1) It'll be very interesting to see over the next year if I'll maintain my perfect track record calling sector bubbles...
In my 27-year career, I've publicly and loudly named seven of them. I was not only proven right in every case, but in five cases, I nailed the peak within a matter of days. And all seven sectors declined by at least 80%.
Here's my track record with links:
- Internet: Three of my Motley Fool columns – Valuation Matters (February 7, 2000), Should Warren Buffett Call It Quits? (April 3, 2000), and Cisco, Apple, and Probabilities (October 16, 2000)
- Housing: My 61-slide presentation on Why We Are Still in the Early Innings of the Bursting of the Housing and Credit Bubbles – And How to Profit From It (May 7, 2008) and my 60 Minutes appearance (December 12, 2008)
- 3D printing: Two Business Insider columns – Whitney Tilson Says 3D Printing Stock Is Going To Plunge 90% (January 6, 2014) and Here's the Hilarious and Brutal Chart That Whitney Tilson Sent Out to Explain the Crash in a Big 3D Printing Company (February 5, 2014)
- Alternative power: My e-mail on shorting hydrogen-fuel-cell maker Plug Power (PLUG) in 2014 and Pulling the Plug on this Fool-Cell Maker (January 21, 2021)
- Cryptocurrency: My e-mail on calling the top of the bitcoin/cryptocurrency madness (December 16, 2017)
- Cannabis: My e-mail on Tilray (TLRY) and heaping scorn on the stock in a Yahoo Finance video (September 19, 2018)
- Meme stocks: My e-mail on my 25-stock "Short Squeeze Bubble Basket" (January 27, 2021)
2) In Friday's e-mail, I wrote:
Last week, I laid out the case for why I think the AI boom has turned into a bubble and is starting to burst – just like the Internet bubble did in March 2000.
But this will be nothing more than a hiccup for the best companies that generate gobs of cash, like Meta Platforms (META), Amazon (AMZN), Alphabet (GOOGL), and Microsoft (MSFT) – all of which I analyzed on Monday.
At the same time, it's very bad news for junk companies with access to nearly unlimited quantities of cheap debt, like CoreWeave (CRWV). Just as I predicted, the stock is imploding, which I discussed on October 15.
Tesla (TSLA) and SpaceX (SPCX) CEO Elon Musk partially agrees with me...
In an X post over the weekend responding to investor and podcaster Chamath Palihapitiya, he wrote that hundreds of "neoclouds" (cloud providers for AI workloads) will blow up:
I think he's right. But we differ on what happens next...
He thinks AI-infrastructure spending will continue to soar, which is why he's making a huge bet with SpaceX by building new data centers.
In contrast, I think the neoclouds will be the first domino to fall as the AI bubble bursts (see further discussion below).
Regarding Musk's post, one of my friends commented:
He's probably half telling the truth. With the other half, he's going to build 10 to 15 gigawatts of compute next year and is trying to scare off potential customers from buying compute from anyone but him.
3) My big call on the AI bubble has generated plenty of disagreement from – and insightful discussion with – my friends and readers. So I wanted to share some of those debates...
Reader Billy D. wrote:
Even if margins go down at the model layer (OpenAI and Anthropic), all these competitive models – Gemini, Meta, xAI, open-source – need computing power to deliver AI. If usage goes up at lower margins, it shouldn't really matter to the chip/data center companies.
Demand is massively higher than supply right now. Musk just said xAI is getting a less than one-year payback on its data centers. Everything is sold out into 2028. There may be an overbuild eventually, but we don't appear to be close to that.
Remember the Internet bubble took seven years, which would take us to 2030, and this appears to be even more supply-constrained than that was. So I'm anticipating a longer up-cycle than that.
So I think you're wrong on this, despite the current volatility and October drawdown (a not-uncommon seasonal pattern). I could be wrong though. Who knows. No one has ever seen anything like this before...
After I sent him my track record of calling bubbles, he replied:
The last five examples on this list had absolutely zero fundamentals associated with them. Total speculation.
This up-cycle has genuine revenue and earnings growth, and valuations are much lower than during the dot-com boom. There are also physical constraints that may prevent overbuilding ahead of demand, unless demand totally evaporates, which I don't see (I guess that's your thesis).
According to Blackstone (BX) COO Jon Gray, the comparison with the telecom bubble is flawed because, in that case, capacity was built in anticipation of demand (which took much longer to materialize) versus today, when "you're building capacity and you're hoping you can keep up with demand." And he notes that "at our portfolio companies, demand on models, spend on models, is up 20-fold year-on-year."
How do you reconcile a payback of less than a year on AI data centers at today's spot pricing with a bubble forming? That means demand could theoretically decrease by 80% (to 20% of current levels), and the payback on AI data centers would still be less than five years (20%-plus return on invested capital). I just don't think we're anywhere close to an overbuild.
To be clear, I'm not long on neoclouds because I don't like their business model. I just think the cloud giants and semis still have a lot of profitable growth ahead.
As I responded to Billy, I think it's interesting that he and others are seeing unlimited demand, whereas I'm seeing the peak of a bubble. That's because the oxygen fueling it – unlimited cheap debt and equity capital to even the worst companies – has started to be cut off.
The collapse of these companies is the sign of the first phase of a bubble bursting, as we're seeing with CoreWeave.
The next phase is lying and denial. Remember Federal Reserve Chairman Ben Bernanke saying in May 2007 that the housing crisis was "contained" and "limited to subprime"? Keep an eye out for similar statements...
4) One of my friends disagrees with me and remains bullish on stocks like Micron Technology (MU) and Nvidia (NVDA).
He accuses me of not being able to do "second-grade math" because the stocks are "obvious buys" at 4.2 times and 13.3 times analysts' consensus estimates for 2028, respectively.
As I replied to him, the question is whether the companies reach those estimates (or anywhere close)...
My Friday e-mail addresses how there are plenty of questions about actual end-user demand. And there's so much circular financing, circular valuation, and circular demand that it makes my head hurt.
As for their low price-to-earnings (P/E) multiples, it's important to understand that, when analyzing cyclical companies and sectors, the P/E multiples are typically the lowest at the peak of the cycle because that's when earnings are at a cyclical peak.
Since it was founded in 1978, Micron had been the very definition of cyclical for 47 years. Now, over the past two years, it has had a growth spurt – as all cyclical companies do. But this one has been more massive than almost any other company in history.
So will Micron revert to being a cyclical company, or has it morphed into a growth company? I would answer by quoting George Santayana: "Those who cannot remember the past are condemned to repeat it."
5) The core of my belief that the AI build-out is a bubble is rooted in skepticism that genuine demand (external to the bubble) is or will be anywhere close to what's being built. And therefore, when the spigot of cheap cash is turned down (as I believe is happening right now), the massive demand and high prices will crash.
It's hard to track how much major companies outside the AI sector are spending on AI. For example, I looked into big companies like General Motors (GM), Walmart (WMT), and Eli Lilly (LLY) and found no public disclosures.
Many of the big banks, law firms, and consulting firms report $1 billion to $2 billion in annual AI spending, but it's not clear how much they're spending externally. For example, the world's largest law firm, Kirkland & Ellis, announced a $500 million investment to build a custom, proprietary AI platform, including an initial $100 million budget for 2026.
And what about spending data for the millions of smaller, private companies? I would love to see disconfirming evidence to my thesis – that major, non-tech, mainstream companies are paying billions of dollars for AI. That would help persuade me that there will be some sort of acceptable return on the trillions of dollars being poured into the AI-infrastructure build-out.
Here's a new report from Apollo Chief Economist Torsten Slok that underscores my concerns: No Signs of AI Payoff Outside the Tech Sector. Excerpt:
Despite massive investments and widespread hype around artificial intelligence, the technology has delivered zero financial payoff to corporate America outside of tech. Revenue per worker is rising for the Mag 7 but is flat for the S&P 493, see the first chart below. Profit margins are rising for the Mag 7 but are flat for the S&P 493, see the second chart below. And there are still no signs of AI boosting productivity growth, see the third chart below. With no impact on the bottom line, mainstream corporate America is pulling back and starting to view token spending not as a source of growth but as a cost center that weighs on margins.
VC-backed companies are spending more and more on tokens, but mainstream corporate America does not need the latest model and is unwilling to pay for it. As long as agentic AI and token spending are confined to a small part of the economy, the macroeconomic effects of AI will be very limited, see the fourth and fifth charts below.
The question looking ahead is whether, even as models become commoditized, mainstream corporate America can turn increasingly cheap intelligence into higher revenue growth and higher profit margins. That may be harder than it sounds. AI may be a great calculator, but in business, intelligence is mostly judgment. Judgment depends on context, and the answer varies with which human or which model you ask. Perhaps most importantly, AI does not understand accountability, trust, culture, unwritten rules, ethical trade-offs or fairness, all of which shape reputational risk and ultimately drive revenue growth and profit margins.
The bottom line is that the AI story will be decided in the S&P 493, not the Mag 7, and so far, there is no evidence that revenue growth or margins are improving outside the tech sector.
Take a look at these three charts from the article:
These tech/investor bros inside the bubble are rapidly adopting AI... and spending massive sums to do so.
But I don't think the rest of us – both businesses and individuals – are following in their footsteps nearly as fast as they think. And that's coming from someone who's probably in the top 1% of AI adopters, as I pay monthly for Claude and regularly use Google Gemini.
It took the better part of a decade before there was widespread corporate and public adoption of the Internet. I knew it was going to be huge – but I was also able to see an obvious bubble.
Human beings never change... The same story is playing out again, but on steroids. And there will be a lot of pain in the AI sector when the bubble bursts.
But the strongest companies will not only survive but thrive... And I don't just mean the cash-gushing tech giants like Meta Platforms (META) and Microsoft (MSFT).
In fact, five little-known stocks could give the Magnificent Seven a run for their money – all because of an AI "superchip." And for early investors, 1,000% returns could follow.
My friend Jeff Brown, founder of Brownstone Research, has identified the company behind this AI breakthrough. And I'll be joining him to introduce it this Wednesday, October 28.
You don't want to miss this special presentation. To reserve your spot, click here.
Best regards,
Whitney
P.S. I welcome your feedback – send me an e-mail by clicking here.