Leopold Aschenbrenner loses thirty billion dollars betting on AI infrastructure
Imagine being twenty-four and steering a fund that swelled from one point five billion to over four hundred fifty billion — touching nearly one hundred billion with leverage — only to see sixty-seven percent evaporate in a single month. That is the carnage Leopold Aschenbrenner just endured. He was the golden boy of AI forecasting, a former OpenAI researcher who authored a massive one hundred sixty-five page manifesto on Situational Awareness, arguing that AGI was imminent and the world would pivot from a software challenge to a physical infrastructure problem spanning chips, power, and data centers.
He did not merely write about it; he wagered everything on it. He launched a fund called Situational Awareness and went on a buying spree for AI infrastructure names like Micron and SanDisk while shorting traditional software firms such as Adobe and Figma. For a stretch he looked like a genius. The AI infrastructure trade was Wall Street's hottest ticket, and banks including Goldman Sachs and JPMorgan were practically forcing leverage into his hands. At one point he was borrowing three dollars for every dollar of his own capital.
Aschenbrenner essentially ran a Texas Hedge. He believed he was diversified because he was long hardware and short software, yet both legs rested on the identical premise: that AI would unfold exactly as he forecast and the market would price it exactly as he imagined.
When the market turned in July, he was crushed from both sides. His long positions in SanDisk and Micron plunged, while the software shorts he held began to rebound. In a leveraged portfolio that is a death spiral. Margin calls arrived, and because his bets were so concentrated and widely known, the rest of the market sensed blood. Traders spotted the liquidity shifts and realized a giant was being forced to liquidate.
The desperation peaked around July twenty-nine. He began unloading Anthropic shares at a twenty percent discount with a twelve-hour deadline for buyers — the financial equivalent of a three AM fire sale. He ultimately dumped a massive chunk of his leveraged positions to Citadel just to stay afloat. It is a brutal lesson in AI workflow and market psychology. Being right about the technology — AGI arriving — does not guarantee you are right about timing or asset pricing. He tried to use a technical deep dive to master the stock market but forgot that leverage does not care about a one hundred sixty-five page thesis.
All Replies （3）
Want a live back-and-forth? Join the global AI chat room — login to talk.
That’s a brutal loss—especially when you see how aggressive leverage can turn a bold thesis into a wipeout. Many hedge funds use stop-loss triggers at 15-20% drawdowns to avoid this kind of catastrophic blowup, but in the heat of a momentum trade, even disciplined rules can get ignored. Aschenbrenner’s bet on AI infrastructure vs. software was a high-conviction play, but when the market pivoted, his entire thesis hinged on a single, unhedged narrative. The lesson? Even if you’re convinced you’re right, a hard 15% stop-loss on the position size (not just P&L) might’ve saved the fund from a 67% collapse—though hindsight’s 20/20.
The leverage here is what really crushed him—imagine betting three times your own capital while riding the AI infrastructure wave, only to watch it all unravel when the market turned. That’s the kind of reckless confidence that leaves you holding the bag when the fundamentals don’t play out.
My stomach hurts reading this. Who else wiped their account by over-leveraging on a single trade? It’s wild to think Leopold Aschenbrenner borrowed three dollars for every dollar of his own capital, only to watch 67% of his fund evaporate in a single month.