San Francisco Daily 360

collapse
Home / Daily News Analysis / 5,000 investors still copy the AI trader who blew up

5,000 investors still copy the AI trader who blew up

Aug 07, 2026  Twila Rosenbaum 21 views
5,000 investors still copy the AI trader who blew up

More than 5,000 retail investors have decided to keep following the investment strategy of Leopold Aschenbrenner, the tech entrepreneur whose hedge fund nearly collapsed in July. Despite a devastating 67% loss at Situational Awareness, Aschenbrenner’s hedge fund, these investors have continued to mirror his public stock portfolio through an app called Autopilot. The decision has paid off so far, largely because the copycat approach does not use leverage.

Autopilot builds portfolios based on the disclosed holdings of well-known investors, fund managers and politicians. It lets ordinary retail investors track the trades and holdings of high-profile figures, adjusting their own portfolios when new regulatory filings are published. Aschenbrenner became one of the app’s most popular strategies shortly after his fund launched in March. By early last week, about $32 million had flowed into the Aschenbrenner strategy, according to Brian Schardt, Autopilot’s chief executive. When the crisis hit, Autopilot gave users a choice: stay with the strategy or leave. Between $3 million and $5 million left. Roughly $28 million, held by more than 5,000 people, remained.

The collapse of Situational Awareness

Aschenbrenner, a former OpenAI researcher, launched Situational Awareness earlier this year. The fund was built around a bold thesis: that the world is entering a period of explosive growth in artificial intelligence, and that the companies building the physical and digital infrastructure for that boom will generate enormous returns. He gained a following for his writing on AI and his conviction that the technology will reshape the economy. That conviction translated into a concentrated portfolio of AI infrastructure names, often held with borrowed money.

The strategy worked for a while. Autopilot’s version of his portfolio was up 53.4% since March, even after the July losses. But the leverage that magnified those gains also made the fund extremely fragile. When markets turned, the losses were not just large; they were violent. In July, Situational Awareness lost 67% of its value.

The collapse was driven by margin calls. When the value of borrowed positions falls below a certain level, brokers demand more capital. Aschenbrenner was unable or unwilling to meet those demands, so he was forced to sell most of his public holdings at a 10% discount to Citadel Securities, the firm run by billionaire investor Ken Griffin. A fire sale of that kind locks in losses and removes any chance of waiting for a rebound. It also creates a clear difference between the original fund and the investors copying it.

Why the copycats did better

The people copying Aschenbrenner through Autopilot have done significantly better than Aschenbrenner himself. While the fund lost 67% in July, the Autopilot version lost roughly 30%. Since hitting a low last Wednesday, it has bounced 35%. The gap comes down to one thing: leverage, or the lack of it.

Autopilot investors do not borrow, short, or trade derivatives. They simply mirror disclosed holdings and update when new filings appear. That means they did not face margin calls when the market dropped. Their portfolios fell, but they were not forced to sell at the worst possible moment. They were able to ride through the volatility, and when the same stocks rallied after Citadel bought Aschenbrenner’s book, they captured the rebound. That bounce may not fully erase the pain of the drawdown, but it explains why so many users stayed.

Schardt put it simply: “Ours is not leveraged, which means we don’t take the same risk that he took, which is better.” Unleveraged, the copy falls less in a rout and rises less in a rally. It is a less exciting ride, but for many retail investors, that is exactly the point. The idea of copying a brilliant fund manager is appealing, but copying his risk profile is not.

Still an all-in AI bet

The Autopilot portfolio is still a concentrated wager on AI infrastructure. It holds 14 stocks, with about 30% in the neocloud company Nebius and roughly 12% each in CoreWeave and Bloom Energy. Nebius, based in Amsterdam, provides cloud computing services targeted at AI workloads. CoreWeave is another cloud provider that has become a major player in the AI infrastructure boom. Bloom Energy builds fuel cells that can generate electricity, a crucial piece of the puzzle for energy-hungry data centers.

These names had a rough July. Concerns about how much big tech companies are spending on AI, combined with jitters in Asian markets, triggered a sell-off in the sector. Highly leveraged positions made the sell-off worse in Aschenbrenner’s fund. For the Autopilot investors, the losses were uncomfortable but survivable.

When Citadel bought Aschenbrenner’s public holdings, the same stocks rallied. That was partly because the market interpreted the transaction as a signal that sophisticated institutional investors saw value in the beaten-down names. It was also a simple supply-and-demand effect: a large forced seller had exited, removing downward pressure. Everyone who had held on through the crash was rewarded.

The portfolio remains vulnerable to the same swings. AI infrastructure stocks are volatile, and the debate over whether big tech is overbuilding AI capacity is far from settled. A concentrated portfolio of 14 names, with a third in a single stock, can move sharply in either direction. The 5,000 investors still in the strategy are making a conscious bet that Aschenbrenner’s thesis is right, even if his execution failed.

The next disclosure

The copying is unlikely to stop soon. Autopilot tracks quarterly 13F filings, which are public disclosures that large institutional investors must submit to the Securities and Exchange Commission. Because Aschenbrenner’s next filing is due by 14 August, followers sat tight through the crash. There was nothing to do until new information arrived, and for many, there was no reason to panic anyway.

Autopilot says it will keep copying the fund as long as it trades. The app’s model depends on the continuous flow of public filings, and Aschenbrenner remains a compelling figure for retail investors who want exposure to the AI boom without having to pick individual stocks themselves. If his next 13F shows that he has rebuilt positions in the same names, the copycat portfolio will likely follow.

There is also the possibility of a bigger development. If Anthropic, the AI company backed by Amazon and Google, goes public this year, one of Aschenbrenner’s biggest private bets would land in reach of the copycats too. Investors who are currently limited to his public stock holdings would be able to mirror his exposure to one of the most closely watched private AI companies. That could draw even more money into the Autopilot strategy.

Aschenbrenner’s next move

Aschenbrenner is not done. Since the collapse, he has wired $400 million into a Sequoia-backed company, according to reports. Sequoia Capital’s partners have said he “is not going anywhere.” That suggests he remains committed to his AI thesis and is willing to put fresh capital to work even after a brutal setback.

The move is a reminder that a trader’s public setbacks do not necessarily end his influence. Aschenbrenner has built a personal brand around the idea that AI will transform the world, and that brand has survived the fund’s implosion. For the thousands of investors still mirroring his portfolio, the story is not about a single bad month. It is about the broader direction of the AI trade.

There are real risks in that approach. Aschenbrenner’s own fund proved that leverage can destroy even the best investment thesis. The copycats have no leverage, but they are still exposed to concentration risk, volatility risk, and the possibility that the AI infrastructure buildout fails to generate profits. The next few months will be a test for both Aschenbrenner and the 5,000 people following him.

Autopilot has no plans to stop offering the strategy. As long as Aschenbrenner remains an active investor, his followers will have access to his disclosed bets. For now, neither are the 5,000 people trading his shadow.


Source:TNW | Insider News


Share:

Your experience on this site will be improved by allowing cookies Cookie Policy