Leopold Aschenbrenner once seemed to see further than almost anyone else. The 24-year-old former OpenAI researcher published a 165-page essay in June 2024 that laid out a startling forecast. AGI could arrive by 2027. Superintelligence would follow soon after. Trillion-dollar clusters would power the transformation. And the United States faced a national-security race against China that few yet recognized.
He called the essay Situational Awareness: The Decade Ahead. The title captured the gap he perceived. A few hundred people in San Francisco and the AI labs grasped the trend lines. Everyone else remained blind. “By 2025/26, these machines will outpace many college graduates,” Aschenbrenner wrote. “By the end of the decade, they will be smarter than you or I; we will have superintelligence, in the true sense of the word.”
Investors listened. They handed him money. His hedge fund, also named Situational Awareness, launched with roughly $225 million from backers including Patrick Collison, John Collison, Nat Friedman and Daniel Gross. Returns exploded. The fund posted gains of 439 percent for the year through June 2026 and 1,551 percent since inception, according to Futurism. Assets swelled past $13 billion early in 2026 and hit a peak of $45 billion this month before the collapse.
Then July arrived. AI-related stocks tumbled. Semiconductor names plunged. Aschenbrenner’s concentrated bets turned against him with brutal speed. The fund lost 67 percent for the month. Margin calls mounted. By late July assets had shriveled to around $10 billion. CNBC detailed the fire sale. The fund offloaded most public equity holdings to Ken Griffin’s Citadel at a discount in one of Wall Street’s largest and swiftest transactions of its kind.
“We let you down this month,” the fund told clients in a letter reported by Futurism. Aschenbrenner had earlier described the sell-off as a “particularly good time to add funds,” per the Financial Times. When fresh capital failed to materialize and losses deepened, he began calling investors frantically. Some could not reach him. “Leopold just stopped taking calls,” one told the Financial Times.
The reversal carries layers of irony. Aschenbrenner built his reputation on the claim that he and a small circle possessed genuine awareness of where AI headed. Markets, it turns out, demanded a different kind of awareness. One attuned to leverage, crowded trades and sudden shifts in sentiment. The fund had borrowed heavily. It ran with a skeleton crew of just four investment professionals and eight employees total, according to a regulatory filing cited by Futurism. Concentration in names such as SK Hynix, Micron, Sandisk and CoreWeave amplified the damage when those stocks dropped more than 35 percent in July.
Short positions in software companies that many feared AI would displace also hurt. Adobe rose instead of falling. The thesis that AI infrastructure would be the bottleneck proved directionally sound. Execution met the unforgiving arithmetic of borrowed money and a broad tech rout.
Aschenbrenner arrived at OpenAI in 2023 to work on the superalignment team. He focused on keeping future systems under control. The lab fired him in 2024 amid controversy over alleged leaks, a dispute he has rejected. Months later the essay appeared and spread through Silicon Valley, Washington and finance circles. It read like a manifesto. Compute scaling at 0.5 orders of magnitude per year. Algorithmic gains at another 0.5. “Unhobbling” that turned chatbots into agents. Another jump like the one from GPT-2 to GPT-4 would deliver AGI by 2027. From there hundreds of millions of AGI systems could automate research, compressing a decade of progress into a single year and producing superintelligence.
“The AGI race has begun,” he declared. “We are building machines that can think and reason.” National-security forces not seen in half a century would awaken. The U.S. might face an all-out race with China’s Communist Party. Or worse. Labs needed to lock down secrets. Government would have to get involved by 2027 or 2028. Superalignment remained unsolved.
The ideas resonated because recent progress had already surprised most observers. Trend lines that looked absurd a few years earlier kept proving accurate. A small group inside the labs had called the advances correctly. Aschenbrenner positioned himself as their interpreter to the outside world. “There are perhaps a few hundred people, most of them in San Francisco and the AI labs, that have situational awareness,” he wrote. “A few years ago these people were derided as crazy—but they trusted the trendlines.”
Wall Street treated the essay as something close to a prospectus. Money flowed. The fund bet on power, optics, storage, data centers and related plays. It later added shorts in semiconductors as the thesis evolved, according to analysis in a May 2026 Substack post. Early performance looked brilliant. Some holdings delivered returns of 100 percent to 800 percent. Capital poured in from limited partners impressed by the marks.
But markets rarely reward vision without discipline. Aschenbrenner had no prior professional investing experience. He operated the fund with minimal staff. Leverage magnified both the upside and the downside. When the crowded AI trade reversed, the structure cracked. Citadel stepped in as buyer of last resort. The fund retained its private stakes, including one in Anthropic, yet the public book largely changed hands in days.
Reports since the meltdown have mixed tones. Business Insider described it as the humbling of a wunderkind while noting that many of Aschenbrenner’s longer-term predictions on AI progress now appear remarkably accurate. The New York Times portrayed a young man who crisscrossed the country raising capital, received a cooler reception in New York where some viewed him as a one-hit wonder, yet still assembled billions. During one meeting a prospective investor pressed him on what would happen if AI failed to develop as expected. Aschenbrenner, the investor recalled, simply believed it would all work out.
That belief survived the summer’s losses. Aschenbrenner kept his stake in Anthropic. He has signaled continued conviction in the infrastructure buildout. Yet the episode exposes a tension at the heart of the current AI boom. Technical insight does not automatically translate into investment skill. The people closest to the labs often see the future most clearly on technology itself. Markets test other qualities: risk management, position sizing, liquidity awareness.
Few dispute that compute clusters continue to grow. Power demand rises. Model capabilities advance. The broader arc Aschenbrenner sketched may still unfold. But timing, leverage and market psychology matter. A 67 percent drawdown in a single month reminds everyone that even those who claim special awareness can lose sight of immediate realities.
Conversations on X in recent days reflect the schadenfreude and the caution. Users noted the timing overlapped with Aschenbrenner’s wedding. Others stressed that AI itself was not the problem. Leverage and volatility were. One post summarized the lesson bluntly: markets care less about hype than liquidity.
The fund’s rapid ascent from $225 million to $45 billion in under two years stands as one of the more extraordinary episodes in recent hedge-fund history. Its equally swift contraction offers a case study. Vision matters. So does structure. So does humility before forces larger than any single forecast.
Aschenbrenner once warned that most of the world lacked situational awareness about AI. The events of July suggest that awareness must extend beyond technology to the mechanics of finance itself. The people building these systems may indeed shape the decade ahead. Yet capital markets operate on their own unforgiving logic. Both truths now sit side by side, more visible than before.
WebProNews is an iEntry Publication