A 24-year-old fund manager's annual return is 24 times higher! His AI investment portfolio targets the "most scarce resource"

Former OpenAI researcher Leopold Aschenbrenner doubled $225 million to $5.5 billion through his fund. He said the bottleneck in AI development lies in electricity, not chips or models.

Former OpenAI security researcher, and German man, only 24 years old, Leopold Aschenbrenner, turned $225 million from the fund he managed, Situational Awareness, into $5.5 billion in less than a year. While Wall Street money poured into AI models and chipmakers, he spotted a blind spot the market overlooked: electricity. By making precise bets on the infrastructure needed to address AI’s high power consumption, he achieved astonishing investment returns.

After leaving OpenAI, he pivoted into the AI investment market

After leaving OpenAI, Aschenbrenner wrote a 165-page report, asserting that general artificial intelligence (AGI) will arrive sooner than everyone thinks, and that the final winners will not be companies with the strongest AI models, but companies that “control electricity.” To this end, he set up a hedge fund called “Situational Awareness LP” and poured $875 million into buying fuel cell company Bloom Energy.

This week, Bloom Energy announced that it has signed a major fuel cell deal with Oracle for 2.8 gigawatts (GW), sending the stock soaring 15% after hours, and the book value of Aschenbrenner’s stake instantly surged to nearly $2 billion.

Portfolio fully revealed: Long on infrastructure, short on traditional IT industries

Reports indicate that his investments followed the “electricity first” logic:

• Bloom Energy (BE): Invested $875 million to buy this fuel cell company. The technology enables data centers to generate power on-site directly, without relying on outdated power grids. Benefiting from the 2.8 GW order signed with Oracle, the stock surged, and the book value of his holdings has already climbed to nearly $2 billion.
• CoreWeave (CRWV): Invested $700 million in this leading AI cloud computing infrastructure provider, locking in scarce infrastructure resources.
• Infosys (INFY): He heavily shorted this major Indian IT outsourcing player, expecting AI coding agents to completely destroy traditional IT outsourcing business.
• Intel (INTC): Used leverage operations through Intel call options, earning several multiples of returns during the 53% rebound in the stock price.
• Core Scientific (CORZ): Holds 10% equity. This former Bitcoin mining company is transforming its existing power facilities into AI data center colocation sites.

The power-hungry monster behind computing power: Electricity usage doubles year after year

Aschenbrenner said that looking back to 2022, the computing cluster that trained GPT-4 consumed about 10 megawatts (MW) of electricity, at a cost of about $500 million. However, AI compute demand is expanding at a pace of about half a quantity order each year, meaning that the electricity needs of the largest training clusters will double every 12 to 18 months.

By 2024, the power consumption of the largest computing clusters has reached 100 MW, equivalent to 100,000 high-end graphics processing units (GPUs) running simultaneously. Now, in 2026, the leading training clusters require as much as 1 GW of continuous power, which is equivalent to the electricity generation output of a large nuclear power reactor.

Whoever controls power controls the future of AI

He estimates that by 2028, the electricity consumed by AI training will skyrocket to 10 GW—larger than the electricity generation of many U.S. states as a whole; by 2030 it will reach 100 GW, consuming as much as 20% of the United States’ current total electricity generation overnight. This is only the electricity for “training” models; if you add the “inference” compute power actually used by the public, the electricity consumption becomes even more difficult to imagine.

However, over the past decade, the total electricity generation in the U.S. has only grown slightly by 5%. Now reports from everywhere about a severe shortage of transformers and data centers that can’t be built are proof that the power grid can’t support it. This is also why he dares to make a heavy bet on Bloom Energy: the real bottleneck in AI development is not chips or software, but whether humans can produce enough electricity.

• This article is reprinted with authorization from: 《Chain News》
• Original title: 《24-year-old fund manager’s annual return is 24x! AI portfolio targets the “most scarce resource”》
• Original author: Co2