Author: Long Yue, Wall Street Insights
A 24-year-old young man raised $45 billion on Wall Street with a long essay predicting how AI would change the world, and exploded by 439% in the first half of this year. However, during the recent one-month plunge in the AI sector, he lost the myth he had built over two years in one fell swoop.
On July 30, a piece of news circulated on Wall Street:
A hedge fund with only 8 employees sold its entire public market holdings—approximately $16 billion in stocks—to a counterparty in a single package transaction over the past 36 hours.
The buyer was Citadel, one of the world's largest hedge funds, founded by Ken Griffin, with $71 billion in assets under management.
The fund forced to sell off is called Situational Awareness (SA), founded by Leopold Aschenbrenner.
It is worth mentioning that Aschenbrenner's fiancée, Avital Balwit, is the Chief of Staff to Anthropic CEO Dario Amodei. This AI crash and his forced liquidation happened just before their wedding.
And just weeks ago, he was the most sought-after name on Wall Street and in Silicon Valley.
Silicon Valley's 'AI Prophet'
Leopold Aschenbrenner, German, 24 years old.
Entered Columbia University at 15, graduated top of his class at 19. After graduation, he joined OpenAI's "Superalignment" team, focusing on making superintelligence align with human values. In 2024, OpenAI fired him for "improperly leaking internal information." He denied this, saying he had only sent a largely non-confidential planning document to external researchers for feedback, and the real trigger was an internal memo he wrote criticizing OpenAI's safety measures.
After being fired, he did something many didn't expect—he wrote a 165-page long essay titled "Situational Awareness."
The essay's core thesis was: AI development is far faster than the outside world imagines; by 2027, AI will be able to conduct AI research autonomously, and humanity will face true superintelligence. All of this requires explosive expansion in chips, memory, data centers, and power infrastructure.
The article caused a sensation in Silicon Valley. Some called it "the most important article in a decade." Podcast host Tim Ferriss called him "the Nostradamus of AI" (similar to a prophet, seer).
In 2024, with approximately $225 million in seed funding, he founded the namesake fund SA. Investors included Stripe co-founders Patrick and John Collison, former GitHub CEO Nat Friedman, and institutions like Jane Street.
With no professional experience, he directly began managing billions of dollars. The fund's size rapidly swelled, peaking near $45 billion in early July 2026.
The entire fund had only 8 employees, with 4 being investment professionals.
'AI Stock God' Bets Big on AI with 4x Leverage, Gains 439% in Six Months
SA's investment logic was very straightforward: since AI requires massive computing power and infrastructure, heavily invest in companies providing that computing power and infrastructure.
Long positions: SK Hynix (Korean semiconductor/memory), Nebius (AI cloud platform), SanDisk (storage chips), Micron (memory chips), CoreWeave (AI cloud computing), Bloom Energy (energy).
Short positions: Shorting traditional software companies like Adobe—with the logic that AI would disrupt them.
Simultaneously, the fund amplified returns by borrowing from banks (i.e., leverage) at a leverage multiple of approximately 4x.
This structure had a feature: banks held the physical stocks, while SA synthetically obtained economic exposure and leverage through "Total Return Swaps" (TRS). The benefit was multiplied gains; the downside was that if stock prices fell, banks would demand margin calls.
From 2025 to the first half of 2026, this strategy worked astonishingly well.
By the end of June 2026, SA's net return rate reached 439%. It was reported that the fund had generated cumulative returns exceeding 2000% since inception. The fund's size grew from less than $1 billion initially to a peak of about $45 billion in early July.
Some early investors saw returns as high as 200% this year.
Getting Hit on Both Ends Within a Month
However, the good times didn't last. In July, AI stocks began to fall.
The market began questioning: Can the high valuations of AI infrastructure be sustained? Massive funds flowed out of AI hardware stocks.
SA's long positions were hit hardest:
-
Nebius fell about 48% from its June high, erasing about $35 billion in market cap
-
SanDisk plunged over 56% in just over a month
-
CoreWeave, Micron, SK Hynix all fell over 35% this month
Meanwhile, the software stocks SA was shorting (like Adobe) rose against the trend, causing losses on the short side as well.
With both long and short legs losing money, the 4x leverage magnified the losses. Banks began issuing margin calls.
This is what Wall Street calls a "margin call": you borrowed money to buy stocks, the stocks fell, the bank demands you post additional collateral, or they will force a liquidation. It's the same logic as a margin call in A-shares.
SA's prime brokers included Bank of America, Goldman Sachs, and JPMorgan Chase, all simultaneously involved to help the fund manage the margin calls or orderly reduction of positions.
Interestingly, on July 24, Aschenbrenner wrote in a letter to investors: "There are times we specifically point out that now is a good time to add to positions—if you've been waiting." The postscript invited investors to add capital on August 1.
36 Hours, One Night Auction
The crisis evolved rapidly within 36 hours.
Late night on July 29, SA urgently reached out to multiple potential buyers, seeking to sell over $10 billion in stocks to stabilize the situation. This news was widely reported by media such as the Financial Times and Bloomberg.
Bidders included: Millennium Management (a global top multi-strategy hedge fund), Jane Street (one of SA's investors).
Millennium made an offer but considered the portfolio too risky and was unwilling to pay a high price.
Ultimately, Citadel won.
In the early morning of July 30, the deal was completed. Citadel took over SA's entire public stock holdings—both long and short positions—valued at approximately $16 billion, at a significant discount. This was one of the largest emergency stock transactions in Wall Street history.
After news of the deal completion spread, the Nasdaq 100 index surged over 3% that day—the market breathed a sigh of relief: SA wouldn't need to dump more shares into the market.
This wasn't the first time Citadel had done such a thing. Historically, it had taken over assets of companies like Enron at low prices during crises. In Ken Griffin's own words, one reason for Citadel's sustained success is: "Experience. Wisdom earned through losses and pain. My leadership team, we've been through many of the market's toughest moments together, learned many painful lessons. But that makes us more effective investors during times of turmoil and crisis."
Citadel was founded in 1990. SA was founded in 2024.
What Remains
After the transaction, SA's assets under management plummeted from a peak of $45 billion to about $10 billion, a drop of over half.
But Aschenbrenner wasn't completely wiped out.
SA retained all its private investments, worth about $10 billion, the most important piece being equity in Anthropic, valued at approximately $5 billion. SA will continue operating as a private investment firm.
Whether these private assets can be realized depends on the valuations of Anthropic and other companies when they eventually go public.
This remains an open question.
Incidentally: Aschenbrenner's fiancée, Avital Balwit, is the Chief of Staff to Anthropic CEO Dario Amodei. SA was previously rumored to be selling its Anthropic stake, but an SA spokesperson denied this.
Where He Went Wrong
The most thought-provoking question about this incident is: Was Aschenbrenner's judgment about AI wrong?
Not necessarily.
The long-term demand logic for AI infrastructure is still widely acknowledged by many institutions.
The problem lay in the portfolio structure.
On X, someone quoted Ken Griffin's previous description of why fund managers fail: "Your portfolio is highly concentrated, with enormous positions, yet you cannot clearly articulate the competitive advantage behind holding these positions."
This was precisely SA's problem: heavily concentrated in a single theme (AI infrastructure), employing 4x leverage, and betting both long and short on two sides of the same narrative. Once market sentiment shifted, there was no buffer.
This structure could generate a 439% return during an uptrend, but could also blow up the entire public portfolio within weeks during a downturn.
As someone summarized: "His judgment about AI wasn't wrong. He was wrong in how he constructed the portfolio."
The Other Side of the Myth
Aschenbrenner's story is one of the most extreme samples of this AI bull market.
One article, one narrative, eight people, $45 billion, a 439% return—and then, within a month, brought down by margin calls.
This is not the end of AI, nor the end of Aschenbrenner.
But it clearly illustrates one thing: Being right about the direction doesn't mean you'll survive until that direction materializes.
