FREE NEWSLETTER
BY NOW, YOU’VE probably heard the story of the 25-year-old wunderkind Leopold Aschenbrenner. After graduating as valedictorian from Columbia University at age 19, he worked for FTX, the crypto trading firm, then found his way to OpenAI, where he worked as a researcher for about a year, until mid-2024.
In the months after he left OpenAI, Aschenbrenner wrote a 165-page paper titled “Situational Awareness,” in which he detailed his views on the future of artificial intelligence. The paper was full of dramatic pronouncements—“the exponential is in full swing now,” he wrote—and ended up being shared widely online.
Capitalizing on that attention, Aschenbrenner established a hedge fund to make bets on the AI economy. He named the fund Situational Awareness, and at first, things went remarkably well. In its first two years, it grew to $45 billion in assets as he correctly identified some of the biggest beneficiaries of the AI build-out, including memory chip maker SK Hynix, fuel cell producer Bloom Energy and AI infrastructure provider CoreWeave. The fund also shorted traditional software company stocks, betting that AI would pressure their business models. And Aschenbrenner invested in some private companies, including Anthropic, the developer of Claude.
For a while, these bets worked out extraordinarily well. In its first two years, the fund reportedly gained 1,000%. The rest of Wall Street began to follow him closely. In a June profile, The Wall Street Journal wrote that his fund’s regulatory filings “are studied like scripture.”
But earlier this summer, both of the trends Aschenbrenner had been betting on reversed at the same time. Fears that AI infrastructure spending was becoming unsustainable led many of these stocks to fall 30% or more. And the traditional software stocks that Situational Awareness had been betting against—companies like Adobe and Salesforce.com—began to rebound, with some rising 20% or more.
Those reversals alone would have been a problem, but it turned out that Situational Awareness had also been borrowing on margin to increase the size of its bets. According to estimates, it was leveraged up to 400%. That led lenders to begin closing in.
It got even worse from there, when the fund’s high profile began to work against it. As it attempted to sell positions to reduce its debt, it got trapped. Because of the size of the orders it was placing, and their concentration among AI stocks, other traders were able to guess that Situational Awareness was the seller. That spooked investors, leading others to sell, thus compounding a downward spiral. In a letter to investors, Aschenbrenner compared it to a bank run.
Over the course of the next few weeks, as the fund’s assets dropped from $45 billion to just $10 billion, Aschenbrenner found himself with few options. At the end of July, he announced that the fund had sold virtually its entire portfolio of publicly-traded stocks to the investment firm Citadel. Because the positions were so large and thus difficult to sell on the open market, Situational Awareness was forced to sell them at what was reportedly a significant discount.
This story might not necessarily seem relevant for individual investors. But there are, I think, several conclusions to draw from this episode.
First, and perhaps most important, it’s a reminder that risk management should always come first. After so many years of market gains, it would be easy to become complacent. But it’s precisely when the market is doing so well that investors should be diligent in considering rebalancing.
This story also reminds us of the importance of diversification. To be sure, Situational Awareness made mistakes, but it also got one very important thing right: It was diversified. Though it had to conduct a fire sale of its publicly-traded holdings, it still holds a multi-billion-dollar stake in Anthropic. Without that, it might have faced total liquidation. The lesson: We should never go too far out on a limb with any investment idea.
British economist John Maynard Keynes was famous for his observation that, “markets can remain irrational longer than you can remain solvent.” In other words, for an investment to be successful, it needs to be correct and correct over the right timeframe. In an ironic twist, in the few weeks since Situational Awareness offloaded its holdings at a discount, many have rebounded. If it had been able to hang on a little longer, the fund might have been able to avoid the situation it was forced into. The lesson: Liquidity is important. This is one of the many reasons I recommend that individual investors avoid private funds—because an asset really only has value if you can sell it when you want to, or need to.
The Situational Awareness story also teaches us something about the narratives that surround the stock market. Because of the number of variables involved, it’s all too easy for market observers to paint virtually any picture they wish. And since no one has a crystal ball, no one can say that anyone else is necessarily wrong at any given time. Concerns about “circular” deals in the AI ecosystem have ebbed and flowed over the past few years, as have worries about the impact of AI on traditional software companies. The lesson: We should be careful to never worry too much about the news of the day because it’s often just that—today’s news, only to be replaced by a potentially different narrative tomorrow.
There’s an easy comparison between the events at Situational Awareness and the failure in the 1990s of the hedge fund Long-Term Capital Management (LTCM). Both got off to a fast start, both involved leverage and both were run by extraordinarily impressive individuals. At LTCM, two of the founders had Nobel Prizes. But ultimately, IQ doesn’t guarantee success. Nothing does. And that, I think, is another key lesson for investors to draw. In managing our personal investments, we should always look for ways to maintain a balanced, center-lane approach.
Adam M. Grossman is the founder of Mayport, a fixed-fee wealth management firm. Sign up for Adam’s Daily Ideas email, follow him on X @AdamMGrossman and check out his earlier articles.
