The phrase 'situational awareness' implies knowing where you stand. Leopold Aschenbrenner, the 25-year-old founder of the hedge fund that bears that name, apparently knew where the AI revolution was headed. He did not know where his own portfolio could go.
The record is public. According to CNBC, Aschenbrenner's fund peaked at $45 billion in assets — a remarkable figure for a manager his age with, by most accounts, limited prior investing experience. The Wall Street Journal reported in June that the fund was up 1,000 percent in the two years since it opened, including 270 percent in the first five months of 2026. Podcaster Tim Ferriss called Aschenbrenner the 'Nostradamus of AI,' a man 'as close to clairvoyant as you could possibly be.' Investors studied his SEC filings to mirror his positions. His early backers included Patrick and John Collison, co-founders of Stripe, and Daniel Gross, who leads Meta's AI efforts.
The thesis was coherent, if aggressive. Aschenbrenner had published a 165-page paper in 2024 predicting that AI models from companies like OpenAI and Anthropic would arrive faster than most expected — and on that point, he was right. From that intellectual foundation he built a concentrated portfolio: long on data center builders like CoreWeave and Nebius, long on chipmakers like Micron Technology and VanEck Semiconductor, and short on software companies like Salesforce and Intuit, which he believed AI would render obsolete.
To amplify the returns, he took on an enormous amount of debt. On Wall Street, the term for this approach is 'going full Kelly' — constructing a portfolio aimed at maximum possible return with no meaningful allowance for the risk of loss. Cliff Asness, founder of AQR Capital Management, which manages roughly $254 billion in assets, put it plainly when asked about the episode: 'For the future, let's file the phrase 'You never go full Kelly' under the heading 'Things of which you should be situationally aware.''
In the span of two weeks in mid- to late July, the AI stocks Aschenbrenner held fell more than 30 percent. The software firms he had shorted rose by a similar amount. The debt he had used to magnify his gains now magnified his losses. The fund lost roughly $35 billion. The word 'hedge,' as the piece's author Joe Nocera notes, had rarely been such a misnomer.
Say it plainly: this was not a failure of the AI thesis. It was a failure of risk management — or rather, the deliberate absence of it. A leveraged, undiversified, one-directional bet on a single macro narrative is not a hedge fund. It is a wager. The distinction matters to anyone whose capital was inside it.
Follow the incentive, not the press release. When a young manager posts four-digit returns in a short window, the question is never only 'how did he make so much?' It is always also 'how much could he lose, and how fast?' The two questions are inseparable. The record of this episode suggests that the investors who pored over Aschenbrenner's SEC filings to copy his positions did not ask the second one loudly enough. Free markets are efficient at allocating capital — but only when participants price risk honestly. Leverage without a floor is not investing. It is speculation dressed in a prospectus.
