A spectacular reversal at one of Wall Street’s most closely watched artificial intelligence hedge funds has delivered a blunt warning to investors chasing China’s fast-moving AI trade.
Situational Awareness, founded by former OpenAI researcher Leopold Aschenbrenner, lost 67 percent of its portfolio value in July 2026 after several heavily owned AI-related stocks fell sharply. The fund subsequently sold most of its public-equity portfolio to Citadel as falling prices, leverage and tightening collateral requirements placed its positions under increasing pressure.
The episode does not prove that artificial intelligence is an investment bubble or that the sector’s long-term growth story has ended. It demonstrates something more immediate: an investor can correctly identify a transformative technology and still suffer devastating losses through excessive borrowing, crowded positioning and inadequate liquidity.
That lesson is particularly relevant in China, where funds and retail traders have been pursuing many of the same semiconductor, cloud-computing and AI-infrastructure themes.
Situational Awareness Rose at Extraordinary Speed
Situational Awareness launched in 2024 and quickly attracted attention for its aggressive bets on companies expected to benefit from rising demand for AI computing.
The fund’s portfolio included businesses connected to semiconductors, memory chips, cloud infrastructure and data centres. Public filings showed positions in companies such as Broadcom, Intel and CoreWeave, while reporting also linked it to SK Hynix, SanDisk and Nebius Group.
The strategy initially produced exceptional returns. The fund gained 439 percent from the beginning of 2026 through the end of June, turning Aschenbrenner into one of the most prominent figures in the AI investment boom. High-profile backers reportedly included Jane Street, Stripe co-founders Patrick and John Collison, and technology investors Daniel Gross and Nat Friedman.
The rapid rise encouraged a perception that deep knowledge of AI development could be converted directly into superior investment performance. However, the fund’s dramatic July reversal showed that understanding a technology is different from managing a leveraged portfolio through a violent market decline. Details of the portfolio sale and earlier performance were reported by Reuters in its account of Citadel’s purchase.
Leverage Turned Falling Stocks Into a Crisis
Hedge funds frequently borrow money from prime brokers to increase their market exposure. Leverage can multiply profits when share prices rise, but it also magnifies losses when the same positions fall.
A fund with no borrowed money can often wait for a stock to recover. A leveraged fund may not have that option. Lenders require collateral, and when asset values fall, they can demand additional cash or securities. The fund must then raise capital, reduce positions or sell assets quickly.
Situational Awareness faced pressure to raise new money or unwind its holdings. It chose to sell most of its public-equity portfolio, with major banks helping facilitate the transaction with Citadel. The transferred assets included the portion of the portfolio financed through broker leverage.
Aschenbrenner told investors that worsening prices and declining liquidity made it progressively harder to keep the portfolio within its risk limits. He compared the dynamics to a bank run in which vulnerability created further vulnerability. The fund removed all leverage after the episode.
This is why leverage can make a temporary market decline permanent. Even when the underlying companies remain commercially strong, a fund forced to sell cannot benefit from a later recovery.
A 67 Percent Loss Did Not Erase Every Earlier Gain
The scale of the July decline was severe, but the fund reportedly remained approximately 80 percent higher for 2026 because its first-half gains had been so large.
That detail is important because the episode was not a conventional bankruptcy in which every asset disappeared. Reuters reported that Situational Awareness retained a portfolio worth roughly $10 billion after the transaction, including private investments such as its stake in Anthropic.
Nevertheless, the fund’s investor letter acknowledged that it had come too close to permanent capital impairment. It also admitted that the firm had let investors down during the month.
The distinction between surviving and managing risk well matters. Remaining positive for the year does not make a 67 percent monthly drawdown acceptable for most institutional investors. Such a decline can destroy confidence, trigger withdrawals and prevent a manager from continuing with a strategy long enough for the original thesis to succeed.
Chinese Investors Were Chasing Similar AI Trades
The warning resonated in China because the same AI supply-chain stocks had helped generate unusually strong returns for Asian hedge funds during the first half of the year.
Chinese and Asia-focused managers had increased exposure to memory-chip producers, server suppliers, data-centre companies and businesses expected to benefit from the expansion of AI infrastructure. Many of these positions became crowded as funds attempted to capture the same growth story.
The reversal was already affecting regional managers before the Situational Awareness sale became public. WT China Fund reportedly lost 17 percent through July 17 after gaining around 120 percent during the first six months of 2026. Keystone Investors’ hedge fund declined approximately 12 percent after rising 63 percent during the same earlier period.
Other Asia-focused managers also experienced losses as AI hardware shares fell and momentum trades reversed. A detailed overview published by The Edge Singapore noted that SK Hynix and Kioxia had each fallen about 28 percent during July at the time of reporting.
Crowded Trades Can Fail Even When the Story Is Strong
AI demand may continue growing for years, but that does not guarantee that every company connected to the sector will justify its valuation.
When many investors hold the same stocks, the trade becomes vulnerable to a sudden reversal. One disappointing earnings report, concern about cloud-company spending or change in semiconductor demand can prompt several funds to reduce exposure simultaneously.
Selling then creates additional selling. Falling prices increase volatility, prime brokers demand more collateral and leveraged investors liquidate positions. Buyers recognise that funds are under pressure and may wait for lower prices, further reducing liquidity.
Shanghai-based analysts quoted by the South China Morning Post focused on two lessons: borrowed money can turn ordinary volatility into forced liquidation, and entering a fashionable theme late may mean buying just as earlier investors are preparing to exit.
The technology can still succeed while the investment performs badly. The internet transformed the global economy, yet investors who purchased highly valued technology shares near the 2000 peak often suffered years of losses.
Expertise in AI Is Not the Same as Portfolio Management
Aschenbrenner developed a reputation through his understanding of advanced AI, computing requirements and the possible economic consequences of increasingly capable models.
That expertise may have helped identify companies positioned to benefit from the AI infrastructure boom. It could not eliminate market timing, liquidity, valuation and financing risk.
Portfolio management requires more than selecting promising businesses. It also requires deciding how much capital to place in each position, how correlated the holdings are, how much leverage can be tolerated and what happens when markets move sharply in the wrong direction.
A portfolio containing chipmakers, data-centre operators and cloud suppliers may appear diversified because it includes several companies. In reality, all of those holdings may depend on the same assumption: continued growth in AI investment. When that assumption is questioned, every position can decline together.
The Bigger Lesson for Chinese Retail Investors
Retail investors may not operate a hedge fund, but many can reproduce the same risks through margin accounts, leveraged exchange-traded products and concentrated purchases of popular technology shares.
The danger increases when recent performance is treated as evidence that the strategy cannot fail. A stock that has doubled can appear safer because its business story has attracted widespread support, even though the higher price may make it more vulnerable to disappointment.
Investors also need to distinguish between company fundamentals and the price paid for those fundamentals. A leading semiconductor manufacturer can continue increasing revenue while its share price falls because the market had expected even faster growth.
China’s AI sector may produce valuable businesses, but enthusiasm alone cannot determine which companies will generate sustainable profits. Competition, export restrictions, capital expenditure, energy costs and rapid technological change can all affect long-term returns.
The AI Boom Has Not Ended, but Easy Profits May Have
Situational Awareness remained optimistic about the underlying companies in its portfolio even after removing leverage and selling most of its public holdings. Some market participants have also argued that forced liquidation could mark a temporary low rather than the beginning of a prolonged AI collapse.
No single fund can determine the future of the entire sector. The sell-off may eventually look like a short-term correction within a much larger technological expansion.
Its warning is still difficult to dismiss. Strong conviction does not replace risk control, and a correct long-term forecast cannot protect a portfolio that is unable to survive short-term volatility.
For Chinese investors, the most valuable lesson is not to abandon AI. It is to separate belief in artificial intelligence from blind confidence in every AI-related stock. Concentration, leverage and crowded positioning can turn a promising investment thesis into a forced sale at the worst possible moment.
