How fast things have changed.
It took one month for the AI trade to go from the most profitable trade of the decade to a record-breaking drawdown, this time not on liquidations but on fundamentals. Whatever social media says, fundamentals did change, and we covered it many times this month.
Despites this rapid change of trend, we reacted fast: booked gains that would be gone by now, cut broken supports that kept falling since, and rotated into stronger sectors.
There’s always a bull run somewhere, the job is to be standing in it. The lessons of the past months saved us from a brutal performance. This month’s lessons will improve our returns for years.
Performance Review
The portfolio hurt. But far less than it could have, and is still performing extremely well given the pain out there.
2025 Returns: +39.5% vs S&P500 +16.4%
2026 YTD Returns: +70.5% vs. S&P500 +10.6%
Since inception (Jan-25): +139.3% vs. S&P500 +29.3%
The next few days matter, but the volatility and uncertainty today are unprecedented, so making plans is complicated. The best approach is to focus on the setups and tune out the noise, leave it to others.
AI is here to stay
That’s a fact. The question is: at what price?
My thesis from two weeks ago that funding was the real bottleneck and would hurt the whole AI chain is materializing. Every name I cut, in profit or not, is trading lower today, most down double digits like Nebius ~15% or Applied Opto down a staggering ~30% since I sold.
This is why I sell losers. Applied was a very large position, and holding it on conviction alone would have destroyed my portfolio and years of returns, even though my conviction was - is, extremely strong. But conviction doesn’t matter during a sector-wide sell-off. An expensive but necessary lesson.
The truth is that fundamentals changed. It’s hard to accept it today, and yet pretty simple to understand, whether the market is right or wrong.
Guidance and growth projections don’t matter if the AI data centers don’t get built. They don’t get built if they can’t be funded. And funding is extremely hard to find right now.
ARR and forward multiples don’t matter if they can’t be reached, and the market is pricing that they won’t. AI will remain, buildouts will continue, but that doesn’t mean stocks get rewarded. Fundamentals can improve and stocks be sold, this isn’t mutually exclusive… Sadly.
Acceleration is the only thing the market has paid for in two or three years; cash generation and stability didn’t matter. So if the one sector that had acceleration slows down, everything gets sold, conviction or not.
And it’s slowing down.
What comes next?
Very hard to forecast, so let’s stick to what we actually know.
Spending continues, at least for those with access to funds. Hyperscalers are diluting like never before: Amazon, Meta, Google and Microsoft all issuing new bonds, Meta pairing them with an ATM offering. AI will keep being built and these names will get tailwinds again. No one knows when, that’s the point. You could be liquidated ten times before then.
Macro is a mess, on two fronts we’ve covered.
Iran. Supposedly about to escalate into nuclear strikes and to stop immediately from one day to another, depending on who’s reporting, the moon cycle, and the alignment of the planets. Nobody knows, nobody will until official news lands, so don’t waste your time or money trying to predict it.
Mid-term elections. Trump lost a lot trying to “fix” the world in two years; the Iran war could be the straw that costs him Congress - but who knows. Votes are today, and the outcomes are wildly different: a Republican Congress could mean escalation in Iran - or not; a Democratic Congress could mean regulation on AI buildouts, a gridlocked government with Trump vetoing everything - or not. Any outcome could be positive, negative or neutral for the AI trade and the market.
The only guarantee is volatility. And uncertainty. Which ever outcome.
But the only truth in the market is price.
The importance of structures
If you’re still bullish AI and mostly invested there, one question: do you know how rare V-shape recoveries are at this point? Very. They exist, they happen. But they’re unlikely, and investing in the hope of one will cost you your wealth over the years, even if you pull it off once or twice.
Structures matter, and the current ones are ugly - disgusting really. They remind me of what cost me so much six months ago on a space name - right before its bounce I’ll admit. But I don’t invest for bounces.
So let’s look at ASTS, the space leader with a 1.7y bull run from breakout to top.
The chart tells the story on its own. You could add the June 12th SpaceX IPO, two weeks after the top - maybe a coincidence maybe not.
What matters is the structure: constant higher highs and higher lows held by the W50, then one breakdown in May 2026 that flipped the whole dynamic. After it came a dead-cat bounce and relentless selling, with the stock now 49% below the first close under its W50 in late June, and 74% below its ATH.
Let’s compare it with Nebius.
We’re already down ~15% since the first close below its W50. I’d expect a bounce back to the W50, and a rejection there. Like for ASTS. It’d be great news if it weren’t, if we could bounce and hold. But structures rarely lie, and rejection should be the base case. A bullish V-shape recovery shouldn’t.
The similarities between both names - and both sectors, should be concerning.
Many still doubt price action, but most who do are down a lot on Nebius and buying more. They’ll get a bounce - they always do, and use it as confirmation bias to hold, straight into the most likely scenario: rejection into lower prices. Maybe not. Maybe they’ll get lucky and see that once in a lifetime V-shape recovery. But the odds are against them, and investing that way, a moment comes where you lose everything.
Price action isn’t an exact science, but it maps demand zones for buyers and sellers; building around those zones raises your odds. No voodoo, just the visualization of what everyone is buying and selling. If 90% of participants sell a structure with a 90% success rate, you’d better not buy it.
I love Nebius and believe they’ll do great things; I closed it up 337% dollar-weighted and 589% from my first purchase, in a little over a year. This isn’t a bearish post, it’s grounded, methodical thinking. I wish I were bullish. But I expect a bounce and a rejection, and the only bull case is a perfect macro and geopolitical alignment. I can’t invest on that.
Not while other names and sectors are handing us perfect setups.
