Leopold Aschenbrenner and the Collapse of Situational Awareness: How a Young AI Prodigy Lost Billions and Was Rescued by Citadel (The Lesson of Icarus)

In the ancient Greek myth, Icarus, son of the master craftsman Daedalus, receives wings made of wax and feathers. His father warns him: do not fly too low, lest the wings grow heavy with moisture, and do not fly too high, lest the sun melt the wax. Icarus, however, becomes intoxicated by the sensation. The freedom, the speed, the feeling that limits no longer apply to him drive him ever higher. The wax melts. He plunges into the sea.

The myth is not merely a story about a disobedient child. It is an archetype. It shows what happens when ambition exceeds measure, when success creates an illusion of invulnerability, and when the experience that teaches fear is absent. The same ambitions-to fly higher, faster, farther than everyone else-have repeated throughout human history: in conquests, in revolutions, in technological booms, and in markets. Today the latest proof is a young manager who, in less than two years, turned an essay about AI into a hedge fund with tens of billions of dollars under management, only to be forced within a few days to sell almost everything to Citadel.

Who Is Leopold Aschenbrenner - The Biography of the 24-Year-Old AI Prophet

Leopold Aschenbrenner was born around 2001-2002 in Germany into a family of physicians. He finished high school early, entered Columbia University at the age of 15, and graduated as valedictorian at 19 with degrees in economics and mathematical statistics. He declined a place at Yale Law School. He worked briefly at the Future Fund of FTX-the philanthropic arm of Sam Bankman-Fried-before the exchange’s collapse. He then joined OpenAI’s Superalignment team under the leadership of Ilya Sutskever and Jan Leike. In April 2024 he was dismissed. He himself claims the reason was a memo to the board concerning security and possible industrial espionage, rather than a leak of information, as the company maintains.

Two months later he published the 165-page essay “Situational Awareness: The Decade Ahead.” In it he predicted that superintelligent AI could arrive by 2027 and described with near-mathematical precision how computational power, algorithmic improvements, and energy requirements would multiply by orders of magnitude. The essay went viral. It was shared from Silicon Valley to Washington. Aschenbrenner seized the moment and founded the hedge fund Situational Awareness LP. Seed investors included Patrick and John Collison (Stripe), Nat Friedman (former CEO of GitHub), and Daniel Gross. The fund launched with a few hundred million dollars. Aschenbrenner had no prior professional experience in asset management.

By mid-2026 assets under management exceeded $20 billion, and at the July peak they reached approximately $45 billion. Net returns from the beginning of the year through the end of June stood at 439 percent. From inception the figure exceeded 1,000-1,500 percent. Investors included Jane Street. The fund became one of the fastest-rising stories on Wall Street.

Table 1: Key Figures of Situational Awareness LP (2024-2026)

The Strategy of Situational Awareness: AI Infrastructure, Concentration, and Excessive Leverage

The strategy was thematic and highly concentrated. Aschenbrenner did not bet primarily on the model creators such as OpenAI or Anthropic (although he held a large private stake in Anthropic). He bet on the infrastructure-the “picks and shovels” of the AI boom: energy, computing power, optical communications, storage, data centers, and Bitcoin miners redirecting capacity toward AI. Long positions included companies such as SK Hynix, Sandisk, CoreWeave, Nebius, Bloom Energy, Broadcom, and Intel. Short positions were taken against software companies he believed would fall behind. Options were used. Leverage reached approximately four times.

The objective was clear: to capitalize on “trillion-dollar clusters” and the geopolitical race for AI. The risks, however, were enormous. Concentration was extreme. Leverage turned any adverse move into an existential threat. Liquidity in some of these names was limited. Aschenbrenner had never managed money through a genuine financial crisis or a prolonged bear market. His entire career had unfolded in an environment of almost continuous technological ascent after 2020.

Table 2: Illustrative Key Portfolio Positions and the Effect of 4x Leverage (July 2026)

What Happened in the Final Days of July 2026: AI Stocks Collapsed and Margin Calls Hit

In the final days of July 2026 one of the fastest and most dramatic endings in the modern history of hedge funds unfolded. Situational Awareness-the fund of Leopold Aschenbrenner, which at the beginning of the month managed approximately $45 billion in assets and had generated a net return of 439 percent from the start of the year through the end of June-confronted a sudden and brutal reversal. Within roughly six trading days the public equity portfolio was almost entirely sold, and the fund’s value contracted to approximately $8-10 billion. The loss for July alone reached about 67 percent. The fund survived, but only thanks to an emergency transaction with Citadel.

To understand precisely what occurred, three elements must be examined simultaneously: what was happening in the market for AI-related equities, what Aschenbrenner had done with the portfolio, and what unfolded in the final 36-48 hours before the deal.

Why AI Stocks Collapsed in July 2026

In the first half of 2026 the AI theme had been the primary driver of nearly all gains in technology stocks. Investors had been willing to pay ever-higher valuations for any company connected to computing power, memory, energy, or data centers. The hyperscalers (Alphabet, Microsoft, Amazon, Meta, Oracle) were spending hundreds of billions of dollars annually on infrastructure. Chip and memory manufacturers-Nvidia, SK Hynix, Samsung, Micron, Sandisk-recorded record demand. Companies such as CoreWeave and Nebius, which provide AI cloud services, and former Bitcoin miners redirecting capacity toward AI, rose by tens and hundreds of percent.

Toward the end of June and the beginning of July, however, sentiment began to shift sharply. Several factors accumulated at once.

First, investors started to question the sustainability of the enormous capital expenditures. Alphabet raised its 2026 capital-expenditure guidance to $195-205 billion and, for the first time since its 2004 listing, reported negative free cash flow for the quarter. Tesla also warned of a “massive” year of capital spending. Meta and the other hyperscalers continued to increase budgets. The market began to ask: when, and whether, these investments would return as profits and cash flow. Free cash flow across the hyperscaler group was under pressure. Valuations of many AI infrastructure companies already embedded an almost perfect scenario of endless demand.

Second, news emerged of progress by Chinese AI models and chip-manufacturing equipment. This fueled fears that American companies might not maintain their technological lead for as long as the market had priced in, and that returns on the enormous U.S. investments could prove lower.

Third, the macroeconomic backdrop deteriorated. Oil prices rose amid geopolitical tension. Expectations appeared that the Federal Reserve might remain more hawkish. On 27 July a strategist at Citadel Securities even published a note forecasting a possible rate hike. Although the Fed kept rates unchanged at its meeting, the tone and dissenting votes heightened nervousness. Long-term yields rose, and risk assets came under pressure.

Fourth, and especially important, significant leverage already existed in the market. Many hedge funds and speculative portfolios had accumulated concentrated long positions in the same AI infrastructure names. When prices began to fall, forced selling became self-reinforcing. Liquidity in some of the smaller and more speculative names evaporated rapidly.

The result was brutal. The Philadelphia Semiconductor Index fell more than 20-28 percent from its June peak and entered a bear market. South Korea’s KOSPI, heavily dependent on SK Hynix and Samsung, plunged by roughly one-third. Individual names in which Situational Awareness was heavily exposed-Nebius, Sandisk, CoreWeave, SK Hynix-lost between 35 and more than 50 percent from recent highs within weeks. At the same time some software companies against which the fund had bet on declines held up better or even rose, further worsening the result.

What Exactly Leopold Aschenbrenner Did and Why It Led to a Margin Call

Aschenbrenner had constructed the portfolio almost entirely around a single thesis: AI would require vast quantities of computing power, energy, memory, and infrastructure, and the companies supplying these “picks and shovels” would capture the largest gains. He did not hold primarily the model creators, but rather memory manufacturers, AI cloud providers, energy companies, and former Bitcoin miners redirected toward AI. The portfolio was highly concentrated. According to public data and reports, key positions included SK Hynix, Sandisk, Micron, CoreWeave, Nebius, Bloom Energy, and others. There were also significant short positions against software companies he believed would lag.

The most critical element, however, was leverage. The fund employed borrowed capital from its prime brokers (Goldman Sachs, JPMorgan Chase, and Bank of America) in amounts that, according to various sources, reached approximately four times equity. This meant that a 25 percent decline in the core positions could wipe out 100 percent of equity. Aschenbrenner relied on the conviction that his thesis was so powerful, and the rise of AI so inevitable, that the market would not turn sharply against him. He had no experience with prolonged bear markets or forced liquidations. His entire professional history had unfolded in an environment of almost continuous technological ascent.

On 24 July, when the decline was already under way and his positions were losing value, Aschenbrenner sent a letter to investors. In it he noted that the fund had achieved a 439 percent net return through the end of June and more than 1,500 percent since inception. He acknowledged that he had not been immune to the latest technology sell-off, but stated that the sell-off created an “especially good opportunity to add capital.” At the end of the letter he invited investors to submit new subscriptions beginning 1 August. Some investors later said the letter had reassured them. In reality the fund was already under mounting pressure from the banks.

After 24 July the situation deteriorated rapidly. Prices of the core positions continued to fall. The prime brokers began demanding additional collateral-classic margin calls. The fund lacked sufficient free capital to meet the requirements without selling positions. Market liquidity in these names had already deteriorated. Attempts to sell gradually in the open market would have triggered an even steeper collapse and still larger losses.

Aschenbrenner and his team initiated urgent discussions with potential buyers-other hedge funds and investors-regarding the sale of parts or the entire public portfolio. The conversations were chaotic. Different terms were offered to different parties. At the same time the possibility of selling a portion of the Anthropic stake was discussed, but that transaction ultimately did not materialize. Late on the evening of 29 July and into the night toward 30 July the pressure reached a critical point. The banks insisted on an immediate reduction of exposure. The fund had no time.

In the early hours of 30 July Aschenbrenner spoke directly with Ken Griffin. Citadel agreed to purchase a large portion (according to various reports-the main leveraged portion or nearly the entire public equity portfolio of approximately $16 billion) at a discount of around 10 percent. The transaction was executed as a block trade before the market opened. Other funds, including Millennium, had also reviewed the portfolio, but Citadel was the fastest and largest buyer capable of absorbing such a volume of risk at once. After the deal Situational Awareness was left with approximately $8-10 billion in assets-primarily private investments, including the significant stake in Anthropic-and without leverage.

In the letter to investors at the end of 30 or 31 July Aschenbrenner wrote: “We let you down this month.” He acknowledged that the fund had come closer to permanent impairment of capital than was acceptable. He cited a loss of approximately 67 percent for the month, yet noted that the fund nevertheless remained up roughly 80 percent year-to-date. He promised to eliminate all leverage and to learn the lessons.

Table 3: Timeline of Events Surrounding the Collapse of Situational Awareness (July 2026)

The failure was not due to an incorrect long-term thesis on AI. Many of the companies in which the fund had invested continued to report strong demand. The failure resulted from the combination of extreme concentration, excessive leverage, and lack of experience with rapid, violent market reversals. When prices fell sharply and liquidity disappeared, leverage transformed a manageable correction into an existential threat. The attempt to raise new capital in the midst of the crisis failed to arrive in time. The attempt to sell positions gradually proved impossible without even greater losses. Only the emergency block sale at a discount remained.

This is the classic mechanism by which young, confident managers with brilliant upside track records and no scars from previous crises collide with the reality of markets. The leverage that yesterday multiplied profits today multiplies losses and removes control. Aschenbrenner had flown too close to the sun-with too much leverage, too concentrated a portfolio, and too little experience of what happens when the market refuses to cooperate.

The Principal Lessons from the Collapse of Situational Awareness

Youth. Arrogance. Excessive risk. Excessive leverage. Lack of financial discipline. And above all-lack of experience. Aschenbrenner had never witnessed a genuine financial crisis in his professional life. He had not managed money through 2008, 2011, 2018, or 2022. He had seen only ascent.

In this business the old saying from the front lines applies: beware of the gray-haired. That is where the young usually depart early. Experience is not theory. It is muscle memory of how markets can reverse in hours, how liquidity vanishes, and how the leverage that yesterday made you a genius today destroys you. Icarus flew high because he could. So did this young manager. The sun, however, always waits.

The story does not end here. The fund continues to exist, albeit in reduced form and without leverage. Anthropic remains. But the lesson has already been written. Ambition without measure and without the scars of previous falls remains the most dangerous combination in the markets.

Frequently Asked Questions about Leopold Aschenbrenner and Situational Awareness

1. Why exactly did Citadel buy the Situational Awareness portfolio rather than another fund?

Citadel under Ken Griffin possessed the scale, liquidity, and willingness to absorb an enormous volume of risk on short notice. The firm manages approximately $71 billion and has long experience with “rescue” transactions during market stress. Other players, including Millennium, also reviewed the portfolio, but Citadel was the fastest and largest buyer capable of closing the deal before the market opened on 30 July.

2. Is Aschenbrenner’s long-term thesis on AI infrastructure wrong?

Not necessarily. Many of the companies in which the fund had invested continued to report strong demand for chips, memory, and computing power. The problem was not the idea itself but the manner of execution-extreme concentration, excessive leverage, and the absence of buffers for a sharp market reversal. The thesis may prove correct over the long term, yet leverage turned it into an existential risk in the short term.

3. What remained in the Situational Awareness portfolio after the Citadel transaction?

After the sale of the greater part of the public equities the fund retained approximately $8-10 billion in assets. The bulk consists of private investments, including a significant stake in Anthropic. Public positions were almost entirely cleared, and all leverage was eliminated.

4. Could Aschenbrenner have avoided the forced sale?

In theory, yes-if he had maintained substantially lower leverage, if he had diversified more broadly, or if he had succeeded in raising fresh capital sufficiently quickly. In practice the combination of a rapid decline, evaporated liquidity, and margin calls from the three main prime brokers left little time. The 24 July letter describing the sell-off as an opportunity failed to attract the necessary capital in time.

5. What is the most important lesson from this episode for investors?

That brilliant upside performance and intellectual confidence do not substitute for experience with real market crises. Leverage multiplies both gains and losses. Young managers who have never lived through a prolonged bear market or a forced liquidation frequently underestimate the speed with which liquidity can disappear. In this business gray hair is often more valuable than the boldest forecasts.

Important notice

This article is for information and education only. It is not financial advice, an investment recommendation or professional advice of any kind. The views expressed are the author’s personal opinions and are not recommendations to buy, sell or hold securities, assets or investments. Investing involves substantial risk, including the possible loss of all capital. Past performance is not a guarantee of future results. Each reader should conduct independent research, consult a qualified financial adviser and weigh financial circumstances, risk tolerance and investment objectives before taking any decision. The author and borsite.net accept no liability for any loss arising from the use of the information in this article.