저자: FinTek채널진행자 · 외국인
요약 보기
마이클 버리가 엔비디아·팔란티어에 대해 약 10억 달러 규모(옵션 기준, 실제 투입은 수천만~1억 달러 추정)의 숏 베팅을 한 것으로 소개. 근거는 AI 기업 간 자금 순환(엔비디아→오픈AI→오라클→엔비디아), 테크 capex가 닷컴버블 수준 근접, 클라우드 성장률 둔화. 반론으로 제번스 역설, 팔란티어의 실질 고객 기반, 엔비디아 매출·이익 성장을 제시. 진행자는 AI 개별주 숏은 하지 않고 연금 포트폴리오를 S&P500·소형 성장주·해외주식으로 분산하겠다고 밝힘.
다듬은 글
Well, it's finally happened. Michael Bur is back, and he has just made his first public announcement in years, breaking his 2-year Twitter silence with a huge $1 billion bet against AI stocks Nvidia and Palunteer. So, if you don't know who he is, Michael Bur was one of the few investors who predicted the housing bubble before the whole market collapsed in 2008.
He ended up profiting over $700 million for his investors and over $100 million personally off that one trade. But now his latest big short is focusing squarely on the AI market. So if you've been following the market at all recently, you've seen the rise in AI stocks like Nvidia, which is up more than 13 times in the past 5 years to a nearly $5 trillion valuation.
While the AI software company Palunteer has seen a similar 1,200% increase to a valuation just shy of half a trillion. In fact, AI and tech stocks have risen so quickly that they now make up more than a third of the entire S&P 500's value. Something that investors like Warren Buffett and Michael Bur have been warning about for years.
But Michael Bur in particular is interesting since he's correctly called every last major stock market crash. He shorted the dot bubble in the early 2000s. He famously predicted and profited from the subprime mortgage crash that kicked off the Great Recession. He even made a small fortune shorting Kathy Wood's Arc Innovation ETF during the tech crash in 2021.
Some people love him, some people hate him, but you can't deny this guy has made a lot of money off of his market predictions. But after deleting his Twitter in 2023, Bur went silent on the markets, at least publicly. But that has now changed. He recently reactivated his Twitter and tweeted out three charts just one day after the SEC released this document, the 13F filing for Michael Bur's fund, Cyan Asset Management.
It's like a cheat sheet for everything Bur has been buying or selling. So, let's see what he's been doing. So, we can see Michael Bur currently holds shares in a few companies from last quarter. Brooker Corp and Molina Healthcare, a biotech and healthcare stock, Lululemon, and a small position in SLM, a new consumer banking stock.
He also made a call bet on oil stock Hallebertton, and on the pharmaceutical stock Fizer. But what caught my attention in this report were Bur's two biggest bets against Nvidia and Palunteer worth up to 186 and $912 million respectively. I mean, he's putting some serious money against these companies.
But why these companies? And why now? Normally, these trades are delayed by up to 45 days according to SEC rules. But based on Bur's recent tweet, we can be pretty confident that this is a bet he is actively making right now. So, let's look at the three biggest reasons Bur may have made this bet based on the data in his first tweet in two years.
And then we'll look at two risks that any investor should watch out for before directly copying Bur's trades here. Now, one nuance here. Even though those numbers add up to a billion dollars in total, that represents the total amount that Bur could earn on this trade. In reality, he made this bet using options and so only a portion of that is actually cash that left his hand to do this.
meaning it's more likely that he has a few tens of millions to maybe a hundred million dollars of his money on the line. But it is a full-on bet because this is an options trade. If these stocks fail to drop in price, Bur could be out the entire amount that he bet. So for us regular investors, it's a sign of Bur's conviction in this short.
So the first big hint we get at why Bur made this bet comes from this chart that he tweeted out. It shows the so-called AI money goround where AI companies are just spending money on each other to create this fake growth. So look at this. Nvidia invests $100 billion into Open AI. This was a promise by Nvidia earlier this year to put $100 billion into OpenAI as a venture investment.
And this money has helped push OpenAI to a now $500 billion valuation. Then we see OpenAI sign a $300 billion deal for Oracle's cloud computing, driving up Oracle's stock price by more than a third overnight. Then to complete the circle, Oracle puts tens of billions of dollars into Nvidia's chips, and that order helps boost Nvidia's valuation.
You have a bunch of money moving around the market, but is it really creating new value, or is it just inflating the net worths of the biggest shareholders? That seems to be the question Bur is asking. And this whole setup has led to some pretty fair criticisms that AI's growth is not organic and it may even be a bubble.
Now, there's been a lot of talk about an AI bubble, and I don't want this video to be all doom and gloom. So, in addition to presenting Bur's main reasons for betting against the AI market, I'll also give a little bit of push back on each of Bur's main reasons because there is a perspective on the other side and I don't want to be all doom and gloom, even though I do agree with Bur on most points.
In fact, according to a survey by Bank of America, 54% of investors believe that we are in an AI bubble right now. And we'll see in a second a few other reasons that Bur is pretty worried that this bubble might burst relatively soon. Now, one caveat here. While there is a lot of money flowing around inside the AI market, there's also a lot of money entering the market from outside.
Take Palunteer for example. They have a bunch of paying customers that are not AI companies, 742 of them, in fact. And that's not including their government clients. Now, that doesn't mean there's not an AI bubble that's being propped up by other AI companies, but it does mean there is some value here.
The question is, how much should it be valued? And are we entering a situation similar to the dot bubble where value was purely based on eyeballs rather than dollars made? But Bur also gave us some hints that he doesn't just see this bubble limited to AI stocks. And like we saw after the.com bubble in the early 2000s, we could see a crash in the AI market spread to the entire stock market and even the global economy.
So, as we talk about the broader effects that AI is having on the stock market, I first need to talk about a different topic, which is AI spam, a rapidly growing problem. There's been a shift in how people talk about AI recently. Even in some of these videos, if I misspeak or I have a typo on screen, some people ask if I am AI, which I'm not.
as far as I can tell. But AI is getting harder to spot. People just don't know, especially maybe older relatives. And now even roocallers are starting to use AI voices of sometimes real people that you know to trick you into buying something or sending them money. But if you ever wonder how AI spam calls and emails end up getting your information in the first place, they've probably bought it.
There's a whole market out there that none of us know about where your information is being bought and sold. like they know your name, your age, your email, your address, your phone number, even your browser history is out there. All of this information is being vacuumed up and sold on these massive lists.
So, how is all this legal? Well, because technically we have the right to demand that our information be taken down from any of the sites that use it. Here are some of the laws technically, but it is super timeconuming and there are literally [snorts] hundreds of websites out there that have your information.
But all that brings me to this video sponsor, Incogn, which is a tool that I am realizing more and more needs to exist. Because what they do is they go out and they take you off these lists. You don't have to do anything. You sign up for Incogn, you give them permission to act on your behalf and they go out and find all the lists with your information and send these legal takedown requests for you.
And if the brokers push back or new ones spin up, Incogn keeps on pushing until your information is deleted off these lists. So, I signed up for Incogn over a year ago when they sponsored a previous video, and it has changed everything. I can actually use my phone again. 90% of the calls aren't spam.
My email's gotten more under control. And this part's really cool. I get to track the exact progress. There's this dashboard that shows how many requests they've sent out and how many have been successful. And month after month, my name and information is disappearing off of more and more of these lists.
I just feel less weight on my shoulders that my name is on some people search website where they can just find my address or they can use my browsing history to affect my credit score. So, if you too want to take back your privacy, there's a link in the description and a QR code. Clicking it helps support this channel and it gets you a big discount off the service.
It's 60% off with the annual plan for Incogn which is the plan that I use so that they constantly monitor and keep on taking my name off these lists. I'm really glad Incogn exists. I think you'll find them useful. incogn.com/fintech. And with that, let's dive back into Bur's biggest short against AI stocks.
So, the second chart that Bur just shared was this one. It's a chart of the technology and media capex over time. And Bur seems to be implying that this chart could be used to predict the next big market crash. Capex, or capital expenditure, is just a measure of how much money companies spend to buy or upgrade physical equipment like a data center.
But what's important about this metric is it's a long-term investment that usually isn't paid off super quickly. And we are currently seeing a surge in capex spending on technology, especially in AI data centers. But to understand why this surge might be a problem, we need to look back at when we last saw data like this.
Tech company spending on capex has now reached a level similar to those of the dot bubble when hype and speculation over the new internet companies push valuations to ridiculous levels. And after that bubble burst, we saw a lost decade in stocks where the stock market failed to even beat inflation for over a decade.
And a big cause of this was overinvestment where companies would put too much money in investing into infrastructure and data center capacity based on super hyped up growth projections that were not realistic at all. And some people say what happened in the dotcom bubble is exactly what's happening right now.
Companies are overinvesting in data centers. Now the thing is if you were running an AI company, you do want to invest for the future. The question is how much growth do you expect in the future? And if you spend too much on that growth and the growth doesn't appear, how do you pay back all that investment that you put into something that nobody wants anymore?
This is what happened to telecom companies in the 2000s. After investing billions in building up fiber optic networks and data centers, the demand never really showed up. And so we saw vast tracks of unused infrastructure, plunging prices, and the downfall of many high-valued companies. And the numbers today are pretty insane.
So the so-called hyperscalers, Amazon Web Services, Google Cloud, Microsoft Azure, their annual spending on data centers has now passed $320 billion per year. And they don't show any signs of stopping because AI companies keep demanding more and more power for their AI models. And so from Bur's perspective, all of this money is going into investing in new, more powerful AI hardware.
But if the demand for that never shows up, these companies could be in for a world of hurt. So that's a fairly concerning parallel to the dotcom bubble. So now, let me give a little bit of push back here in terms of what some other investors are saying about this investment because some investors believe that the current push for AI data center investment is actually a really good thing.
And one of the big push backs I've seen here is an economic principle known as Jevans paradox. An economic principle that dates back to at least the industrial revolution when they were studying how coal was used to make energy. The paradox goes that as a technology makes a resource more efficient like how much energy you can get out of coal.
You would expect that you need less coal to meet demand. But in reality, demand for that resource often ends up increasing. And this is because when you make something efficient enough, new entrance enter the market and they find new use cases for this more efficient resource. And some investors argue that could be what's happening right now in the data center space.
So even though AI servers and training are becoming more efficient, we're seeing increased demand for those things because new entrance are just jumping into the market, which could mean that current investment here isn't the result of overinvestment and hype. and it's more of a short-term squeeze on companies trying to catch up with all the demand that's out there.
And I think that might be true, but there's another reason to bet against the AI market that Bur didn't mention explicitly, but is kind of running underneath all of this conversation around AI stocks, and that is debt. We can put this as reason two and a half, but how debt is feeding into the AI market is really important context for anyone to understand before we move on to Bur's third reason, as well as the two big risks that investors could see if they tried to copy Bur's trades directly.
See, a lot of people argue that the biggest difference between the dot bubble and the AI bubble of today is that the.com bubble was driven mostly by debt. Internet companies were simply borrowing cheap money in order to reinvest in growth even though they weren't making any money. But modern AI stocks look pretty different.
The balance sheets of the biggest AI companies still look pretty solid. And most of them, at least in their financials, don't seem to be taking on a ton of debt. This is mostly coming from their profits. At least that's how it looks on the surface. But there's some evidence that these companies are taking on a lot more debt than anyone realizes.
Some AI companies are just moving the debt into specially written contracts that avoid having the debt show up on their balance sheets while in other cases the debt is shifting to investors who are borrowing money in order to invest in these companies. This is a chart of margin debt. So the amount of debt taken to invest in stocks and it is currently sitting at well over $1 trillion.
Meaning that if the market fails to continue to rise, those investors will need to pay off that debt in some way or they could default on it, which could trigger a cycle of selling to pay off debts that continues to go on over and over and over again, which is kind of familiar to Bur because this is similar to what happened during the 2008 housing collapse.
Hidden debt that was destabilizing the market. But with all that said, debt is fine as long as the market continues to grow or these companies only invest as much as there is demand. But if growth slows or there's less demand than these companies are planning for, that's when things could get a little messy.
So, how likely is growth to slow? Well, Bur shared a pretty good hint at what he thinks about this in the third chart that he shared. And then after that, we'll look at the two biggest risks for any investors who are looking to copy Bur's strategy here. So, the third chart that Bur shared in his tweet looks like this.
In black, we can see the average cloud growth rate of Amazon, Google, and Microsoft from 2018 to 2022. and it was pretty high. But in red, we can see the same cloud growth rate from 2023 through 2025. And Bur seems to be pointing out that the growth rate is slowing, at least as a percentage. This is a big deal because cloud computing is the backbone of the AI market.
Who do you think is buying up all those Nvidia GPUs? Because up until now, the more computers and power you have to train and run your AI models, the more powerful those models can become. And so Bur by showing that cloud growth is actually slowing down might be implying that the so-called AI demand is not there to justify the valuations of these companies since it's clearly not enough to keep their growth rates up.
In other words, this could be a sign from Bur that AI growth is not going to be as high as the market thinks. Now, I think that is a totally reasonable way to read this data. But I also think you can read it in another way. Yes, we have seen cloud growth slow from 2023 through 2025, but there is a fairly long lag time in how long it takes to build a data center.
I mean, it's a lot of physical stuff that you have to set up. And here's that chart of infrastructure investments from earlier. Remember, this is money that tech companies are investing into setting up things like data centers. And there is clearly a dip in spending from 2023 through 2024, with the really big rise only happening more recently.
And that most recent rise probably isn't showing up in this chart. So if we looked at this same chart in 3 years, we may see growth rates start to tick back up. But I think regardless of what happens with cloud growth, this chart is a good reality check. And that as much as AI companies are hyping endless growth, not every indicator supports that.
But okay, Bur clearly believes that AI and especially Palunteer and Nvidia specifically are overhyped. But what should we do as regular investors? Well, let's take a look at two big risks with just copying Bur strategy directly and then I'll also share my personal approach for dealing with a potential AI market crash.
So, our first big risk is very specific to how Bur is trading these stocks and that is by using options. So, back during the 2008 housing crash, Michael Bur made his first big short, betting that the entire US housing market was on the verge of collapse. But what's often left out in that story is Bur actually first made that prediction way back in 2005.
And over the 3 years that followed, he nearly lost everything before the bubble finally burst in 2008 with his investors revoling and trying to pull their money out before the crash ever happened. And I think there's an important lesson there if you want to short the market without billions of dollars in investor funds backing you because shorting the market is not exactly the same thing as just buying the market in reverse.
Shorting also involves timing. So, you can correctly predict that a stock is going to crash, but unless you get the timing perfect, you can still lose money with shorts. We'll see how the second risk ties into this, as well as a potentially better strategy for dealing with this. But there's a phrase that captures this idea pretty well.
The market can remain irrational longer than you can remain solvent. And that is why the options market can be very risky. Even if you're confident that a stock should drop, sometimes the market just behaves irrationally for a very long time. But the second risk is a little bit broader and I think it gives us a little bit more actionable advice for what we can do.
So Bur's bet here requires that Palunteer and Nvidia drop in price. And while AI valuations have absolutely ballooned, and we've even seen plenty of tech CEOs, including OpenAI CEO Sam Altman, call AI a bubble directly, but it's not 100% clear that that bubble applies to every AI stock. Tech investor Ross Gber or Gber, he had a pretty interesting take on this.
He said about big tech that their valuations aren't out of sync with their business. Investors just aren't used to seeing numbers that big and their profitability is actually pretty insane. So basically, for every fake AI startup out there that makes no money and is selling AI blenders, there's also a company like Nvidia.
Yes, Nvidia's stock price is up 1,200% in 5 years, but their revenue is up around 700% and their net income, basically their profit, is up 2500% in that same time. Is the stock expensive? Sure, they are absolutely factoring in for future growth. But it's not entirely irrational. It's just that some investors do think it will continue to grow.
And Fed Chairman Jerome Powell even spoke out against a direct comparison between AI stocks and the.com bubble, saying, "AI stocks actually have earnings, unlike the dot companies." It's not the strongest endorsement, but hey, he's not one for hyperbole. But all that being said, what am I doing with my money?
Well, I don't currently hold shares in Nvidia specifically, but I also don't plan to short them because I am not confident in my own ability to time the market like maybe Michael Bur can. On Palunteer's side, I'm actually working on a dedicated deep dive into that company, applying my background as a cloud engineer to figure out what they actually do and what they should be worth.
So, I'll offer my specific thoughts on that company in that video. But, in general, I don't believe that AI companies that do have a genuinely good balance sheet that are profitable and are expecting future growth necessarily have to crash. But, I do think they are assuming a lot of future growth. And if that future growth doesn't show up, that could be a warning sign to investors that it's time to get away from those stocks.
Or if you don't think that growth will appear today, like Bur, I can understand not wanting to play the game in AI. But personally, I plan to handle the AI risk by splitting my retirement portfolio, which is the money that I need to live on when I retire, into one-third S&P 500 ETFs, 1/3 small cap growth ETFs, and one-third into international stocks.
So, even if the big AI companies drop, I still have other parts of the market to balance out my portfolio and hold it up through diversification. That's the best strategy that I've found based on the data I've seen. But even that isn't bulletproof. I recommend checking out this video about Warren Buffett's recent warning about the market that goes far beyond AI stocks.
Comments 0