Organized & Compiled by: Depth Tide TechFlow

Hosts: Josh Kale; Ejaaz Ahamadeen

Podcast Source: Limitless Podcast

Original Title: Forget NVIDIA| This 24-Year-Old's $4.5B Bet on AI's Real Problem (Leopold Aschenbrenner)

Broadcast Date: March 4, 2026

Key Points Summary

Recently, everyone has been talking about Leopold Aschenbrenner—24 years old, with a $5.5 billion AI hedge fund, the son of the U.S. stock market. However, most discussions remain at the level of "he is amazing" and "he made so much money," with very little content that truly dissects his holding logic.

Two months ago, the Limitless Podcast did an episode analyzing his 13F report line by line:

Why he liquidated his Nvidia shares, why he put 20% of his position into a fuel cell company, why he bought up a bunch of Bitcoin mining companies, and why he shorted Infosys. At the time, this episode had almost no discussion. Looking back now, the judgments made then have largely come true, and it's worth going back over.

Highlights of Opinions

On Leopold Aschenbrenner's Investment Performance

• “Last year he managed $1 billion… Today, just a year later, that $1 billion has grown to $5.5 billion.”
• “His fund was established at the end of 2024, with an initial size of $255 million. In just six months, his fund outperformed the S&P 500 index by 8 times.”
• “He wrote a 165-page article titled ‘Situational Awareness’. In this article, he essentially predicted that we would reach General Artificial Intelligence (AGI) by 2027.”

Shift in Investment Paradigms: From Chips to Infrastructure

• “He sold Nvidia, Broadcom, TSMC, and Micron. These are all major AI infrastructure companies.”
• “By the end of 2025 or early 2026, he believes the market will have fully reflected the value of GPUs.”
• “He has shifted his focus to major bottlenecks that investors have not yet paid sufficient attention to—energy and infrastructure.”
• “The existing power grid was designed for humans, not to meet the enormous AI demands we face today. This is where his current investments lie.”

Core Holding: Bloom Energy

• “Bloom Energy is currently his largest investment, accounting for 20% of his entire portfolio… He has built a huge position in this company, amounting to $855 million.”
• “Bloom Energy has developed a device known as an oxide fuel cell… that can directly convert natural gas into electricity usable by data centers. It is modular and can be deployed quickly.”
• “They have a backlog of orders worth $20 billion. Revenue is projected to grow by about 34% in 2025, and they expect a further 40% growth in 2026.”
• “If you use products like Bloom Energy's natural gas turbines, you do not need to rely on the power grid at all. You only need to install it next to the AI data center.”

Infrastructure and the “Shortcut” of Bitcoin Mining

• “Leopold has significantly invested in CoreWeave. He has made the largest leveraged investment in core GPU infrastructure and energy supply.”
• “He has invested in many Bitcoin mining companies… because these companies have the two key elements needed to build AI infrastructure: land and electricity.”
• “He acquires these companies to obtain their licenses and access to the power grid. Typically, obtaining these licenses can take months or even years.”
• “It's like taking over a bar that already has a liquor sales license instead of applying for a new license and waiting for years, which is a very clever 'shortcut'.”

Short Logic and the End of IT Outsourcing

• “He holds a short position on a specific company, which is Infosys… Their business model relies entirely on providing cheaper labor than in Western countries.”
• “He realized that these models are now strong enough not only to automate simple jobs but also to handle some very important IT processes, leading to his massive short position on this company.”

Investment Philosophy: A Return to the Physical World

• “Companies that rely solely on software will find it very difficult in the future. His shift is not just around building architecture, but investing in the physical world, such as manufacturing, factories, energy, and infrastructure.”
• “These are areas that cannot be constructed through AI but require manpower, licenses, and legislation to realize hardware and infrastructure.”
• “Energy is the only resource that everyone cannot supply enough of… everything revolves around one core: powering the future.”

The Young Investment Prodigy Leopold Aschenbrenner

Josh Kale:

There is a guy named Leopold Aschenbrenner, who is 24 years old this year. We reported on him in an episode last year when he was only 23, managing $1 billion and focusing on investing in emerging cutting-edge AI concepts and technologies. And today, just a year later, that $1 billion has grown to $5.5 billion.

This guy, much younger than both of us, just achieved a breakthrough performance, making more money in the AI field than any other fund in the world. More importantly, AI is the hottest market right now, meaning the competition is extremely fierce. So it's clear that this guy named Leopold is doing something different.

Just last week, his latest 13F report was released, and we finally got a glimpse into his recent trading activities. So next, we will carefully analyze these documents to see what this person did to catapult the funds he manages from $1 billion to $5.5 billion.

Insights from the 13F Report

Ejaaz Ahamadeen:

He accomplished these achievements in just 12 months. His fund was established at the end of 2024, with an initial size of $255 million. In just six months, his fund outperformed the S&P 500 index by 8 times, growing to $2 billion. Since we last discussed his third-quarter fund report on the show, his fund has grown another $1.5 billion. So, he can now be said to be in a time of explosive growth.

He is very young and has made a significant shift, but it all aligns with what he calls his “Bible”—a 165-page article titled “Situational Awareness.” In this article, he essentially predicts that we will reach General Artificial Intelligence (AGI) by 2027. In this grand article, he details his vision of how the AI revolution will unfold. His predictions have been almost entirely correct, as he successfully foreseen the GPU infrastructure boom, and now he has proposed a very important shift, which we will delve into next.

Shift from Chips to Infrastructure

Josh Kale:

I think the entire investment philosophy is shifting from chips to infrastructure. What we see on the screen is very interesting. He created a document with Claude that will guide us through the entire change record from last year to this year. Perhaps we can start with the assets he liquidated, as these positions are quite significant, including Nvidia, where he sold $300 million worth of put options in a single quarter.

Ejaaz Ahamadeen:

You will see that many of the stocks he sold are very popular companies, and many people are currently investing in these companies. So the question arises, why did he sell off $1 billion worth of stocks from these companies? He sold Nvidia, Broadcom, TSMC, and Micron. These are all major AI infrastructure companies.

He made a profit on his sale of Nvidia stock; he held $300 million in put options, meaning it is likely that he profited from the decline in Nvidia's stock price over the past few months. So the question is, why did he do that?

In his 165-page paper, he mentioned that by the end of 2025 or the beginning of 2026, he believes the market will have fully reflected the value of GPUs. These values primarily come from companies like Nvidia and Broadcom that manufacture these chips and then stack them up for AI labs like OpenAI and Anthropic to train their models.

And now, he has shifted his focus to a major bottleneck that investors have not yet paid sufficient attention to—energy and infrastructure. Currently, a major issue many AI labs face is: First, they have too many GPUs; second, the existing power grid was designed for humans, not to meet the enormous AI demands we face today. This is where his current investments lie.

Selling Nvidia Put Options

Josh Kale:

Seeing him sell Nvidia put options and completely exit his investment in Nvidia is quite interesting to me. Because when I chat with friends or talk to regular people on Wall Street, Nvidia is the company everyone talks about, and it’s the biggest investment target. And seeing him turn his back on Nvidia, I think it proves once again that he is always one step ahead, always able to foresee future trends rather than dwell on past hot spots. In his view, the future focus is on infrastructure, shifting from chips to information-based strategies.

This might be where we can dive deeper into his new investments because these are the stocks you should be following. These are the assets he currently holds, and he believes will grow in the future. If his judgments are correct, we should be able to see considerable returns from them. So what investments did he add this quarter?

Ejaaz Ahamadeen:

Here is a very neat portfolio chart that categorizes all of Leopold Aschenbrenner's investments according to the AI technology stack. We can see the investments categorized into power generation, real estate and facilities, computing and hosting, connectivity, storage and memory, chips and wafers, etc.

Actually, I want to add something to what I just mentioned. I noticed that he made a very clever trade with Intel (Intel). He sold off his existing shares while still holding a huge long position. This way, he released liquidity and directed funds toward other companies. The main company he invested heavily in is in the power generation sector, called Bloom Energy. This company was almost unknown about three months ago, but they specialize in manufacturing power turbines for AI data centers.

He has built a huge position in this company, amounting to $855 million. Although it shows $876 million here, the report states it is $855 million.

Bloom Energy: Power Innovator

Josh Kale:

Bloom Energy is currently his largest investment, accounting for 20% of his entire portfolio. This is completely unrelated to the chip sector and is a completely different direction. I researched their business and found it to be quite interesting.

Bloom Energy has developed a device known as an oxide fuel cell, which is an advanced technology that generates power from natural gas on-site. Typically, when natural gas is delivered to a data center, it needs to be heated and cooled through turbines, which is a very clumsy energy production process. Bloom Energy's "fuel box" can directly convert natural gas into electricity usable by data centers. It is modular, can be quickly deployed, and doesn't seem to face supply shortages. As far as I know, they plan to produce 2 GW of electricity this year.

This is a very interesting energy play. I've been looking for the "NVIDIA of the energy sector"—that is, the "chip manufacturer of the energy field." I have not yet found a company that completely matches that description, but perhaps Bloom Energy could become that company.

Ejaaz Ahamadeen:

I also reviewed their recent financial reports since they are a public company. They have a backlog of orders worth $20 billion. Revenue is expected to grow by about 34% in 2025, and they anticipate another 40% growth in 2026, indicating that their demand is exceeding supply.

You mentioned oxide fuel cells. Their natural gas turbines are particularly attractive because they do not require reliance on the existing power grid. As I mentioned earlier, the current grid is under a lot of pressure because humans need energy, and AI data centers also need energy, leading to rising energy prices in areas where AI data centers are located. If you use products like Bloom Energy's natural gas turbines, you do not need to rely on the power grid at all. You just need to install it next to the AI data center to obtain electricity at an efficient cost for training or inference of your GPUs and data centers.

Companies like Broadcom and CoreWeave will require this energy, especially super large-scale cloud service providers and AI labs. It reminds me of the game Civilization, I don't know if you've played it, but it feels like you move infrastructure and energy production facilities to your small settlement to drive its development, and what happens here is quite similar to that scenario.

Josh Kale:

It is evident that energy shortages do not exist, the issue lies in who can produce the most energy. They certainly have a very large backlog of orders, but the question is whether they can produce enough products to meet those orders? Manufacturing capacity becomes a key issue here. In many such investments, we are entering a "nuclear" world, where manufacturing truly becomes an important field. I would love to dive deeper into this in the future to see if they really have the ability for mass production. But for now, this is undoubtedly a very important investment area, making up 20% of his portfolio. So, what other notable positions are there in his new investment portfolio?

Ejaaz Ahamadeen:

He has also increased his investment in CoreWeave by about $300 million. Imagine being an AI lab, you need GPUs. But purchasing GPUs from companies like Nvidia is only part of the job. Deploying those GPUs into rack servers, providing power supply, doing tech engineering support, and maintaining GPU servers and cooling systems is a completely different matter. So you can outsource this work to a company called "new type of cloud service provider," which is CoreWeave, they specialize in handling these matters.

Broadcom also provides similar services to some extent, but CoreWeave is a smaller company originally focused on GPU gaming services, and has now transitioned to specifically serving AI. Leopold has significantly invested in CoreWeave. In the third quarter we discussed before, he had already invested $500 million, and now he has added another $300 million. Now his total investment in CoreWeave may have reached $800 million, but the story goes deeper. He also holds about 10% of CoreWeave's major supplier, Core Scientific, which specializes in providing energy grid construction services for CoreWeave.

If you consider the betting strategy in this investment, Leopold may have made the largest leveraged investments in core GPU infrastructure (like CoreWeave's new type of cloud services) and energy supply (such as Bloom Energy), which are the two main positions held in his current fund.

Bitcoin Mining

Josh Kale:

I find it interesting that he has begun to hold enough shares in these companies to become an activist investor, someone who can actually influence the decision-making of these companies. I find this very intriguing. As I studied his portfolio, aside from power generation being the obvious direction, I noticed that he added the most positions related to real estate, adding about 10 positions related to real estate, which are related to Bitcoin mining.

What we are seeing now is that he has invested in many Bitcoin mining companies. This seems a bit strange and somewhat irrational. After all, the cryptocurrency market isn't thriving, and Bitcoin hasn't performed well. Why is he buying these Bitcoin mining companies? The reason is that these companies possess the two key elements needed to build AI infrastructure: land and electricity.

What does Bitcoin mining require? It requires vast amounts of energy and enough space to place GPU racks. While Bitcoin mining has not completely declined, it is clear that these companies' real estate resources and power resources can yield a better risk-return ratio. It looks like he is betting that these Bitcoin mining companies will either sell their land use rights and licenses or directly transform into AI data centers.

Ejaaz Ahamadeen:

It is important to clarify that his interest in these companies is not for mining; he is acquiring these companies to gain their licenses and access to the power grid. Usually, obtaining these licenses takes months or even years. This is also why we see companies like Meta, Microsoft, and OpenAI announcing compute partnerships worth $1.4 trillion, but these collaborations have not yet fully translated into the models they rolled out. This is one reason why the supply of GPUs always lags behind the latest generation, because they cannot timely acquire these licenses.

Leopold instead acquires these smaller companies that already have licenses, bypassing the entire licensing process. He completely strips away these companies' crypto services and repurposes them for specifically training AI models, becoming the infrastructure providers for these AI labs. It's a bit like taking over a bar that already has a liquor sales license instead of applying for a new one and waiting for years; it’s a very clever “shortcut.”

AGI and Market Trends

Josh Kale:

One of the things I admire most about his investment philosophy and the process of seeing these ideas validated over the past year is its simplicity and efficiency. For example, Bitcoin mining companies obviously have licenses and energy, and it is clear that every AI company needs these resources. So why isn’t everyone buying these companies? I think it’s because these ideas are too simple, and many people are blocked from investing. But time and again, his simple ideas have proven to be correct.

Will Leopold's prediction about achieving AGI by 2027 also be correct? Will we really achieve AGI by 2027?

Ejaaz Ahamadeen:

To validate this prediction, we opened a prediction market on Polymarket, predicting whether OpenAI will announce that it has achieved AGI before 2027. As it stands now, when Leopold proposed this fund, many people were skeptical about his prediction, but the probability in this prediction market is now 13%. So it looks somewhat remote. His investment philosophy may be correct, but the timeline might be slightly off.

This probability is indeed low. However, I must say that he was initially criticized for this paper, as many people thought his viewpoint was too bizarre and unrealistic. About 50% of people believe AGI will be achieved in the next few months, while others think it won’t happen until 2030. Leopold is the only one to predict 2027, and currently seems to be the closest to being correct.

He predicted the importance of GPUs before the GPU boom occurred. Now, he is making predictions before the energy infrastructure boom arrives. So I think he is still leading in this regard.

However, his portfolio does not only consist of long positions; he also holds a short position on a specific company, which is Infosys, a company focused on IT outsourcing primarily based in India. Their business model entirely relies on providing cheaper labor than in Western countries (like the United States or Europe). In simple terms, it’s “outsource all your administrative IT work to us, and we’ll handle it for you.”

I believe his bet here is based on trends he has observed. He has seen the rise of products like Claude Code and GPT Codex 5.3 and realized that these models are now strong enough not only to automate simple jobs but also to handle very important IT processes, leading to his massive short position on this company.

I think this is one of his deeper investments and more aligned with the trends we are currently seeing, seeing him boldly put his money where his viewpoints are.

Bull Markets and Bear Markets

Josh Kale:

We can discuss the reasons for a bull market and a bear market. When you enter such a portfolio, what are some criticisms or cautionary points? The first thing that comes to mind is that this investor is only 24 years old; I am not sure if he has the experience that many other investors have, and to some extent, this could be an advantage, but at some point, will this advantage crumble?

Another thing that concerns me is that the investment philosophy of this fund resembles a single thematic bet. If the growth rate of AI infrastructure and related expenditures slows down, or if the macroeconomic environment changes, every position in this portfolio could be subjected to downward pressure. There is hardly any hedging space here. So this strategy does have some potential vulnerabilities, but at the moment, all signals indicate that this fund's performance will continue to rise.

Ejaaz Ahamadeen:

If you look at some of the most renowned investors of our time, their success has never hinged on how much money they made in a particular year or quarter, but on whether they can consistently deliver stable returns year after year, decade after decade through compounding growth. Leopold's start has been astonishing, with his performance far exceeding the average level of hedge funds in any industry, not just in AI, but he will still need to prove himself over a longer timescale; time will tell.

I just want to say that this person who was once fired by OpenAI has profound insights into the future of AI development, and has made the boldest predictions, and he is the only one so far who has been nearly accurate in all predictions. He has put considerable effort into that 165-page paper, full of confidence in his viewpoints, and as it stands now, everything is paying off for him.

Will the future change? It might. But you can view these reports and investments as real-time tracking tools for his insights on bottlenecks in the AI race; I want to emphasize this point. Initially, his fund's investment philosophy focused on GPUs. He believed that GPUs would become a demand hotspot, and the market underestimated this opportunity. Now his viewpoint is that this opportunity has been fully priced in by the market and that the next bottleneck he sees is shifting toward energy infrastructure.

Look at Elon Musk, he is launching data centers into space. Why? Because the sun provides more energy. And companies like Google, Meta, Broadcom, and Nvidia are all investing in data centers or data center infrastructure to gain access to the power grid. And he is simply putting money into areas where this demand exists; I think that is a smart move.

Josh Kale:

I recently read a great article by Naval, and its core idea is that relying solely on software companies will become very difficult in the future, as it is now very simple to develop and generate custom software. I think his shift is not just around building architecture, but investing in the physical world, such as manufacturing, factories, energy, and infrastructure. These are areas that cannot be constructed through AI but require manpower, licenses, and legislation to achieve hardware and infrastructure, which I think is the direction for future development.

Energy is the only resource that everyone cannot obtain sufficient supply of. Whether it's power generation or real estate investment, everything revolves around one core: powering the future. In the last earnings report season, only a few companies like Google, Amazon, and Nvidia have committed $650 billion in capital expenditures, which indicates that a large amount of capital will be invested in solving this issue, and his portfolio is clearly positioned to capture all of this upside opportunity.

Ejaaz Ahamadeen:

Yes, he has indeed made some investments that you might consider high-risk. For example, unless you are very familiar with the energy infrastructure field, many people may not have heard of Bloom Energy at all. But this company can be seen as a first-tier or even top-level energy company, especially in the portable energy space. He pieced these clues together, believing that the power grid cannot support current demand, and so decided to invest in this company. He put in money with extremely high conviction. We are talking about him committing nearly one-fifth of his entire portfolio to this one target.

This is a highly concentrated, high-risk, high-conviction investment approach. But if it succeeds, that’s why his portfolio has achieved returns of 4.5 to 5 times in a year and a half. We have to give him respect for increasing $1 billion to $5.5 billion in just a year; it’s simply incredible.

The Future of Leopold's Investments

Josh Kale:

Overall, it is astonishing that he has achieved such accomplishments, and his latest shift from hardware to infrastructure to energy seems directionally correct and holds great promise. If you agree with his investment portfolio, then this may be a worthy opportunity to watch. Of course, this is not investment advice, it’s just this person’s portfolio; however, it does look very promising and may perform exceptionally well this year.

Josh Kale:

I am also very curious about what our audience thinks. I want to know if you consider our investment analysis to be professional level, if it reaches Leopold's standards, or if you think we are completely wrong and have missed some obvious stories.

Ejaaz Ahamadeen:

Do you know what I want to know? I want to know what you think is the best stock this year.

Josh Kale:

Yes, Leopold has bet on Bloom Energy. I want to know, what is your Bloom Energy? What have we missed that needs to be known to achieve 5 times growth again this year?