The 24-year-old genius investor Leopold Aschenbrenner, who has recently been all over social media, has made another move. He started his fund just over a year ago, and his assets have surged from $1 billion to $5.5 billion—so fast that it makes people question whether they’re even living in the real world. The key is that his latest investment logic has completely flipped: stocks in the chip sector that everyone was chasing before—he’s now selling them off.

First, let’s talk about what he’s dumping. Nvidia, Broadcom, TSMC, Micron—these former AI-infrastructure stars. In just a single quarter, he got rid of $300 million worth of Nvidia put options and directly made a profit. Why do this? His 165-page paper, **“Situational Awareness,”** explains it—he believes that by the end of 2025 or the beginning of 2026, the market has already fully priced in the value of GPUs, and the Nvidia stock price is basically reflecting the ceiling.

The problem is that he sees the next bottleneck the market hasn’t noticed yet. The real dilemma facing today’s AI labs isn’t a lack of GPUs, but a lack of electricity. Yes, that’s as blunt as it sounds. The existing power grid is designed for humans and simply can’t handle the appetite of AI data centers. That’s exactly where his new bet lies.

He puts 20% of his portfolio—$855 million—into a company called Bloom Energy. What does this company do? It produces oxide fuel cells that can directly convert natural gas into electricity usable by data centers. The key point is that their backlog of orders has reached $20 billion, revenue growth is expected to be 34% in 2025, and they’re forecast to grow another 40% in 2026. His logic is straightforward: energy is the true scarce resource in the AI era.

Besides energy, he’s also heavily doubling down on infrastructure. He added another $300 million to CoreWeave, bringing his total investment to nearly $800 million. This company specializes in GPU deployment, power supply, and cooling systems—handling the gritty, labor-intensive tasks that serve as AI lab infrastructure outsourcing. But most interesting of all is how he operates with Bitcoin mining companies.

At first glance, it feels a bit strange—why invest in Bitcoin mining companies? But if you think it through, it becomes clear. These mining companies have two things: land and power permits. Normally, obtaining these permits takes months or even years, but he instead directly acquires these companies, instantly bypassing the entire process. Then he turns those resources into AI data centers. It’s like taking over a bar that already has a liquor sales license, rather than applying for one from scratch. Brilliant—truly.

He also shorted Infosys, a company that relies on providing cheap IT labor outsourcing. The reason is also simple: models like Claude and GPT are strong enough now to automate complex IT processes, meaning the traditional outsourcing model’s business is over.

What’s fascinating is that the entire investment logic reflects a major shift—from pure software back to the physical world. Things like energy, manufacturing, and infrastructure can’t be produced by AI on its own; they require real money, genuine permits, and actual legislative support. Those are the true future moats.

Of course, some people question him because he’s only 24, has limited experience, and his portfolio is overly concentrated in this single theme of energy infrastructure. If the pace of AI spending slows down or the macro environment changes, the whole portfolio could be hit. But based on current signals, each step he’s taken is being validated. He correctly predicted the GPU boom before, and now he’s positioned himself ahead of the energy infrastructure boom.

Worth noting: giants like Google, Amazon, and Nvidia pledged $650 billion in capital expenditures in the previous earnings season, all aimed at solving energy and infrastructure problems. This guy’s portfolio is clearly already set up to profit from that. He turned $1 billion into $5.5 billion in just over a year—an almost unheard-of growth rate in the hedge fund world. Whether it can keep going still needs time to prove, but judging by his logic and execution, this young man really is doing something different.