Former OpenAI Researcher Leopold Aschenbrenner’s AI Fund Liquidation: How Leverage Amplified the Crisis?

TradingKey - Once deemed one of Silicon Valley's most iconic AI investors, he has now become the most closely watched victim in the current AI market rout.

According to reports, Situational Awareness, an AI-focused hedge fund founded by former OpenAI researcher Leopold Aschenbrenner, has recently been hit hard by a sharp correction in AI-related stocks. Under leverage pressure, it was forced to sell a large amount of its public market holdings and seek new financing.

According to people familiar with the matter, the fund has been in continuous communication with existing investors and lenders in recent days to secure new funding, while also offering some investors the opportunity to acquire part of its assets. However, under mounting liquidity pressure, the fund ultimately chose to sell most of its public equity portfolio.

Multiple media outlets subsequently confirmed that Citadel, one of the world's largest hedge funds led by Ken Griffin, has taken over its roughly $16 billion public equity assets. The transaction, completed in less than 24 hours, is seen as one of the largest emergency stock asset transfers on Wall Street in recent years.

Just a month ago, the fund was still the brightest star of AI investing on Wall Street. Now, as the market enters a deleveraging phase, its sudden reversal has become a microcosm of the concentrated release of AI trading risks.

Why 439% Returns Reversed So Quickly

Leopold Aschenbrenner founded the fund after leaving OpenAI in 2024, naming it after his published essay 'Situational Awareness.' The paper boldly predicted the rapid arrival of AGI in the coming years and triggered a global AI infrastructure race, quickly establishing him as one of Silicon Valley's most watched AI thinkers.

After leaving OpenAI, he translated this thesis directly into an investment strategy, making massive bets on the AI supply chain, including infrastructure companies in computing power, cloud computing, memory chips, power, and data centers.

This positioning once yielded spectacular returns.

According to a July letter to investors, the fund's net return reached 439% as of the end of June this year, significantly outperforming the vast majority of its Wall Street peers, and attracted support from high-profile investors including Jane Street, Stripe co-founders Patrick Collison and John Collison, as well as Nat Friedman and Daniel Gross.

However, high returns also masked high risks. According to people familiar with the matter, to further amplify returns, Situational Awareness had long used leverage for its investments. While leverage helped the fund rapidly expand its gains when the AI sector was rising, losses were similarly magnified during the recent broad tech pullback.

Over the past month, global AI trades experienced a sharp correction. AI infrastructure sectors—ranging from GPUs and HBM memory to data centers, power equipment, and cloud computing—saw widespread and significant pullbacks, with many previously high-flying hot stocks plunging by dozens of percentage points in a short period.

As one of the most steadfast bulls on AI infrastructure, Situational Awareness bore the brunt of the hit.

According to the fund's latest portfolio disclosures, Oracle ( ORCL ), AMD ( AMD) have both seen cumulative declines of nearly 20% recently; Nebius ( NBIS ), Sharon AI ( SHAZ ), Bloom Energy ( BE ), and SanDisk ( SNDK) and other AI-related stocks have posted even steeper drops.

Meanwhile, some short positions in software stocks previously established by the fund also failed to effectively hedge the overall losses.

As asset values continued to shrink, prime brokers began raising margin requirements. People familiar with the matter said institutions including Goldman Sachs ( GS ), JPMorgan ( JPM ), Bank of America ( BAC ), and Citi ( C) subsequently assisted the fund in asset disposals to avoid a larger-scale, concentrated sell-off in the public market.

Ultimately, the majority of the public equity assets were acquired in block by Citadel, saving Situational Awareness from the risk of further forced liquidation.

Market participants believe this was also a key reason behind the subsequent notable rebound in the AI sector—as the largest potential seller exited the market, investors' concerns over liquidity risk eased.

Founder Remains Bullish on Long-Term AI Opportunities

Despite suffering a heavy setback, Aschenbrenner has not changed his long-term outlook on the AI industry.

In a letter sent to investors on July 24, he admitted that the fund failed to escape the recent market correction, particularly the significant volatility in Asian tech stocks. However, he believes the current pullback has instead created the most attractive positioning window in more than a year.

The letter noted that if investors have been waiting for an opportunity to increase their allocation, now is a moment worth considering. The fund is also allowing investors to add capital before August 1, hoping to bolster its capital base.

Aschenbrenner also expects that the AI industry could still see new catalysts in the second half of the year, including Anthropic's potential IPO and further progress in the commercialization of generative AI, both of which have the potential to reinvigorate market sentiment.

AI Trade Entering Deleveraging Phase?

The experience of Situational Awareness may also reflect that the AI trade is undergoing a deep correction.

Over the past year, global capital has continuously poured into the AI infrastructure sector—spanning GPUs, HBM memory, cloud computing, data centers, and power equipment—sending valuations climbing rapidly. Meanwhile, many institutions have also utilized leverage to expand their investment scale.

However, when stock prices suffered consecutive declines, highly leveraged strategies began to expose concentrated risks. Forced position reductions further exacerbated market volatility, forming a classic deleveraging cycle.

Nevertheless, in the long run, most institutions have not dismissed the development trend of the AI industry. Citadel's willingness to acquire large-scale equity assets also reflects that major institutions remain optimistic about the long-term value of AI infrastructure. It is just that after a rapid rally, the market is searching for a more reasonable valuation level.