Compiled & translated by: TechFlow TechFlow
Speakers: Josh Kale, Marketing, AnthropicAI; Ejaaz Ahamadeen, former Product Manager, Coinbase
Podcast source: Limitless Podcast
Original title: Leopold Aschenbrenner says "No More Stocks!"
Broadcast date: June 17, 2026
Key points summary
Leopold Aschenbrenner, considered one of the world's most aggressive AI investors, has short NVIDIA, ASML, and Oracle in the open market with a nominal position of approximately $9 billion, while simultaneously shifting funds towards deeper AI infrastructure and model assets such as power, memory, data center networks, and Anthropic. The two hosts believe this doesn't necessarily mean the AI bubble has burst, but rather signals a shift in infrastructure trading from a "chip-first" to a "energy, network, and data center construction-first" approach. This is especially true after NVIDIA recently completed a $25 billion bond financing round and Anthropic's valuation has been inflated; the market implications of this assessment are rapidly amplifying.
Summary of key viewpoints
Leopold's core trading logic
"The classic 'shovel-selling' trade in AI is already too crowded, and Leopold's recent position changes convey this signal."
"His assessment is not that AI infrastructure has peaked, but rather that certain layers within the infrastructure stack, especially semiconductors and traditionally popular stocks, are already overcrowded."
"If the question becomes, where will the funds go next? There are two answers. The first and most direct is that they will flow to the next real infrastructure bottleneck, namely, electricity, memory, and data center networks. The second answer is that mysterious investment that was only revealed a few weeks ago."
"His bets have always been very infrastructure-oriented, investing in both optics companies and power-related companies."
"If he's cautious about NVIDIA, the funds will go to things like power and memory; at the same time, he also wants to invest directly in the 'mining' itself, instead of just buying 'shovels,' and Anthropic is his favorite mining platform."
Signals from NVIDIA's Funding
"The issue isn't whether NVIDIA will continue to make money, but why a company with extremely high profit margins and plenty of cash on hand would need to borrow another $25 billion from outside."
"If a company is simultaneously making large-scale stock buybacks, drastically increasing dividends, and borrowing money in the same month, it's clearly not borrowing because it's short of funds. A more reasonable explanation is that it's seeking cheap capital, and the financing methods in this AI boom are undergoing a slight shift."
The next wave of AI infrastructure dividends
"The real bottleneck is no longer just GPUs, but power, memory, data center networks, and the ability to actually build these things."
"Even if you raise a lot of money, you can't build data centers quickly enough, expand memory chip production capacity sufficiently, or immediately expand the power grid, power lines, and related infrastructure. There aren't enough people on the ground, and approvals, regulations, and various procedures are also holding you back."
"Whoever has the capability to build the data center will make the money."
Optical modules, copper and optical fibers
"As GPUs become larger, copper wires will get hotter and lose more energy, resulting in poor efficiency. Fiber optics will then become the next upgrade direction."
"In high-bandwidth, short-distance transmission scenarios, copper is almost the only material that people truly want to use. Only when it becomes unsuitable, such as due to excessive distance or heat, will it be switched to fiber optics. Therefore, the market demand for a combination of copper and fiber optics is currently very strong."
"Copper futures have been performing very strongly recently, essentially because everyone needs it. It's the most crucial basic material for short-distance, high-bandwidth transmission, while optical fiber is the next step."
Copper remains the most critical material for short-distance, high-bandwidth transmission, but as the distance increases and the heat becomes too high, it must be replaced by optical fiber.
"The next wave of funding will go to infrastructure companies that don't sound sexy."
Why is energy the safest bet?
- "I've always been bullish on energy because even if AI demand slows down, energy itself will remain a global necessity, and that demand will only increase."
- "The single trend that will continue to rise regardless of the circumstances is our demand for energy, electricity, and power, and these are the companies I'm most willing to long in long-term."
- "What I most want to follow are the companies Jensen invests in, and whose investment logic overlaps with Leopold's. So, the company I'm closest to copying right now is Marvell."
- "The best long-term holdings are not necessarily the hottest chip companies, but rather the power infrastructure sectors that are unavoidable regardless of the macroeconomic environment."
Leopold's AI Portfolio
Josh Kale:
Leopold Aschenbrenner, a 24-year-old who specializes in AI investment, is now widely regarded as the world's leading AI investor. Rumors suggest his fund's nominal holdings have exceeded $20 billion. A month ago, when we read Ejaaz's post, his fund was only $13.7 billion, meaning it's essentially doubling every quarter.
We've obtained some significant new developments in his recent investment moves. In the last episode, we discussed his portfolio, and the most surprising point was that he was short a company almost everyone is familiar with: NVIDIA, the world's most valuable company and the hottest topic in AI. Many people couldn't understand why he would make a short position of over $9 billion in such a company.
We've now obtained a new clue that might explain this. NVIDIA is raising funds, and through bond issuance. On the surface, this seems illogical. Why would a company as large and highly profitable as NVIDIA need to raise another $25 billion in cash, something it just completed? Today, we want to discuss Leopold's investment portfolio to understand why he's made so much money, what he's looking at next, and what this funding round for NVIDIA really means.
Ejaaz Ahamadeen:
Let me give you some background first. Leopold Aschenbrenner was a researcher at OpenAI. About a year and a half to two years ago, he raised a fund. The initial size was not very large, I think it was about $200 million, but according to his most recent 13F filing, the fund's public holdings are now worth $13.7 billion.
So the market is naturally eager to know exactly which positions he has invested in, what his core investment logic is, and where his next big deal will be made. To understand this, we must first know that a month ago, Leopold was actually very optimistic about the entire AI sector, especially the "shovel-selling" logic, which refers to GPUs like NVIDIA and upstream hardware suppliers.
But about a month ago, the market realized he wasn't so bullish on the semiconductor sector. He remained optimistic about the real bottlenecks like memory and power, and might also be bullish on emerging cloud vendors, but he was conspicuously bearish on NVIDIA, the world's most valuable company. More specifically, he had a total of about $9 billion in short positions in several companies considered core beneficiaries of AI infrastructure, including NVIDIA, ASML, and Oracle.
The logic short NVIDIA
Ejaaz Ahamadeen:
This news has caused concern among many, who fear that the AI bubble is about to burst. After all, on the surface, NVIDIA's GPUs are still selling well, and demand doesn't seem to be weakening significantly. So where exactly is the problem?
We later uncovered several new clues, the most important of which was that NVIDIA had just raised $25 billion through bond financing. This means it wasn't simply using its own cash reserves, but rather adding additional leverage. So the question arises: why would one of the world's most profitable, highest-margin, and strongest cash flow companies need to borrow $25 billion from outside?
Josh Kale:
Moreover, they initially planned to raise only $20 billion, but it eventually expanded to $25 billion, with subscriptions exceeding three times. In the last episode when we discussed this portfolio, we mentioned not to worry about a bubble, because although these companies have huge capital expenditures, their revenues are also high enough that they can theoretically support their expansion with their own balance sheets.
But this is the first time since 2021 that NVIDIA has explicitly sought financing off-balance-sheet rather than directly using its own cash reserves. I recall it currently has around $12 billion in cash on hand. Putting all this together creates a strange tension: on one hand, Leopold is short, and on the other hand, NVIDIA seems to have unlimited cash and profits, yet it's still issuing debt. So what exactly happened?
NVIDIA Bond Financing Breakdown
Josh Kale: Ejaaz, could you break this down for us? Because this isn't a typical financing deal, it's a bond issuance. Ultimately, NVIDIA now has another $25 billion on its balance sheet, and the interest rates look quite low.
Ejaaz Ahamadeen:
I'll present both explanations. NVIDIA already had approximately $13.7 billion in cash on hand, meaning it could easily spend its own money. So why seek external financing? The simplest analogy is buying a house. Many people, even if they have the cash, will still choose to take out a loan because they can use their own capital for other things, and if the cost of borrowing is low enough, it's actually more cost-effective.
The interest rate environment has been unfavorable in recent years, but if you're NVIDIA, one of the world's most valuable and sought-after companies, you can borrow money on very favorable terms. This $25 billion bond financing has maturities ranging from 2 to 30 years, making it almost like very cheap money, with interest rates approaching the yield on US Treasury bonds.
Moreover, this funding round was approximately four times oversubscribed, meaning there was $85 billion in market capital vying for the $25 billion allocation, giving NVIDIA virtually free choice in choosing investors. According to NVIDIA's official explanation, this was primarily a financial arrangement to repay and refinance some existing debt. Google did something very similar a few weeks ago and again in February of this year. Therefore, you can certainly accept this explanation and view it as a financial optimization.
However, it's also hard to ignore the other side: in the past month and a half, NVIDIA, Amazon, Google, and several other hyperscale cloud providers have almost all been raising external funding. Some have issued bonds, others have sold stocks. Leopold's view may not be entirely without merit; could this be a signal that the bubble is starting to burst and the house of cards is beginning to shake? However, if you only look at the financial structure, this doesn't yet clearly point to danger.
Josh Kale:
That's my view too. $9 billion shorting NVIDIA is a truly massive position. But in our research, we also noticed something else: on May 18th, NVIDIA's board authorized an additional $80 billion in share buybacks and increased the dividend from 1 cent to 25 cents per share, a 25-fold increase.
If a company simultaneously engages in large-scale stock buybacks, drastically increases dividends, and borrows money in the same month, it's clearly not borrowing because it's short of funds. A more reasonable explanation is that it's seeking cheap capital, and the financing methods in this AI boom are undergoing a slight shift. Everyone wants to participate in these capital operations, and NVIDIA realized that raising funds through bond issuance was even cheaper than other financing methods, so it simply went ahead and did it. At least for now, NVIDIA itself is still doing quite well.
Why did he adjust his portfolio?
Josh Kale: That brings us back to another question. What is Leopold thinking? Why did his judgment change? The stock chart you just showed also illustrates that NVIDIA's recent performance hasn't been particularly strong, but it hasn't been terrible either. It's still the world's largest company with a market capitalization approaching $5 trillion, and a 7% drop in a month is nothing compared to the soaring performance of other AI stocks.
Ejaaz Ahamadeen:
I don't think NVIDIA will disappear. Its GPUs, including the CPU product line launched just a few weeks ago, are expected to perform exceptionally well. The demand for AI products is currently in exponential excess, and the core machine suppliers that can truly meet this demand are primarily NVIDIA.
However, I do feel that the classic "shovel-selling" trade in AI is already too crowded, and Leopold's recent position changes convey this signal. Looking at his recent 13F chart, you can see that his short positions are clearly biased towards the semiconductor sector, such as NVIDIA, ASML, Oracle, and several other infrastructure-level companies.
At the same time, he also heavily invested in memory, power, and new cloud technologies. This suggests that his judgment is not that AI infrastructure has peaked, but rather that certain layers in the infrastructure stack, especially semiconductors and traditionally popular sectors, are already overcrowded.
If the question becomes where the funds will go next, there are two answers. The first and most direct is that they will flow to the next real infrastructure bottleneck, namely, electricity, memory, and data center networks; the second answer is that mysterious investment that was only revealed a few weeks ago.
Anthropic positions were accidentally exposed.
Josh Kale:
This is the most surprising thing for me. I only found out yesterday after you told me, and my first reaction was that it was impossible. Could it be that Leopold's "Situational Awareness" fund actually has 20% allocated to Anthropic? There are rumors that this company accounts for about one-fifth of Leopold's fund's holdings. The Wall Street Journal and several other media outlets have said so, and people very close to the deal have also confirmed this.
This became a completely unexpected card in his portfolio. Because 13F filings only disclose publicly traded holdings, not private equity, and Anthropic happens to represent a large portion of unlisted shares. This is precisely why people began to understand why the outside world valued his portfolio at $20 billion.
If 20% of the fund is held by Anthropic, and he invested in it around early 2025, the returns for that year would feel like seven years have passed for Anthropic. This change would significantly alter our understanding of his entire portfolio.
Ejaaz Ahamadeen:
Yes. He first invested in Anthropic through private channels or a fund around March 2025, at which time Anthropic was valued at approximately $60 billion. Now, based on the latest valuation round, it has been valued at $965 billion.
This equates to a nearly 15-fold increase. According to the calculations shown in our program today, the value of his most recent 13F disclosure of his liquid portfolio was $13.7 billion. If we add the Anthropic position reported in the Wall Street Journal, it would increase by approximately $7 billion, bringing the total fund assets under management to $20 billion.
How outrageous is this? Even Bill Ackman, a top investor with three or four decades of experience in the market, has a Pershing Capital fund of roughly $20 billion. Leopold has only been in this game for a year and a half, and he's only 24 years old with almost no real investment experience.
But he made some incredibly astonishing predictions, and what's crazy is that he had actually written all of this down in advance. A year and a half ago, when he launched his fund, he published a 65-page article on AI called "Situational Awareness," which explained the entire logic almost completely, including how funds would rotate from semiconductors and some infrastructure sectors to other bottleneck constraints. Now the market is developing along this line, which is truly amazing.
The next wave of infrastructure investment
Ejaaz Ahamadeen:
So this also tells me where the money will flow next. If he is cautious about NVIDIA, the funds will go to places like electricity and memory; at the same time, he also wants to invest directly in the "mining" itself, instead of just buying "shovels". Anthropic is his favorite mining platform.
Josh Kale:
This certainly looks like a new trend, and he's still one step ahead of most people. For the past 12 months, everyone has been searching for the bottlenecks in AI, exploring rare metals, memory, RAM, and so on; the market chased each of these areas. These assessments weren't entirely wrong, because that wave of market activity did indeed occur.
However, looking at the current landscape, valuations of those sectors that were once considered bottlenecks are gradually becoming more reasonable. People have a better understanding of these companies' business models, market potential, and future revenue, so much of their value has already been priced in. In the next round, we're more concerned with where the money will continue to flow.
You just mentioned land, electricity, rocket casings, and physical infrastructure, and this direction seems correct. Because if we think about what AI is truly most important about, the answer increasingly seems to be the ability to build physical infrastructure. Look at xAI, or more accurately, look at SpaceX, which is already publicly listed; its core revenue doesn't come from the rockets themselves, but from building AI infrastructure.
Looking at its recent deals with Anthropic and Google, the value they bring exceeds the combined value of Starlink, Starship, and the entire satellite business. There is clearly enormous demand and immense value involved. So the question becomes: who can actually build these things?
SpaceX is clearly an answer. Its stock price reached $230 in after-hours trading last night, corresponding to a valuation of approximately $3.1 trillion. We'll be dedicating a week to SpaceX because its recent surge has been truly phenomenal. Having just completed the acquisition of Cursor, its valuation has now reached $3 trillion. Elon Musk has earned more in a single day than Warren Buffett earned in his entire career.
Who will reap the next round of benefits?
Josh Kale: What we're interested in is which companies are best at building this kind of hardware infrastructure, at developing those "machines that make machines." Considering Leopold's direction and the overall trend, we think funding will move in this direction next. So, Ejaaz, which companies will actually benefit from this rotation?
Ejaaz Ahamadeen:
Many of these will be infrastructure companies that don't sound sexy. Marvell has been a frequent name mentioned in the past month. A few weeks ago at Computex in Taiwan, Jensen Huang stated directly on stage that it would be the next trillion-dollar company.
Just three months before he made this statement, NVIDIA had invested $1.5 billion in Marvell. I'm starting to wonder if this counts as insider trading or market manipulation, because after he said that, the stock price jumped 70%.
I think it's actually quite easy to directly conclude that AI infrastructure has peaked right now; but if you compare it to historical financial crises, such as 2008, the current round hasn't fully exhibited the characteristics of high leverage, financial engineering, and systemic manipulation.
There are two key differences. First, the products these companies are making today are genuinely being bought by people. Neither the dot-com bubble nor the financial crisis saw such solid, real demand. Second, constrained by the laws of physics, we can't actually leverage indefinitely right now, because the entire system is limited by manpower and infrastructure capabilities.
Even if you raise a lot of money, you can't build data centers quickly enough, expand memory chip production capacity sufficiently, or immediately expand the power grid, power lines, and related infrastructure. There aren't enough people on the ground, and approval processes, regulations, and various procedures are also holding you back.
So I actually think this gives investors an advantage. Since you already know that the hottest chip and shovel-selling deals are overcrowded, the money will flow to power, data networks, companies like Astera Labs, and then to other related sectors. What you really need to think about is when these contracts will start to materialize, when these wafer fabs will actually be built, when SpaceX's rockets will be able to launch AI satellites, and even when they can start using solar energy to train AI models.
The timeline dictates the betting pace. At least, that's the framework I use for investing, though this isn't investment advice. I see it this way because over the past year and a half, we've witnessed firsthand how funds have flowed from general AI stocks to semiconductor and infrastructure trades.
Josh Kale:
If you continue looking at this portfolio chart, you'll find that this story is clearly written into his holdings structure. By category, what's his largest allocation? It's electricity and energy. Next is memory, followed by cloud computing and GPU miners—the most tangible infrastructure.
He wants to own shares in emerging cloud providers like CoreWeave, as well as in miners who have already switched to cloud computing power. What he wants is this physical infrastructure, because he believes it's the real bottleneck. As you just mentioned, there are many more detailed aspects involved, such as actual construction, hardware manufacturing, and the data center construction itself, all of which are extremely difficult.
If there's a major bottleneck, it might be the approval and licensing processes. So who's solving these problems? SpaceX wants to move data centers into space, and Tesla wants to use humanoid robots to solve the manpower issue. But both of these are still far off. In the short to medium term, there are actually many untapped opportunities, and that's precisely where Leopold is betting.
Advantages of optical modules and optical fibers
Josh Kale: I'd like to add a detail we haven't discussed in detail yet. For those who want to dig deeper and find more excess returns, many clues are actually hidden in optics and the underlying technology stack. Ejaaz, you've been researching this recently, could you talk about your thought process?
Ejaaz Ahamadeen:
If you look at the listings on his screen, CoreWeave and Iron are basically top-tier new cloud service providers. Simply put, they're somewhat similar to Amazon Web Services, except that AWS provides cloud services to internet companies, while these companies provide ready-made GPU infrastructure to AI companies.
They handle everything for you – GPUs, networks, deployment – so AI companies don't have to worry about the underlying infrastructure and can directly train models and access computing power. CoreWeave and Iron have been among his largest concentrated positions since he first established them, and they have also brought him the highest returns.
It's also worth noting that he still holds these two companies in his largest positions today. This suggests that, in his view, the deal is far from over. Furthermore, he has also privately invested in Core Scientific, a company that can help unlock CoreWeave's infrastructure supply capacity. In a sense, he's adding another layer of leverage to CoreWeave.
Beyond these, look at companies like Coherent and Lumentum; they're essentially suppliers of fiber optics and optical connectivity. To put it simply, communication between semiconductors and GPUs traditionally relies on massive amounts of copper wire. The problem is, as GPUs grow larger, copper wire gets increasingly hot and loses more energy, becoming very inefficient. Fiber optics, in this context, becomes the next upgrade. It can transmit data faster, is more cost-effective, and allows companies providing inference and training computing power to make more money. So you'll find that he's consistently betting on very infrastructure-oriented things, investing in both optical companies and power-related companies. It might not sound glamorous, but in my view, this is where the money is truly flowing.
Josh Kale:
Copper is also very interesting to me because I've only recently realized how crucial it is for short-range data transmission. In many high-bandwidth, short-distance transmission scenarios, copper is practically the only material that people really want to use. Only when it becomes unsuitable, such as due to excessive distance or heat, will we switch to fiber optics. Therefore, the market demand for the combination of copper and fiber optics is currently very strong, which is why observing this copper trade is so interesting. Copper futures have been performing strongly recently, essentially because everyone needs it; it's the most crucial foundational material for short-range, high-bandwidth transmission, while fiber optics is the next step.
Thinking about it from a deeper level, the materials science field has always been fascinating. At the very bottom of all the layers, it's about what raw materials are most essential for achieving intelligence. Copper is one, lithium is another, and there are many others. We really should dedicate a separate issue to materials. Perhaps before Leopold even reaches that level, we'll be able to see the next wave of change.
Josh Kale:
If we keep going down the stack, we can even go directly to a copper mine to see how these things are made. But going back to the core judgment, I think the next round of rotation will indeed shift from those seemingly smaller bottlenecks to the truly difficult things, namely hardware and the construction of large data centers.
Whoever has the capability to build data centers will make the money. We've already seen how much SpaceX has earned from the overwhelming demand for data centers. Whoever can bring more data centers online faster, and whoever can provide enough power and GPUs, will make the most money. This is essentially what Leopold is betting on right now.
Has a bubble appeared?
Josh Kale: In summary, we don't think we've entered a bubble burst phase. Leopold's positions seem more like rotation than a complete retreat. So, should we continue to follow his lead?
Ejaaz Ahamadeen:
I admit that when I first saw his 13F, my initial reaction was, "This guy is actually bearish on the world's most valuable company, with demand booked until 2029? That's outrageous!" But now, seeing this round of financing, I'm starting to think that if NVIDIA continues to take on external debt, and may even sell shares in the future, and if this trend continues, then Leopold might be right again.
If that's the case, his fund might eventually surpass the world's top traders and best investment funds. He's truly been winning consistently, which is hard to deny.
Josh Kale:
However, there's another important point. His entire life has almost exclusively been about long; he's never truly experienced the test of large-scale selling. As we mentioned earlier with Bill Ackman, achieving a 30x return and surviving in the market for 30 years are two different things.
If he can truly maintain this growth and learn when to sell, how to manage risk, and how to protect himself with hedging, that would be even more formidable. We are already beginning to see the beginnings of this capability. That $9 billion short position wasn't actually achieved through direct short with $9 billion in cash, but rather through options and leverage; it wasn't a naked short position. In any case, this matter is worth continuing to observe.
Energy is the core bet
Josh Kale: If you had to pick one stock from his entire portfolio that you would most like to buy, which one would you choose?
My own answer is energy stocks. I've always been bullish on energy because even if AI demand slows, energy itself remains a global necessity, and that demand will only increase. Even without considering AI, we still need more energy and more electricity. Companies like Bloom Energy, which can improve power supply and transmission capacity, are the area that excites me most, because they are the most like a hedging bet. The single trend that will continue to rise regardless of circumstances is our demand for energy, electricity, and power, and these are the companies I'm most willing to long positions in for the long term.
Ejaaz Ahamadeen:
My answer is a bit of a cheat. What I really want to follow are the companies Jensen invests in that overlap with Leopold's investment logic. My closest target right now is Marvell. Although it's not a publicly owned company by Leopold, it perfectly aligns with his betting direction in fiber optics and electricity, and Jensen has already invested $1.5 billion of his own money in it.
I've observed a trend: whenever Jensen invests in a company through NVIDIA, whether it's Intel, CoreWeave, or something else, the stock price almost always goes up. So, my current portfolio is roughly around this level. I also hold some CoreWeave myself, because both Jensen and Leopold are extremely bullish on it.
Josh Kale:
Marvell has risen 270% in the past six months. This may really be a good rule of thumb: if someone like Jensen, or even someone with huge influence like Trump, publicly says they're going to buy a certain stock, you should probably take a serious look.
Past experience has repeatedly shown that such signals often have a very high probability of materializing. Cases like Intel and Marvell demonstrate that, on the one hand, these signals clearly indicate the market's understanding of its message, and on the other hand, the market has the power to influence these companies' outcomes. Therefore, this market movement has been truly frenzied.
I hope it continues. And judging from the current situation, it certainly seems likely to continue. At least for now, we remain bullish and optimistic, and will continue to make judgments based on daily changes.
Josh Kale: Is there anything you'd like to add regarding the Leopold portfolio update?
Ejaaz Ahamadeen:
I'd really like to hear what those who are skeptical think. If, after listening to our analysis, you feel we're completely wrong, or that we've misunderstood something, please feel free to point it out.
Yesterday, I stared at NVIDIA's $25 billion funding news for a long time, initially intending to nitpick. But from a purely financial perspective, it does make sense. Why not borrow such cheap, virtually risk-free money? Borrowing other people's money to expand is clearly more reasonable than selling your own equity, because this way you retain more future earnings.
