Leopold Aschenbrenner's biggest positions are getting hit again.
Since October highs: SanDisk down 19%. Micron down 14%. CoreWeave down 13%. Nebius down 11%. SharonAI down 30%.
SanDisk and Micron alone are 55% of the fund, per his Q2 13F.
In July, a drawdown like this forced him to sell his book to Citadel.
Now look, Aschenbrenner is arguably one of the sharpest minds in AI right now. Former OpenAI researcher. The guy who wrote the Situational Awareness paper that half of Silicon Valley passed around like contraband. His thesis on where AI is going is probably right.
And none of that saved him or matters, because markets don’t care about reality.
Because being right about what to own is a completely different skill from knowing how much to put in and when to get out. The ideas were fine. The process around them is where it broke.
55% of the fund in two correlated semiconductor names. No mechanical exit when the thesis stalled. The concentration worked beautifully on the way up and became a trapdoor on the way down.
This is what separates the legends from everyone else, by the way.
Buffett's first rule is don't lose money. Then back your best ideas heavily.
Druckenmiller has said openly that what matters isn't how often you're right, it's how much you make when you are right and how little you lose when you're wrong.
Rokos sizes by conviction and cuts the moment the thesis breaks, which is a little more shaky than others, but he likely has a tonne of macro-focused indicators in his modelling.
Tepper buys big when everyone else is panicking, but only once risk and reward are clearly in his favour.
None of them wins by predicting better. They win because the process around the bet is completely clear.
The overriding point here is this.
You are almost certainly making a version of the same mistake at a smaller scale.
Maybe not 55% in two stocks. But 20% in one, with no predetermined point where you'd admit the thesis is wrong.
Or a position that ran 40% and you watched it give back every penny because there was no exit trigger, just a vague feeling that it had further to go.
Or three holdings in the same sector that you think of as diversified but would all gap down together on the same headline.
You won't feel it until the drawdown is already in your account. By then the damage is done and you're sitting there, hoping the candle turns green, telling yourself you're a long term investor when really you just don't have a plan.
Aschenbrenner had a brilliant thesis and it still wasn't enough. Your thesis is probably less brilliant (no offence, mine probably aren’t either, apart from my bets on infrastructure). So the process matters even more.
I built a quick way to check yours.
The Process Score is 11 questions about how you actually invest. Takes about four minutes. Your report scores you across five layers (Filter, Enter, Size, Manage, Exit), names the one habit costing you the most with real numbers from real stocks, and gives you one thing to fix this week.
If you'd like to find out whether your process would survive a drawdown before it actually hits, instead of finding out the way Aschenbrenner did...
David
P.S. after the process score you’ll be asked to book a call. Do it. I want to talk to you about the future of AI and where it’s going. But tomorrow, we also have a group call on how to hedge political risk in the UK. Click here to add to your calendar >