Maybe you've had this experience lately: You're riding high, feeling good about yet another record high in stocks, and then someone pops up on TV to kill the vibe. "It's an AI bubble," they say. "It's like 1999."
This actually happened to me last week (before President Trump's new tariff threat sent stocks tumbling on Friday). I'd just read some convincing research arguing against a bubble, only to have my good mood ruined by a TV segment saying the opposite. Tired of the whiplash, I decided to plant my flag in the "not a bubble" camp. (Read the story here.)
Imagine my surprise when the most pushback I got came from my own boss, Steve Russolillo, BI's chief news editor and frequent author of this newsletter on Sundays.
Steve is worried that we are in an AI bubble — one that could get bigger and eventually pop more spectacularly than the dot-com boom and bust.
We decided to sit down and hash it out.
Valuations
Steve: Joe, I read your piece with great intrigue, but I must say the old-school thinking in me shudders when Wall Street firms start using non-traditional metrics to justify a rally — as both Goldman and Morgan Stanley did.
Conversely, a tried-and-true valuation indicator — the Shiller P/E ratio, which dates back to the 19th century — is at a frighteningly high level, above 40. It was higher during the dot-com bubble. Ignore the ratio at your peril: it accurately projected market tops in 1929 and 1999-2000, and also flashed red before the mid-2000s housing crash.
Joe: I agree the Shiller P/E is alarming! But I also think it's failing to account for a few key attributes of the companies leading the market. If you adjust valuation measures for profit growth, cash flow, and profit margins, parallels to the dot-com era weaken significantly.
And yes, Wall Street loves to create new metrics to paint the story they want to tell. But in this case, the adjustments paint a more accurate and modern picture of corporate health.
Company quality
Joe: The companies driving the AI revolution are just stronger. On average, they have better cash flow, operate more efficiently, and are more profitable. That's certainly the case for the absolute biggest names responsible for moving the market: Nvidia, Microsoft, Amazon, etc.
Steve: There's no denying the dominance of the biggest companies. In fact, they're a little too dominant for my liking. The Mag 7 stocks make up more than one-third of the S&P 500. The concentration risk in the market is enormous — it's rare that so few companies make up such a large percentage of the overall index. Any stumble in even just one of these companies could drag the market down, and fast.
A circular economy
Steve: New deals in the AI space are announced almost daily nowadays. Hundreds of billions of dollars are being thrown around, with a growing chorus of investors and analysts wondering about the circular nature of the deals — and their sustainability. "If anyone decides to pause and ask, 'What's our real economic return here?' it could be a big problem," Jim Chanos, the guy who shorted Enron, warned recently.
Joe: I must admit, OpenAI's position at the center of so many of these deals does make me a bit uneasy. Especially about companies like Oracle and CoreWeave, whose fates are now completely intertwined with it. But a recent Bank of America note put my mind at ease, forecasting that just 5% to 10% of spending by 2030 will be vendor-financed.
If you ask me, warnings of an AI bubble are becoming an unsustainable bubble of their own.
Steve: OK, this is getting too meta for me.
Where do you stand on the big AI bubble debate? We'd love to hear your thoughts. Please email jciolli@bjinnox.com and srussolillo@bjinnox.com.