Editor's Note: The Verge's commentary "The loss of Situational Awareness" not only discusses leverage but also points the finger at Silicon Valley's worship of the "young genius + grand AI narrative." The article argues that Leopold Aschenbrenner quickly gained influence, capital, and support from celebrity investors with a lengthy piece on AGI, but technical prowess, clout within the industry, and asset management ability are not interchangeable. When a small-scale, inexperienced team begins managing assets and leverage exposure that are not commensurate with their abilities, the risk is systematically amplified.
The author's key argument is that this crisis has exposed a typical organizational failure: individual prestige and industry endorsements have replaced scrutiny of investment capabilities, grand narratives have overshadowed risk control, and limited management experience has been rapidly magnified by external capital. The real problem with the fund may not necessarily be a misunderstanding of AI's long-term direction, but rather a lack of basic constraints on leverage, liquidity, concentration of holdings, and tail risks. The more compelling a viewpoint is, the more an organization needs to establish checks and balances to avoid transforming "conviction" into irreversible risk exposure.
The significance of this matter lies in the fact that AI investment is entering a phase where high capital injection, high valuations, and crowded trading coexist. The long-term demand for cloud computing, chips, and model companies may continue to grow, but once market expectations, funding conditions, or delivery pace change, the first to feel the pressure is often not the industry trend itself, but the asset prices built on high valuations and leverage. For investment institutions, true "situational awareness" is not about believing in a certain future earlier than others, but about being able to control risk while betting on the future and ensuring their survival.
The following is the original text:
I'm no finance expert by any means, but now I do have one piece of advice I can offer to those in the finance industry: when naming a hedge fund, never choose a name that will sound particularly ironic after the fund blows up.
For example, don't call it "Long-Term Capital Management"; don't call it "Amaranth Advisors" — a name derived from the amaranth flower, symbolizing eternal life. And certainly, do not name the fund "Situational Awareness." It sounds almost like "Hubris, Inc."
In any case, the hedge fund named "Situational Awareness," founded by a former OpenAI employee, with the founder only 24 years old, has a key focus on artificial intelligence. After enduring weeks of poor performance in AI stocks, the fund has sold most—according to some reports, all—of its publicly listed stock portfolio to Ken Griffin's Citadel.
Now, this is the situation that all of us have already "perceived."
You might recall earlier this week I mentioned that the market has become particularly nervous about AI-related risks. Now it seems we have finally found the institution that was swimming naked when the tide went out.
How bad is the situation?
According to CNBC, the fund was valued at $45 billion in early July. After selling assets to Griffin, it is now left with only $10 billion.
Previously, the record holder for the largest trading loss was Archegos Capital Management. According to The Wall Street Journal, the firm lost $8 billion in a period of 10 days in 2021. If the above figures are eventually confirmed, "Situational Awareness" will have lost three times that record in AI investments.
Every detail of this disaster is more absurd than the last.
"Situational Awareness" has a total of only eight employees, with four being investment professionals. CNBC writes, "According to regulatory filings, the fund's largest holdings as of the end of the first quarter included Nebius Group, SanDisk, Micron, and CoreWeave. The price of these four stocks has all fallen by more than 35% this month."
In the coming days, we will probably hear more news, especially from those Wall Street professionals who are taking advantage of this group of witty trading counterparts.
This investment firm is basically going to become an AI think tank
How did things come to this?
The name of Situational Awareness LP comes from a series of articles on machine intelligence, but with rather shallow content. The articles are written by the 24-year-old mastermind behind this fund—Leopold Aschenbrenner, whose name is also quite uncommon.
He wrote, "We are creating machines that can think and reason." This statement precisely reveals that he has no idea what "thinking" really means.
He also wrote: "By 2025 or 2026, these machines will surpass many college graduates. By the end of the century, they will be smarter than you or me; we will usher in a true superintelligence. In this process, a national security force not seen for half a century will be unleashed, and shortly after, the 'Project' will launch. If we are lucky, we will engage in a full competition with the CCP; if unlucky, a full-scale war will break out."
Part of me really wanted to start mocking his every assertion sentence by sentence, beginning with the first line of the article - "You can always see the future first in San Francisco." - but I decided to restrain this impulse.
This series of articles forms the theoretical basis of the hedge fund. Ultimately, its logic is: General Artificial Intelligence really exists (laugh), and it will arrive by 2027 (laugh to death). Therefore, the whole point of this fund's existence is to pour as much money as possible into AI stocks and then make a fortune.
"This investment firm will basically become an AI brain trust," Aschenbrenner said in a four-hour podcast interview with Dwarkesh Patel. A four-hour podcast, the favorite medium of expression for Silicon Valley elites.
He said: "Our understanding of the situation will far exceed those in New York managing funds. Our investment performance will definitely be outstanding. But equally important, this situational awareness will help us understand what is happening, become a rational voice in public discourse, and qualify us to advise others."
I cannot describe how deeply the terrible charts in these articles have impressed the 'certain type of Silicon Valley male'.
Axios once enthusiastically praised this series of articles, calling it "a useful and eye-opening comprehensive review of the high-level Silicon Valley discussion." Shav Vimalendiran, co-founder of the AI company SAMMY Labs, also wrote that Aschenbrenner's insights "have broadened my perspective as an AI practitioner beyond just the technical aspects." SAMMY Labs is an AI company trying to simplify legal texts.
Do you know who else loved this article? Ivanka Trump. She called it "a brilliant and important piece."
Why are these seemingly serious individuals investing in a 24-year-old's first hedge fund?
The 'situational awareness' backers include Stripe co-founders Patrick Collison and John Collison, as well as Meta AI executives Daniel Gross and Nat Friedman.
The fund's research director, Carl Shulman, previously worked at Peter Thiel's Clarium Capital. Later on, even Jane Street invested money. Jane Street is a prominent Wall Street trading firm and a place where many young effective altruists, including Sam Bankman-Fried, like to join.
The Wall Street Journal wrote in June of this year: "Jane Street's investment in 'sentiment analysis' is particularly noteworthy because the company rarely entrusts capital to external fund managers."
So why would these seemingly serious individuals invest in the first hedge fund managed by a 24-year-old? My most reasonable guess is that Aschenbrenner's investors rely on his accumulated social reputation and network endorsement.
Social proof is perhaps the laziest and most potentially disastrous way to vet someone. Just ask any Theranos investor; or go ask those who once entrusted their money to Bernie Madoff.
However, for Silicon Valley, this vetting approach seems to be good enough.
At the age of 17, Aschenbrenner was dubbed an "economic prodigy" by Tyler Cowen, a renowned libertarian economist in certain Silicon Valley circles. According to Fortune, Cowen's Emergent Ventures has also granted him funding. Aschenbrenner has also published articles in Works in Progress, a publication funded by Stripe. While at Columbia University, he co-founded the school's effective altruism club.
In 2021, Aschenbrenner, at the age of 19 and representing Columbia University graduates, joined the FTX Future Fund. This is the charity arm of the cryptocurrency exchange FTX. Later, Sam Bankman-Fried's fraudulent activities were exposed, leading to the collapse of FTX. His colleagues at the fund included the "philosopher king" of the effective altruism movement, William MacAskill, and Avital Balwit, who later became the chief of staff at Anthropic.
After leaving there, Aschenbrenner immediately joined OpenAI's super-alignment team. Fortune, citing several former colleagues, stated that he was "inept at handling politics and interpersonal relationships," and described him as arrogant and sharp.
《Fortune》 reporter Sharon Goldman wrote in a profile: “Several researchers also mentioned that at a holiday party, Aschenbrenner, during a casual group conversation, directly told then-CEO of Scale AI, Alexandr Wang, how many GPUs OpenAI had. In the words of one, he ‘just blurted it out’.” Both Wang and Aschenbrenner deny this ever happened.
Later, Aschenbrenner was fired from OpenAI for leaking internal information, although this incident was unrelated to the Scale AI matter. Two months after the dismissal, he published the series of articles titled “Perception of the Situation.”
Clearly, the first and second rule of business is not to get liquidated
Sharp-eyed readers may have noticed that Aschenbrenner had no prior experience in fund management on his resume when he started the hedge fund.
They may have also observed that Aschenbrenner's entire work history consists of a few months at FTX and approximately a year at OpenAI. They may wonder: did the Collison brothers, Gross, and Friedman have so much money to spare that they would fund this upstart?
The answer is clearly yes.
Tax the rich! In the still-idyllic year of 2024, Aschenbrenner described the fund's investment strategy in this way: “Clearly, the first and second rule of business is not to get liquidated.” He told Patel, “Timing is crucial. The betting order on the road to AGI is actually incredibly important. People underestimate this.”
In the coming days, we will probably have a fuller picture of how “Perception of the Situation” collided with a wall and went up in flames. But for now, it's roughly like this: hedge funds often amplify their investment scale through borrowing.
Therefore, if you truly believe that AI represents the future, as Bloomberg's Matt Levine wrote, you would not “put yourself and your investors 100% into the AI craze.”
You would put in AI craze an amount equal to 300% of the fund's size. This fund once claimed a return rate of 439%. But borrowing that can amplify gains will also magnify losses.
If you borrow money to bet on AI stocks, and those stocks start to plummet — as they have recently — the person who lent you money will issue a so-called "margin call," requiring you to top up your loan collateral.
According to the Financial Times, Situational Awareness initially held conference calls with investors and lenders, attempting to raise more funds. It even offered some investors the opportunity to buy parts of its portfolio.
Subsequently, Situational Awareness sold a large portion of its public stock assets to Citadel, owned by Griffin, this morning. The Financial Times reports that the size of this portion of the public stock portfolio originally amounted to $16 billion.
You have to manage the overall risk exposure very, very carefully
Publicly traded stocks are typically the most liquid part of any investment portfolio. According to the Financial Times, Situational Awareness still holds some undisclosed assets, including $5 billion worth of Anthropic shares. CNBC states that Situational Awareness is also in discussions to sell this stake, "but it's not clear whether the transaction has been completed."
In 2024, Aschenbrenner said, "You have to manage the overall risk exposure very, very carefully. If you expect these crazy events to happen, then just as surely, in real life, there will be some crazy things that you didn't anticipate."
One thing he may not have anticipated is that General AI simply won't arrive — at least not by 2027.
Aschenbrenner also said, "A friend jokingly said that this investment firm served as a perfect hedge for me personally. Either AGI comes in this decade, and my human capital depreciates as a result, but I can turn it into financial capital; or AGI doesn't arrive, the fund performs poorly, but I'm still in my twenties, and still smart."
Yes, Aschenbrenner is still in his twenties for sure!
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
