원문 보기
Michael Bur has a way of shaking people awake. And his newest warning about a massive AI bubble is one of those moments that makes you stop and think, "All right, what exactly is he seeing that everyone else is missing?" Because when the same guy who predicted the 2008 collapse says the market is repeating old mistakes just wrapped in newer, shinier technology, you can't help but pay attention.
Right now, it feels like AI can only move in one direction up. Stock charts are hitting levels nobody expected. Social media is buzzing with overnight millionaires. And there's this pressure that if you aren't putting your money into AI stocks, you're basically choosing to be left behind. But Bur isn't swept up in that excitement.
He's stepped back and is breaking down in detail why he believes the AI story is running far ahead of reality, especially on his Substack where he lays out the numbers behind his concerns. That question sits at the heart of everything. Is AI truly the unstoppable revolution it's being sold as? Or are we watching early clues of a bubble forming, the same way people ignored warnings during the late 1990s?
Are companies pouring billions into data centers, chips, and AI infrastructure because real demand is exploding or because they're terrified of missing out? And if Bur is right, what does that mean for you if your savings, retirement account, or trading portfolio is already tied to these soaring tech names?
This video breaks the whole thing down in the simplest possible way. I am going to unpack exactly what Bur is seeing, why he thinks the numbers don't line up with the hype, why is AI repeating.com history, and most importantly, if you already own AI stocks or want to invest in this space, what should you actually do with this information?
To answer that, we first need to understand who Michael Bur really is and how his mind works. Michael Bur's story doesn't start with hedge funds or Hollywood. It starts in hospital corridors, long overnight shifts, and a mind that refuses to stop questioning how things really worked. Before anyone on Wall Street knew his name, he was a medical doctor at Vanderbilt and a resident at Stanford, spending his daylight hours treating patients and his late nights reading financial filings for fun.
He posted his analysis online, and those posts were so sharp that professional investors started following him long before he ever thought of managing money. That curiosity eventually pushed him to launch Cyan Capital in 2000. From the first day, he showed he wasn't built like other investors. While the.com bubble was turning every internet idea into a billion dollar stock, Bur was in the footnotes, studying real numbers instead of chasing hype.
When valuations stopped making sense, he shorted several overvalued tech names and made money as the bubble burst. That willingness to stand alone defined him long before anyone called him the big short guy. And of course, 2008 sealed his reputation. Bur didn't make a huge bet because he felt something was wrong.
He made it because he read loan tapes line by line and found the cracks the entire banking system was ignoring. Those trades earned him around $100 million personally and more than 700 million for his investors. He wasn't guessing. He was early, precise, and stubborn enough to trust the numbers even when the world said he was crazy.
Of course, he's not perfect. Some of his post208 warnings didn't play out, but people still listen because of how he thinks. He digs into accounting details, incentives, and history in a way most investors never do. When he raises a red flag, it's because the numbers pushed him there. And that's why his warnings about the AI boom deserve attention.
Now, when Michael Bur says the AI boom feels like a bubble, he isn't attacking the technology itself. He's talking about the financial story being built around it, a story that, in his view, looks almost identical to the pattern that led to the dotcom collapse. To him, a bubble isn't defined by excitement or innovation.
It's defined by money moving faster than real demand, expectations outrunning reality and valuations built on the assumption that growth will continue in a straight line forever. And when he looks at the AI economy right now, those signals are flashing everywhere. Start with the sheer scale of spending.
Bur points to what he calls the public horsemen of AI. Microsoft, Alphabet, Meta, Amazon, and Oracle. Together with Nvidia and startups like Open AI, these companies are on track to pour more than 1 trillion dollars into AI chips, data centers, networking equipment, and cloud infrastructure. That number alone should feel familiar.
During the telecom boom of the late '90s, companies spent billions laying fiber optic cables across the country. But when the bubble popped, analysts discovered that less than 5% of that fiber capacity was even being used. The technology was real. The demand wasn't. Bur fears AI is heading down the same path.
Massive supply buildout, but unproven real world usage. Then there's Nvidia, the company at the center of the AI explosion. In two years, its stock has surged roughly 10 to 12 times and its market is over $4 trillion, making it for a moment the most valuable public company in the world. Bur compares this directly to Cisco during the dot era.
Not because the businesses are identical, but because the pattern of investor expectation is Cisco was the backbone of the internet in the 1990s. As the world rushed online, its stock didn't just rise, it exploded, climbing several thousand% over the decade and becoming the most valuable company in the world by early 2000.
But when expectations finally snapped, Cisco's share price collapsed by more than 80%. And even though the company remained profitable and continued growing, the stock took decades to recover. To bury, today's AI ecosystem shows the same cracks. too much money, too much confidence, and too little proof that demand can justify a trillion dollar expansion.
And that's before looking at the accounting issues that worry him even more. Bur's biggest concern with the AI boom isn't just the spending or the hype. It's the math underneath the earnings. And the piece he keeps emphasizing both on his Substack and through the filings he analyzes is something most people never think about.
Depreciation. It sounds boring, but in Bur's view, this one accounting choice is creating billions of dollars in profits that don't actually exist. And those inflated numbers are helping drive the AI bubble higher. Here's the simplest way to understand it. Imagine buying a phone you know you'll replace in 3 years, but telling yourself it will last six.
On paper, the phone looks cheaper because you're spreading the cost over twice as long. But in reality, nothing changed. You're still going to replace it in 3 years. That gap between the paper version and the real version is exactly what Bur thinks is happening inside the biggest AI and cloud companies.
Over the past few years, the giants behind AI have stretched the useful life of their servers and network equipment, which is the accounting estimate that determines how quickly those assets lose value. Alphabet increased server life from about 3 years to 6 years. Microsoft did the same, 3 years to 6 years.
Amazon went from roughly four years to 6 years before adjusting to 5 years. Meta pushed its useful life assumptions to around 5 1/2 years, telling investors it would cut its 2025 depreciation expense by about $2.9 billion. And Oracle extended its estimates from 5 years to 6 years. These changes make expenses appear smaller today, which makes profits look bigger.
But if the hardware really becomes outdated in 2 to 3 years, as it often does with Nvidia's rapid chip cycles, those stretched timelines can distort reality. Bur estimates that between 2026 and 2028, the major AI hyperscalers could be understating depreciation by roughly 176 billion. that inflated earnings by more than 20% for companies like Meta and as much as 26.9% for Oracle.
In other words, a huge portion of the profits investors are celebrating may not reflect economic reality. And Nvidia, despite not manipulating its own depreciation, sits right in the middle of this. Its explosive growth depends on customers whose earnings may be puffed up by accounting choices. If those customers pull back, Nvidia's demand story gets shakier.
And that to Bur is one of the most dangerous pressure points in the entire AI boom. If Bur is right about the way the AI boom is being built, then the real risk isn't that AI fails. It's that the expectations around AI are so stretched, so aggressively priced in, and so heavily supported by inflated earnings that even a small slowdown could trigger a major unwinding.
And history has shown more than once that when expectations break, prices don't drift gently, they snap. And recently, we've already seen hints of that same fragility in today's market. Nvidia, the face of the AI boom, dropped almost 5% in a single trading session, dragging the NASDAQ and S&P lower with it.
That wasn't caused by bad earnings or a crisis, just a small shift in sentiment. When a stock that big pulls back that fast and the entire market wobbles with it, it shows how tightly valuations are stretched and how quickly expectations can start to unwind. The cleanest parallel is the dot era. Companies didn't collapse because the internet wasn't real.
Technology changed the world, but valuations were built on fantasies, straight line growth, limitless demand, and spending that assumed customers would show up forever. When reality settled in, many must-owned tech names fell 70% to 90%. Today's AI story has that same tension. Big tech as a whole is pouring in a trillion dollars into AI infrastructure, servers, networking, cloud capacity, and endless racks of Nvidia chips.
And many of the companies doing this are reporting profits that, according to Bur's estimates, may be inflated by 20 to 27% because of stretched depreciation assumptions. When you combine massive capital expenditure, record valuations, and earnings that might be overstated, you get a market that doesn't need bad news to break.
It only needs reality. From here, there are three broad paths. The first is a sharp correction. The moment demand slows or earnings disappoint, valuations could reset fast, 40%, 50%, even 70% isn't unusual when bubbles unwind. The second path is the slow bleed. Stocks that drift sideways or decline gradually for years while real earnings try to catch up to the hype.
And the third is selective survival. A few giants win long-term, but dozens of AI plays fade because they were never built on sustainable demand. And Bur's own trades make it clear which outcome he thinks is most likely. His Scion filings showed over $1.1 billion in notional put option exposure, about $912 million against Palanteer, and roughly $186 million against Nvidia.
Those positions made up nearly 80% of his fund. You don't take concentrated asymmetric bets like that, unless you believe the market is priced for perfection, and perfection won't last. If you already own AI stocks, this is the part of Bur's warning that matters most. Not the headlines, not the predictions, but the simple question of what his analysis means for your portfolio.
And before anything else, it's important to say this clearly. His view doesn't automatically mean you should sell everything or run for cover. Instead, it's a prompt to rethink how much risk you're carrying and whether that risk fits your goals, your timeline, and your comfort level. Start with the numbers Bur highlights.
If companies like Meta and Oracle have earnings boosted by 20 to 26% because of stretched depreciation estimates, then some of today's valuations may be sitting on profits that won't hold up when accounting normalizes. If the biggest AI builders are spending more than $1 trillion on infrastructure, and that pace slows even slightly, the ripple effects will hit Nvidia, cloud providers, and every stock tied to the AI ecosystem.
And if Nvidia's 10 to 12 time surge in 2 years was fueled by hyperscaler demand that may not be sustainable, then anyone holding AI heavy portfolios should at least recognize how much perfection is already priced in. But Bur's message isn't panic. Ask yourself, if my main AI stock dropped 50 or 60%, would my plan survive?
If my portfolio has become 40, 50, or even 70% tech or AI, was that intentional? or did those names simply balloon? One of Bur's most useful lessons is his focus on position sizing. His own put option bets, more than $1.1 billion in notional exposure, nearly 80% of his fund, weren't wild swings. They were controlled risks.
For everyday investors, that translates to something simple. Big ideas don't require big bets. They require smart ones. So, if you already own AI stocks, the takeaway isn't fear. It's making sure your exposure matches your goals and remembering that hype can lift prices quickly, but reality always returns.
If you're watching the AI boom from the sidelines while everyone else talks about life-changing gains, it's completely normal to feel like you're missing something. FOMO or fear of missing out isn't a weakness. It's human. Nvidia has climbed 10 to 12 times in 2 years. AI giants have crossed trillion dollar valuations and every headline suggests artificial intelligence is about to reshape entire industries.
It's hard not to feel behind, but this is where Bur's perspective becomes useful. His message isn't meant to scare you. It's meant to help you think clearly. Prices aren't the same as progress. AI might transform the world, but that doesn't mean every AI linked stock at today's valuation will be a great investment.
The story can be huge, but the math still has to make sense. A big part of Bur's caution comes from the possibility that much of today's AI demand is circular. Big tech spends billions on AI. Startups raise money to build with AI. Those startups then pay big tech for cloud and compute. Revenues rise, but some of the demand is really just money circulating inside the ecosystem, not sustainable customer needs.
We have seen this pattern in every major bubble. Early demand looks explosive until it doesn't. Even if you do want exposure, which is completely reasonable, the key is avoiding FOMOdriven decisions. You don't need to buy at all-time highs or chase whatever is trending. There's a calmer path. Build slowly.
Wait for pullbacks. Diversify and invest based on your goals instead of hype cycles. At the end of all of Bur's analysis, the warnings about inflated earnings, trillion dollar spending, recycled demand, and stocks priced for perfection, his message is simple. Think for yourself. His value isn't that he might predict every bubble.
It's that he refuses to let hype or headlines replace real thinking. In a market where narratives move faster than facts, that mindset matters more than ever. Bur has always approached markets the same way. question assumptions, study the data, and protect the downside before chasing upside. Whether it was the.com boom, the housing bubble, or today's AI surge, he looks for the gaps between expectations and actual economic behavior.
And right now, with big tech spending on AI infrastructure and reporting profits potentially boosted by 20 to almost 30%, he believes that gap is widening. Even though the market felt a wobble, smart money is still buying. Click the video on screen to find out the six stocks Smart Money is buying right
댓글 0