HomeArticle

25 years old, forced liquidation, lost 100 billion yuan before the wedding

投资界2026-08-02 09:44
It has more twists and turns than any movie.

At the end of July with warm market sentiment, global AI giants finally ushered in a sharp rally. In particular, SK Hynix's surge of over 28% brought a long-lost refreshing feeling to the market.

Unfortunately, that genius stock god failed to survive this market swing.

A few weeks ago, 25-year-old Leopold Aschenbrenner was still the widely talked-about AI prodigy on Wall Street. His hedge fund Situational Awareness LP, which heavily bets on AI-related stocks, soared all the way in the first half of the year riding the AI market boom, with a return rate exceeding 400%.

No one expected that the global AI market correction in the following July caught the highly leveraged Situational completely off guard. On July 30, Situational lost nearly 200 billion USD from its peak total size of about 450 billion USD, and was then forced to sell most of its public positions at a discount.

"We have let you down this month," Leopold apologized in a letter to investors.

Overnight, the fund shrank by 100 billion RMB. For Leopold, this moment is probably busy and full of mixed emotions: this weekend, he is holding his wedding with Avital Balwit, chief of staff to the CEO of Anthropic.

From Peak to Liquidation

The 25-year-old Genius Stock God Was Sniped

Leopold indeed has the qualifications to be hailed as a Wall Street genius trader: with only a few employees, he expanded a fund to a size of 3 trillion RMB in just two years.

Back at the end of 2024, Leopold, a researcher who was fired by OpenAI, chose to start his own business and established the hedge fund Situational, going long on energy and computing power infrastructure required for AI development while avoiding the crowded software application layer.

His investment map also expanded accordingly — regulatory documents show that as of the end of the first quarter of 2026, Situational's largest positions include Nebius, SanDisk, Micron and CoreWeave, as well as heavy positions in clean energy company Bloom Energy, SK Hynix, Oracle, AMD, while shorting software companies such as Adobe. In addition to these public stock positions, Situational also holds shares in Anthropic, chip startup MatX and data center enterprise Fluidstack.

In the first half of this year, the skyrocketing boom of the global AI industry was obvious to all, pushing Situational to the peak of public attention.

According to the Financial Times, Situational disclosed in a letter to investors in early July this year that the portfolio return in the first half of 2026 was as high as 439% — far exceeding the average level of Wall Street. Another CNBC report shows that the fund's AUM once climbed to 450 billion USD in early July.

The harsh reality is that making money in a favorable market trend and making money through a storm are two completely different capabilities. And heavy losses often occur right after people make easy money in a favorable market.

A closer look at the above heavy-weight AI industry companies held by Situational shows that they have been on a downward trend since late June, and the decline lasted until the end of July. For example, Nebius, SanDisk, SK Hynix, Micron and others all saw pullbacks ranging from 20% to 60% during this period.

On July 24, Leopold, who could barely hold on, sent a letter to investors: this is the best buying window since the beginning of 2025, and invited investors to inject new capital on August 1.

But Leopold never got the capital that could help him survive the storm.

On July 30, Situational was forced to liquidate all its public market stock positions, and the buyer was Citadel, also from Wall Street — this institution founded in 1990 has an AUM of about 710 billion USD. It is reported that within 24 hours, Citadel decided to buy most of the remaining 160 billion USD public market stock positions of Situational.

"The story of old Wall Street players swallowing young ones has staged again." In addition to Citadel, quantitative giant Jane Street, top hedge fund Millennium and others have also entered the closed bidding circle to discuss taking over the assets held by Situational.

After the sale to Citadel, Situational is expected to retain a private company investment portfolio of about 100 billion USD, including the equity of Anthropic.

It is thought-provoking that actually late on Wednesday, Situational had reached a 35 billion USD deal to sell Anthropic's equity to an investor consortium led by Greenoaks Capital and HSG, but the deal was finally cancelled on Thursday morning.

Failed by Leverage

The collapse of the masterwork of such a legendary Wall Street figure seems quite tragic in the current anxious market atmosphere.

But did Leopold really lose money?

Situational's investor letter shows that despite the heavy losses in July, the year-to-date return rate is still about 80%. According to previous reports from the Wall Street Journal, the fund's year-to-date return rate as of the end of May was about 270%.

Although the profit has shrunk significantly, Situational is still profitable. It is a pity that if he could have waited for one more day, Leopold would have earned far more:

The day after Situational's sale, the US stock market opened, and most of the underlying assets in its investment portfolio surged collectively. SanDisk, CoreWeave and Bloom Energy rose by 25.99%, 21.51% and 26.49% respectively. At the same time, SK Hynix's stock price increase also expanded to 30% on Friday, once hitting the limit up.

It is undeniable that the long-term direction of AI infrastructure that Leopold is bullish on is correct. Unfortunately, Wall Street does not reward correctness, it only rewards those who survive. One of the important reasons why Leopold failed to survive until dawn is that he bet too aggressively.

"When you are fascinated by a narrative, such as that AGI has arrived or is coming soon, some people will feel that the entire financial system is no longer important, and it is better to pull the leverage to the maximum. But people on Goldman Sachs' trading desk will say that these people are crazy, this is just the stock market," a Wall Street investor commented.

It is reported that Situational used leverage as high as four times — this strategy can rapidly expand the AUM in a unilateral bull market, but once the asset side encounters a black swan event or insufficient market confidence, it will quickly trigger margin pressure.

Unfortunately, in the past July, the AI sector that Situational heavily bet on just saw a sharp global correction. Goldman Sachs said in a prime brokerage report sent to clients recently that as of July 28, Asian fundamental long-short funds fell by an average of 18.6% this month. To make matters worse, the software stocks that Situational shorted rose in the opposite direction almost at the same time.

Caught between two sides, "At this leverage level, his liquidation was doomed, there is no other outcome," a market insider sharply commented.

It is worth mentioning that the 25-year-old Leopold probably has no time to review this storm carefully for the time being — this weekend, Leopold is about to hold his wedding. According to a Fortune magazine report last October, Leopold is engaged to Balwit, the only direct subordinate of Anthropic CEO Dario Amodei.

Revere the Market and Stay Restrained

After witnessing this fall, the question lingers even more: is there a bubble in AI investment?

Almost since the advent of ChatGPT in 2022, such concerns in the market have never stopped. The direct reason is that in the past few years, global capital has launched a vigorous round of value revaluation around AI:

From chips, servers, optical modules, data centers to storage, the market value of international giants such as Nvidia, Broadcom, Micron Technology has continuously refreshed records. Looking at the domestic market, from domestic large models, computing power infrastructure, robots to smart terminals, players including Zhipu, Changxin Memory Technologies, "Yizhongtian", Unitree Robotics continue to ignite the market boom, giving birth to companies with a market value of over a trillion RMB.

This surge is very reminiscent of the Internet craze more than 20 years ago. The experience people learned from that period is that every round of technological revolution is accompanied by a capital boom, and then it will inevitably face the cruel test of profit realization.

For example, the current market has begun to worry about the return on AI's huge capital expenditure.

Throughout July, giants including Amazon, Alphabet, Microsoft, Meta all released a highly similar signal: continuously raising full-year capital expenditure, with almost all funds flowing to AI servers, data centers, self-developed chips and storage production capacity. According to media statistics, the full-year capital expenditure plans disclosed by major giants in this July are close to 800 billion USD.

Of course, from another perspective, such a large-scale capital investment further demonstrates confidence in the AI revolution.

But from capital investment to technological breakthroughs, and then to profit realization, the time required is often longer than enthusiastic investment expectations. Therefore, the cycle of this round of AI investment will definitely be longer than many people imagine.

Perhaps Leopold did not misread the direction and had enough optimism and faith, but in addition to these, the ability to safely go through the ups and downs of the cycle is equally important — just as Wall Street never lacks bold traders, nor does it lack veteran traders, but there are no bold veteran traders on Wall Street.

This article is from the WeChat official account "PE Daily" (ID: pedaily2012), written by Feng Yuchen, and authorized for release by 36Kr.