works, publicly.

So I share the diagnostics, the successes and mistakes, to strengthen a system that’s performing well, but isn’t yet perfect. This is what today is about: the last month, the mistakes I made, and the measures I’m putting in place so they don’t repeat. I’ve identified two clearly.

To talk numbers: my portfolio was up ~160% YTD at the top in May. It’s up ~60% today. That is NOT normal, it shouldn’t have happened. A 60% YTD is still a massive outperformance over the S&P ~8%, and it should be seen as an excellent year - which it is. But the subject here isn’t the returns. It’s the process that let me give back so much in such a short timeframe.

I’ve diagnosed my own mistakes before - notably overtrading, which I’ve corrected; I haven’t done it since. That one was about me, today is about improving the system itself, to protect my net worth in riskier environments like today’s.

More Wars, More Fear

A quick word on what’s been happening.

Two main reasons for the violent sell-off in risk these last weeks. First, AI hardware was extended and took most of the drawdown while most were heavily invested. Second, the framing changed: few of us expected the Iran/US conflict to start again while peace frameworks were already on the table. And yet, here we are.

You know how this works. Black one day, white the next. But the risk is different now; the market is extended, so investors aren’t in the state of mind they were six months ago. They’ve made tons of profit and intend to keep it. We’re in a “better to sell first than last” environment, where risk isn’t a short-term opportunity but a place to pull profits from and wait.

So how do you avoid giving back a year’s gains in a month?

Below, I break down where I went wrong; the two specific mistakes that cost me most - including the one I estimate cost me half of my drawdown, plus the rules I’m putting in place so they do not repeat and how I’m managing my book as the market shifts.

Let’s get into what I’ll do differently, and what I’m doing now.

Read more

Market psychology 101.

Those trades are now crowded. Everyone and their grandmother knows the current bottlenecks and the best proxies for return. The market is a place of probabilities, and you want to store your money in the assets with the highest potential. Those are rarely crowded places.

We are exactly where we were in September last year for neoclouds.

GPUs and compute were the obvious trade; we could never have enough, and the market priced them to perfection, until the trade got crowded, everyone understood the thesis, and latecomers got punished. Fast forward to today: only the greatest names bounced - Google and Nebius, while the lesser ones didn’t - CoreWeave and Oracle.

Being late on a right thesis is just as good as being wrong.

I'll use Nebius as an example to illustrate what I expect. Everyone here knows the company inside out, so I'll focus on market mechanisms.

The stock topped early October ‘25, shortly after the game-changer deal with Microsoft, in an atmosphere of stupid bullishness. Everyone talked about the name, everyone wanted to buy, and price targets were in the four digits. The trade was crowded.

Then the correction came, violent, down 50% in a month. It retested its W50 rapidly and went through months of consolidation, never losing that W50, before breaking out again.

This is what I expect for photonics.

Violent retests - where we are now. Weeks of consolidation. And breakouts later, for the best assets only - which are only a handful. Meanwhile, liquidity will rotate to the sectors we talked about, just like it went to energy/industrials at the end of 2025.

Chances are you’ll do fine if you hold great assets and accept the volatility — but you’d better have a strong stomach, and accept cutting those positions if they go south. The market won’t be forgiving on diamond hands holding the wrong bottlenecks.

Last week’s article went over this rotation and the pockets of strength on the market.

• Catching the space retest, with Planet Labs up 20.7% and ASTS up 15% since.

• Looking at Amazon with a 4.5% bounce.

• SaaS names like Qualys, Zeta and Tenable up 20%, 9%, 27% respectively in a week.

• Looking at healthcare, which performed well - especially Abivax, which received the confirmation expected the day after my article and shot up 50% in a week.

The breather is in AI hardware - including power hardware. It feels violent because the sector is overcrowded, so everyone talks about it. But other sectors are behaving really well, and we were positioned for it; the portfolio is down less than 1% this week despite a violent pullback in key sectors.

So the real question now is how to manage this retest.

Below, I break down exactly which names I’d buy, why I’m going as high up the supply chain as possible, and the one signal I need to see before I deploy the cash I’m still holding. I’ll also walk you through my book; what I’m accumulating and why, what’s on the chopping block, and the single name I’m about to size up.

Read more

But today isn’t about tech. It’s about an unloved, underestimated, disregarded sector.
Undercrowded, with a massive AI tailwind.

Who Buys Healthcare?

That’s a great question. If you did, you probably underperformed these last months. But things are changing.

First, the market is heated. I said yesterday it was time to slow down, to let breathers happen and prices deflate for this year’s winners. They ran a lot and now need a minute. No reason to sell those positions, but plenty of reasons to look elsewhere.

Second, artificial intelligence is becoming so powerful that governments now have to control the public release of new models - whether it is normal in a democracy or not. And one industry will thrive from having such a data-processing tool: research. Healthcare R&D, of course.

If you haven’t seen “The Thinking Game” and are curious about the early years of AI and the people who built it, go watch it. It tells the story of DeepMind’s founding founder, and how they worked for two decades to build what we have today - getting bought by Google along the way. They tried many things, but eventually found the two best subjects to test AI.

1. Chess - and later as it got too easy, Go.

2. Protein folding - when Go got too easy.

I’ll let AI explain protein folding, as I’m not a biologist - rest assured you don’t need to be one to make money.

Protein folding is how a chain of amino acids becomes a working 3D protein. It works because the chain naturally bends, twists, and packs into the shape that is most stable in its environment. That final shape matters because a protein’s function depends on its structure. If folding goes wrong, the protein may misfunction or clump together.

It is a major AI training problem because predicting the final 3D shape from the amino-acid sequence is extremely complex. The challenge comes from the huge number of possible shapes and interactions.

AI is meant to process massive amounts of data and propose solutions at a speed unthinkable less than a decade ago. Protein folding was solved and Deepmind’s team received a Nobel Prize in Chemistry for the latest version of their model. The benefits for healthcare are hard to quantify… but “game changer” about covers it.

So the real question is: what is AI’s limit in healthcare?

Hard to say. For most sectors, the bottleneck is the lack of specific data. That’s never been the case here, where the issue was enough compute to process it. That’s why healthcare should see a transformation like never before in the years to come.

But we know this doesn’t necessarily translate into market gains, because liquidity flows to the highest potential, not the best fundamentals. For years it’s been AI hardware and infrastructure, but the market is waking up to the rest. This last earnings season showed SaaS names weren’t as endangered and could generate cash from AI, and their stocks started to react.

But… isn’t healthcare even more impacted by AI?

Market Mechanisms

I’ll repeat things I’ve written many times - sorry to those reading it for the 100th time, but they matter.

You don’t need to be an IT engineer to invest in Nvidia. You don’t need to be an astrophysicist to invest in SpaceX. You don’t need to be a microbiologist to invest in healthcare research. The only thing you need is to understand why the market would get excited about it.

And the market is always excited about one thing: unlimited potential.

You know the common denominator of Micron, Nvidia, Palantir, Coherent or Nebius? The market couldn’t - and still can’t, anticipate their market size, revenues or future margins, because demand for a disruptive technology is impossible to anticipate.

The market would rather push stocks to “stupid” valuations, not because investors believe Palantir “deserves” 200x sales, but because they’d rather assume it earns it than miss out. If someone told you “this is the next Google,” would you care about the price you paid? No. You’d buy. Because you’d rather buy expensive than miss out.

Bubbles aren’t born of stupidity. They’re born of the impossibility to anticipate.

Now back to healthcare and AI. What’s the market for a company that cures a type of cancer? Degenerative diseases? Auto-immune ones? Or, less morbidly, menopausal symptoms, hair loss, and countless other “comfort” treatments?

I’m not necessarily one of those who believe humans will cure every disease on earth thanks to AI, but I am one of those who believe AI will accelerate research, not just in time and expenses, but in outcomes.

And that creates a situation where anticipating revenues, margins and market size becomes extremely complex for the market. A situation where investors are almost forced to create a bubble, because they’d rather be in it than out of it.

A 2025 study reported that using the AI system Transpara as a second reader in breast screening increased sensitivity by 8.4% and found more interval and future-detected tumors.

Another concrete case is clinical workflow automation: Chelsea and Westminster NHS Trust began piloting AI-generated discharge summaries for clinician review to speed patient throughput and reduce admin burden.

Palantir Foundry pilot: the trust reported a 28% reduction in an inpatient waiting list, theatre utilization improvement from 73% to 86%, and longer booking lead times (from ~6 to ~17 days) after using Palantir’s Foundry to unify scheduling, rostering, and waiting‑list data for operational decisions.

For drug discovery, the clearest real-world signal is that AI-designed or AI-optimized therapeutics had progressed into human trials by 2025–2026, showing that AI is already moving candidates into the clinic rather than staying in the lab.

In the rest of this article, I’ll cover the sector-wide breakout I’m seeing across 50+ charts, the proxy I’m using to play it with the least risk, and four de-risked names I’d actually buy. Plus the real risks (this is still biotech), my exact buys, and the portfolio changes I made this afternoon.

An entire sector is waking up after years asleep.

Read more

Why such a difference between my content’s portfolio and my own? Because I have been overtrading, and I am sure many of you also have.

Volatility and noise are the worst combination, they trigger your emotions and reduce your focus; nothing matters anymore but to make up the loss of the last day or week, and doing so ends up making even bigger losses. Mistakes, on mistakes, on mistakes, until your spirit - and portfolio, breaks.

This is a classic pattern in investing and we are all guilty of it. The good thing is that we can work on fixing it and while my own mistakes were made earlier this year, this last week has been great. I stuck to my system.

Reviewing last month’s plan, most was spot on - getting rid of space names before the SpaceX IPO and trimming core hardware names like Soitec. There was one mistake: believing in SaaS. Those gave a pretty violent retest this month and went on to chase traders stop losses. The core thesis behind those names is the right one, but we have to give time for price action to develop - some are interesting today.

Short term, I see profit taking on AI hardware, a healthy leg up on energy and SaaS. I intend to play this by trading SaaS names - already started with Elastic and Oracle, hold energy names, and accumulate hardware when they reach my buying targets - which Founding Members have access to.

On the index, the important data is that buyers stepped up at the D50, once again, on both the S&P and Nasdaq.

But this isn’t a beautiful candle, the wick went to test that W21 and wicks have the bad habit of being retested. And evening was sold. The most important time of the market is the last hour, when buyers and sellers fight to decide the value of an asset before the market closes. When sellers overwhelm buyers for days in a row, the consensus is overall bearish, it means the market isn’t ready to bid higher just yet. Daily timeframes are really short, not what I use to make decisions, but it is worth noting.

So far, we are holding key supports, but without any fireworks. I wouldn’t be surprised if we were to start some choppy months with more pain in extended sectors.

Anticipation is Everything

The markets always focuses on the next wider narrative. For months, it has been the war in Iran but rapidly moved to the coming resolution, as both countries proved over the weeks that the conflict wouldn’t be long. So the market anticipated its conclusion, anticipated oil prices going back down and supply chain reopening. Because it had to happen, soon.

Now, it has. So the market focuses on the next narrative, inflation - as expected. The market always anticipates, you can’t buy the news, the market prices in anticipation. Inflation is expected, so the market prices it, now.

The Fed didn’t help with a hawkish tone and the removal of key forward data - which were mostly bullshit but the market used them as a support for projections. And BoFA added another layer with a note the market read as “inflation will kill us all” while it was more like “be mindful of the potential inflation”. So let’s start by putting some color on that.

What BoFA said is that they expect three rate hikes this year, for a total of 75bps. No one can predict this, but this is their view and they have every right to share it. But instead of panicking, let’s focus on what we know, can control, and our system.

1. Interest rates aren’t as important as they used to be. They create panic and hurt some businesses, but with the constant and accelerating fiscal deficit, higher rates are only taming slightly liquidity inflows, not stopping it.

2. Investing is about liquidity rotation; money goes to the highest potential returns. With higher interest rates, where are those? Not real estate, not cash, defensives or gold and silver, as both are falling after a massive speculative run. Meanwhile, private companies are developing the most deflationary technology ever created. So, where are the highest potential returns?

3. We have a system. We don’t know, we can’t know how the market will react. We buy strength and cut weakness; we don’t need to anticipate the market’s volatility spikes. If our stocks are weak, we sell. Strength, we buy. The in-between doesn’t matter.

To me, there is nothing else to it. I can comment but can’t anticipate. What I can do is manage my book, and I do so by following my convictions and price action.

Micron Earnings

There isn’t much to comment on, Micron is the new Nvidia; memory is the new GPU, and we already know and said many times that this trend of personalized hardware won’t slow down any time soon so more memory means more connectivity, more GPUs, more bandwidth…

The market focused on GPUs, focuses on memory, and will focus on another vertical tomorrow because eventually, they all matter and all grow in volume - and prices, eventually becoming the next chokepoint. No one can accurately forecast where this is going just yet, which means the market will continue to price in the best case scenario as it would rather create a bubble than miss out.

So this quarter was what anyone could have expected: a banger quarter with the confirmation of everything we’ve heard for months. I am not sure why anyone was surprised to be honest while every hardware management told us this would happen.

The quarter was amazing and the market responded to it, without much fireworks after such a run and anticipation, but it didn’t sell the news so it means it still can be surprised - which is a good thing as expectations were sky high.

Those stocks are going much higher, it’s just hard in terms of risk management to size a position here, and yet I could bet that most of those who bought on Friday will outperform the market by next month. But I won’t, because it doesn’t fit my system.

Overall Appreciation

Fundamentals are as strong, but the market is giving us some signals.

The next weeks will be different from the last. This isn’t a time to accelerate, but to slow down. It actually was a time to slow down the moment I shared this article about how close we were to the breather, but I - and probably you, didn’t. I saw that the market was starting to shift against my trades but kept going, thinking the next one was the good one.

The reminder of this article will look at strength and weakness pockets, review my own positions, watchlist and my short/medium term plan especially when it comes to liquidity rotation and inflows - as I still have some cash left.

I remain a buyer but keep in mind that some sectors might be starting their struggles, which could go on all summer after amazing performances this year. Some trades are crowded, and crowded by the wrong population while some other names, not less important, are completely deserted. Those are the ones I’ll be buying.

Read more

What many investors forget is that stocks down 40% in a few days or weeks are often still up double digits year-to-date, and triple digits over the last year. Uptrends aren’t up only, especially on growth and narrative names. They’re a continuity of higher highs and higher lows, which can be stomach-wrenching enough to make you give up if you bought at the wrong time.

This article goes over a sector with exactly those characteristics: darlings turned into stinky socks from one week to the next, that few want to buy now that they’re down 40% in a month, while still up 100%+ over the last year.

I’ll cover the sector, the best setups, fundamentals, and my personal buys.

Read more

You don’t need a lot of positions to make a killing. You only need to go heavy on a handful, and hold them until your thesis plays out or invalidates. Those who held Nebius, Silicom, Soitec and others were rewarded for doing nothing but buying at the right time. It’s that simple.

Today, I bought a new one.

A stock I’ve had on my radar for months, and written about already. It took weeks and patience for the breakout and the retest to happen, but we’re finally there. I opened a large position this morning, as I always do for core positions, and sent an alert to paid subscribers just after. The rest of this article covers why: the company, its narrative, its fundamentals, and how I plan to trade the name.

Read more

I’m closing the Founding Member promotion after 147 of you chose to trust me for the months ahead. I thought 50 would be a successful launch, but we are once again in the #10 best sellers… This is way above anything I’d hoped. The best way to express my gratitude is to work twice as hard at making it worth your investment, so I will. My goal isn’t to simply share trades with entries and exits, but to improve the way we think about the markets, build investing systems that work for us, learn to make money by ourselves.

We’re here to learn together.

A few words on the structure moving forward. Nothing changes for free and premium subscribers. The first keeps access to the macro and fundamental/narrative portion of each write-up; premium keeps access to all current formats, including the detailed investment strategies and targets.

Founding Members have access to the chat where I share comments and alerts on swing/core positions, to my buying list updated and shared every weekend. They now also get dedicated write-ups for the swing plays, which used to be in the chat but were impractical. And I’m working on giving Founding Members daily access to my screener data through a website, so they can run their own research on the same base I do. Not ready to share it yet, but it’s coming.

I’ll use this write-up to quickly share my view, which is fairly positive. The S&P 500 and Nasdaq both bounced perfectly on their D50 - key medium-term trend, and reclaimed their D21 during the week, on volume. Buyers stepped up big, and we apparently keep buying the dip regardless of macro. Infinite optimism.

Still, I’d remain cautious. This week is K. Warsh’s first FOMC, and we know inflation is back, confirmed by the latest CPI and PPI prints. The market is pricing no hike, which seems right: inflation comes from softening past tariffs and oil pressure that looks close to resolution; though I’ll believe that when traffic is fully restored through the Strait, not before. The keynote will matter.

So caution, volatility should be strong but could go both ways. The U.S. market and the AI trade remain the best place for returns and therefore the best place for spare cash. That’s why it’s been ripping for so long. Don’t call the bubble until we have clear signs of it, and no, one highly anticipated IPO isn’t it. Not yet.

Now, the Founding Member swing plans.

Read more

Today is the last day of the launch promotion, the final chance to become a Founding Member at $350/year, fixed. From tomorrow, this subscription will be $500/year.

Founding Members get the full write-ups, Substack chat with market updates and real-time buying alerts and my personal buy-target spreadsheet. Plus new content next week: my favorite swing setups, built on price action and option flow.

Everything is built around one system which combines stock picking and liquidity attribution and has outperformed for two years straight, publicly.

The alert portfolio, started April 7th and based on Substack’s chat alerts, is up 64.54%.

The core equal-weighted portfolio, built on my write-up theses and buy targets, is up 102.57% YTD.

The value here is quantifiable and verifiable. Not every trade is a win, but over time, the system generates alpha. I’m not here to play guru, tell you what to do, or show fake numbers. I’m here to share the journey honestly - my failures and my wins, both essential to learning the most important skill there is: how to make money with money. I don’t have all the answers. No one does; anyone saying otherwise is lying.

This is the last 24 hours at $350. The discount won’t be offered again.

Whether you join today or not, nothing changes; the content keeps coming, I stay available and I am extremely grateful for your time, trust and attention. We’re all after the same thing: not just returns, but the skill behind them.

The grind never stops.

Most investors are currently focused on the AI hardware upgrade cycle, from memory to photonics, passing by networking and more - and they are right. The inference optimization is a key vertical nowadays and I’m part of it with some of my holdings. But energy remains at the heart of every economic activity: without it, nothing happens, and without it, all that new hardware will sit idle in a powerless data center.

Of all the bottlenecks you hear about on social media, energy is the least mentioned, yet the most important. And just like hardware, it has to go through an entire innovation cycle, because it isn’t optimized for AI today. We need new infrastructures as much as - more even, new networking systems.

And that’s what we’ll talk about today.

• The current situation and the need for innovation.

• The new electrical systems design.

• How to capitalize on it.

As this is the second write-up on energy, I’ll redirect you to the first one for the overall introduction: the growing usage of electricity, the massive consumption from data centers that will only accelerate, the current issues around how and where electricity is generated and transported, and the core problem for data centers - the fact that they need more energy, today.

That first write-up focuses on the need for energy, not the electrical architecture of data centers. It covers why we need so much, and how data centers can produce or find more. This second write-up focuses on how they plan to optimize it for AI.

Current Situation & Problematic

Electricity was my worst nightmare in engineering school, and yet here we are.

Today’s electrical infrastructures were built years ago for low voltage and stable loads, for a good reason: that was the need at the time. The standard model was 480VAC with as many AC/DC conversions as needed - if that was Chinese you need to read that first write-up; it was the safest and cheapest, with nothing pushing against it. That means tons of copper wires and various hardware in the path which take up space that could hold more compute.

Complex infrastructures, hard to scale and upgrade.

These infrastructures remain the norm as nothing changed since. Data centers and enterprises consumed more electricity, but volume alone wasn’t enough to change the architecture. What would force a change is if we consumed differently, and that’s exactly what’s happening with AI.

Compute racks - notably because of GPUs, have a different need. They’re more aggressive, volatile, and power hungry than other industries. A cluster’s consumption can jump from 10% to 90% in the blink of an eye, putting extreme tension and heat on the power chain.

Compute densities that once averaged 10–50 kW per rack are now routinely exceeding 120 kW, with projections approaching 1 MW per rack in the near future. This dramatic increase is driven by AI servers hat pack high-performance cores into tightly integrated architectures for ultra-low latency communication, creating unprecedented demands for power density.

On electricity 101, you learn about three metrics to understand electricity. Power - the total delivered, equals current - the flow of electricity; and voltage - the pressure moving the electrons.

As racks require more power, engineers have two levers: increase current or voltage. With data centers, the solution was to stay at low voltage - the safest and cheapest option. But low voltage means high current, and that comes at a price: the more current passes through copper, the more the metal heats. If that rings a bell, it’s because we have the exact same problem inside the hardware; that’s what photonics is trying to solve with light - the second biggest narrative today.

For electrical infrastructure, the last lever available is to raise voltage, keeping the copper from heating, because that heat is complex and expensive to manage. So they came up with two ideas: ±400VDC (400V Direct Current) and 800VDC. Both are different architectures, requiring different hardware, with different potential.

Today, the latest Nvidia GPUs can still run on 48VDC - current architectures. Problems start with the next version of Rubin: a single GPU will always function on 48VDC, but an entire rack won’t get enough power to run at 100% capacity. The next generations will require at least ±400VDC, which is enough to power ~1MW racks - most of them nowadays, while multi-MW racks will require 800VDC - most of them tomorrow.

Raising voltage raises end-to-end efficiency - fewer AC/DC conversions, less copper, less heat, etc... Over 150% more power moves through the same copper at 800VDC. Much easier infrastructure to manage and maintain, just complex to engineer.

The future vision centralizes all AC-to-DC conversion at the facility level, establishing a native DC data center. In this approach, medium-voltage AC is directly converted to 800 VDC by large, high-capacity power conversion systems. This 800 VDC is then distributed throughout the data hall to the compute racks. Architecture streamlines the power train by eliminating layers of AC switchgear, transformers, and PDUs. It maximizes white space for revenue-generating compute, simplifies the overall system, and provides a clean, high-voltage DC backbone for direct integration of facility-level energy storage.

Here is the blueprint of Nvidia’s 800VDC.

±400VDC is the safer of the two, the one we know how to install, already in use in a few rare data centers - one of them Google’s. 800VDC relies on hardware that’s still being developed and tested today; it isn’t a ready-to-deploy solution. Everything has to be built from scratch, just like for AI hardware, which is exactly why the opportunity is so large. We’re living through a moment where entire hardware architectures have to be created to meet the demand of a brand-new technology. From scratch.

And you can’t migrate from one to the other; the two infrastructures are completely different by design, so data centers have to be built with that in mind from the start. Today’s choices have to account for future compute: even if the real need for 800VDC isn’t for right now, buildouts will be designed around it to anticipate. It’s the only architecture that can efficiently power multi-MW racks, and those are coming.

Timing’s the main risk behind these investments, but also the opportunity. The market knows and anticipates the move to 800VDC, but has no confirmations yet. Most expectations point to 2027 at the earliest and we could easily have any financing issues, delays, or changes in design by then. But the buildouts will happen and under the right conditions some of these names will deliver multiples over the next few years.

Those two architecture are the future, a necessity to improve efficiency and reduce costs. They aren’t a maybe, but a when.

Hardware

Solid-State Transformers (SSTs)

SSTs are the central piece of hardware managing the AC/DC transition, replacing the many components that handled it before. Connected to the grid, BESS, or BTMs, software-optimized to deliver or convert AC/DC depending on the need.

They are the central black box, filled with hardware doing the heavy lifting.

Power Semiconductors - SiC & GaN

SiC and GaN are wide-bandgap materials replacing traditional silicon to manufacture power semiconductors - transistors, as silicon is inefficient at these voltages. It’s still sufficient for ±400VDC, though even those mostly use SiC already. These materials are used to manufacture MOSFETs and HEMTs.

Both transistors control voltage, raising or lowering it wherever needed. MOSFETs, with a much higher voltage tolerance, usually handle extremely high voltage at the dangerous points of the circuit, while HEMTs handle the lower voltage with extreme precision, like the final delivery to the GPUs. They’re the maestros of the circuit, always delivering exactly what’s needed, not more, not less.

DC-DC Conversion - Bus & Point-of-Load

The “tubes” inside the data center that manage the power, integrating MOSFETs and HEMTs. A single GPU doesn’t run on 800V, so voltage and current have to be fine-tuned for each component, just like irrigating a field takes tons of water, but you don’t dump it all on one crop, you spread it across the whole field. SiC and GaN manage the voltage; Bus and PoL are the converters that put them to work.

Bus is the larger step-down, taking the power upstream and bringing it close to the racks, while PoL is the surgeon, delivering the last centimeters with extreme precision to the hardware.

Busbars, DC Breakers & Connectors

Safety. As I mentioned, working at 800VDC is like nothing we’ve dealt with before, and safety is actually one of the reasons deployments are slowing down.

Busbars are rigid bars, made of copper and/or aluminum, that conduct the power, replacing cables. In 800VDC those busbars get smaller and lighter while needing better insulation for safety and liquid-cooling - since heat is still present.

DC Breakers are safety switches, meant to shut down all or part of the infrastructure when needed, and 800VDC changes how they work. In AC, voltage drops to zero many times a second, so cutting it mechanically is “easy.” But at 800VDC the voltage is constant and extreme, and cutting it would generate heat in the form of plasma - you really don’t want that. So instead of a mechanical switch, 800VDC uses Solid-State Circuit Breakers, SiC semiconductors that can freeze the current in microseconds.

Connectors are what they are, the pieces linking the bus/PoL to the racks. And once again, at 800V, you want safer connectors than the ones used at 48VDC.

Power Supply Units & Racks

The final piece of the puzzle: the famous racks, where the hardware gets plugged in and interconnected. Same story, at 800V you need a safer architecture and power supply units inside the racks. Those can also be optimized as a lot of hardware isn’t necessary anymore - notably AC/DC converters and components now handled higher up the chain.

Heat management

Every unit of power pushed into a rack comes out as heat, and airflow isn’t enough to cool the hardware. This can’t be avoided. It’s the same constraint thousands of people are working on inside hardware: how to remove the most heat in the least space. Liquid cooling is the most efficient option, since it moves far more heat than air for the same volume, and it can run right inside the racks - which is why they usually are designed together.

It comes with a straightforward but specific plumbing. Coolant Distribution Units pump liquid to the cold plates inside the racks - a simple metal plate sitting on top of the GPUs (or any hardware), running a closed loop of liquid that pulls the heat out, gets cooled down, and comes back for another turn. The difficulty is making cold plates thin enough but completely leak-proof; a single leak could destroy an entire rack - so these are safety-critical parts.

There’s also research on immersion, where the hardware is dipped directly into a non-conductive liquid. Higher performance, but not quite there yet - and not directly tied to the electrical infrastructure either way.

Volatility Management

Probably the most important vertical, and the one everyone seems to forget. A GPU cluster doesn’t draw power smoothly, it jumps from 10% to 90% in a blink and drops back just as fast; an electrical nightmare. The power chain likes it steady and stable, not volatile, spikes that could pull energy meant for somewhere else, creating lags or overloading some equipment.

The fix, shown in Nvidia’s 800VDC blueprints, is to place a buffer close to the racks in the form of batteries the clusters can draw from. It sits close to the source with no impact on the global power chain: the battery charges itself from a stable input and releases the spikes the cluster needs, making the volatility invisible to the rest of the chain.

The grid only ever sees a smooth GPU load, while the other side of the battery is a constant storm. This hardware is usually called BESS, and it comes in two parts: the battery, and the software optimizing the spikes and the recharge. Storage used to be optional, now it’s shifting from its original purpose into an optimization role. It’s not there to deliver missing power; it’s there to manage the chain.

And this whole system is built into the 800VDC architecture.

In Brief

This is the reasoning and the architecture behind the new 800VDC infrastructures that will become the new normal for AI data centers.

Data centers connect to the grid/any behind-the-meter source: batteries, generators, or renewable production through their Solid-State Transformers, the single point of AC/DC conversion, which increases efficiency. From there, MOSFETs and HEMTs inside the Bus converters carry the power into the data center, and Point-of-Load deliver it to the final hardware, inside racks interconnected by busbars and connectors, and kept safe by a Solid-State Circuit Breaker “kill switch”. The whole architecture is liquid-cooled, and volatility spikes are absorbed by BESS, the buffer that lets the rest of the circuit stay stable.

Those are the most important and impactful changes made to today’s data centers to power the next generation of GPUs and compute hardware in the safest and most efficient way.

The next part of this write-up will cover the opportunities tied to this new architecture, and how to capitalize on them over the next few years: from the stable giants to the more speculative pure-plays, and how to play them short, medium, and long term. We know the basics, now here’s how we turn them into profits.

Read more

But it had to happen. We expected it, I wrote about it less than a month ago. The index and many individual stocks were extended, we had made too much money too fast, and eventually most had to take profits and trigger a real selloff.

I felt like writing this weekend. First because I need to, second because I believe most of you would like to read some words. My personal portfolio is down 28% this week, which is a lot. I am guilty of the same mistake, always: overtrading. I always take too many side positions when the market gets excited and end up with some poorly timed entries. When that’s done on margin, days like these hurt. A lot.

I’m not here to complain, but to learn; that comes through writing. This Substack isn’t like others. I am not perfect, I am not here to tell you everything I do is perfect or that everything is great, that I have all the answers. I am here to share my journey toward being successful in the markets, so we can all learn, reflect on our mistakes, improve our systems, and succeed.

Losing so much money is also, somehow, a blessing. It means I was where I had to be over the last months, as most of the drawdown came from last month’s winners. I should have managed my liquidity better these past weeks but I was down a lot only because I was up even more. There are improvements to be made, yes, but also successes: the right stock picks, the right sizing. Important not to minimize those, as they are massive improvements compared to a year ago.

My portfolio is up 110% YTD. The alert portfolio is up 54.84% and the core portfolio 93.92%. Important to keep perspective, this isn’t a bad year, just a bad week, and we still have months to go. As my portfolio compounded significantly - up ~300% over the last two years excluding deposits, the numbers are starting to feel real in a different way. Losing ~$10,000 a few years ago felt manageable. When a weekly loss is counted in yearly salaries… it hits differently.

Never waste a good crisis. Churchill was right.

The lessons are manifold and I’ll work hard to apply them. I know exactly what I have to do, now I have to engrave it in my brain, build habits that stop the impulse to always take that one extra trade. Improve.

I do not have the perfect recipe yet. I only have the blueprint.
We’re building the rest together.

In this write-up, I’ll zoom out, justify this price action, think through what comes next, review key fundamentals and news, and conclude with my plan for the days ahead.

Markets, Selling & Fundamentals

Now, let’s focus on data. What happened?

My only answer is: nothing. Every asset class sold off, every sector sold off. Crypto is down crazy, gold and silver lost key supports, leaders saw double-digit drawdowns. Global selloffs usually happen for two reasons:

1. Investors flee to safe assets out of fear, but that requires a reason, which I don't see, and would trigger inflows on safe assets. Gold was down 5% in a week.

2. Classic liquidity cascade; most participants decided enough was enough, took profits, and triggered margin calls, stop losses, and every mechanism that comes with peak FOMO and retail optimism.

On the negative, the Strait of Hormuz remains closed, some strikes were apparently exchanged, negotiations stalled, and we are waiting on Warsh’s first rate decision mid-June. I’ve flagged this as a potential negative catalyst, though the market had already priced in no changes which should be what will happen.

So beyond wide profit taking after months of up-only and an extended market, I struggle to see any reason for our actual long-term uptrends to stop here.

So, what’s next?

Market-wise, nothing is certain. The S&P and Nasdaq were both extended, needed to breathe, and are breathing. Both broke their D21, their short-term uptrend indicator on higher volume, meaning lots of organic selling with few buyers stepping in.

Can you blame profit taking after 34.75% return in 45 sessions, and some patience before buying back in? I can’t.

The problem is that when buyers don’t step in to hold a key level, it means more sellers than buyers exist at that price. If nothing changes, Monday’s dynamic shouldn’t be different. If I didn’t buy potatoes today, I won’t buy them Monday if nothing changes. That said, the last two years showed markets can change their minds overnight and without any reason. Buy the dip has been the mantra for a long time. But statistically, more short-term pain is the more likely outcome.

The healthy targets are both W21, 7% lower on the Nasdaq and 4.3% lower on the S&P. And honestly, that’s what you want. Portfolios shrink more, yes, but liquidity flushes, overleveraged players learn a lesson, and healthy capital can re-enter at better prices to reinforce a longer, healthier uptrend.

Fundamentally, I believe the verticals I follow are stronger than they were a week ago.

Earnings this week from Credo, HP and Broadcom told the same story, again and again. AI demand isn’t just for off-the-shelf products, it’s for customized, optimized hardware that let hyperscalers deliver tailored compute to their clients.

Demand for XPUs and networking is simply insatiable. During the quarter, bookings for AI semiconductors were over $30 billion against the $10.8 billion we shipped.

Our visibility runs all the way to 2028 right now. Three months ago, I can tell you our visibility run pretty much 2027. Today, it runs to 2028. We're talking about capacity as measured by gigawatt power that are way ahead of what we have expected, say, six months ago.

Broadcom Earning Call

There's a reason when you've seen ASPs increase as much as they have over the last year, that customers are still saying, 'Hey, we want to buy. We're going to start new programs. We're willing to pay the higher prices because we clearly see that there's an ROI.'

When we talk to our customers, they're clearly at the early days of their Agentic AI journey. They see that they want their data to be more on-prem. They want to do more of the compute on-prem. We're seeing demand at the higher end of our platforms, so more memory, more compute power.

HPE Earning Call

Shocker.

On cybersecurity, both Palo Alto and CrowdStrike beat expectations and confirmed AI as a tailwind. Rubrik also impressed at its scale. This entire earnings season delivered one message: inference requires more hardware, more optimization, and AI is finally showing up in SaaS bookings and earnings.

I struggle to see how that leads anywhere but one direction: more.

Until now, infrastructure builders spent their cash flow, cash reserves and took debt, step by step. The only logical conclusion from those steps taken through two years is that they found ROI. If one were spending without reason, others would have held back. None did.

This week they took one more step: dilution. Google announced an $84.75B equity program. Meta announced plans for tens of billions.

I need to take a minute to explain the magnitude of this.

For decades, the Mag7 were the best investment; they were cash cows returning profits to shareholders through buybacks and dividends. Hundreds of billions went back to them, excluding price appreciation. They made so much cash they had nothing better to do with it than give back, please investors.

Google’s equity program tells you management now believes raising liquidity by any means necessary for AI buildouts is more important than decades of shareholder coddling. I’m not sure the market will like this, but it’ll have to make its peace with this decision. This is the new paradigm, join it or don’t.

This is a very powerful signal.

And it wasn’t the only one. Berkshire Hathaway participated for $10B to this capital raise. Buffett’s company, even if he’s no longer making the calls. The symbol matters, and the fact that they have ~$380B left in cash, meaning this could be a first injection. This is the first major institution, external to the AI ecosystem, injecting capital into AI buildouts. Until now, hyperscalers relied on VCs, cash flow, debt. Today, an outside institution is funding AI, and could open the door for many others.

Like… This one.

Today, AI is seen as a tool to increase profits. Eventually it will be seen as critical infrastructure for energy efficiency, medical research, public systems optimization, and maybe more... There is a case where states participate in buildouts or support through other levers as the technology becomes a national interest. Trump's administration already took stakes in companies like Intel, I see no reasons for this trend to stop.

Lastly, we learned Google contracted SpaceX for $920M/month of compute over ~3 years, guaranteed for one. SpaceX is the only company left with available compute following its migration of Grok to Colossus II, freeing Colossus I.

I’ll conclude this fundamental parenthesis here.

Looking at my portfolio is rough, but writing this section is exciting. Every company involved in AI is telling you they’ll spend more or make more - or both, and what everyone called overspending last year will turn out to be the beginning of even more spending. Every analyst and management forecast 2027 CapEx increases each passing month. Google opened Pandora’s box with dilution. It is very hard now not to expect Microsoft, Amazon and others to follow, while external participants are also helping to fund those buildouts…

This is the state of AI demand. Even while stocks go down.

On the premium section, I will go over the opportunities and liquidity attribution as the market is shifting and it is important to follow, identify the best narratives/assets, as you understood my overall assessment of the situation.

Yes, stocks are falling. But it looks like only so they can bounce back harder.

Read more

But a sauce alone leaves you starving. It’s only worth something served with the right ingredients, and cooked to perfection. Serve it with rotten ones and you’ll ruin the dish anyway.

In investing, the secret sauce is the system, and a system is only as good as the person following it. Take away the ingredients - discipline, commitment, research, work… and you’re left with an empty plate filled with some weird liquid.

This write-up covers my system - how I built it and why, and everything I think matters in stock picking, from how the market actually works to the act of picking itself and taking profits. I’m still years from calling myself successful, but that’s only a matter of time, and it will be because of that constantly improving system.

Here’s what you can expect.

1. Stock picking concepts

2. The importance of systems

3. Markets and price action

4. Sizing and dilution

5. My personal system

6. Trimming and selling

Stock Picking Concepts

Notice I call my activity stock picking. Not investing, trading or speculating.

I want to get away from those words; social media has turned them into labels with very personal definitions. I don’t care about the name. I care about finding the best returns. Call me a trader or a speculator, that’s fine, as long as performance follows.

This is what stock picking is to me. I’m not buying the companies I love or the ones that “look cheap”. I follow companies for their fundamentals and narratives, but I buy the stocks I believe have the best chance to deliver the most returns.

This has nothing to do with timeframes or methods. If I believe a stock has what it takes, I’ll hold it forever - I’ve held Palantir for a year and a half with great pleasure. Focusing on returns doesn’t mean shortening timeframes, it means leaving personal bias behind.

Returns are what we’re here for at the end of the day.

The Importance of Systems

Systems matter because investing on opinion alone doesn’t work. It might, for a while, with some luck and a healthy market, but it fails over decades as you will eventually be wrong big time. Every super investor has a system; it’d be foolish not to have ours.

Druckenmiller sizes up aggressively on high risk-reward after a deep review, famous - with Soros, for “buy first, study later”. Buffett and Munger buy value below their assumed intrinsic value and popularized “moats” - which is misused 99% of the time nowadays. Lynch focuses on growth, categorizes stocks, and ignores volatility.

These are systems. Different from one another, but all successful because each fits the investor’s personality/objectives. Look deeper and you’d find checklists and methods running through every step - picks, sizing, selling.

If you don’t have a system, you’re probably doing it wrong, or haven’t been at it long. Either way, that should change, or you’ll make mistakes. I made mine this year with Nebius - a costly mistake that happened only because I ignored my system. I shared more over the months, because diagnosing them is how you improve.

A system shields you from yourself.

Take the last few years. A tariff war, real wars, oil supply disruptions, covid, rate hikes, inflation, and countless other events. If you tell me none of it touched you emotionally or made you question your portfolio, I’d say you’re lying. And if I asked how many of you rushed a pick, to buy or sell, I’m sure the guilty rate would be 100%, and I am also sure most of those transactions would be considered mistakes today.

Systems prevent that. Follow one, run the checklist, and you can tell yourself.
No. This shouldn’t be done, so I won’t do it.

Markets & Price Action

Many invest, but rare are those who stop to ask what the markets actually are and how they work. So we’ll start there.

The markets are a place where individuals trade certificates of ownership, a tiny slice of any publicly traded company. That, everyone knows. Where it gets trickier is what the market actually looks at.

A stock’s price comes down to two variables.

1. A financial metric, which reflects the company’s ability to generate cash. That’s what the market craves: safe, consistent cash generation. It’s factual, reported every quarter during earnings season.

2. A multiple, the market’s trust in the company’s future cash generation. Volatile, but usually stays within known ranges for each sector.

Most investors look at earnings and P/E, but plenty of others work. A company’s value is those two multiplied together - the market’s verdict on what the business is worth.

But that value isn’t today’s, it’s tomorrow’s. A company reports its current results; the market prices its future. If Apple delivers the most amazing quarter but warns sales will fall 10% this year, the stock drops, it doesn’t rise. Apple is a great company today, but its short-term future is in danger; that’s what the market prices.

Multiples move according to the market’s confidence in future cash generation. Lots of potential or stability? High multiple. Risky or unstable? Lower one. That’s how and why stocks move. The market looks at three inputs to decide on multiples, in order.

1. Global liquidity; is there enough money around to buy more stocks and is the world in a stable enough situation.

2. Narratives; what happens tomorrow for this company and its business.

3. Fundamentals; how trustworthy is this company and how strong is its business.

Every company is different. You can’t judge a hyper-growth tech firm the way you judge a utility. Right metric for the right business, the source of disagreements, but the principle holds for every stock in the market.

From here, a company has two levers to increase its stock’s value.

1. Increase cash generation

2. Expand its multiple

The first is straightforward: a healthy company grows, improves margins, prints more cash. If the multiple holds - stable narrative and liquidity, the stock goes up. Simple.

The second is where the magic happens. Less science, more sentiment. But sentiment follows rules too, with specific triggers.

• Accelerated Growth: when revenue starts growing faster, the ceiling disappears. The market prices in optimism through higher multiples, because it doesn't know where the limit is. Better to anticipate a huge success than miss out on it. Humans are optimistic by nature; this is growth investing 101.

• Margin Expansion: more cash from the same revenue, and the market rewards that efficiency with a higher multiple.

• Increased Safety/Trust: a company that can generate cash safely for decades earns a "trust premium," because the market can forecast its future.

That’s how the market works, and how to avoid value traps - companies that look cheap but have no path to expand margins, accelerate revenue, or earn the market’s trust. 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 and cash generation, creating a double appreciation: rising cash and expanding multiples.

That’s what stock picking is about.

Price Action

The most underrated data source there is, ignored by investors who think their own opinion is enough. It isn’t. Have you ever asked yourself what a chart actually is?

Take any data set you can think of, they are always represented visually, in diagrams or charts, to read them more easily and spot trends. The lines are the data. Price action is the same thing; the data behind it is every transaction happening each second.

The market is a consensus built from millions of interactions, and price action is that consensus made visible. A chart tells you who wants what and at what price. It shows which group is in control. That matters for one reason: you only make money when your stock goes up, so that’s where you want to be.

I focus on stocks going up, or stabilizing, because I'm here to make money, not to hold a bias and hope to be proven right.

Ranges, Breakouts, Retests & Averages

Those are the four concepts you need to read price action, the only ones. It’s far less complicated than most make it look. I rely on them for two reasons.

• Human behavior: price action literally shows what investors are doing. Price up, buyers are overwhelming sellers. Price down, the reverse.

• Self‑fulfilling prophecies: these patterns have been watched for centuries. People expect certain reactions at certain levels, act accordingly, and the pattern reinforces itself.

Ranges are the first sign selling pressure has exhausted itself. The stock stops falling because buyers hold the line, supply and demand finally match. The bleeding has stopped, but it doesn’t mean the uptrend is starting.

Ranges can last weeks or years. The market rewards cash generation; until a company shows clear acceleration, it has no reason to pay up. So the range drags on, that’s how value traps form.

Then come breakouts; where it gets interesting. Buyers overwhelm sellers and agree to pay more. Sellers have vanished. The chart won't tell you why, but it tells you the narrative is changing. The stock starts printing higher highs and higher lows; buyers in control, paying up for the same asset.

Here’s an illustration.

Lots going on, so let's break it down.

• Sector 1 is pure disinterest. The stock keeps falling, nobody wants to buy higher. Buying here wastes time and liquidity.

• Sector 2 is seller exhaustion: a higher high in August after two near-identical lows, then a third barely lower in October. Sellers are tired, buyers show up at the same price point for months in a row.

• Sector 3 is the first positive sign; the stock doesn't retest its late-2022 low. Buyers stepped in early, impatient, pushing price higher in January. They're clearly willing to pay more. The stock is now attractive, not enough for a large position, but worth interest and a test buy.

• Sector 4 seals it, especially landing after earnings. The chart won't tell you that but it shows the market is ready to pay up: buying higher at the lows, pushing higher at the highs. We know the market wants more of it.

Aggressive accumulation happens on retests and moving averages, levels everyone watches, the self-fulfilling prophecies that keep the trend alive. The pattern is always the same: ranges from hesitation, breakouts, higher highs/lows, retests as the market revisits the breakout.

Retests are simple human behavior. A stock rips to new highs, investors take profits, sellers briefly outnumber buyers, price cools back toward the breakout. If the trend is strong and fundamentals healthy, buyers step right back in as the price is back to a previous consensus everyone wanted to buy a few weeks earlier.

Who passes on that?

Every higher high after the breakout above is marked with a green line, and every push came back to or close to it. As long as buyers step in at those levels, demand holds. That’s when we buy, until demand fades.

In On Running’s case, that run took 88 weeks and returned 151%, from the first retest in June 2023 to the top in January 2025. Buy more aggressively at Sector 3 and it’s 98 weeks for 222%. When I say I have no problem holding for years, I mean it.

The second prophecy is moving averages - smoothed-out price action. Three are watched religiously: the 21, 50, and 200 (red, green, grey in my charts). Again, they matter only because everyone looks at them.

Take TransMedics. For six months, buyers stepped in every single time on the W50, printing a series of clean higher lows. Coincidence?

There are no coincidences in a place where millions interact for hours a day. There’s only information, and it’s our job to read it. In TransMedics’ case it ended badly, the stock broke down. Not an issue: when it did, I closed the position. Not because I stopped believing in the company, but because the market was telling me buyers were not stepping in, so the stock was no longer the best vehicle for returns. I left.

For the better, it fell more than 40% after.

Stock picking isn’t about feeling. It’s about being where returns are made.

Every stock has a “primary pattern”, the one most participants watch. For some it’s breakouts, for others moving averages. The job is to spot it and use it to be aggressive at the right time, which has nothing to do with timing the market - which means trying to buy the bottom or sell the top. This is impossible to do and I do not even try to. What I try to do is to catch the middle portion of a run, when the market gave enough confirmations for probabilities to be in my favor so I can go big, and enjoy the run until it shows some slowing down.

Money is made when stocks go up. Everything else is noise.

Why don’t I buy downtrends?

A downtrend is the market’s way of telling you something’s wrong. Sometimes it’s an overreaction. More often, the market sees something you don’t.

Buy one while being wrong and you’re holding degrading fundamentals. Markets don’t care about “cheap”; this is a personal concept which means nothing. A stock can fall far below any analyst target on nothing but lack of interest.

I’ve shown that averaging down is often a waste of time, energy, and sanity, with a concrete example. Watching your position shrink doesn’t help you sleep, and patience gets you the same average cost without the stress.

Buying a downtrend is pretentious. You’re assuming you know better than the market’s consensus - you won’t 99% of the time. Even if you were right, nothing says the market will agree any time soon. Could be months. Could be years. Meanwhile your position keeps shrinking, the risk keeps growing, and you underperform the market while sleeping badly.

Following price action reduces risk. Let the market tell you you’re right.

Position Sizing

We call ourselves stock pickers, but picking isn’t the important part. Sizing is.

Concentration is key. I hold the fewest positions possible; always under ten, ideally under five. I can do so because my system is extremely selective. I have a crystal-clear selection pattern: why the stock should appreciate, at what price I would buy it, why it would be better than another... Not many stocks meet such criteria at the same time.

I’m not always right, and things rarely play out exactly as expected, but I don’t need them to. I only need to trust my opinions and then follow my system to the letter, buying large positions when my conditions are met, and closing one when my thesis or my system breaks - as I did with Duolingo and Hims.

I like to say we aren’t here to play Pokémon; holding 10 or 20 names is a waste. The goal is the best ones, not the most.

Imagine a $100k portfolio, ten positions of $10k each. One returns 100% in a year, your portfolio is up 10%. You’d be better off putting 20% into a “safer” name expected to return 50%, and pulling that same 10% with far less risk.

Many believe a small moonshot will make them a fortune. It won’t. The game isn’t to find a 10-bagger, it’s to be heavy in one. If I expect a stock to return 100%, I go big. If I’m scared to go big, my conviction isn’t strong enough, maybe I shouldn’t own it at all, and should grow another position with less upside but more size.

You won’t always pick the #1, but you’re still better off with a 30% position in the #3 top performer than a 5% position in the best.

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

Diversification & Dilution

Most investors think diversification means owning different stocks. Wrong. It means owning assets that aren’t driven by the same narratives, uncorrelated names that balance each other’s swings.

Own both IREN and Nebius, or ASML and TSMC, and you aren’t diversified, you’re diluted. These pairs move together, so it’s poor allocation. Put your capital on the best risk reward, then find real diversification elsewhere. Hold a healthcare stock alongside AI compute, energy, and AI hardware and you’re mostly there; all stocks share some baseline correlation through liquidity, but these won’t move in the way the pairs above do, because the narratives differ.

A handful of positions is easier as there aren’t that many great opportunities per sector at the same time. We tend to scan many names inside one sector and have the urge to hold them all, but they’re correlated. That’s like collecting only fire-type Pokémon. Go get that water badge…

When you reach the league, your team is six Pokémon, different types, the best of each, because that’s what wins. Best in each category, with a plan for how to use them, when, and which to sacrifice. Everything planned to win.

Clear selection. Clear strategy. Clear focus.
Stock picking is like fighting the League.

My Personal System

In a lot of situations you know something could happen, but you have no idea if it will. That’s where a system matters, it should drive your actions; not your read on the world or geopolitics, which is probably as flawed as mine; nothing wrong with that, these are genuinely complex subjects.

A system is personal. Traders sell once their targets are hit. Some investors never sell. Some sell on fundamental changes, others on technicals. There are no rules for what a system should look like, but everyone needs one. To buy, hold, and sell.

I run two types of position, treated differently:

1. Core positions, bought at a healthy price after market’s confirmations.

2. Swings, shorter-term, bought to extract a pre-defined performance.

My system is straightforward and goes for both types of positions.

A clear selection process, through price action screeners.

• Strong fundamentals and narratives only - sector selection is key.

• Within an uptrend - above the weekly 50, higher highs and lows.

• Reasonable valuation based on my assumptions (for core positions).

A clear buying process.

• Breakouts on volume.

• Breakout retests.

• Weekly 50.

A clear selling and trimming process.

• Trim on extensions - RSI heating up + extension from key averages.

• Sell on three conditions only:

  • the thesis is broken,

  • the stock falls below identified support / W50,

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

That’s me. It doesn’t have to be you. What matters is having clear conditions for every transaction made in the markets. None should be based on personal bias.

Trimming and selling

By now, you guys know everything I know about the markets, price action and my system. The only thing missing is trimming and selling, for which I will use some concrete examples for both types of position.

Swings

My swings always come with a crystal-clear plan. I already know the narrative and the fundamentals, I buy on a key support, and I know where I want to sell. I usually run them on margin, so my focus is risk reward, always minimizing the first.

I don’t care if a name ran 2,158% and trades at 181x sales, if I can risk 1% of my net worth for a high-probability 10% on margin, it’s a good deal. I’m not building a position for the next decade, just extracting a defined performance for a defined risk.

Take my AXTI swing. I bought it as I follow photonics and know the fundamentals and narrative behind each of its leaders - AXTI being one of them. I wouldn’t swing a name I don’t know.

Since its bull trend started, volume constantly accelerated and price never lost its daily 50 - green line, which makes it a logical entry. Statistically, buyers and algos should step in there, and they did.

Entry set, position on margin - which carries more risk than a cash-secured one, so a stop loss follows: 5% flat below entry. And since I didn’t want to overdo it, I set a take profit just below the all-time high, at $69.

That’s a swing: a clear entry on narrative, fundamentals, trend, and support with a non-negligible probability of success; a tight stop because capital preservation is the #1 rule on margin; and a profit target to keep greed out. Yes, sometimes - like here, the name breaks out right after you sold, leaving you behind.

But this trade, bought on margin (cash I didn’t have in the account), returned 64% in six sessions while risking 5%. Better to cut a margin position after a great trade than let it run, fall into greed or FOMO, and watch it reverse and burn you. Margin is dangerous - use it only with a clear plan.

Selling is part of the plan, not an adaptation to price action. Every entry comes with an exit, and there’s no coming back. The trade works or it doesn’t. No in-between.

Core Positions

I treat core positions differently, because of how they’re bought: healthy valuation at the start of an uptrend. Their run should span months at least, through plenty of ups and downs which, as long as key levels hold, I’ll gladly stomach.

Selling happens for three reasons:

1. My thesis is broken; what happened with Hims. The reason I owned it stopped being true, so I let it go and moved on.

2. The uptrend breaks; even if my thesis holds, the market isn't buying it anymore. Usually a sign of weakness, for any number of reasons, from the market's stupidity to mine (maybe I missed something). I'd rather not take the risk; I always respect the market.

3. I find a better risk reward elsewhere.

As long as my thesis/key levels hold, I have no reason to sell, even if the stock looks “expensive.” Like “cheap,” this word means nothing. Palantir was called “expensive” past $40, ran to $207, and broke its W50 at $140.

What were the reasons to sell earlier? System-wise, none. Psychologically? The stock had run hard, traded at 100x sales and a dumber P/E, social media was screaming sell. Yet the company kept accelerating revenue and margins, and the multiples followed. Those who trimmed at $40, me included, felt pretty stupid missing a fortune-maker.

Selling shouldn’t come from bias, whether external conditions, pressure, or the urge to lock gains because “it ran too much.” Trimming is fine, but it also needs rules. I watch three indicators:

1. First, my extension alert, those small red flags on the chart, a homemade signal built on a few conditions. Not perfect (none are), but a good read on when things get hot.

2. Second, volume. A great stock goes up on high volume, or explosively on low volume. Either a lot of buyers are paying up, or there are no sellers at this price. Both mean the will to hold beats the will to sell.

3. Lastly, momentum, the pace of the move; it gauges the strength of the trend. Over the months a stock naturally draws less interest and momentum slows. If price keeps rising on falling momentum, it's quietly losing its aggressive buyers, and will stabilize or decline.

Back to Palantir. Ignore the first red flag on the far left, it was triggered during the breakout which naturally comes with an extension. The ones in the middle fired on growing volume and accelerating momentum: an extension fueled by real demand. The fifth fired on a weaker momentum spike but still-healthy volume.

There’s a case to trim on those flags even while volume and momentum are healthy. My buying alerts triggered ~$10 so a first trim after an 8x on a clear extension is reasonable I’d say.

It gets interesting at the ~$180 flag in August 2025. Up ~70% from the prior flag, momentum no longer following - the move is slowing, and volume steadily falling. This time all three indicators agree. From here the odds of big further returns are statistically low; it might be time to rotate. Trim or sell is each person’s call, but the system is clear: something must be done.

Across the whole run, Palantir gave two system-driven windows to trim, then the signal to close when it broke its W50 around $140. Sitting on your hands while the position compounds is the hardest thing to do in this market, and the most important, so you don’t cut compounding early and for no reason.

Following the system, the position started between $10 and $15 (breakout or retest), trimmed around $90 and ~$185, and closed ~$140.

How many who bought in the ~$15s actually pulled those returns? Almost none. Most sold out of fear, pressure, the “too expensive” or “unsustainable” narrative, or some other psychological reason. The exercise is easier after the fact, sure. But we have to learn after the facts, this is how we build systems, with knowledge.

Two things keep great stock pickers from life-changing money: they don't size, and/or don't let winners run. It’s also why I run concentrated positions and cut losers fast, freeing liquidity for compounders. There aren’t many Palantirs out there; owning one with size and holding it is enough to change your life. Which is exactly how I should have behaved, instead of trimming at $40 on personal bias.

We learn from our mistakes.

The Taxes Situation

This depends on where you live, but at the end of the day we’re all taxed. In Europe, most of us pay a flat tax on profits, triggered when we sell. That belongs in the selling process.

If you want to trim a core position with no fundamental change - one you’d buy back lower, do the math for your tax system first. The bigger the gain, the bigger the drawdown you’d need before buying back for the trim to be worth it. Here’s a small table assuming a 30% flat tax.

If the pullback you expect isn’t deep enough to buy back meaningfully more shares post-tax, you’re better off just holding.

Concrete Examples

Conclusion

One of my latest write-ups goes through concrete examples, from identifying assets with price action, volume and narrative to buying it and trimming/closing. It is the logical next and final read to understand my system and thinking from A to Z, so I will redirect you to it directly instead of copying it all here.

That’s everything. To close, once more: systems are personal, and only as good as your ability to follow them. I’ve been guilty of breaking mine too many times, and it has always cost me. Over the long run a good system will make you miss opportunities, but it’ll keep you out of far more traps and mistakes, and that is what improves returns.

The hardest part is letting the frustration of a missed opportunity go. Letting the ego go when you’re sure you can do better, hold a little longer, because you just know it’ll work out. The truth is it doesn’t matter. The market is always full of opportunities, executing perfectly on a few matters more than catching one or two more.

Systems are personal. Find what works for you, and follow it.

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’ll start with what matters: Thank you. Over 100 of you subscribed in less than two weeks, we reached the Substack financial top 10 and Best Seller top 50 for several days this month. It all means a lot to me and I do not take your trust lightly.

This motivates me to do more - at least better. I am working on a few more services, including a homemade screener to improve my screening and potentially share it with premium subscribers so you can leverage the data I use by yourself. A few weeks from being ready.

To review the numbers since this Substack passed premium, my equal-weighted core portfolio is up 110.61% YTD and 196.38% since inception in early 2025. My personal portfolio, which only leverages the information I share here, is up 169.91% YTD.

And most importantly, my alerts equal-weighted portfolio is up 71.6% since launching in early April. Most Founding Members have already recouped their subscription two weeks after paying for it - probably even already recouped the next ones.

This is what I focus on every day. Not just returns, but a clear system to identify the right sectors, names, setups, and share the convictions to concentrate and hold through uptrends.

For those who aren’t aware or haven’t taken the step yet, know that the launch promotion runs until June 13th, with the Founding Member plan at $350/year. Past this date, prices will increase and no equivalent promotions will ever be proposed. Here is the future organization and pricing.

The New Structure:

• Free: Macro updates, quarterly reviews, the thesis portion of all write-ups. (The specific investment strategy, trade executions, swing plays, and chat are now reserved exclusively for paid tiers).

• Premium ($350/year): Full access to all write-ups and monthly plan, including the exact investment strategy and trade executions. (This tier does not include Substack chat, alerts, or future services).

• Founding Member ($500/year): The complete, unrestricted service. You get full access to everything in Premium plus my buy targets spreadsheet, Substack chat, real-time alerts, and future services in development.

If you feel like jumping in, the next two weeks are the time to do so. Now that this is out of the way, let’s review the month, and plan for the next one.

Last Month’s Plan

I shared three verticals worth focusing on last month.

1. AI Hardware, like Soitec, Global Foundries, Marvell, Silicom, AsteraLabs, ARM, Credo, MU, Sandisk or Nebius. All of them are up 30% minimum since. AI hardware was the golden source of returns, but that might change.

2. Space, Intuitive Machines and RocketLabs, up 80% and 77% respectively. This was the upcoming sector with amazing charts and rising narratives without much momentum yet.

3. Defensives with SolarEdge, Venture Global and Intrepid Potash, the first up 71% in a month while the two others are flat or slightly down.

Bull trends make you feel like a genius; it’s important to keep our feet grounded but also important to realize that this isn’t a bull run, but sectorial bull trends, and that identifying those trends is already a victory. This is the stock picker’s market I was expecting, and it will continue being for the months to come. In those conditions, liquidity concentrates, so do returns, and performance comes with sizing.

Our focus has to be on the few perfect set-ups only, so it is important to tune off the noise and focus. There is always another trade but it doesn’t matter as reducing size will lead to lower total net worth performance, while that is what we are aiming for.

Large returns only matter if they are on large positions.

Macro Review

It looked like another long and eventful month, but the macro situation did not change. The Strait of Hormuz closed for three months now. A deal is apparently awaiting Trump’s final approval but it’s hard to know what is exactly happening so I wouldn’t count on anything until we see ships crossing the strait for a few days without issues.

As shared in my previous reviews, this blockade leads to global inflation and it starts to show as both PPI and PCE came hot. We cannot deny anymore that inflation is back. The only question is whether it will be a rapid spike or a more sticky price increase.

No one has the answer but the market is not yet worried. I believe it expects the Strait to reopen without much real impact as most companies had stockpiles and made do until now without much impacts.

The Fed is supposed to share its decision on interest rates on June 17 and the market is expecting rates to stay put; oil has fallen significantly as traders price in a deal to reopen the Strait

So no, the market isn’t really worried. But we’ll need to keep an eye on it.

I’ll do a brief parenthesis on the “but my stocks aren’t impacted by inflation either way”. Sure. But that doesn’t matter because investors and funds are. The markets are about liquidity not fundamentals so if inflation spikes, investors will withdraw cash because they need it or would rather anticipate a drawdown than live through it. This is what drags stocks lower, not fundamental impacts. I am not saying this will be the case, but it’s worth keeping this in mind.

Earnings Review

If you’ve read them, you know what I’ll say now. If you haven’t, just know that all of them confirmed my expectations. My thesis for months was that AI would need optimization at every layers of the stack, from hardware to software all the way down to token consumption; that is exactly what is happening.

Worth noting that the market is learning to trust again as both Anthropic and OpenAI continue to raise funds, grow revenues, ARRs, with the first even talking about potential profits as soon as this year.

This new technology is creating many new winners generating massive returns which is why analysts call it a bubble. New technologies require everything to be built from scratch so as many new verticals emerge at the same time from the ground, it gives the impression of stupid speculation. It isn’t.

GPUs aren’t the heart of demand anymore. Hardware capable of extracting the maximum performance from those GPUs is. Everything is possible and everyone places their bets. Only a handful will succeed and become tomorrow’s hardware giants, but until the winners are clearly identified, the market will look at every possible alternative and anticipate its success.

What matters for us and any responsible investor is to be aware of the situation and responsible enough to know that a moment will come when we will need to get out to keep those hard-earned returns. We will have signs; our job will be to accept them and not let the bullish atmosphere cloud our judgment.

My Mistakes

I wanted to take some time to share some of my mistakes this month. I know that when writing a premium Substack, you are expected to be infallible and only talk about the wins. It’s bad for business to talk about mistakes. But I started this Substack as a personal journal meant to help me - and it has. Writing helps me accept, visualize and improve. If I cannot write about a company, I know I shouldn’t buy its stock. If I write about my mistake, I can diagnose and correct it.

So I will continue to write about them, because it helps me and because I am sure it can help you. I made a handful of big ones this year, many this month as the market got crazier, noisier and triggered some FOMO, something I thought I had under control… Let’s get through them.

1. Nebius. I’ve talked about this one already, as I sold the stock on a personal bias, not following my system. I should still be holding it today, I don’t.

2. Intuitive Machines and Rocket Lab. Two swings with controlled and safe setups I chose to reduce my size on at the lows, just before the pump we’ve seen, for no reason but discomfort with my margin while the market was shaky for a few days.

3. SolarEdge. This one was played out perfectly, but instead of going in heavy on the latest retest as the stock respected my system for months, I chose to hold my current position and used my liquidity on a swing trade which returned nothing.

4. Fluence. I believe in this name. But I still had FOMO and bought outside of my system. The position is healthy but I should have been more patient on my first purchase, closer to key averages.

Those are the mistakes. The diagnostic is wider; psychological. There is a clear pattern: I didn’t stick with my plans/system, because I was either uncomfortable with my portfolio or felt I could do better with my gut than my system. I was wrong.

• Trades should be held until successful or not, regardless of the market’s situation. I have a clear plan when I buy, with controlled risk, and only take great setups. As long as the plan is clear, it should be held.

• Core positions > Swings. If I have the opportunity of building a large core position at a great price, I should. Most of my yearly returns came from those, and most of my mistakes came from attributing liquidity to swings instead.

• Only perfect setups are worth buying. There are hundreds of opportunities at any point in time in the markets, we can’t catch them all, so it will always be better to build large position on the best risk reward setup.

This is what I’ll need to work on. Fewer transactions. Perfect setups only, great isn’t enough. Scaling sizes. Sticking to the plan from A to Z. There are no limits to what one can achieve in the markets, as long as one accepts to learn.

That’s what we’re doing, and will continue to do.

On the premium part of this write-up, I will go over my actual portfolio, latest transactions and views of where we are headed in term of sectorial rotation and which core/swing position I am looking at with concrete examples.

Read more

Soitec’s revenues are down 34% YoY. Most would say this is a dying business and its stock is not worth looking at, and yet… Well, they’re wrong. Earnings and financials look behind, they do not matter anymore as the markets price the future, not the past - which is why only guidance matters during earnings.

Soitec is up ~290% since I shared my thesis and made it my biggest position a few months ago, it is up ~20% more today while dropping a -34% YoY decline. And this quarter confirmed my thesis, that the future is bright and that the market should continue to push its stock.

Are Fundamentals Weak?

No. Soitec is at the heart of photonics. Optical semiconductors need SOI wafers and Soitec owns the most efficient technology to manufacture them cheaply and at scale. And while we think about data centers today, the truth is that SOI wafers will also be present in every hardware leveraging AI with a need for low latency in the future, which means robotics, EVs, 5G hardware and smartphone, and many more…

Many of the five billion AI device by 2030 will be always on and extremely sensitive to active and passive power consumption. We also expect AGI to drive new needs for connectivity with more data volume to be exchanged, lower latency, higher band, lower power consumption and higher integration.

So in conclusion, for both infrastructure and edge applications, we are targeting every layer of the AI stack with our expanding portfolio of engineer substrate.

This is a pretty large market until the end of this decade.

But yes, if you look at past revenues only, you will not consider Soitec. Rapid growth slowdown, negative margins, unprofitable - EBITDA profitable though with positive FCF. What’s to like here?…

Well, the narrative.

We have had all the confirmations we needed through this earnings season to have the quasi-certainty that photonics are the future and are about to become the most demanded piece of hardware in data centers whether within Nvidia’s GPUs or Arista’s XPOs. Any pluggable today and CPO tomorrow will include SOI wafers.

Customers come to us for prototypes and move from prototype to qualification faster and faster.

We expected around 30% CAGR but we are now expecting more, but we have to be humble in this industry, I will not give you a firm number.

[For photonics] we have new customers, in Singapore, starting prototypes from scratch. And we have more and more in the last months

This should add enough context to dig a bit deeper and you will find out some interesting things.

First, Soitec grew revenues 24% QoQ, exceeding their guidance of 20%. This is a lot, and doesn’t happen without tailwinds. Their photonics vertical was responsible for $100M of revenues FY26, which based on previous quarters indicates a healthy sequential acceleration - hence demand as the previous quote confirmed.

Photonics are still a new technology, at least applied to those use cases and are only starting to be tested now over the last few months, haven’t yet started to ramp up as confirmed by all hardware manufacturers those last weeks. Yes demand is rapidly growing but scaling such hardware requires time, and even more time for it to be represented in earnings deck.

This partly explains the “low Edge & Cloud AI YoY growth” as the technology needs to scale but there is more to it. Soitec stopped the production of its Imager products so the YoY comparison is unfair as including previous volume, discontinued today. Growth stands at a healthy 19% excluding this vertical.

This is the real signal about AI demand for SOI wafers.

This evolution reflects positive volume and product mix dynamics, notably in photonics-related activities, as well as the contribution of FD-SOI in Edge AI applications. Strong Photonics-SOI growth momentum continues, with revenue rising above $100 million in FY26, earlier than initially anticipated, marking an important step in Soitec’s expansion into high growth AI data center architectures.

Just like Global Foundries and just like we anticipated when I bought in Soitec months ago, the legacy verticals hurt its profile and will make the company looks like a dying one instead of what it is really, a transitioning one. Smart money might stop at this or not be able to invest yet as the profile doesn’t meet their requirement, but it doesn’t mean the narrative isn’t here.

Those are temporary setbacks until photonics become large enough to generate enough growth , or other verticals stabilize - which doesn’t seem to be the case. Management is dealing with this period as they should, with sane financial decisions to reduce spending, focus on its balance sheet and its production towards future core products: AI & Cloud photonics.

They confirmed that manufacturing capacity from those legacy verticals could be re-allocated towards other hardware, under more demand - understand photonics here, at low cost and short timeframes. In plain English: slow demand for other verticals means they are ready to use it to scale photonics more rapidly.

We can scale by reallocating fungible SOI capacity already installed without years of capacity ramp and qualification.

This decline is legacy demand might end up being a blessing.

So if you focused on financials without looking at the context, sure, this quarter looked shit. But when you add context, fundamentals and market trends, well… A shitty quarter becomes the confirmation of a bull case. It will take time to materialize, but the market will continue to anticipate it.

Because that’s what markets do, they anticipate.

For now, I’ll stay out, but let’s see.

Nvidia

Today’s report will be short but important as the king reported this week, with another banger, which shouldn’t be a surprise. I’m not sure who had bearish expectations to be honest, after everything we’ve heard these last weeks...

The most important data this quarter is what I’ve been highlighting for weeks now: as inference arrives, GPUs aren’t the heart of AI demand anymore, networking is - and everything that can help GPUs achieve their task without idle time.

Data Center compute revenue was a record $60.4 billion, up 77% from a year ago and up 18% sequentially. Data Center networking revenue was a record $14.8 billion, up 199% from a year ago and up 35% sequentially.

Data center networking revenue of $15B nearly tripled year over year. InfiniBand has had a very strong quarter growing more than 4x year over year driven by deployments of our next-generation XDR technology. Spectrum-X, our end-to-end Ethernet platform purpose-built for AI, is now larger than all Ethernet network peers combined."

That was the thesis on AsteraLabs a months ago and it played out wonderfully as the stock is now ripping at new highs, and is the actual reason for buying Arista last week on the W50 retest, as shared in the Substack chat. Those two companies are at the heart of the networking demand and leaders in their products, so as Nvidia confirms the direction of hyperscalers’ spending, it also confirms my thesis on both.

The bottom line is always the same but I’ll repeat it: inference requires more than raw compute, it requires a complete optimization and that comes from other constructors; this is the narrative behind photonics, Silicom, and so many others.

This earnings season proved this. Nvidia confirmed it. And analysts don't see it stopping any time soon.

With analysts now forecasting hyperscale CapEx to exceed $1 trillion in 2027 and agentic AI beginning to proliferate all industries, AI infrastructure spending is on track to reach $3 trillion-$4 trillion annually by the end of this decade.

I have covered everything you needed to be convinced on this narrative which remains the strongest today and you guys now my view on the subject: extremely bullish, as long as the market loves the sector.

Small note on SaaS: as agentic AI scales, their products also will. I’ve been patient enough for long but any switch in sentiment on those stocks will have to be bought because this is where the next round of performance will be - this and robotics.

Since the advent of ChatGPT, we have witnessed mainstream AI transition from one-shot inference to reasoning and to now agentic. AI is no longer a nice to have. AI is now a necessity for enhancing productivity across all industries and roles.

Tokens are now profitable, so model makers are in a race to produce more.

Sigma Lithium

Sigma reported earnings late last week; the thesis continues to play out. The most important information being the repayment of 21% of their debt and the confirmation of their credit line and pre-payments to fund their expansion. This is what makes the play so interesting as financials improve, so more confirmations are valuable for the market and its perception.

We have repaid 75% of our short-term bank trade debt in the last year. Those were the trade finance lines that we used to finance our operations throughout the last two years. They've gone from $90 million down to $13 million only. That is a staggering 75% reduction. Not just that, in the last quarter alone, we managed to decrease these lines by 46%. That's a quantification of our ability to generate cash.

Everything was good, with healthy growth, improving margins and cash generation partially due to an increasing commodity price as I expected when I wrote the thesis. Demand is growing and their ESG product is attracting demand at a higher price point while having a lower production point than competition. Management confirmed their guidance of 240,000t/year for 2026. And that’s before they triple their production by the end of 2027.

We had one more great news as Sigma developed a “new” business: lithium fines. Fines are smaller lithium particles whose sizes do not meet the threshold to be manufactured into cells. Most of the times, those are thrown away, but Sigma figured out a way to purify them and… sell them, thanks to their high qualit. This is a net increase in revenues at no added cost as the mining is already done, which yielded $21M this quarter - not negligible, helping to improve margins and growth. This wasn’t a one-off as there is demand for up to 350,000 tons from their buyers.

On the stock itself, the market sold the stock for two reasons, in my opinion.

1. It ran a lot and this quarter “only” confirmed the core case - more the bull case to me well. The stock got expensive, traders took profits, end of story. No big deal.

2. A preliminary demand was filed by a local judge against Sigma’s operations, which he considers “a massive violation of human dignity” due to the potential impacts it has on neighboring cities - noise, tremors, waste piles... This is just a demand, not a ruling, and management released a statement saying they will defend their case and are operating within all regulatory frameworks. Noise.

This was a good moment for traders to deleverage and give the stock the retest it needed, right onto my entry point as shared in the chat. This quarter was in line with my thesis and as demand for lithium isn’t slowing down, especially not “cheap” ESG compliant lithium, I continue to be bullish on the company and am glad to be in.

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 is what a winner looks like. This is what we buy, hold and accumulate, and do so as long as it keeps winning.

The million-dollar question is how to find them? If it were easy, everyone would be rich so it obviously isn’t, but many winners share the same characteristics, meet the same criteria. That’s what we’ll talk about today.

The Winners’ Trinity

Most winners share three key characteristics.

1. A positive narrative within an accelerating sector, with clear catalysts.

2. Improving financials with a path to even more improvements.

3. A growing buying volume on the stock.

Lots of companies meet these criteria at one point in time. But only a few can hold them through months or years, with constant improvements and demand. Those are the ones we look for and even if some will fail - maybe most, every winner starts with those three criteria. So this has to be our starting point, filtering comes later.

So, what about Nebius? Well, it of course met the three criteria - and it still does.

First, narrative. If we were to remember the environment in early 2025, the narrative was to buy as many GPUs as possible to train the best model. Nvidia was king, LLMs were only starting to be real, with no commercial usage yet nor a clear idea of what could be done with them. AI equaled Nvidia.

DeepSeek came and showed the world that training wasn’t only a matter of GPUs, but also a matter of infrastructure and optimization. This changed the perception of the market around two things.

1. Nvidia’s GPUs are necessary, but not self-sufficient.

2. Open-source LLMs will be a thing to boost optimization.

The third key development was usage; we’d need a bit more time to confirm this but it seemed evident that AI was here to stay and would generate ROI to those using it, which meant a growing demand for inference - an unknown beast yet. The thesis was clear, it would be bigger than training.

This is the environment in which Nebius arrived, with an understanding other players didn’t have yet. Their core business wasn’t GPUs, but optimized compute. They built vertically integrated infrastructure for training/inference, optimized for open-source models. It seems obvious today, but it wasn’t early 2025. Their focus was on quality and cost-competitiveness.

This is what set Nebius apart. This, and their financial position.

Second, financials. Nebius got “lucky”. Being an entity of a Russian company forced to be dissolved, they started with a head-start in the form of an extremely efficient and running data center in Finland, and ~$3B in cash plus other business and equities - ClickHouse, Avride, TripleTen…

As their focus was optimized compute, it meant cost-competitiveness. Management focused on cheap energy, racks, hardware... Everything built within Nebius was and still is optimized for cost, which allowed them to be competitive, attract, and grow faster than competition with a control on their margins.

Strong balance sheet, potential for growth acceleration, control on margins, within an accelerating narrative and demand at the start of inference.

Lastly, stock demand. This might have been what lost many investors on Nebius as they didn't understand that the ticker used to be the Yandex ticker, and only became Nebius post split, late 2024. The chart would look disgusting without knowing this.

From this moment, volume has constantly accelerated with stable prices as those who knew bought. I buy on supports, and found Nebius early 2025, so my first purchases were in the mid $30s, while my big purchases were in the low $20s during the April tariff flash-crash.

If you know me, you know I only buy uptrends, stocks above their W50 but again, this chart includes Yandex which means its averages are including previous price action, which has nothing to do with Nebius. I bought the low of the range on a company with an extremely strong narrative, strong financials, with growing volume. The perfect purchase would have been on the $50 breakout assuming we were to follow my rules, and that would have been enough to yield massive returns.

What Then?

Then, nothing. When you find a winner, you hold, occasionally trim, buy on retests, and wait. The hardest in investing is always watching your money compound without doing anything. There is a playbook for this.

The stock gave you two extensions the last year and a half, and two retests. This was when you trimmed, and bought back, with the third extension being given right now.

The hardest is holding as you fear for your profits shrinking.

So far, my system returned 816% on Nebius.

This was how to identify Nebius, early. It met the holy trinity then, and held it since. It wasn’t the only one the last two years, nor the last one as we will see later. Here are other examples I participated in, while you could find many, many more.

Palantir.

Narrative - Artificial Intelligence started to be used with the most obvious use case: to help companies optimize their operations. Palantir’s AIP was released in April 2023 and meets this exact need, focused on governmental contracts and expanding into commercials.

Financials - Strong Balance sheet with ~$4B of net cash, improving operating margins, reaching profitability and a clear path to cash generation acceleration as demand for AIP was rapidly accelerating at higher margins.

Stock demand - The first post-AIP earnings set the story and stock demand exploded, with a massive breakout on volume. This is where the buy happened, at breakout or in the weeks after during the horizontal stabilization, constantly confirmed by earnings.

The rest is history, with four large extensions meant to trim, three clear W50 retests and ~1,200% returns over the 18 following months, at the top. I detailed the entire ride and how to optimize returns with a clear system on my selling playbook.

That system returned 815% on Palantir in 18 months.

Hims & Hers.

Narrative - GLP-1 pills are the new miracle pills and a telehealth company is setting up compounding to meet the current shortage, meeting FDA requirements. Their product is a perfect market fit but their application is also optimized for consumption, ease and discretion.

Financials - Flat balance sheet but rapidly improving margins and cash generation as usage of their platform increases in term of users and spending - pushed by GLP-1 pills. The potential is as strong as the demand and the shortage length.

Stock demand - Without surprises, the stock anticipated and broke out late 2023, with volume increasing after the May announcement. Volume looks small in the screenshot as it compares to what came after, but it was huge early 2024 compared to previous volume: buying pressure was crystal clear.

The $14 retest offered a perfect trinity setup. Narrative was strong, growing; financials were healthy and improving and stock demand was accelerating. One clear trimming signal at the top, two other retests to increase the position, and an obvious close to me ~$62 as the narrative broke.

My system returned 353% on Hims in ~12 months.

As I said, there were many more out there over the past years, especially with AI. But we shouldn’t try to catch them all. Our job as stock pickers is to catch a few of them, hopefully the best ones, with large sizes. Those three names were enough to generate wealth as long as taken with big sizes. You didn’t need other trades to make a killing.

Running towards all new shiny objects doesn’t accomplish anything.

My sole job with this Substack is to identify this trinity, to share them from the core narrative to their financial, including stock demand. To play them and to return as much as I can with a crystal clear system, from stock picking to buying and selling.

Today, I own six positions that meet that trinity.

Read more

The Current Market

I’ve been very bullish over the past weeks, and extremely bullish on AI hardware over the past months, making the sector my biggest position. But we are reaching some high momentum now, in terms of extension and short-term performance.

The S&P 500 is up 16.57% in a month and a half, the Nasdaq is up 26% in the same period. The first trades ~7% above its W21, 9.1% above at the top of the week, and the second ~12% above. This kind of extension is where retracements/stabilizations happen. I was the first one claiming for weeks that this was one of the most hated rallies I’ve seen and that we shouldn’t try to anticipate its outcome, but those words were written two weeks ago now and the situation is a bit different for me.

This is one of the most hated rallies I have seen. Don’t fight it, don’t try to assume it can’t run higher or longer. It can. The game is different when liquidity concentrates into a few names or narratives, and a hated rally can last longer than any bears would have assumed. It will end, no doubt here, but what we need to do is to be ready for when it does, not try to guess when or why.

This is where we think about turning down risk depending on our situations - and I’ll describe mine after.

Extensions aren’t enough to trigger a breather, even at this level. The market needs a catalyst, and even if it ignored it until now, it has one: inflation. Been talking about it for months now. It was - and still is, inevitable.

The Production Price Index jumped 1.4%, the biggest jump since covid’s supply chain disruption. Logical, as we are going through a supply chain disruption again, centered on oil/energy in general.

The PPI reflects price changes for manufacturing goods. Say I produce wooden chairs, I need wood and some other materials; the PPI will reflect the price increase of those goods. PPI usually is a pre-indicator to global inflation as if I, as a producer, have to pay more to manufacture my chairs, I’ll probably sell my chairs at a higher price to compensate.

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, which has no/very limited exposure 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 steps 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.

So without any surprise to me, PPI is increasing, and prices should follow.

To be honest, there are no certainties yet as a temporary inflation could be absorbed by manufacturers who’d accept some hit on their margins in order to keep volume. Either way, cash generation declines for them.

But what matters in the markets is perception, and perception this week was that inflation is back, for real, with the 10Y back close to yearly highs, and that will raise concerns around rate cuts - and potential hikes.

We’ve talked about Kevin Warsh, the new Fed chair that was sworn in this week to take over Fed FED after J. Powell.

The market tanked following his nomination as investors worry Warsh ends up head-to-head with Trump and refuses to cut rates, which is what Trump wants.

I think Warsh will walk in and cut rates. Why? Because it’s the only reason Trump put him there. If Warsh wanted independence, Trump would have found someone else, someone more malleable.

The second problem is that obeying Trump creates even bigger issues as it means the FED independence is gone. If Trump - or any government, controls both monetary & fiscal policy, it’ll push America toward disaster: the last step necessary for collapse. It would take years, but it would be unavoidable as politicians abuse both frameworks and crumble the currency for political (and personal) gains. On the good side, assets would appreciate if that were to happen, but I am not sure what would be best.

The Fed next rate decision will be in June, all eyes on what Warsh will say for his first minutes, but as the Strait is still closed and we’re already seeing the accumulation of a year of tariffs and energy disruption… This kind of disruptions don’t disappear in a week, but in months from the moment flows restart - and they haven’t.

We have another variable with Trump meeting with Xi this week. Not much news yet but this renewed relationship could change lots of things for the world’s economy. The only valuable comments I’ve heard so far concerned agricultural imports.

Back on inflation, the only companies benefiting from this kind of environment are commodity sellers - usually, and we’ve been talking about lots of them over the last months so you guys should have options in terms of accumulation.

This is the market’s situation today. Extension, clear sign of inflation and fear of it. But that is only macro, and the market isn’t only affected by macro.

The AI Dichotomy

This is where things get extremely interesting. As the earnings season is over, we can all reach one clear - and unsurprising, conclusion.

AI is bigger than we originally thought, is generating more cash than we originally thought and will trigger more spending than we origin… You get it.

I’ve written four different earnings reports and covered ~50 companies, and all said the exact same thing: that they didn’t have enough compute, need more hardware, infrastructure, energy, and had so many clients fighting for their services that they ended up with pricing power.

The interesting part is that those companies’ final products are services, not goods; I’d go further as to call them deflationary services. Sure, to deliver those services, they need hardware and infrastructure whose prices are rising due to inflation and supply issues, but those companies have the cash to accept price hikes. Look at Google or Meta. When your core business is advertising, cash generation doesn’t change; clients rotate but cash still flows.

In brief. Companies with massive and constant cash generation can buy hardware at a higher price to sell a deflationary service to companies going through an inflationary spike, so they want to buy it.

So… Why would the AI cycle slow down, even if inflation spikes?

That’s from a business perspective. The issue with stocks is that they are impacted by liquidity. Investors sell during inflationary times not because they do not believe in the companies but because they need cash/are afraid of the market’s reaction/anticipate a correction. Better be the first one selling than the last.

It’s impossible to predict what will happen when it comes to the AI sector. But with potential inflation spikes, AI hardware seems to be the only sector that could thrive, so my conclusion remains the same as months ago: why touch any other sector than AI hardware and defensives?

Impossible to know what will happen, but it is possible to plan for the different potential outcomes. As long as I have a system and a plan, I’ll do well.

My Situation

My portfolio is up 141.20% YTD running at ~60% margins - I have cash outside of my broker so more like ~45% in reality.

Portfolio’s well balanced with 65% tech/risk and 35% energy/materials. This is what diversification should look like, I’ve written about this, you guys should go read this article if you haven’t.

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.

I am now in a situation where I’d rather lock in gains and reduce margin to none, even maybe raise some cash. When you hit triple digit returns in five months, you don’t want to take too much risk if conditions aren’t perfect, even with a diversified portfolio. I’d rather chill a bit until the market gives clear directions.

We could have a healthy consolidation now. If so, I’ll have time to buy on breakouts. If we were to continue higher, I’d still be part of it without margin, with the certitude that a breather is coming and that slowing down isn’t a mistake. And if we were to have that breather, then I’ll be ready to go back all-in on key assets at great prices.

So for me, now is about reducing risk and getting ready for the next leg up, whether it happens after a rapid consolidation, or a deeper breather.

Now let’s dig into the specifics; my plan for each of my positions and the ones I want to buy/accumulate, at what price and priority.

Read more

I’ll also take a minute to remind you that there is a launch discount for those interested in the premium content. A 30% discount for the yearly Founding Member plan, giving you access to all current and future services, at a fixed price.

The promotion will end on June 13th.

Now, let’s dig into what matters: earnings!

AI & Compute

Nebius dropped another banger, without surprises after the last months’ news - their Meta partnership and a new $2B financing round from Nvidia, which made me buy back into the stock after my terrible selling mistake.

The cherry on top for this quarter was the above-expectation EBITDA, which means not only their compute service is growing rapidly but it also is generating cash, and this is what sets Nebius apart from most of its competition: its financial efficiency and balance sheet - which was the largest part of the bull case even back in early 2025.

We also learnt that they honored their commitments to both Microsoft and Meta - which is important as their contract with them and their potential expansions depends on deliveries and timings.

This financial position comes from their history but their cash generation comes from their focus on efficiency and vertical integration, confirming once that optimization of the entire compute chain is key. I write the same thing every week now…

Our platform is most efficient when we own the full stack... from bare metal to multi-tenancy to inference to agentic and more.

Then come the classic comments about supply constraints and clients fighting for any compute they can get which means pricing power, at least outside of their core clients which usually come at preferential/fixed rates. Nothing new.

The demand is broadening across industries. Today, we typically see several customers competing for every GPU we bring online.

Strong market demand is translating into pricing gains in our latest deals.

As for the source of demand? inference, with two new contracts with Revolut and Monday this quarter - big players.

Inference is the fastest-growing segment, new segment, in our stack... Token Factory is our primary inference product now.

I could go further on how excellent the quarter was but I don’t think I need to. The bottom line is that everything is either great or better than great and that demand for compute is through the roof, which means there is a need for more of it - you heard that a lot lately. More compute also means more infrastructures, which means more datacenters, more hardware and more energy - we talked about this a few days ago.

Raise our 2026 CapEx guidance to between $20 billion and $25 billion, which is up from our prior range of $16 billion-$20 billion.

Contracted capacity already exceeds 3.5 GW, far surpassing the goal of 3 GW we set for the end of the year. Our execution gives us the confidence to raise our contracted power guidance to more than 4 GW by year-end. We have now secured two gigawatt-scale sites in the US. We broke ground at our Missouri location yesterday; today, we are announcing a new site in Pennsylvania where we have secured power for a deployment of up to 1.2 GW. Owned capacity now accounts for more than 75% of our contracted power. We expect capacity added in Q3 to significantly expand our footprint. We continue to expect 800MW to 1GW of connected power by year-end.

We build very efficiently... what we do around creative use, you know, interesting technological ways of heat reuse.

The difference between contracted and connected power is what is online today and what they plan to bring online in the medium term. Management is telling you they plan to triple/quadruple their online capacities in the next years. And every other hyperscalers/neoclouds said the same.

We’ll see on the premium part the investing implications, and they are as big as a bull could have expected in terms of hardware demand - but not only.

I made a mistake on Nebius. I corrected it and called the stock an obvious buy on the $110 retest. This turned out to be a great decision as this purchase is now up 100% in two months. I hope you guys are still in, and killed it with this trade.

I continue to believe that Nebius is the best neocloud on the market and has room to go much higher in terms of stock price as long as balance sheet and EBITDA margins remain healthy. As long as they can prove they will generate cash from their business, the stock can climb, and is worth holding.

Tower Semi

The photonic fab pure play did great, this is excellent for us and our Global Foundries position which is still considered to be #2 by the market but still comes with massive capacity within a rapidly growing demand.

There are no doubts on the actual demand for optics, what is interesting is the most demanded product which isn’t what social media believes to be - as hinted at by Arista Network and AsteraLabs last week.

We do see extra dense pluggable optics, XPO, being led by Arista with the aim to extend the served generations of pluggables. Forum panel presentation and near package optics, eventually also co-package optics, emerging and coexisting with pluggables for the next several years.

We are already seeing strong demand for NPO products in 2027. Given this strong customer traction, it's our expectation that Tower SiPho will continue to lead in these new optical form factors. NPO is likely to ramp over the next several years and precede a significant ramp in CPO for our primary customers.

Pluggables are not going away at all. Pluggables will stay extremely strong at least through the 2030.

I understand the massive traction behind CPOs, but the real traction should be behind optics globally. Sure, it is almost the same, but not exactly, and it changes our stock picking as some CPO names are key to optics, others are key to CPOs. That’s a big difference in which to buy and which to give time to, despites the hype.

Either way, demand for optics is accelerating, requiring fabs expansions financed in part by clients’ early commitments - which highlights their confidence in their need.

We continue to strengthen our alignment and partnerships with our photonics customers through the execution of long-term customer commitments, contractually representing $1.3 billion revenue in 2027, with significantly larger valued contracts for 2028, backed by approximately $290 million in prepayments already received from our largest SiPho customers.

This is really the only important take-away of Tower: optics demand is growing.

Cisco

I won’t go over the entire quarter but it is important to talk about the resurgence of the 2000’s bubble king who’s finally adapting to AI. This was probably just a matter of time, but it finally is happening.

Given the strong demand, we now expect to take AI infrastructure orders of approximately $9 billion from hyperscalers in FY '26... we've had just probably more design wins and more success than we expected at the beginning of the year.

The overall take-away is positive: it means that buildouts continue and that one more player is necessary when it comes to rack networking connectivity, that the market is so large that there is enough room for multi-20%+ growth players.

Energy & Batteries

Venture Global delivered what I expected in today’s environment: a double beat, a massive 59% YoY growth and an accelerating demand for U.S. LNG as the commodity is disrupted with the Strait of Hormuz blockage.

The real signal is that this disruption isn’t just changing short term dynamics, it is changing long-term dynamics as clients know that this situation could/will repeat and secure 5-year long contracts from U.S. providers, and therefore Venture Global.

Venture is taking advantage of the situation, selling its future capacity at a premium to clients with high demand for safe and regular production over the next years. The cost of a disruption like today is higher than a higher price point for the commodity, so it’s all beneficial for both parties - at least from today’s point of view.

This led VG to raise FY26 guidance from $5.2–$5.8B to $8.2–$8.5B of EBITDA, showing the market that growth leads to cash generation, and the market loves that.

So, all is great - except for Europeans.

On the battery side, EOSE gave two “positive” signals this quarter.

First, they secured financing. When you generate $57M revenues for a gross loss of $44M and burn ~$150M in a quarter, you kinda need financing, especially to please the market. Their problem is lack of scale, as batteries are expensive to produce and require factories to do so, and investing in factories is also expensive.

That’s why they set up a joint venture called Frontier Power USA with Cerberus, a private equity firm which signed a 2GWh order for EOSE’s batteries.

Second, this financial agreement implies that Cerberus has a final buyer for this capacity, they wouldn’t put their equity and cash on the table to own part of EOSE through a joint venture and order for 2GWh of batteries without a client. As to who that is - or if they really have it, we have no idea. This is where speculation starts, the bet is that they have AI data center demand - as the order is for their Turbine-X product. That order would start in 2027 for a three-year length. For a bit more perspective, this deal is for 2GWh over the next 3-years while EOSE’s total production capacities are ~2GWh in 2025.

I stand by my conclusion on EOSE and others that I shared on my detailed review on datacenters' infrastructures.

Intuitive Machines

The stock ran a lot in anticipation of the earnings, now up ~45% since entry was shared a month ago. There were no surprises on these earnings, with all new contracts already announced, including one yesterday with the U.S. force for space surveillance.

Like Firefly, they subtly announced that they weren’t expecting new NASA contracts right now, but also confirmed the agency’s wish to accelerate missions’ pace - which is what matters. They also confirmed that IM-2 was on track and that IM-3/4 are 90% complete, no delays expected, and shared a record $850M backlog. Lots of positives with the confirmation that both hardware/communications are now generating healthy revenue stream, which is a great bonus.

The global space narrative is pushing higher, as expected, on the continuous need to control this new region for resource and geopolitical reasons. I see no reasons for this to slow down in the near term, meaning I see every reasons for contracts to continue flowing and holding our positions, at least for now.

China & Asia

Alibaba’s reports wasn’t that bad but I will hold onto my previous opinion as to why I sold my position.

My thesis was simple: AI revenues would rapidly increase as Alibaba is a vertically integrated provider - the only equal to Google, within one of the most populated countries of the world and its politics are trying to shift towards consumerism. One can see the bull case rapidly here, use consumption and advertising cash generation of your core business to fuel and accelerate AI spending.

But e-commerce in China became a competitive nightmare and Alibaba was forced to invest in its retail platforms to maintain growth. Sure, AI services revenues are rapidly growing and generating cash but in the meantime, global cash generation is falling off a the cliff and potentially slowing AI investments.

Now, don’t get me wrong. I believe Alibaba is an excellent company and maybe an excellent opportunity. But the market is punishing every stock with lower cash generation, I don’t see it rewarded a Chinese name with decreasing cash generation. I think my current positions all have a much higher upside than Alibaba short/medium term, so I’d rather buy more of those.

The potential is still here. But we’ll have more opportunities if Alibaba were to really become Google’s equal and trigger comparable returns. There’s no rush for now.

I’ll also rapidly comment on both Sea and JD as I posted my view on Sea and it made lots of people unhappy.

The bottom line, as sad as it is, is that every e-commerce/fintech or “everything app” player is investing in their growth while consumption is not trending up which means the market is doubtful on the return on those investments.

The only one this quarter who did well and increased its margins was JD, proving that past investment increased margins and cash generation. Guess what happened? Stock went up. Still wouldn’t touch a stock within an extremely weak sector, but the reaction tells you what the market wants, as usual: increased cash generation potential.

You can use every metric you want, from revenues growth to low multiples compared to historical ones, the conclusion remains the same. Today and for the short/medium-term, those companies will be spending, reducing margin and cash generation, for a return that is impossible to anticipate. This situation is comparable to SaaS - who aren’t spending but on who the market doubts future profitability.

That’s all it is. We are officially in a stock picking market, where everything doesn’t go up anymore and if you pick uncertain cash generation names with no immediate tailwinds, you’ll get bitten. You can of course accumulate and believe it will work out long term, you’ll probably be right when it comes to those two. It’s just not what I do nor would want to do.

To each our methods.

Other Interesting Earnings

I also want to rapidly talk about Hims and On Running.

Hims is finally showing what I was concerned about ~9 months ago when I closed my position. The issue with hypergrowth stocks is that it is very easy to look back and assume the future will be like the past, which is what so many did. The future isn’t the past.

Not gonna be hiding, this is a “I told you so” commentary. It doesn’t really matter what we believe, when the data goes against our thesis, we get out. This is how we invest and how we avoid becoming bag holders. You have no value added building a community of diamond hand that end up losing money.

Hims will be an example I will use for years, having lived through the up and down and all the stages of this stock. The company will continue to operate and is legit, not going against that - at least until proven to the contrary. As for the stock, there’s nothing for us there anymore. Hims was a GLP-1 company. Not sure it can transform itself into something else, only time will tell.

On Running is just a guilty pleasure as I love the company and their products. We’re looking at a very healthy company with declining growth, logically as those products aren’t the most appealing to consumers lately. Sure, On is a premium brand but premium buyers aren’t accelerating their purchases…

Hopefully, a time will come when those products will be in high demand again but as for many other sectors lately, we’re better off far away from them. Focusing on what works, on strength, but still keeping an eye on great companies.

Now, let’s pass to the premium part and the comments on how we can use these information to our advantage in terms of investing.

Read more

Electricity & Datacenters

This won’t be a surprise if I were to tell you that data centers run on electricity, right? Hardware must be plugged into a power source, while GPU consumption is usually higher than average hardware consumption. As a rack will include dozens of GPUs and a data center dozens of racks plus cooling systems, ventilation, etc… Well, an AI data center will consume more than a traditional data center.

It’ll actually consume a sh*t ton of electricity, pardon my French.

If you want some comparison, a U.S. family of four consumes ~13,000 kWh per year on average, slightly more than 1,000 kWh per month, and they do not consume 24/7 as kids go to school, parents to work and all of them sleep at night - keep this in mind as it’ll create a problem.

Hardware doesn’t sleep, it is on 24/7. Today’s AI data centers have a need for triple-digit MW and we are already building GW data centers. Today’s assumptions for AI data centers are between 175 and 200 TWh per year - not just including AI but all data centers, which is ~5% of total U.S. generation or ~13.85M families of four.

With the new GW monsters, some assumptions talk about doubling to tripling this consumption by 2030, consuming between 12% and 17% of total U.S. electricity.

This is our first problem: we do not have enough. If you wonder how that’s an issue, I’ll frame the situation differently.

Electricity generation is limited. If AI data centers take it, someone else doesn’t have it. When there is a fight for a commodity, it usually goes to the one who pays the most for it, increasing the commodity price for everyone else. And Mr Dupont won’t be able to compete with Microsoft, so global electricity prices will rise due to demand/supply market mechanisms.

AI isn’t the only reason for electricity costs to rise, let’s not generalize, but it is part of it and will grow in proportion as the new monsters are built.

Now, we have to talk about the grid. I covered the topic and how that works here, you should take a quick read to understand this concept - it’s important. In a nutshell.

Going a bit further into electricity now. The grid is the name given to the electrical system of a city or a country, from producers to end-users. It is an incredibly complex piece of engineering, constantly hanging by a thread as its production has to equal consumption at any time. The entire infrastructure could overheat & burn if this weren’t the case - or at least be damaged.

As I shared earlier, data centers are online 24/7 but the grid doesn’t have the same demand 24/7. The need for electricity fluctuates depending on the month but also the hours, with a base load - constant demand, and variable consumption - households or temporary needs like climatization or cooking.

This is our second problem, electricity generation has to fluctuate during the day and increase at peak load while grids are meant to generate more than this peak load, having new giants plugged to it, adding massively to the base load, might reach production limit and create supply issues. This kind of is related to our first problem, so let’s call it a first and a half issue.

The second half of this issue would be that the U.S. grid hasn’t been upgraded for decades and uses inefficient/outdated infrastructure, for production and transport. And you guys know we don’t like inefficiency when it comes to AI lately, where everything must be optimized.

Because most data centers are plugged into the common grid, they use the same infrastructure and compete for the same electricity as Mr Dupont, and this is a major problem, one which data centers try to resolve for two main reasons.

1. They want access to reliable and cheap electricity without external impacts, to be able to manage their operations as they see fit.

2. Political pressure is growing as it cannot be tolerated for data centers to consume more than 10% of U.S. production and impact households’ prices.

Our third problem comes from electrical engineering and the AC/DC currents. I won’t go into the details, but the overall concept is pretty easy to understand - as is the problematic it creates.

The difference between alternating and direct current is how they flow; the first alternates directions based on frequency while the second is flowing straight from A to B. What matters to understand is that DC is more stable and optimized for electronic usage - data centers need DC. AC used to be easier to manipulate and transported over long distances, and our infrastructure was built using it and is too expensive to change now, which leaves us a legacy system while new technologies like HVDC could replace it, but won’t - that’s political.

Current can be converted between AC/DC with specific hardware – rectifiers/inverters, but there is a tax to it in the form of lost energy which is estimated to be between 10% and 15% by specialists. Every time current is converted from one form to another, energy is lost. When we talk about GW data centers, a 10% conversion loss cannot be tolerated in terms of added costs – we are talking tens of millions of dollars easily, not including the systems required to cool the heat generated by this loss.

In a nutshell, the problem is simple: we do not generate enough energy today to meet AI’s needs, current infrastructure is old and inefficient and has to be optimized just like compute has to be. This can be fixed in three ways.

1. Generate more electricity.

2. Store energy outside of peak load.

3. Build auto-sufficient systems for AI data centers.

But our biggest problem is different and justifies the accelerating spending on new energy solutions today. It isn’t technical, but temporal. Data centers need an efficient solution now, today. Not in two years. They need to power their expansions which aren’t being built at a bureaucratic pace but at a private company pace, and cannot wait for grid upgrades.

And that’s what we are going to talk about today. The narratives, the technical, the opportunities and how to capitalize on them, right now.

Read more

Tons of information, but only valuable information - and mostly bullish.

The Hardware Saga

As usual, I will treat hardware names as one as they are all part of the same narrative, one of continuous demand for optimized architectures to improve GPU computing capabilities.

To spoil it, this week confirmed what we knew from the first and second earnings’ week. Demand is not slowing. No point in doubting anymore.

AI Core Infrastructure

Let’s start with core AI hardware first, with a quote from Arista’s management, to put you at ease on the current state of hardware demand.

The supply, however, is a slightly different and opposite tale. We are experiencing industry-wide shortages across the board, be it wafers, silicon chips, CPUs, optics, and of course, memory that I referred to last quarter, coupled with elevated costs to procure these. I think the supply chain problem is not a one or two quarter phenomena. We now think it’s a one or two year phenomena.

This quote is already extremely bullish, and SMCI confirmed this shortage in their own earning call. When all key actors tell the same thing, we should listen. Their knowledge is thousands of times better than ours…

Now, earnings were great but the stock still tanked, for two reasons.

1. As components are in shortage, price increases could hurt margins. The market doesn’t accept lower margins lately, for any company, not even AI hardware stocks apparently.

2. They shared they’d continue to focus on XPO for scale-up while the market is focused on photonics and CPOs - not the perfect narrative.

   Within a rack and scale-up, there’s still a number of choices. I think within short distances of 2 m- 3 m, you’re still gonna see a lot of co-packaged copper. I think XPO in terms of density will be another alternative. I don’t rule out open CPO as well over there if they’re really looking to maximize their density in a minimum amount of space. I think XPO will be particularly prevalent in scale-out and scale-across, and will be one of the choices in scale-up.

   Arista is one of the key giants in the industry so again, we should listen. If they say XPOs are in high demand and will be the most demanded hardware until 2027, before CPOs, they know better. Also worth noting that Astera Lab’s management said the same thing and confirmed growing demand for XPO. If necessary, XPOs are copper-to-optic pluggable while CPOs are optic-to-optic ones.

But earnings were excellent; management increased growth guidance and confirmed that scale-out/across was healthy and growing, while they’d work on scale-up from 2027 with copper solutions first, optic later. This isn’t bearish on photonics, but we have to keep in mind that the technology will take time to be finalized and scale, and shouldn’t rule out copper until then.

Arista is down 19% since earnings and buyers aren’t really stepping in. There is a case to buy the W50 ~$130 for a trade. The chart is wonderful but we are trading at a high valuation with declining volume and margins concerns - which seem a bit overblown but the market is always right, so…

This cannot become a core position, but a trade is possible with a tight stop-loss.

On CPUs, AMD and ARM confirmed that more of them will be needed to orchestrate GPUs compute for inference.

I personally was a bit bearish on AMD as the narrative was that their GPUs would overtake Nvidia as a cheaper alternative. Always called this bullshit, but I missed the bigger picture: CPUs. AMD had everything to be a massive player within this narrative, and I missed that.

The take-away from this quarter is simple and comparable to what I’ve been writing almost every day for months now: AI infrastructure is focused on customization as GPUs need support to be used at 100% of their capacities.

Data center is now the primary driver of our revenue and earnings growth, and as AI adoption scales, demand is increasing not only for accelerators, but also for the high-performance CPUs that power and orchestrate those workloads.

Inferencing and agentic AI are increasing the need for server CPU compute, as these workloads require additional CPU processing for orchestration, data movement, and parallel execution, in addition to serving as the head nodes for GPUs and accelerators. As a result, we are seeing both stronger near-term demand and deeper engagement with customers on long-term capacity planning.

At our Financial Analyst Day in November, we outlined the server CPU market growing at approximately 18% annually over the next three to five years. Based on the demand signals we are seeing today and the structural increase in CPU compute requirements driven by agentic AI, we now expect the server CPU TAM to grow at greater than 35% annually, reaching over $120 billion by 2030.

While we are still in the early stages of the AI infrastructure cycle, the pace and scale of deployments we are seeing today reinforce both the magnitude and durability of the opportunity ahead. As inferencing and agentic AI deployments scale, they are fundamentally increasing compute requirements, driving both larger scale accelerator deployments and significantly more CPU compute.

You know, it’s very hard to call exactly, but we certainly see the movement where in the past the CPU to GPU ratio was just as a host node, you know, in like a 1- to- 4 or 1- to- 8 configuration, now changing and getting closer to a 1-to-1 configuration; you can even imagine if you get lots and lots of agents that you could have more CPUs than GPUs.

AMD expects its CPU branch to grow 70% YoY. This quarter was also bullish on Silicom as AMD confirmed a growing demand for FPGAs for inference architectures.

Design win momentum grew by a double-digit percentage year-over-year with billions of dollars in new wins across markets, reflecting the continued expansion of our embedded business from a primarily FPGA-focused portfolio. Our semi-custom engagements also expanded in the quarter as data center, communications and other embedded customers leverage our broad IP portfolio and high-performance expertise to build differentiated solutions.

This confirms my bull case, although Silicom already received an inference PoC order so the bull case was already confirmed, but more confirmations never hurt. And we had one more with ARM as management confirmed that SmartNICs - not just FPGAs, were also accelerating in demand for AI inference purposes.

It’s hard not to be overly bullish on this inference PoC and Silicom.

ARM estimate CPU demand will quadruple by 2030, confirmed their architectures will be part of all hyperscalers’ next designs and that demand for their AGI CPU doubled in a few months to ~$2B run rate for the next year.

Market’s problem was the same as for Arista; management confirmed they secured the components for ~$1B of orders only and that the rest would come at a more expensive price point - if they were to find supply, while the transformation from 100% licensing to manufacturing will reduce margins long term. Extremely bullish on the market’s dynamics, but if you do not have the components to meet demand and reduce margins by doing so…

I chose to close my trade with a nice ~20% gains post earnings as ARM is valued for perfection and the promises of a quadruple demand volume for the next three years is great, but execution remains a question mark at this valuation.

Lastly, Lattice also reported great numbers. I look at it as a proxy for Silicom as they are a pure FPGA play, key component of Silicom’s SmartNICs. The quarter beat expectations with record revenues, growth and margins, and highlighted the bull cases for FPGAs with different partnerships; inference but also the future bull case: robotics.

The story is centered around AI data center and compute because of the current bottlenecks FPGAs and SmartNICs can – and will, help with, but robotics will be built around FPGAs. It won’t be here to help, it will be designed around them.

Coming back to partnerships, Lattice announced one with Nvidia’s robotics programs and TXN for the same reason. If those constructors need FPGAs, even if not sold within NICs by Silicom, it means that demand is real and others will need complete and personalized hardware including FPGAs.

Photonics

Global Foundries, one of my core holdings. The quarter showed what I wanted: healthy enough core business with an explosive vertical, its AI data center business growing 32% YoY and expected to grow ~100% FY26 - including its SiGe business which is oversubscribed until 2027. The market continues to treat it as a slower giant because of its other verticals, which aren’t that sexy.

I believe those verticals are what makes Global Foundries a great pick. Management is optimistic on their trajectory for 2026, except for their smartphone vertical - although they highlighted a growing demand from AI wearables which could balance lower smartphone demand and is a great indicator for my Nordic Semi thesis. Second, those verticals are the source of cash generation allowing expansions without dilution, an advantage compared to Tower semi.

The company is winning designs, 50% more wins than last year at the same period, is oversubscribed on photonics/SiGe demand for data centers, forced to increase CapEx to meet a growing demand for higher-margin products… This is a great position to be in, having the cash to expand while expecting a ~100% growth from an accelerating vertical on which management announced its new CPO technology, not only meeting clients’ expectations, but exceeding them.

We shouldn’t expect explosive returns, but slow and steady gains as core business bounces while higher-margin photonics grow in volume. Guidance came at $1.76B Q2-26 and management recommitted on its $1B run rate from photonics by FY28, up from ~$200M last year and planned to double this year.

Last interesting note, management had a $500M buyback plan and bought back for $400M worth of shares, almost consuming entirely their program. This shows what management thought about valuation – just like I did.

On optics, the signs are crystal clear: supply is just not keeping up with demand.

Coherent reported accelerating growth and margin expansion, hence pricing power.

On track to double internal InP output by year-end and more than double again by 2027.

Applied Optoelectronics reported the same. Revenues grew 12.5% sequentially with a 154% growth for their datacenter products, guidance is planning a 19.2% sequential growth for Q2 and management confirmed this trend should accelerate.

Looking ahead, we continue to anticipate a strong volume ramp of our 800G products starting in Q2 and we anticipate sequential revenue growth throughout this year, with significantly larger growth expected starting in Q3 as additional capacity comes online.

And Lumentum said they were short of supply of everything and are fully booked for what they have.

Those three report confirmed two things.

1. The increasing demand for optic hardware to scale scale-out and scale-across.

2. The increasing demand for optic hardware for scale-up through photonics.

3. A crystal clear supply shock for most of the required components, from InP substrates (Axt and IQE are two great names to buy in this vertical if given the opportunity) to lasers, and more.

AAOI’s management expects this supply shock to continue until at least mid-2027 and plan to ship ~900,000 monthly 800G/1.6T optic transceivers then, from ~100,000 today with most of the new production within U.S. fabs - a great bonus for hyperscalers who do not want to rely on geopolitics and complex supply chains.

This is the size of the demand and the infrastructure expansions happening to meet it.

Few bonus, AAOI management also talked about reaching profitability early this year which means not only accelerating growth but also improving cash generation. This is the double re-rating situation we’ve been talking about and the golden standard for investing. This is what we want to own, even if in AAOI’s case it will take a few more quarters to materialize.

For the broader market, this indirectly confirms the growing demand for Soitec’s wafers and Global Foundries’ capacities to manufactured finished CPOs.

As for market reactions, most names ended up green after some volatility; keep in mind that most of them ran 1,000%+ and need to breathe a bit, while being priced for nothing less than perfection - and more. The narrative just strengthened after this week and I’ll buy any dips on the key names of those verticals - if given.

I also had a look at Fabrinet (hardware packaging company) and On (semiconductor fab), both reported excellent earnings. The first one is expanding its capacities to meet increasing demand - optics are the star of the show while CPOs haven’t started to scale. The second yielded 30% sequential growth on its AI vertical and is expecting to double revenues this year, following TXN’s example as AI infrastructures expand. Also interesting to note a growing demand for storage related hardware, which plays into my SolarEdge thesis and its B2B new business - more on this a bit later.

Compute Providers

Few words on CoreWeave and Iren who follow the path of hyperscalers with the same trend of demand outstripping supply, especially for CoreWeave and its Google-like RPO growth, reaching ~$100B with an average contract length ~5 years.

Gone are the days - or should be, where the market doubted compute demand or clients’ capacity to pay; this backlog isn’t just OpenAI, it includes Meta, Microsoft, Jane Street, Mistral, Perplexity, Anthropic… Some concerns still exist but the market should open its eyes on compute demand and financial commitments, especially as token price is getting cheaper and usage is accelerating as clients find new economically viable sources of consumption. Consumption isn’t slowing, on the contrary.

The market continues with the same narrative though, “neoclouds don’t make cash”. Margins decrease and explosive RPOs mean more spending, not less, which comes with more depreciation, lower margins, etc…

This is why Meta was punished, why I believed Nebius would also be, what CoreWeave is struggling with today and why I prefer to invest in hardware despite the potential in compute providers and my bullishness on token consumption, and AI at large.

Cash generation is what matters the most for a stock.

Iren is Iren, not my personal favorite as I consider it a great energy/infrastructure player but not a neocloud, like most of the Bitcoin miners reconverting in compute - I might be wrong here but I’ll be fine with this. Still, they received a partnership with Nvidia which helped the stock, while missing on market expectations for both revenues and earnings.

The partnership is classic, Iren will be a privileged partner to Nvidia who will rent part of their compute once installed, plus have the right to buy up to 30M of shares at $70 in the next five years. This is a net positive for Nvidia who get orders for its hardware, compute it needs and in exchange, gets the right to own a portion of the company who took all the risk. Iren will have a guaranteed client for its compute which derisks the investment for the next years, but will still need to fund the hardware as Nvidia isn’t helping here.

Sure the market likes it, and it’s not a bad deal. It’s just clear who wins the deal and who needs it badly. There are better fundamentals out there - Nebius notably.

The Software Saga

In this bloodbath, one irresistible Gaul resists, and his name isn’t Obelix, but Palantir. We have rarely seen such companies on the market outside of AI hardware – even within there aren’t many with such execution.

So I’ll be brief, another quarter, another masterclass.

More acceleration, more cash generation, more deals, more demand, just… more. That’s all there is to say and that shows how important AI enhanced services will be in the future. Can’t forget that as a day will come where performance will be in those assets, not hardware anymore.

But. Because there is a but. The market isn’t reacting, the stock is down 7% since the print and I see excellent reasons for that.

1. The stock front ran earnings. This always happens in growth, and this is why investing based on multiples never works. As the market anticipates, multiples rocket and then stabilize, giving time to the stock to catch up. This is where we are with Palantir. This stock will go higher, but the probability of nothing happening or it going lower first is much higher than others.

2. The market has new shiny toys. Excitement is not in AI models or Palantir anymore, it shifted to new hardware technologies with photonics, CPUs, memory, etc… As I’ve said a few times now, liquidity is thinner and concentrating, leaving the old toys to go towards the new shining ones. Liquidity makes stocks move.

What I know is that a stock not running after a beat and raise revenues means there is no demand for it. However amazing the company is, I’d rather be in other stocks.

To comment rapidly on Duolingo, well… AI services aer still a niche with only a few massive winners. Management said monetization would slow down in exchange for user growth and a focus on retention and acquisition. Can’t say it isn’t working.

But as usual, the market cares about cash generation. It’s great to grow users but if it is just to add more users who do not pay, well… what’s the point? So until cash is generated, the stock won’t see any progress.

Shopify is in a different boat but with the same kind of reactions as the stock sold off 15% after strong earnings. What matters is the future and not the past and guidance is looking a bit slower. Any SaaS with slower growth isn’t worth a premium, this is the market’s current rules.

In a market where Palantir is down 7% after such a quarter, I wouldn’t look at any SaaS as potential plays… Just accept it. Their time will come. The only winners this week were Fortinet and Akamai, two cybersecurity oriented services. Not surprising at all to see cybersecurity being the first sector pushed higher by AI, and I will keep a close eye - and maybe propose some trades, on Palo Alto and Crowdstrike earnings, as those could follow.

Defensives

Starting with SolarEdge which delivered healthy earnings considering Enphase’s results a few days ago - which confirms my choice of picking SolarEdge. Revenues and guidance were in line with expectations with strong demand from Europe, as expected now that interest rates are lower and the Iran situation is pushing energy prices with no end to this in sight. Europeans who are stuck between Russia - fom whom they refuse to buy oil, and the U.S. who overcharges for LNG. Energy prices won’t go down and many Europeans know that.

In the U.S., residential remains slow but the narrative is on commercial and AI datacenters whose energy consumption is through the roof and constantly growing, looking for cheaper and grid-compatible alternatives while their new platforms are meeting interest.

I didn’t see anything wrong with this quarter, nothing which would go against my thesis and the market sold the stock for no specific reason but maybe the EPS miss which was due to a one time bad debt write off, while overall the company’s margins are getting much healthier.

I personally sold more puts on the W50 at $36 and bought a ~70% position in shares, I’d buy more on the W50.

Now I want to talk about Fluence Energy on which I bought a position today pre-market. I invested in SolarEdge partly for their future AI datacenter batteries they are developing as we know the grid cannot support many more data centers and those want - or will be forced, to use alternatives not to impact global electricity prices. Batteries have a great potential to help.

But they need to be adapted to the grid and datacenters, to be the bridge between two different electrical structures which is technically complex, which is why I am bullish on SolarEdge developing their new hardware.

Thing is, Fluence Energy announced yesterday having signed MSAs with two hyperscalers for their batteries and said they were the winner of a 25+ provider study.

The selection process for both of these MSAs was subject to multiple rounds of review. In each case, Fluence was chosen after meeting criteria specific for each customer. In one case, the customer's process began with 26 different best vendors. Fluence was the first to complete all qualifications to sign a global MSA. In the other case, the customer had requirements which made it hard for many competitors to comply with. In both cases, we believe Fluence's understanding of customer requirements, rapid response time, and differentiated products were key in driving this engagement. These MSAs established Fluence as a qualified supplier, positioning us to bid on expected near-term data center projects for both hyperscalers.

MSAs are technical reviews and purchase agreements for their solution, without financial commitments. Meaning their batteries will be bought but we do not know for how much and this commitment isn’t represented in the company’s financials.

Talking about the financials, Fluence increased its backlog by 40% over the last six months to $5.6B, without including those two hyperscalers - while one is expected to sign the financial contract and disclose its purchase by Q3-26. In brief, demand is accelerating without including on hyperscalers.

The company is valued at $3.5B at the time of writing, at 1.3x sales. I bought a position even after a 60% jump in two days. It is scary to buy after such pump; there is a part of speculation since MSAs still have to translate into financial orders, but I have shared many times that I am bullish on batteries for datacenters and this kind of event is exactly my bull case. If my thesis is materializing while the stock is still cheap, I want to be a part of it as it also is plausible that SolarEdge’s batteries were tested and didn’t meet requirements.

It is risky, but the potential is here.

Nutrien also reported earnings on Wednesday and numbers are good. The take-away is simple: demand for fertilizers is growing while supply is tightening and prices are rising. Not great for inflation outlook, but great for defensive names.

We increased production from our low-cost North American assets and positioned our supply chain to reliably supply our customers amid tightening global fertilizer supply and demand fundamentals.

Global potash demand remains strong and we have maintained our previous forecast range for global potash shipments of 74 to 77 million tonnes in 2026. We anticipate relatively tight potash fundamentals throughout 2026 with demand trends expected to test existing global operating and supply chain capabilities.

This is bullish for my Intrepid Potash position at least in terms of market trends, as the company remains more speculative than giants like Nutrien and will require good execution to yield returns. But this at least confirms the global thesis.

The Space Narrative

RocketLab delivered a record quarter filled with confirmations of the growing and accelerating demand for space launch, systems and services, which is my thesis. We are talking about record launches backlog and the largest ever Neutron contract, which is a great confirmation as it is a direct competitor to SpaceX.

I closed my trade opened and shared on the Substack chat today at $98, returning 25% in three weeks.

Firefly also released a correct quarter with good demand, revenues and backlog but the market did not like comments about no NASA contract coming for the next three to six months… Except RocketLab, most of the sector relies mostly on public contracts so no new ones means no increased expectations and no re-ratings.

The bottom line remains positive for the sector as the interest continues to grow and I continue to be bullish, but still less than the AI hardware sector, meaning my liquidity will continue to go there more than anywhere else.

Transmedics’ Fiasco

I shared on my monthly review that I opened a trade on Transmedics; My thesis was that the margin narrative due to fuel prices was overblown and that I expected revenues to beat expectations. I was wrong.

Revenues fell in line with expectations but operating margins fell to 7%.

On a narrative that fuel prices would crush margins and cash generation, any misstep hurts badly, whatever the reasons. In this case, margins decreased as investments continue - on their clinical trials and expansion. Normal and healthy business is being punished. I talked about this potential scenario during my last quarterly review, and it materialized.

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?”.

I called that trade an everything or nothing trade. It’ll be nothing. I sized it to be minimal and not hurt me, so it won’t. I played and I lost, moving on!

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 just in terms of returns, but in terms of significance. It proved that my system works and helps avoid mistakes, it proved that focusing on strength is how money is made, it proved that leaving your personal bias at the door improves your returns.

It proved that this Substack has value.

If you were to look at the last months of content, since I refocused myself in January, I was bearish on 95% of the market and extremely bullish on the few names which had a strong price action and demand.

I highlighted liquidity and stock selectivity issues from the market in February, focusing myself on defensives; proven right in March as most of the stocks I did not want to be in fell off the cliff and are still underperforming today, which didn’t make me more bullish - with reasons. I got a bit more balanced in April as even though I called a potential bear market, I also highlighted that at least one sector was worth buying: photonics.

Everything wasn’t perfect, I won’t claim the contrary. But over the months, my content was crystal clear: focus on the names and sectors which would yield the most returns at the best risk/reward. Following those regardless of my personal opinion led me to share defensive assets until March - while social media were buying falling knifes, and buy the strongest sector then, yielding wonderful returns in April.

Everything is written, black and white.
Everything is timestamped.

I started to quantify returns based on my articles and alerts shared in the Substack chat - which you should join as I share both my long-term and swing plans, creating two equal-weighted portfolios. You’ll find the logic behind both strategies here. I believe this to be the best representation of my stock picking and therefore this Substack’s value.

The long term portfolio, which buys and holds my positions at healthy valuation and price action, is up 57.59% YTD generating 53.64% alpha.

My swing portfolio, which buys interesting assets based on price action and market strength with a pre-defined plan, is up 62.43% YTD, generating 52.40% alpha since opening.

My personal portfolio is up 107.58% YTD at the time of writing, a performance to be proud of for a public bear. Mistakes were made and it could have been better, but the markets are a place of constant improvement, so I’ll continue to improve my writing, my investing and my trading.

Leaps have already happened since I started writing less than two years ago. Many more are coming. There are no limits to what can be done in the markets if one accepts to learn. There are good and bad times, but no real limits. So, let’s keep grinding, improve the system, and always generate more alpha.

Before we start, if you haven’t read those articles yet, I really advise that you do so. I will try to have a consolidated write-up in the coming weeks with my entire investing process revisited, but until then, this is what you should read.

• The Art of Selling

• Systems, Stock Picking & Diversification

• The SaaSpocalypse is Justified

Once again, I started to share some of my swings and price action alerts on Substack chat, and I don’t think all of you are aware/present. So if you are interested in those real-time alerts, this is the place to be and the next step on this service.

Now, let’s review the month, and plan for the next.

Today’s Market

I have no concrete answer on what happened this month - no one does, but I can try to share my thoughts. The market is forward-looking and cares about future cash flow generation. Today’s situation isn’t better than the end of March, it is actually worse. The Strait is closed, oil is not flowing, its price is rising and the consequences are concrete, real and will be long. We’ll talk about this later.

On the contrary, we have seen the first steps of peace talks, and this is what the market focuses on. Stocks sold off as the war started, even before it. Stocks ripped off before its end. Markets’ anticipate. What they see is a conversation between the two main actors. The U.S. was able to reduce Israel’s aggressiveness towards Lebanon while Iran accepted to pause its nuclear program.

The takeaway is that parley is possible, concessions can be made.
They aren’t made yet, but they can, and the market anticipates it.

This is one part of the market’s excitement; the anticipation of a soonish return to normal as mom and dad could discuss their issues like adults and make the best decision for the kids. Seems like the discussion is a bit more complex than the market originally thought, but it is happening.

The second part of the trigger is the market’s realization that demand for AI services and hardware is really strong and will continue, strengthen, and is simply necessary, not because companies like to spend but because they must spend; something I was already sharing months ago, but the market doubted it.

But once again: AI demand is through the roof.

The necessary next step is optimized, more efficient compute, to help providers deliver more with less. This grows revenues and margins by reducing costs & will reassure the markets about the financial soundness of their expansion.

This next step requires scaling up AI infrastructure. The companies providing this hardware - like ALAB, are not impacted by the CapEx risks that Oracle and others suffer from. Their business is simply manufacturing hardware, selling it at high margins and generating cash right away.

Maximizing compute per unit of space & energy is core for any provider’s profitability. Demand for the hardware capable of doing so will grow because providers have no choice but to buy it. They are in a race against their own leverage & need efficiency to generate cash as fast as possible.

Every data set of the last months suggests this take was accurate. Last week’s and this week’s earnings confirm it - and you should read both reviews to understand why AI hardware is the place to be. Hyperscalers need to spend on more efficient hardware to help GPUs compute. We simply don’t have enough, and today’s infrastructure cannot meet demand.

A number of AI companies, including Anthropic, have faced a capacity crunch for computing in recent weeks, leading to price increases for access to AI processors, outages, and rationing.

The market finally realized this and pushed AI hardware names higher, setting a new ATH on Nvidia, a significant move which quarters of jaw-dropping results did not achieve. Very hard for me to be bearish on the sector considering this as I ask myself.

Why would investors buy Nvidia higher today, without any news, while those jaw-dropping quarters didn’t make them blink?

I have two answers. Either they didn’t believe it then and believe it now - my opinion as the market is also pushing new hardware verticals and not just trading winners. Or this is only traders centralizing liquidity on a name before making their final profit.

Tomorrow’s Market

If you look at the market closely, you’ll notice that one risk sector alone is responsible for most of the gains this year: AI hardware. The only other sectors outperforming are defensives - energy or materials. The rest is either flat or down. Selectivity isn’t the sign of a bull market.

From a macro perspective, the Strait is still closed, damages is growing each passing day and the pressure to reopen it must be stronger than ever. It is hard to imagine countries like China not pressing Iran’s government - or whatever is the chain of command in the country, to accept a deal and move on.

Still, the entire world is taken hostage by both Iran and the U.S. and a moment will come - sooner rather than later, where pressure will make one - or both, break. This situation cannot continue or it will push major world actors to take drastic measures which would have fewer consequences than continuing with the blockade.

What is obvious today is that damages will be consequential. I’ve already talked about them, and we must be ready for when they show up. Don’t be surprised when/if inflation picks up and risk assets nosedive as the market anticipates economic slowdowns. The chances of this happening are high after two months of disruption - and counting. Another example – besides the last one about fuel prices in Portugal, is Airlines in Europe canceling some of their flights due to fuel costs. Reserves and future purchases are now depleted or not sufficient to maintain normal business.

I continue to be bearish on most of the market. Nothing changed, or maybe, a lot has changed for the worse, macro-wise; despite the S&P 500 being at an all time high, I see no reason to be less bearish than I was a month ago.

But I know two things for sure.

1. I cannot be 100% confident in my bias. Macro is too hard to anticipate and there is a possibility for nothing to happen, for economies to absorb this disruption or only have limited impacts. Even if my confidence in inflation is sky-high, I cannot know when/if that will happen.

2. If I were right and it were to happen before the end of the year, I still couldn’t predict how the market will react and what the situation will be like. AI hardware names might continue to rocket with massive spending while the rest tanks. Who knows?

So I continue to follow my system, and buy strength.

As long as the assets I look at hold their W50 and bounce on it, with strong narratives and fundamentals, I will buy/hold. I have no reason to behave differently.

The only thing I do differently is liquidity attribution. There is strength in defensive assets and I started pushing more liquidity towards them. Those hedge my portfolio short-term and would yield results long-term if I were to be right. It’s all about finding a balance between concentration, performance and safety.

But always focused on strength.

Portfolio & My Focus

Today, I see strength in three verticals and own 9 positions (Soitec, Global Foundries, ARM, Silicom Ltd, Rocket Labs, Intuitive Machines, VG, Intrepid Potash and SolarEdge). I usually hold 5 positions max but Rocket Lab/Intuitive Machines and SolarEdge/VG have combined the size of one which really makes 7 positions, while my account is running ~50% margin, which justifies more diversification, while 100% of my margin is set up on trades with stop loss, to mitigate risk.

Overall happy with my portfolio, but I will continue to rotate liquidity as the market moves and creates new opportunities, if necessary.

AI hardware, my favorite.

Strength is undeniable and broad. Winners keep winning, with TSM and Nvidia at all time highs after years of constant demand growth and cash generation while new technologies have a clear tailwind as the market is realizing that compute is and will be limited for years by physical constraints – you can’t generate more energy with a snap, the only way forward is optimized compute.

This creates many new narratives, or “bottlenecks”, the new favorite word on social media, and many new opportunities as some technologies must be built from scratch, from optimized energy sources to optimized semiconductors, etc… Everything is new.

I do not think retail investors realize the opportunity in front of us. This is one of the best times ever to make life-changing returns, if you focus on the revolution in front of you… Ignoring this sector today is a stock picker mistake.

Let’s break down the winners and my view of each.

1. Photonics. We’ve been over the technology and its key players, this is a must-read, not to buy today but to be aware of the players and their fundamentals for when opportunities are given. To be ready.

   There are tons of ways to invest in photonics but most of those stocks have run too much to be safe risk/rewards so I’d personally wait for pullbacks, and I would make sure to buy only the good ones - the ones cited in my write-up. Poet taught us that many of them, the least necessary ones, will die slowly and be forgotten. Only invest in those who cannot be replaced. If photonics were to become the new normal, Soitec and GlobalFoundries cannot be avoided. They aren’t the only ones, but those are the right focus.

   As for buying, the rule is always the same: be patient and keep a close eye on the W50. That’s where I’ll buy mine, and that’s where most of my liquidity will be going if given the opportunity - especially in Soitec, but not only. And if there are great swing plays, I’ll share them on the Substack chat.

2. Optimizers. Photonics is a wonderful technology, but it isn’t ready, lots of R&D is ongoing and volume production hasn’t started. This is the future, but AI inference needs to be optimized today. Which opens other verticals.

   1. Engineering and design. Whether for photonics or not, everything starts by engineering. I have two names worth looking at on different verticals. Marvell for its focus on networking architectures, in high demand as bandwidth and latency are the current bottleneck; and ARM which is focused on optimized CPU architectures, also in high demand to orchestrate GPUs.

   2. Hardware optimizers. AI inference needs to be optimized today, that comes with solutions which already exist. My favorite remains Silicom Ltd and their Ditto hardware, but I also need to talk about AsteraLabs which I had to close due to price action, but proved my thesis right a few weeks later. I’d gladly buy back into that one on a healthy pullback. Credo is also on my list, while I also look at current leaders with MU, Sandisk and the likes.

3. Compute providers. Besides hardware, as compute is constrained, neoclouds have a major role to play. I was wrong on thinking Nebius would be punished for slower growth due to physical constraints - although the market is still punishing others for the same reasons, but I was also right to buy back in on price action signals - even if that trade was closed to concentrate liquidity into others a bit later. Both Nebius and CoreWeave are excellent buys in my opinion, on healthy pullbacks.

There are many more excellent names, great buys and great holds, some I haven’t had time to look at and others I know but cannot focus on. I’d love to buy them all but never will, concentration makes returns. Never forget that 80% of your returns come from your sizing, not your picking. You do not need to buy everything, you need to be laser-focused on a few and execute perfectly with large positions.

Putting 50% of your net worth into a safer risk reward which returns 30% will yield more than a 2% allocation into a moonshot pushing 200%. You can pick a 10x and make no money. It’ll give you a great dopamine rush but it won’t make you rich, and dopamine isn’t really going to change your life. If you want dopamine, play sports. Don’t chase it in investing, it isn’t the right place.

All stocks aren’t meant to be bought, and you shouldn’t always run behind the shiny new ones. I made multiple times more money in a month with a large allocation on Soitec than in a year and a half holding Palantir from $17 to $180.

If that doesn’t inspire you to focus on what really matters: sizing, nothing will…

Space, the rising one

I shared my thesis a few days ago so I won’t go over it again, nothing changed since except that most of those names gave strong pullbacks. I did accumulate LUNR and RKLB as shared in my chat, but they now need to confirm and push higher, otherwise those will just be fake breakouts and force me to close my positions.

I have a smaller attribution on this sector, I believe in it and in its potential, but I prefer AI hardware. I only own two names and both combined are worth a full position ~20% of net liquidity; I do not intend to grow it much further as long as the AI trade continues. But I do not want to have all my eggs in the same basket so this vertical gives me an interesting risk diversification.

Defensives, the real hedge

The real diversification is here, with a much bigger portion of my portfolio at ~40% net liquidity, concentrated in VG and Intrepid Potash, with a smaller allocation on SolarEdge.

This isn’t only about my personal bias on macro, this is also because both names have a very strong price action, and did bounce on interesting regions to accumulate. My bias helps but I am still following my system here, with IPI bought on a strong volume breakout retest close to its W50.

I believe those three are the best candidates as many others names like Halliburton or Schlumberger, and defensive ones like Darlings or Smith & Wesson have already returned what I expected them to. I also keep a close eyes on material names, with my favorite being Sigmal Lithium and would gladly buy a pullback, as the stock ran 53% since presented, two weeks ago...

My focus remains on AI, the next step to look up to is the resolution of the Middle East conflict which shouldn’t be long, and will be resolved before the consequences show up on macro, which would give a window for the market to focus on tech earnings and AI buildouts. Until things change. Hopefully, they won’t, but I will not ignore strength in defensive names.

This is the market as I view it today. Price action confirms that the leaders are in AI so that is where most of my liquidity is, while keeping my portfolio balanced with other narratives, including a defensive one.

My Guilty Pleasure

I opened a trade this week, one I shouldn’t have as it doesn’t fit my system. But sometimes, breaking the system can work and I accept to take this risk because of the interesting risk/reward profile, and the position my portfolio is in today.

Transmedics, again. I’ll make the thesis short. The company is excellent, with almost perfect results for a good year now, an excellent Q4-25, and yet the stock is down -18% year-to-date and no one cares about it anymore. I did close my position as I never hold weakness - and was right not to.

So why look at it? Because it is a name I follow since two years now, and one I know very well. The narrative today is that as fuel prices spike, Transmedics’ margins will tumble while management shared multiple times that they were passing most of the added costs to customers. The market is expecting $174.4M in revenues while Q1-26 was record-breaking in terms of flights and Europe is starting to scale; I personally estimate the floor to be ~$172M only in U.S. so revenues should be a beat - or so I believe. Margins won’t hurt in Q1 and management’s comments should help the market understand the fuel pricing dynamic.

The stock is trading close to its lowest multiples ever while getting ready to share what could be its best quarter ever, show Europe expansion, reassure the market on fuel costs and give information about their ongoing trials.

For once, I trust my gut more than the market, but I trust it with a reasonable position in July’s $150 option calls. My bet is that the market realizes its mistake and reprices the stock properly post Q1-26 this Tuesday.

To be crystal clear, if I am wrong, those calls are going to $0.

This isn’t a trade for the faint of heart. This is a highly speculative position based on my knowledge of the company and its dynamics - which I could misunderstand. I am well aware of the situation and accept to take the risks, but this is an everything or nothing earnings play.

Let’s see how that works out.

Looking Ahead

We are still in the heart of earning season and there will be much more to say, with an extremely busy week coming, important for the portfolio: Firefly Aerospace, Palantir, On, Fabrinet, Lattice, Global Foundries, SuperMicro, AMD, Transmedics, AsteraLabs, Arista, Lumentum, SolarEdge, Nutrien, ARM, COHR, CoreWeave, RocketLab, Iren… And those are only the portfolio-impacting ones, not including the ones I have an interest in.

Going to be busy and you can expect one or two reviews this week. I know those are less read but keep in mind that following sectors and competition/partners is key in tracking our positions. I wouldn’t write about earnings if they didn't matter, it certainly would save me tons of time!

To conclude,

This is it for me, there wasn’t much news this month nor subjects to dig. The market is the same, the conditions are the same, the risks and opportunities are the same. It doesn’t always have to be complicated, sometimes the best thing to do is just to continue doing what worked.

So that’s what I’ll be doing. Look at the names I mentioned in this write-up, open the positions on those retesting key supports, rotate liquidity by cutting losers fast if I have some, trim weakening winners... The usual.

With Nvidia pushing higher, I wouldn’t be surprised if this market had more juice in it for new records as liquidity concentrates into the most obvious trade: infrastructure. Once again, with today’s news and dynamics, I do not see reasons not to have most of our portfolios invested in this vertical. If you missed the train or don’t feel comfortable buying, just let it be, wait for your turn. There’ll be more occasions. Just remember for the next times. Why play the game in hard mode? Focus on strength at healthy price levels and let the market do the rest.

This is one of the most hated rallies I have seen. Don’t fight it, don’t try to assume it can’t run higher or longer. It can. The game is different when liquidity concentrates into a few names or narratives, and a hated rally can last longer than any bears would have assumed. It will end, no doubt here, but what we need to do is to be ready for when it does, not try to guess when or why.

I’ll continue to follow my system, share it with you regularly with articles, reviews and chats. Join us not to miss anything!

Another month starts. Let’s kill it.