How Leopold Aschenbrenner's Thesis Became a Hedge Fund

Back in 2024, fresh off his firing from OpenAI, a 22-year-old published a 165-page essay arguing AGI was closer than anyone admitted. Then he built a hedge fund around it. Here's how Leopold Aschenbrenner did it and where his thesis stands today.

Situational Awareness: The Decade Ahead

It may be hard to imagine now, but in June of 2024 when Leopold Aschenbrenner published his set of essays titled Situational Awareness, few people outside of close AI circles had thought hard about the frontier described by the then 22-year-old. However, shortly after its publishment, the essay became a viral sensation. While AI safety researchers noted many of the ideas weren't novel, they nonetheless recognized the powerful articulation with which Aschenbrenner could project these messages to a wider audience.

Aschenbrenner argued broad points on the coming of AI superintelligence. On artificial general intelligence (AGI), or AI that can exceed the capabilities of a PhD-level human researcher, he claimed this would be upon us by 2027. However, he noted that to prepare for the advent of AGI, major infrastructural bottlenecks needed quick addressing within the US, things like energy and compute capacity. Finally, he raised alarm bells on the coming geopolitical race to wield Superintelligence (ASI) as a superweapon, focusing on the need for more intellectual security specifically against China.

Source: Situational Awareness: The Decade Ahead by Leopold Aschenbrenner

While he may have been a young little-known name at that time, Aschenbrenner's credibility as an ex-OpenAI Superalignment researcher meant his arguments were taken seriously by those in the public. His takes weren't just speculation.

Outside of broadcasting his messages on AI, Aschenbrenner at the same time decided to put his money where his mouth was. Just before the release of Situational Awareness, Aschenbrenner founded Situational Awareness LP, a hedge fund which he now heads in making bets directly related to his essays' theses.

Who Is Leopold Aschenbrenner?

Before discussing the contours of Aschenbrenner's investments, it's important to understand more about the eccentric and some would say brilliant individual who captured the attention of government officials and investors alike.

Aschenbrenner grew up in Germany, the son of two doctors and from a very young age already demonstrated superior intellectuality. Attending public school in Berlin, he skipped several class grades before graduating at 15 years old. Of his time spent in public school, Aschenbrenner has noted that he often felt his "weirdness" and "disagreeableness" was heavily discouraged. It's these qualities which he has since embraced and many, including himself, attribute to his success and brilliance.

After high school, Aschenbrenner moved to the US and enrolled at Columbia University, studying economics, mathematics and statistics. He graduated valedictorian of his class at just 19 years old and then left for economics research at Oxford.

From there, Aschenbrenner launched his career in the private sector after joining the FTX Future Fund team (the Effective Altruism project of another prodigy, Sam Bankman-Fried). It wasn't long before the fund was forced to shutter due to the collapse of FTX, but during his stint, Aschenbrenner remarks that he learned the importance of paying attention to a leader's character outside of their achievements.

Following the collapse of FTX, Aschenbrenner moved into work on AI, joining OpenAI's Superalignment team. The work Aschenbrenner did within this group in understanding how to steer and control AI systems smarter than any human directly impacted his essay writing and the thesis which now drives his fund, Situational Awareness LP. His early concerns about security led him in 2023 to write an internal memo arguing OpenAI's measures were insufficient, a memo which he believes led to his ultimate firing from that team shortly after.

From Essay to Hedge Fund

Just before publishing Situational Awareness, Leopold Aschenbrenner started a fund to make bets on his theses, a firm called Situational Awareness LP. Other commentators have noted that these bets are not directly on AI research and deployment. Rather, Aschenbrenner's focus so far has been investing in his convictions about bottlenecks, eschewing a Gold Rush "picks and shovels" strategy during the modern-day equivalent AI boom. This means for his fund, rather than investing in "gold" entities responsible for AI models or chips, Situational Awareness LP might be long the "picks and shovels" or power plants that are needed for running operations.

Outside his theses and his status as an AI savant, some might question whether investors would really trust a finance outsider with huge sums of their own money. So far, this has proven a non-issue. As part of an interview for Aschenbrenner's profile in Fortune, one investor, Graham Duncan, noted that Aschenbrenner reminds him of other famous finance contrarians, explaining that "If you want to have variant perception, it helps to be a little variant." The fund was seeded by big Silicon Valley names like Nat Friedman, Daniel Gross and Patrick and John Collison. In the first half of 2025, the fund returned +47% net of fees. As of Q1 2026, the notional 13F exposure has risen to $13.68 billion as compared to $254.8 million at the end of 2024.

While for some, investing in Situational Awareness LP is investing in Aschenbrenner and a belief that his perceived brilliance will translate into outstanding returns, for others it's an investment in the belief that Aschenbrenner's essays were correct predictions of AI's future.

What specifically has Aschenbrenner's "picks and shovels" approach meant for his portfolio composition thus far? Some have described his portfolio as having four pillars: energy, computing power, optical communications and storage. Ultimately, these areas are all related to the infrastructural bottlenecks that directly feed into the geopolitical AI race and the buildout of the "trillion-dollar clusters" Aschenbrenner was originally concerned about. Some of his fund's long bets include Bloom Energy (responsible for onsite power generation at data centers), CoreWeave (a cloud GPU infrastructure provider), Lumentum (optical communications infrastructure) and Core Scientific (a previous cryptocurrency mining operation that has shifted to data center operations).

And for the most part, Aschenbrenner's theses have played out quite well, with only a few caveats. One such caveat was the "DeepSeek shock" in 2025. While a large part of Aschenbrenner's bottleneck thesis pointed to compute as a major constraint on the further development of AI, the release of Chinese AI startup DeepSeek's highly efficient model demonstrated that it wouldn't just be whoever had the most GPUs that wins the AI race, it could be who uses them most efficiently.

Aschenbrenner, however, has repeatedly acknowledged himself the idiosyncrasies that may arise related to his AGI and more general AI predictions. He's stated that the "sequence of bets on the way to AGI matters." While the "DeepSeek shock" demonstrated how efficient use of GPUs could undermine the need for a rote buildout of training compute, it also demonstrated that at the same time need for inference compute would likely increase. When other investors panicked during that moment, Aschenbrenner and his fund sensed the selloff was an overreaction and bought the dip. In the fund's most recent Q1 2026 13F filing, some puzzled over the vast number of puts on chip designer names like Nvidia, Oracle, Broadcom and AMD. At the same time, others see the report as a cue that Aschenbrenner's thesis updates in real time; he could be hedging his positions or taking the view that chip designers have already priced in years of upside.

Aschenbrenner's Theses in 2026

While Leopold Aschenbrenner's investment theses might demonstrate an ability to change, how have his essays' claims held up since 2024?

On one hand, his prediction that a huge infrastructure buildout was both necessary and imminent has been validated beyond even his original guesses. Aschenbrenner had thought AI investment could hit $500 billion by 2026 and maybe $2 trillion by 2028. Independent estimates from 2025 already expected announced expenditures to exceed $5 trillion in datacenter capital expenditures through 2030. Amazon, Microsoft, Google, Meta and Oracle are projected to spend over $600 billion in capex in 2026 alone.

Source: "Situational Awareness, Two Years Later" by Omer Ansari

On the other hand, the groundbreaking prediction that we could have AGI by 2027 seems to be slightly overexuberant. While his ideas about increases in OOMs of algorithmic efficiency have mostly been validated, researchers have conflicting opinions in 2026 about when AGI may arrive. While companies like OpenAI have stated that they know how to build AGI, other research from researchers like the AI 2027 project predicts we may only progress to AGI around 2029-2030.

Finally, Aschenbrenner's claims about an AI geopolitical race have also been proven largely correct. Recently, researchers have claimed that Chinese AI companies are about six months behind American labs, a much narrower gap than in 2024. On top of the closeness of the race, the current US administration has begun taking the AI threat seriously. In 2025, the Trump Administration launched The Stargate Initiative to create a $500 billion public-private venture for the buildout of physical and digital infrastructure needed to support AGI.

Conclusion

Leopold Aschenbrenner's story is ultimately inextricable from his hedge fund. While the essay might have come first, the capital followed shortly after. This means Situational Awareness LP is not just a collection of stock picks dressed up in intellectual language. It is a published, falsifiable worldview made liquid, one that anyone can read and grade quarter by quarter.

Not everything has landed as written. AGI by 2027 now looks optimistic by a few years, and the compute bottleneck has proven more nuanced than raw GPU counts suggested. But on the questions that matter most for markets, like a historic infrastructure buildout, Aschenbrenner has been more correct than almost anyone in traditional finance. The fund's trajectory from $225 million to $13.68 billion in disclosed exposure reflects that.

Whether AGI arrives in 2027 or 2031, the decade he described in his essay on AI is already underway. The manifesto and the markets are telling the same story.