Recently, 24-year-old Leopold Aschenbrenner has become the most dazzling star in the US stock market - his AI hedge fund managed by him skyrocketed from $1 billion to $5.5 billion in just one year, and within just six months of its establishment, its performance exceeded the S&P 500 index by eight times. Most people are amazed by his youth and earnings, but rarely delve into the core logic behind his holdings. In fact, the limited podcast's detailed analysis of its 13F report two months ago has now been widely implemented, and this foresight in advance layout is precisely the key to his outstanding performance. Today, we will review the investment logic of this' son of the US stock version 'and understand his paradigm shift from chips to infrastructure.

Leopold's investment starting point is with a depth and foresight that surpasses his peers. He once wrote a 165 page paper titled 'Situational Awareness', in which he boldly predicted that General Artificial Intelligence (AGI) would be achieved by 2027, and all of his portfolio layouts revolved around this core prediction. From initially betting on the GPU craze to now shifting towards energy and infrastructure, his investment logic has always been one step ahead of the market, avoiding crowded tracks and accurately capturing opportunities in industry bottlenecks.

###Investment paradigm transition: from chips to infrastructure, escaping the crowded track

Leopold's most disruptive operation is undoubtedly clearing a group of AI chip leaders and shifting their focus to energy and infrastructure. He decisively sold stocks of popular AI infrastructure companies such as NVIDIA, Broadcom, TSMC, and Micron, and even sold $300 million worth of NVIDIA put options (and profited from it). The core judgment behind this is that by the end of 2025 to the beginning of 2026, the market has fully reflected the value of GPUs, the dividends of the chip race have been fully tapped, and the next core bottleneck of the AI industry lies in energy and infrastructure.

His judgment hits the pain point of the industry: the current power grid is designed for human daily needs and cannot bear the huge energy consumption of AI data centers. With the advancement of AGI research and development, the demand for electricity in AI laboratories will grow exponentially, and energy supply and infrastructure construction will become key variables restricting the development of AI - this is precisely the core logic behind his decision to abandon chips and turn to investing in the physical world. This kind of thinking of "escaping the crowded track in advance and laying out the next bottleneck" is the key to his ability to outperform the market.

###Core heavy warehouse: Bloom Energy, betting on the 'hidden champion' of AI energy solutions

In Leopold's investment portfolio, Bloom Energy is the absolute core - accounting for up to 20% and holding $855 million, making it his largest investment target. The reason why this previously unknown company was able to secure its heavy position bet lies in its precise match with the energy demand of the AI era.

Bloom Energy focuses on developing oxide fuel cell equipment, with its core advantage being the ability to directly convert natural gas into electricity that can be used in AI data centers. The product is modular, can be quickly deployed, and does not rely on existing power grids. For AI laboratories, this means that they can completely overcome the energy consumption bottleneck of the power grid, directly build power supply equipment next to the data center, and obtain stable electricity at an efficient cost, solving the energy pain points of AI training and reasoning.

Fundamental data also supports this bet: Bloom Energy's demand backlog orders are as high as $20 billion, with revenue growth of about 34% in 2025 and an expected increase of another 40% in 2026, showing a situation of demand exceeding supply. Leopold sees it as the 'Nvidia of energy', betting on its monopolistic potential in AI energy solutions - after all, AI can replace software jobs but cannot create energy out of thin air, which is also one of the core logics behind his investment in the physical world.

###Taking shortcuts: Bitcoin mining companies+CoreWeave lock in core AI infrastructure resources

In addition to Bloom Energy, another major highlight of Leopold's layout is the clever targeting of core resources in AI infrastructure through investments in Bitcoin mining companies and CoreWeave, taking a shortcut of "overtaking on a bend".

He has significantly increased his holdings in CoreWeave, with a cumulative investment of $800 million. As a "new cloud service provider", CoreWeave's core business is to provide one-stop services such as GPU deployment, power supply, and cooling system maintenance for AI laboratories. When AI laboratories accumulate a large number of GPUs, the subsequent infrastructure construction and operation are the real necessities, which is also the core reason why Leopold values it. In addition, he also holds approximately 10% of CoreWeave's main supplier, Core Scientific, forming a collaborative layout of "GPU operation and maintenance+energy grid construction".

Even more surprisingly, he sold a large number of Bitcoin mining companies, not because he was optimistic about the cryptocurrency market, but because he was interested in the core assets of mining companies: land and electricity. Bitcoin mining requires a large amount of energy and space, which are precisely the core requirements of AI data centers. More importantly, these mining companies already have ready-made power grid access rights and related licenses - which usually take months or even years to apply for. Leopold, by acquiring mining companies, directly bypasses the licensing process and quickly obtains the core resources needed for AI infrastructure, which is equivalent to "taking over a bar with a liquor license instead of applying for it himself", efficiently and accurately.

###Short selling logic: betting on the end of IT outsourcing and practicing the trend of AI substitution

Leopold's investment logic is not only reflected in his long position, but his short position also conforms to the trend of the times - he extensively shorted Infosys, a company focused on IT outsourcing in India.

The core of Infosys' business model is to rely on low-cost labor to undertake IT outsourcing business in Western countries. But Leopold believes that with the rise of AI tools such as Claude Code and GPT Codex 5.3, AI can not only automate simple tasks, but also handle complex IT processes. The advantage of cheap labor will be completely weakened, and the IT outsourcing industry will come to an end. He practiced this judgment with real money and shorted Infosys, hedging market risks and once again confirming his profound understanding of AI trends.

###Investment philosophy: Return to the physical world and bet on core assets that cannot be replaced by AI

All of Leopold's holdings revolve around a core investment philosophy: companies that rely solely on software will face difficulties in the future; The assets of the physical world - manufacturing, factories, energy, and infrastructure - cannot be replaced by AI and are also the core investment direction for the future.

In his view, AI can generate software and automate processes, but it cannot create energy out of thin air, build infrastructure quickly, or obtain scarce grid licenses - these physical assets that require manpower, legislation, and time accumulation are the most core bottlenecks and long-term investment targets in the AGI era. This is also the underlying logic behind his shift from chips (the combination of software and hardware) to energy, land, and infrastructure: energy is a scarce resource that no one can meet, while infrastructure is the prerequisite for the landing of AI. Only by focusing on the core of "providing power for future AI" can we capture the most essential opportunities.

###Controversy and Prospect: Genius Carnival or Long Term Winner?

Despite Leopold's stunning performance, the market's controversy over him has never stopped. Some people question whether the 24-year-old lacks long-term investment experience, and his fund layout is too concentrated, almost a single theme bet - if the growth rate of AI infrastructure spending slows down or macroeconomic changes occur, the entire investment portfolio will face huge downward pressure and lack sufficient hedging space.

But it cannot be denied that every prediction he makes accurately follows industry trends: betting on chips before the GPU boom, laying out Bloom Energy before energy bottlenecks emerge, and shorting Infosys before IT outsourcing declines. His investment portfolio is more like a "real-time tracking tool" for bottlenecks in the AI race, always leading the market by one step.

Nowadays, Google、Amazon、 Technology giants such as NVIDIA have committed $650 billion in capital expenditures, focusing on AI infrastructure and energy solutions, which is highly aligned with Leopold's layout. His success is not only a manifestation of talent, but also a profound insight into industry trends - not chasing hotspots, not blindly following the market, but starting from the underlying logic, finding the underestimated core bottleneck, using high belief and high concentration layout to capture the dividends of the times.

In the future, whether Leopold can continue to create miracles will need time to test - after all, great investors do not rely on a single year's explosion, but on long-term stable compound interest growth. But in any case, this 24-year-old genius's position logic provides us with a new perspective: in the era of AI, the real opportunity may not be in the noisy software and chip race, but in those silent and irreplaceable physical worlds.