You know, I have a story that just caught my attention earlier—did you know there's a 24-year-old making a fortune without buying Nvidia stock?

It's about Leopold Aschenbrenner, a former OpenAI researcher. The strategy behind the fund he launched is truly fascinating. He doesn't buy Nvidia or OpenAI. Instead, he invests only in the physical bottlenecks necessary for AI to actually operate—namely, electricity, chip manufacturing, and data center infrastructure.

Apparently, he's achieved a 61% profit in two months. The main drivers are two stocks. Fuel cell company Bloom Energy has surged 239% since the start of the year. He had invested $875 million in this company's stocks and options at the end of last year, and now it's grown to nearly $3 billion. Then there's Intel. When the stock was at $20, he bought 20 million call options. Last week, it rose to $113, nearly a fivefold increase in less than a year.

The reason for this success can't be fully explained just by looking at his publicly available portfolio. Media outlets like Motley Fool publish multiple articles analyzing his holdings daily, and discussions on Reddit debate whether his strategy should be mimicked. But even looking at the position reports released 45 days late, the market has already moved halfway there.

The real point isn't just that. It's about the source of why he keeps making correct bets. His network makes this obvious at a glance. Among the fund's LPs are two founders of Stripe and Nat Friedman, a product manager at Meta AI. These individuals are directly involved in major decisions like large-scale power purchase agreements and data center expansions every day.

Moreover, the research director is Carl Shulman, a veteran in AI with a background at Peter Thiel's hedge fund Clarium Capital. He's responsible for translating AI insights into actionable trading strategies.

What's particularly interesting is that he's also invested in Bitcoin mining companies like CleanSpark and Bitfarms. These companies are converting mining facilities into AI computing centers. Cryptocurrency mining farms have extensive power supplies and cooling systems, making them among the scarcest resources for AI data centers. By the way, he worked nine months at SBF's Future Fund in 2022.

Another point not to overlook is that his fiancée is the chief assistant to Dario, CEO of Anthropic. Dario has experience working at OpenAI and maintains deep ties with Anthropic as well. He's one of the few people with pipelines to both of the two giants in the AGI race.

I think this is the core of his strength. The AI papers he publishes are based on inside knowledge from OpenAI. His claim that the bottleneck in AI isn't algorithms but electricity and computational power comes from the internal roadmap of the research labs. By having industry insiders as LPs, the fund gains increasingly concentrated information. The more concentrated the information, the higher the precision of investment decisions. This creates a positive feedback loop, making entry barriers for outsiders even higher.

But vulnerabilities exist too. His holdings are highly concentrated, and leverage is significant. As long as the premise of "continuous expansion of AI infrastructure" holds, everything should proceed smoothly. But if AI development slows down or technological breakthroughs bypass energy bottlenecks, the reduction of concentrated positions could happen much faster than the speed of building new ones.

Ultimately, his investment logic is public, and the portfolio reports are accessible. But the stance he took when making decisions and the source of his knowledge can't be replicated. I think this is the greatest asymmetry with significant costs in the modern era.