A 24-year-old's fund bets $9 billion on AI's energy needs

A $9 billion fund managed by a 24-year-old former OpenAI researcher is capturing attention by betting against the mainstream AI trade, loading up on energy and infrastructure stocks while reportedly shunning chipmaker Nvidia. The fund, Situational Awareness, managed by Leopold Aschenbrenner, has built a significant position in fuel-cell company Bloom Energy, a move rooted in the thesis that electricity, not algorithms, is the ultimate bottleneck for artificial intelligence.

"The bottleneck of AI is not in algorithms, but in electricity and computing power," Aschenbrenner argued in a widely circulated 165-page paper. He posits that the power consumption for a single AI training cluster will surge from megawatts to gigawatts, nearing the output of a large nuclear power plant, as the industry races toward Artificial General Intelligence (AGI) by 2027.

The fund's strategy has yielded significant returns, with a simulated portfolio reportedly rising 61% in two months. According to a holdings report, an initial $875 million position in Bloom Energy (BE) stock and options has swelled to nearly $3 billion, as the company's shares rose 239% year-to-date. A similar bet on Intel (INTC) through 20.2 million call options, purchased when the stock was near $20, has also produced outsized returns as the share price climbed toward $113.

This "physical arbitrage" strategy hinges on the idea that while the tech industry understands the impending power crunch, the public market has not yet fully priced the value of companies that can solve it. By investing in the physical constraints of the AI boom—power generation, chip manufacturing, and data centers—the fund is betting on the essential infrastructure that AI cannot function without, a move that could shift focus from semiconductor giants to the energy sector that underpins them.

The Power-First Portfolio

Aschenbrenner's portfolio construction reflects a singular focus on AI's physical needs. Beyond the major investment in Bloom Energy, which provides on-site power generation through solid-oxide fuel cells, the fund has also taken positions in Bitcoin mining companies CleanSpark and Bitfarms. The rationale is that these miners already possess the large-scale power access and cooling infrastructure that are becoming the scarcest resources for new AI data centers.

This contrasts sharply with the prevailing market wisdom, which has poured capital into Nvidia, whose GPUs are the primary engine for training AI models. By liquidating positions in popular tech names to fund these infrastructure bets, Aschenbrenner is making a high-conviction wager that the market is mispricing the foundational layer of the AI stack. The thesis is gaining broader traction, with reports that Microsoft may be forced to shelve its 2030 clean energy targets specifically because of the ballooning power requirements of its AI initiatives.

An Information Arbitrage

The strategy's success may stem from more than just a clever thesis. Aschenbrenner's time at OpenAI's "Superalignment" team provided a direct view of the internal roadmaps and resource requirements for next-generation AI models. His paper, published after he was fired for what OpenAI called "leaking information," can be seen as a public version of the insights gained from that inner circle.

The fund's investors also provide a continuous information pipeline. Limited partners reportedly include the founders of Stripe and Nat Friedman, the former CEO of GitHub, who are deeply involved in Silicon Valley's infrastructure spending decisions. This creates a powerful feedback loop: the fund's returns attract more well-placed investors, whose insights, in turn, sharpen the fund's investment accuracy. This structural advantage allows the fund to act on the known future needs of the AI industry before the broader market has fully digested the information.

For investors, the strategy highlights a new narrative in the AI space. While Nvidia shares trade at a high premium, companies like Bloom Energy, trading at a fraction of the valuation, may represent a different way to gain exposure to AI's growth. However, the fund's highly concentrated and leveraged positions are a double-edged sword. The strategy's success is entirely dependent on the continued, rapid expansion of AI infrastructure. Should the pace of AI development slow, or a new technology bypass the energy bottleneck, the very concentration that fueled its gains could lead to a rapid reversal.

This article is for informational purposes only and does not constitute investment advice.