The story that ran through markets over the weekend was not an earnings result. It was the near-collapse of an AI hedge fund run by a 24-year-old, and on this week’s Switzer Show, Shaw and Partners’ Adam Dawes used it to make a point about how fast the AI trade can turn.
The fund is Situational Awareness, run by Leopold Aschenbrenner, a German-born former OpenAI researcher. According to reporting by CNBC and the Financial Times, it had grown to about US$45 billion, with leverage of up to 400%, before the late-July sell-off in semiconductor stocks triggered margin calls.
On 30 July it sold its entire portfolio of public shares to Ken Griffin’s Citadel in a single trade, and its assets collapsed to around US$10 billion, most of that a private holding in AI developer Anthropic that it kept. The fund has not confirmed the figures.
He then wrote a widely-read essay called “Situational Awareness,” raised money from backers including Stripe’s Collison brothers, and launched the fund named after it. Dawes’s line on the show: “it shows how quickly fortunes are made and lost.”
The damage, Dawes said, came from leverage. Because the fund was geared several times over, “some of these stocks were down three times the percentage that they would have been down normally,” which forced the margin calls. It was the gearing, not the AI theme itself, that undid it.
The blow-up landed in a broad AI and chip rout. The Nasdaq 100 fell about 10% from its record high, and semiconductor stocks fared worse, shedding more than a trillion dollars in value. Six of the “Magnificent Seven” have now reported earnings, leaving Nvidia, which reports on 26 August, as the one the market is waiting on.
Dawes told the show he still thinks the AI trade is “very much in train,” but that it is “long in the tooth,” and that Nvidia’s result will be the tell. Whether the trade has another leg, he said, will take “more than just good results.”
Adam Dawes was speaking on the Switzer Show. His views are his own and are general information, not financial advice.
