AI Hedge Fund Defies Odds: From OpenAI Essays to $20 Billion Success

Betting against the obvious often yields the biggest rewards. That's what 24-year-old former OpenAI researcher Leopold Aschenbrenner demonstrated with his AI hedge fund, Situational Awareness. This fund has turned heads on Wall Street, managing a whopping $20 billion and gaining 270% this year through May. The strategy? Back electricity and computing power over the conventional favorites.

The Numbers Tell the Story

The fund's performance is staggering. If you invested in January, your money would have nearly quadrupled by spring. Unlike most investors jumping on the AI bandwagon by buying into tech giants, Aschenbrenner took a different route. His focus is on the physical constraints of AI, namely, electricity and computing power. He’s not betting on AI companies themselves but on those selling 'shovels' in the AI gold rush. Think energy providers and hardware hosts.

One of his biggest bets is on Bloom Energy, which makes fuel cells for on-site power generation. Alongside, he’s invested in CoreWeave, renting AI computing power, and former mining data centers now repurposed for AI workloads. These positions align with his thesis that the real bottleneck isn't in the chips but in the power needed to run them.

What If the Chipmakers Win?

Here's the twist. While Aschenbrenner bets on power providers, he's betting against the chipmakers. His fund has shorted Nvidia and a selection of chip stocks to the tune of over $3.5 billion. Why? Because chip prices assume perfect conditions, and he argues the real scarcity will be electricity, not semiconductors. But what if Nvidia and its peers continue to dominate the tech space? What if technological advancements in chip efficiency outpace power consumption growth?

If chipmakers advance faster than anticipated or if new energy solutions emerge, Aschenbrenner’s strategy could backfire. Such a scenario would undermine his shorts and possibly erode the fund’s gains. Moreover, if AI spending slows or power becomes more accessible and cheaper, his big bets could see significant drawdowns.

The Anthropic Bet

Aschenbrenner's largest position isn't a stock at all. It's a private stake in Anthropic, the company behind the Claude chatbot. He acquired this position in February 2025 when Anthropic was valued at $60 billion. Fast forward to May 2026, and Anthropic’s valuation skyrocketed to $965 billion after a new funding round. That's equivalent to a fifth of his entire fund, marking a significant concentration risk. What happens if Anthropic's valuation falters or the chatbot market experiences a downturn?

While Anthropic’s pre-IPO valuation suggests potential for more growth, private valuations often lack the liquidity and transparency of public ones. The fund's reliance on this single investment underscores a high-risk, high-reward gamble.

Our Take: Power Play or Pyrrhic Victory?

There's no denying Aschenbrenner’s approach has been lucrative so far. The data is unambiguous. However, the sustainability of relying on electricity as the bottleneck in AI's expansion is questionable. Power constraints may indeed limit AI's growth, but betting heavily against established semiconductor players is risky. If losses hold through the weekly close, initial gains could dissipate.

For now, Aschenbrenner's insights into AI's infrastructure needs have paid off. However, investors should watch closely as the space evolves. Does this fund signal a shift towards valuing the unseen cogs in AI's machinery? Or will his contra-chip strategy prove overly optimistic in anticipating a power crunch?

History rhymes here. Just as some miners made fortunes selling shovels during gold rushes, Situational Awareness may be onto something by focusing on AI's operational backbone. But, much rides on whether these 'shovels' truly outshine the 'gold'.