24 years old, managing nearly billions of dollars, and you're still asking ChatGPT what stocks to buy 24 years old. Not starting a business at the age of 24, not raising funds at the age of 24. He is 24 years old and has managed nearly 9 billion US dollars. What are you doing at the age of 24? Perhaps still struggling with whether to become a ChatGPT member. This person is Leopold Aschenbrenner @ Leopoldasch, a former employee of Openai. After being fired, he wrote a 165 page AI article and then started a fund. Bloom Energy、 Electricity, mining companies Intel call、 He has almost hit the hardest layers of AI infrastructure in optical communication and data centers. The first reaction of an ordinary person must be: What exactly did he buy? But to be honest, it's already too late to ask. The 13F position has a 45 day delay. By the time you see the answer, the part where others truly make money may have already been halfway through. This article is not about who the next Bloom will be. ━━━━━━━━━━━━━━━━━━━━━━ ⚙️ 1、 I'm not buying concept stocks, but the lifeblood of AI Most people start their career in the US stock market with stocks. First see Nvidia rise, then research AI; first see Intel explode, then supplement chip logic; First see the news on Bloom Energy, then start tutoring on electricity. Leopold's logic is reversed. He doesn't start from stocks, he starts from a worldview. This worldview is simple and frightening: if AGI really approaches, then AI is not a technological revolution, but a real-world industrial mobilization. What does industrial mobilization mean? This means that electricity, chips, data centers, land, cooling, licensing, optical communication, storage, and capital expenditures will be re queued and re priced. Computing power needs to run → electricity is needed. Insufficient electricity → requires data center, on-site power generation, power grid, land, and cooling. When the whole market is watching who is the best at making models, he is buying the 'food that models must eat to survive'. The most common mistake that ordinary people make when researching AI stocks is to treat the most powerful companies as the most valuable link. In reality, the opposite is often true. The place that rises the most fiercely is often the one that was originally the least seen but also the first to be insufficient. What ordinary people see is the trend of technology, what they see is a bill of the real world. ━━━━━━━━━━━━━━━━━━━━━━ 2、 Standing at the pinnacle of information that ordinary people cannot enter Leopold's location is already more expensive than that of ordinary people. Who is his LP? The two founders of Stripe. Dealing with the payment flow of a large number of technology companies in Silicon Valley. What companies are expanding, what industries are accelerating, they smell it six months earlier than you. Former GitHub CEO Nat Friedman, the core of the AI product+computing power decision-making circle. Research Director Carl Shulman, Peter Thiel comes from a hedge fund background and knows how to translate AI community cognition into trading. The most outstanding is his fianc é e. Avital Balwit, Chief of Staff to Anthropic CEO Dario Amodei. Who is Anthropic? Claude's parent company. OpenAI's most direct opponent. He has worked at OpenAI himself, his fianc é e is at the Anthropic core layer, and LP is connected to Stripe, GitHub, and the Silicon Valley AGI community. The same news, what you see is news. What he sees are orders, talent, production capacity, policies, capital expenditures, supply chains, and assets at the next level that have not yet been priced. What are you asking: which stock to buy? He asked: Which link on this chain will break first? What is the biggest misconception of ordinary retail investors? I'm too timid, I don't have the courage to fuck No. You have lost the speed of information in front of you. Even more cruel is still ahead. Do you think copying his position can make money? Don't dream. He's not just good at rushing, he's good at hedging, reducing positions, and using options for protection. In the second quarter of 2025, he used semiconductor ETF put options for large-scale hedging. Holding put options on companies such as NVIDIA and TSMC in the third quarter. What you see is' he bought '. What you can't see is: why he bought, when he added, when he withdrew, and what to do if he made a mistake. ━━━━━━━━━━━━━━━━━━━━━━ 3、 What is truly worth learning is not the position list, but how he breaks down his worldview into positions The easiest thing to do across the internet now is to steal his 13F. What did he buy? What is the size of the storage space? Who is the next Bloom? Who is the next Intel? These questions may seem diligent, but they are actually quite passive. Because holding is not a method, holding is just a result. What you see is' he bought '. What you can't see is: why he bought, when to add, when to withdraw, how to hedge against mistakes, which layer of logic has already been priced, and which layer has not been reflected by anyone yet. Even fewer people, like Leopold, stand at the intersection of AI cutting-edge laboratories, Silicon Valley capital, and public markets, turning circle confirmation gaps into transactions. His core methodology can actually be condensed into a few sentences: First, confirm the super trend and don't look for illusions in small trends. Break down trends into physical supply chains, without staying at the conceptual level. Find bottlenecks in the second and third layers, and avoid chasing after leaders who have already been fully priced. Expand your search, organization, and proof capabilities with AI and circle information, without letting the model directly hit your head. Finally, the decision to place bets is based on the understanding of the industry, risk discipline, and position management, rather than letting emotions rush for you. ━━━━━━━━━━━━━━━━━━━━━━ 4、 He's not a god either. He's betting on a super narrative that will backfire Of course, this does not mean he is invincible. I actually think that the more this kind of story is written, the less it can be written as closed eyed worship. His playing style is very focused and sharp. It relies on a premise: AI capabilities continue to advance rapidly, the AGI timeline is generally established, inference demand continues to explode, global capital expenditure continues to tilt towards AI infrastructure, and real-world power, data centers, and upstream supply continue to be tight. But if AI gets stuck in a bottleneck and infrastructure starts to retreat, what about the position? Returning to the beginning, the scariest thing is not that he is 24 years old. He didn't manage nearly 9 billion dollars, nor did he bet on Bloom and Intel. What ordinary retail investors fear is not missing out on a 24-year-old new stock god, but the next time the market re prices, they can only see someone else's answer 45 days later.