The Situational Awareness hedge fund, founded by 25-year-old former OpenAI researcher Leopold Aschenbrenner, sold the bulk of its public equity portfolio to Ken Griffin’s Citadel after a brutal month of AI infrastructure losses , but it deliberately held onto its private stake in Anthropic. In short: leverage forced a public-market retreat, while a still-appreciating private bet on one of the world’s most valuable AI labs was left untouched.
That single decision , sell the public names, keep the Anthropic shares , is the clearest signal yet of how sophisticated AI investors are separating short-term volatility from long-term conviction. Below, we break down exactly what happened, why it happened, and what it means for anyone watching the AI investment landscape in 2026.
What Is the Situational Awareness Hedge Fund?
The Situational Awareness hedge fund is an AI-focused investment firm launched in 2024 by Leopold Aschenbrenner, a former member of OpenAI’s “superalignment” team. The fund built its name , and its early returns , on a simple but aggressive thesis: scaling artificial intelligence would require a massive, sustained build-out of semiconductors, compute capacity, memory chips, and energy infrastructure.
That thesis wasn’t just a talking point. Aschenbrenner turned it into concentrated, leveraged bets on the companies positioned to benefit from the AI infrastructure boom , chipmakers, memory producers, clean energy developers, and cloud providers. For a while, it worked spectacularly well.
The Fund’s Meteoric Rise
Before its recent troubles, the Situational Awareness hedge fund had become one of the most talked-about names in AI investing:
- Returned 439% for the year through June 2026, according to the Financial Times.
- Assets under management reportedly peaked at around $45 billion.
- Attracted early backers including quant-trading firm Jane Street, Stripe co-founders Patrick and John Collison, and Meta executives Daniel Gross and Nat Friedman.
- Built its reputation on essays arguing that AI scaling would demand unprecedented infrastructure investment , a thesis that, for over a year, the market rewarded handsomely.
Who Is Leopold Aschenbrenner?
Question: What is Leopold Aschenbrenner’s background? Direct answer: Aschenbrenner is a German-born researcher who enrolled at Columbia University at 15 and graduated as valedictorian at 19, before joining OpenAI’s superalignment team in 2023 and later launching the Situational Awareness hedge fund in 2024.
He had zero prior trading experience when he started the fund. That detail matters, because it underscores how unusual his rise was , he went from AI safety researcher to managing tens of billions of dollars in under two years, largely on the strength of a widely circulated investment thesis rather than a traditional finance background.
His Departure From OpenAI
Aschenbrenner was dismissed from OpenAI in 2024 over what the company described as an improper disclosure of internal information. At the time, the superalignment team was co-led by OpenAI co-founder Ilya Sutskever and researcher Jan Leike. The fallout rippled across the AI industry: Sutskever went on to start his own AI company, Leike joined rival lab Anthropic, and Aschenbrenner used the moment to launch the Situational Awareness hedge fund.
Why Did the Situational Awareness Hedge Fund Sell Its Public Portfolio?
Question: Why did the Situational Awareness hedge fund unwind its public stock holdings? Direct answer: The fund used leverage , borrowed money , to amplify its bets on AI infrastructure stocks, and when those stocks fell sharply over the past month, the losses were magnified to the point that the fund was forced to sell its public equities to Citadel.
This is the core mechanic behind the story, and it’s worth expanding on, because leverage is the difference between a bad month and a fund-threatening month.
The Role of Leverage
Leverage lets a fund control a larger position than its actual capital would normally allow, borrowing money to buy more stock. When positions move in your favor, leverage multiplies gains , which explains the fund’s extraordinary 439% return. But when positions move against you, leverage multiplies losses just as fast, and lenders can force a sale if collateral value drops too far.
That’s precisely what happened here. As AI infrastructure equities sold off, the fund’s leveraged positions lost value faster than the underlying stocks themselves, triggering a forced liquidation of public holdings.
Which Stocks Got Hit Hardest
Some of the fund’s most damaging losses came from names directly tied to its “AI infrastructure” thesis:
- SK Hynix and Sandisk , memory chip producers, both down more than 30% over the past month.
- Bloom Energy , a clean energy developer tied to data center power demand.
- Nebius Group , a neocloud provider serving AI compute customers.
Public investors grew increasingly worried that massive AI capital expenditures weren’t translating into near-term revenue , a concern that hit infrastructure-adjacent stocks especially hard, and one that directly undercut the fund’s core thesis in the short term.
The Citadel Deal
Citadel, the hedge fund run by Ken Griffin, purchased the bulk of Situational Awareness’s public holdings after the losses mounted. This fits a recognizable pattern for Citadel, which has a reputation for stepping in to acquire attractive assets when leveraged players are forced to unwind. Notably, Citadel’s own portfolio already included some of the same AI infrastructure names, suggesting Griffin , like Aschenbrenner , still expects the sector to recover over time.
After the sale, the fund’s total assets fell to roughly $10 billion, down from around $20 billion in recent months and far below the reported $45 billion peak.
Public Sell-Off vs. Private Holdings: What Stayed and What Went
The table below summarizes the key difference between the assets the fund sold and the assets it kept , and why that distinction matters.
| Asset Type | Examples | What Happened | Why |
| Public equities | SK Hynix, Sandisk, Bloom Energy, Nebius Group | Sold to Citadel | Leveraged positions lost value quickly; forced unwind |
| Private stake , Anthropic | Anthropic shares (valued ~$5B) | Retained | Illiquid, no leverage/margin call risk, expected IPO upside |
| Private stake , chips | MatX (AI chip startup) | Retained | Not part of the leveraged public book |
| Private stake , infrastructure | Fluidstack (AI data center startup) | Retained | Same as above; private valuation, no forced selling |
The pattern is clear: everything that was leveraged and publicly traded got sold. Everything private , illiquid, unleveraged, and not subject to daily market pricing or margin calls , stayed exactly where it was.
Why the Situational Awareness Hedge Fund Kept Its Anthropic Shares
Question: Why didn’t the fund sell its Anthropic stake along with everything else? Direct answer: Private company shares like Anthropic’s aren’t traded on public markets, so they can’t be margin-called the way leveraged public stocks can , and with Anthropic reportedly gaining value ahead of a possible IPO, selling now would mean giving up substantial future upside.
This is arguably the most important detail in the entire story. The Situational Awareness hedge fund’s stake in Anthropic is currently valued at around $5 billion, according to Bloomberg. Anthropic itself was last valued at $965 billion in a Series H funding round in May 2026, and reports indicate the company could go public as soon as October 2026 , potentially at an even higher valuation.
Why Anthropic Shares Are Different From the Rest of the Portfolio
- No leverage attached. The fund’s public positions were juiced with borrowed money; the Anthropic stake reportedly was not, which means there’s no lender forcing a sale.
- No daily mark-to-market pressure. Private shares don’t move with every market swing, so there’s no margin call risk the way there is with publicly traded stock.
- Asymmetric upside potential. With an IPO on the horizon and a near-$1 trillion valuation already on the books, the position could appreciate substantially further , a windfall that could help offset the fund’s public-market losses.
- Strategic conviction. Anthropic represents the kind of long-horizon AI bet that first drew the fund’s early backers, distinct from the shorter-term infrastructure trade that just went wrong.
It’s worth noting that Situational Awareness also retained other private holdings, including a stake in AI chipmaker MatX and a position in AI data center startup Fluidstack, which was reportedly in talks earlier in 2026 to raise new funding at an $18 billion valuation. The common thread across all of these: private, illiquid, and untouched by the leverage-driven forced selling that hit the public book.
The Broader AI Capex Debate Behind the Sell-Off
Question: Why did AI infrastructure stocks fall in the first place? Direct answer: Public investors grew concerned that the enormous capital expenditures being poured into AI infrastructure , chips, memory, data centers, and energy , weren’t yet translating into near-term revenue, which triggered a broad sell-off across the sector.
This concern didn’t appear out of nowhere. For more than a year, hyperscalers, cloud providers, and chipmakers had been announcing ever-larger spending commitments to build out AI compute capacity. That spending was the entire foundation of the fund’s original thesis , the belief that scaling AI models would require a historic infrastructure build-out, and that the companies enabling it would be rewarded accordingly.
For a long stretch, the market agreed. But markets tend to reprice quickly once doubt creeps in, and a wave of investor skepticism about whether AI capex was generating proportionate returns hit infrastructure-adjacent names hard and fast. Memory chip producers, energy developers, and cloud infrastructure providers , precisely the categories the fund was most exposed to , bore the brunt of that repricing.
Why Leverage Made the Damage Worse
It’s worth restating the mechanics here, because they explain why an otherwise defensible long-term thesis produced such a fast, painful outcome. A leveraged fund doesn’t just lose money when its positions decline , it can be forced to sell at exactly the wrong moment, locking in losses that an unleveraged investor could have simply waited out. That’s the difference between “the thesis was wrong” and “the position couldn’t survive the drawdown,” and reporting suggests the latter is closer to what happened here.
Aschenbrenner’s Response to the Losses
Even as losses mounted, Aschenbrenner reportedly didn’t back away from his thesis. In a July 24 letter to investors, he characterized the sell-off as one of the best buying opportunities the fund had seen since early the previous year, and he invited clients to commit fresh capital starting August 1. According to reporting, that appeal didn’t attract the level of new commitments he had hoped for , a reminder that investor confidence can lag behind a fund manager’s own convictions, especially after a steep and highly public drawdown.
What This Story Means for AI Investors
The rise and partial retreat of this AI hedge fund offers a useful case study for anyone tracking capital flows into artificial intelligence. A few takeaways stand out:
- Leverage cuts both ways. The same borrowed-money strategy that produced a 439% return can just as quickly force a fire sale when sentiment turns.
- Public and private AI bets behave very differently. Publicly traded AI infrastructure names are exposed to daily sentiment swings, earnings pressure, and margin calls. Private stakes in companies like Anthropic are comparatively insulated from that volatility.
- Concerns about AI capex are real. Investors sold off memory, energy, and cloud names partly because of growing skepticism that massive AI infrastructure spending is translating into near-term revenue , a debate that’s likely to keep shaping the sector.
- Distressed-asset buyers are circling. Citadel’s willingness to absorb Situational Awareness’s holdings shows that other large investors still believe in the underlying AI infrastructure thesis, even if the timing was rough for the original holder.
- Not all “AI exposure” is equal. A fund can suffer major public losses while its most valuable single position , a private stake in a frontier AI lab , keeps appreciating in the background.
- Fund manager conviction doesn’t guarantee investor confidence. Aschenbrenner’s own belief that the sell-off represented a buying opportunity wasn’t enough to draw significant fresh capital, which shows that reputational damage from a public drawdown can outlast the underlying investment logic.
- Watch where forced sellers’ assets land. When a leveraged fund unwinds, well-capitalized buyers like Citadel are often waiting to absorb the positions at a discount , a dynamic worth tracking for anyone evaluating entry points in AI infrastructure stocks.
How This Compares to Other AI-Focused Funds
It isn’t the only vehicle built around a concentrated AI infrastructure thesis, but its trajectory , a rapid rise fueled by leverage, followed by an equally rapid forced unwind , is a pattern worth watching across the broader category of AI-focused hedge funds. Funds that combine high conviction with high leverage tend to produce the most dramatic outcomes in both directions, and the AI infrastructure trade in particular has proven volatile enough to punish that combination severely over the past month.
Frequently Asked Questions
How much did the Situational Awareness hedge fund lose?
Exact loss figures haven’t been disclosed, but the fund’s total assets fell from a peak of roughly $45 billion to about $10 billion after the public equity sale to Citadel , a decline driven by leveraged losses in AI infrastructure stocks rather than a single disclosed loss figure.
What is Situational Awareness’s stake in Anthropic worth?
The fund’s Anthropic stake is currently valued at approximately $5 billion, according to Bloomberg, and that figure could rise further if Anthropic completes an expected IPO as soon as October 2026.
Did Citadel buy Anthropic shares from Situational Awareness?
No. Citadel purchased the majority of the fund’s public stock portfolio, including holdings like SK Hynix, Sandisk, Bloom Energy, and Nebius Group. Situational Awareness’s private holdings, including its Anthropic stake, chipmaker MatX, and data center startup Fluidstack, were not part of that sale.
Is the fund still operating?
Yes. Despite the forced sale of its public portfolio, the fund continues to operate with roughly $10 billion in assets, retains its private company stakes, and its founder has publicly stated he still believes in its original AI infrastructure thesis.
The Bottom Line
The story of this AI hedge fund is really two stories in one. The first is a familiar cautionary tale about leverage: a young, high-conviction fund manager rode an aggressive AI infrastructure thesis to spectacular short-term gains, only to see those same leveraged positions unravel just as quickly when public sentiment shifted. The second story is quieter but arguably more telling , even in the middle of a forced liquidation, the fund held firm on its private stake in Anthropic, a bet insulated from margin calls and still pointed toward a potential IPO windfall.
For anyone trying to read the current moment in AI investing, that contrast is the real headline: public AI infrastructure trades are volatile and leverage-sensitive, while private stakes in category-defining AI labs remain a conviction bet that even a distressed hedge fund wasn’t willing to give up.