The Birth of a New Stock God in the US Stock Market: Clearing Nvidia, Collecting All 'Trash', How Did 27 Year Old Leopold Make 5.5 Billion with the 'Physical World'? If someone tells you that an AI hedge fund doesn't even have a single concept stock from Nvidia, Microsoft, or OpenAI, it is replaced by fuel cell companies, bankrupt Bitcoin mining companies, and Intel, which has been wildly underestimated by analysts. You will definitely think that the fund manager is crazy about money. But this' madman 'is none other than Leopold Aschenbrenner, a 27 year old former researcher on OpenAI's Super Alignment team. In the past 12 months, his fund Situational Awareness LP has managed a scale of less than $400 million and managed to reach $5.517 billion, a 14 fold increase! With targets such as Bloom Energy, Cipher Mining, Intel and others skyrocketing by over 10% in a single day, Wall Street is finally starting to take seriously this young person's "weird" shopping list. His investment logic, derived from the 165 page "Ten Thousand Words Book," is being brutally validated by reality: the narrative of AI is retreating from the models on the screen to the land and power grid beneath its feet. Take you on an article to unravel the path of this' version of the son 'of the US stock market. 1、 $5.5 billion 'waste recycling station': What did he buy? Opening Leopold's 13F holdings report, the largest holdings are not in chip companies, but in Bloom Energy (accounting for 15.87%, $876 million). This is a company that produces solid oxide fuel cells, which can directly convert natural gas into electricity. Why would an AI fund invest its assets in power plants? Because AI consumes too much electricity. Gartner predicts that the power consumption of AI servers worldwide will nearly quintuple by 2030. And the dilapidated power grid in the United States, with an average lifespan of over 25 years, simply cannot withstand it. Bloom Energy's batteries do not need to be connected to the power grid and can generate electricity 24 hours a day directly next to the data center. Whoever can solve the power outage crisis is the most profitable "water seller" in the AI era. The second largest heavy position: CoreWeave (accounting for 22%, with options and underlying stocks exceeding $1.2 billion). This company used to be a mining farm for Bitcoin. After the collapse of the cryptocurrency industry, they used all their GPUs to run AI and transformed into the largest arms dealer in AI computing power. Leopold is interested in it because it has a massive GPU and deep binding with Nvidia. The third largest warehouse: Intel (accounting for 13.54%, $747 million). When Intel's stock price plummeted and was ridiculed by the entire online community, Leopold crazily bought call options. He is not betting on Intel's technology, but on the national will of the United States - in the technology game between China and the United States, Intel has received nearly $8 billion in subsidies from the US government, and it is the only "biological son" of domestic contract manufacturers in the United States. The craziest shopping spree: a bunch of Bitcoin miners. Core Scientific、IREN、Cipher Mining…… He bought all these mining companies that were either bankrupt or struggling in the bear market. The reason is simple: the most scarce power quotas and sites in AI data centers are all in the hands of these miners! Bitcoin mining sites have long locked in the cheapest long-term electricity contracts in the United States. Overnight, these miners who were originally engaged in cryptocurrency trading have become the biggest landlords of the AI era. At the same time, he decisively liquidated his holdings in star stocks such as Nvidia, Broadcom, and Vistra that were hyped up in the fourth quarter of 2025. His logic is extremely clear: the bottleneck of AI is not in software, but in hardware; Not in algorithms, in electricity; Not in the cloud model, in the physical world. 2、 From the Rebel of OpenAI to the 165 page 'Ten Thousand Word Book' This genius, who was admitted to Columbia University at the age of 15 and became a representative of graduates at the age of 19, originally worked in OpenAI's "Super Alignment Team" to prevent AI from losing control. But he was fired after being accused of "leaking secrets" for submitting a memorandum to the board warning of security vulnerabilities in the company. After leaving, he wrote a 165 page article titled 'Situational Awareness: The Next Decade', predicting that AGI (General Artificial Intelligence) is highly likely to be achieved by 2027. This ten thousand word book is the map of his $5.5 billion holdings. He prophesied in the book: A trillion dollar computing power cluster is about to emerge. Six months later, the Trump administration announced a $500 billion Stargate infrastructure project. Serious power crisis. In order to provide power, Microsoft even restarted the Three Mile Island nuclear power plant, which had experienced a serious nuclear accident that year. If AGI really arrives in 2027, the world will need massive computing power ->computing power needs GPU ->GPU needs electricity ->where will electricity come from? Mining sites and new energy. This is his ultimate investment path of following the clues. 3、 DeepSeek's Night of Critical Strike: Stocking up When Everyone is Afraid On January 27, 2025, the DeepSeek model released by China shocked the world with an extremely low cost (about 6 million US dollars). The whole Wall Street is in a panic: if AI can be solved at a low cost, will the hundreds of billions of dollars invested by American technology giants become foam? Nvidia plummeted nearly 17% in a single day, causing the entire semiconductor sector to collapse. Leopold's holdings are all in infrastructure, so theoretically it should collapse. But when the investor called to question, he only replied in five words: 'Leopold says it's fine.'. ） Because he had already figured it out in the book of ten thousand words: cheaper training costs proved that algorithm efficiency was soaring. The decrease in costs never eliminates demand, but only creates larger and crazier demands for model training. He went against the trend and increased his holdings in panic, and it turned out that the AI sector quickly experienced a retaliatory rebound. Summary: The essence of investment is to find "physical bottlenecks" Leopold's story is not just a story about a young man who became rich overnight. When everyone was staring at how strong the GPT-5 was on the screen, he looked up and saw the chimney of the power plant and the substation of the mine. In the California gold rush of the 19th century, it was not the gold diggers who made a fortune, but Levi Strauss, who sold shovels and jeans. In the AI frenzy of 2026, the most valuable thing may not necessarily be the algorithm code, but the physical world foundation that supports this code from being powered off.