He Was Up 1,000%. Then July Happened.
Situational Awareness ran 4x leverage into the greatest AI rally ever recorded. It took one month for the market to take it all back, and Citadel was waiting at the bottom with a checkbook.
Here is the cleanest one-sentence summary of the entire past twelve months of markets: a former OpenAI researcher turned his AI thesis into a reported 1,000%-plus run, levered it four to one, and lost roughly 67% in a single July.
Leopold Aschenbrenner's Situational Awareness fund was the trade of the cycle on the way up. Long AI infrastructure, long the memory makers, short legacy software. For a year it looked like prophecy. Then Meta announced it would start selling surplus compute, TSMC guided capex to the moon, the SOX rolled over, and everything that made the fund brilliant made it uninsurable.
Per reporting from CNBC and others, the fund was forced to unwind its public book, roughly $16 billion of positions, and the buyer of the bulk of it was Ken Griffin's Citadel. Read that again. The most disciplined risk shop on earth got to buy the most concentrated AI portfolio on earth at forced-seller prices. The house did not just win, the house bought the other player's chips at a discount on the way out the door.
The degen lesson is ancient and nobody will learn it: leverage does not change whether you are right, it changes whether you get to stay at the table long enough to be right. SK Hynix can be a generational company and still cost you everything in four weeks if you are 4x long when the crowd leaves.
Days before the blowup, the fund was reportedly asking investors for more money. That is the part every retail degen should frame and hang above the terminal. It is never different this time. It was not different this time. It will not be different next time, when it will once again feel completely different.