AI Hedge Fund Blowup: $45B to $10B After 45x Leverage Bets

There is a special kind of humbling that only Wall Street can deliver, and this week it was served ice cold to Leopold Aschenbrenner, the 24-year-old prodigy behind Situational Awareness, an AI-focused hedge fund that peaked at $45 billion in assets before collapsing to roughly $10 billion after a cascade of margin calls forced a fire sale of its portfolio.

The fund's sin was not picking the wrong theme. AI is still the hottest trade in markets. The sin was concentration and leverage, reportedly as high as 45 times on positions including SK Hynix and SanDisk. When your thesis is right but your position sizing is suicidal, the market will find you eventually. Earnings season, with its binary surprises, is basically a leverage detector.

Enter Ken Griffin and Citadel, who stepped in to buy a chunk of the distressed portfolio at a discount. This is not the first time Griffin has played ambulance-chaser-with-a-checkbook, and it will not be the last. Situational Awareness's unwind also created collateral damage across AI-adjacent names, amplifying moves that were already volatile heading into a fractured Fed week.

The broader context makes this story more interesting than just a cautionary tale about a young trader swinging too hard. Bloomberg's coverage this week noted that hyperscaler AI revenue itself has concentration problems: UBS estimates OpenAI and Anthropic will represent 48% of Google Cloud revenue in 2026, and 69% of Microsoft Intelligent Cloud's year-over-year growth in 2025 came from OpenAI alone. So the fund blew up betting on AI infrastructure, in a market where AI infrastructure revenue is itself propped up by companies losing money. That is a lot of circular reasoning held together by vibes and venture capital.

To be fair, the bull case on AI capex is not dead. Amazon's AWS just posted revenue jumping over 37% to $42.2 billion, and the four largest hyperscalers have collectively committed trillions to AI infrastructure buildout. The thesis is not wrong. The leverage was just wrong enough to make it irrelevant.

Situational Awareness is now a case study in the oldest lesson finance keeps refusing to learn: being early and being right is worthless if you cannot survive long enough to collect.