This is analysis of a debate, not a prediction. We present both sides and attribute the claims; we do not forecast a crash or a melt-up.

What "bubble" actually means

A financial bubble is when an asset's price rises far above its underlying value, setting up a sharp fall. The reference point everyone reaches for is the dot-com boom: between Netscape's 1995 listing and the March 2000 peak, the Nasdaq-100 rose roughly 1,090%. Since ChatGPT's late-2022 launch, the same index has risen about 140% — a smaller move, but a fast one, and built on companies with real profits rather than the revenue-free startups of 1999. Whether that distinction is enough is the whole argument.

The bull case: the earnings are real

The strongest case against a bubble is that the leading AI companies actually make money. Morgan Stanley's Michael Wilson has framed the rally as "an earnings story, not a multiple expansion one," per Fortune. Tech stocks trade at roughly 25 times expected earnings — the forward price-to-earnings ratio, or what investors pay per dollar of next year's profit — which is high but far below the 58 times reached at the 2000 peak.

Bulls also point to genuine productivity gains: LPL Financial noted U.S. productivity jumped in late 2025 as output rose while hours barely moved — evidence, they argue, that AI is delivering real efficiency, not just hype.

The bear case: concentration and circular money

Skeptics raise structural alarms that go beyond valuation.

Concentration. Apollo's chief economist Torsten Slok warns the S&P 500 is "no longer diversified," with the top ten companies making up around 41% of the index's value, Fortune reported. When a few names dominate, a stumble in any of them drags the whole market.

Spending vs. revenue. The big cloud companies are on track to spend hundreds of billions of dollars on AI infrastructure in 2026, by Goldman Sachs estimates. Analysts at Man Group argue that identifiable AI revenue remains a small fraction of that outlay — a gap they call hard to sustain (the precise revenue figure is contested and moves fast).

Circular financing. Perhaps the most specific worry is "vendor financing," where chipmakers and cloud providers invest in AI startups that then spend the money buying those same firms' chips and capacity — making demand look organic when it is partly the same dollars circulating among a few connected companies. Bloomberg has mapped such arrangements among Microsoft, OpenAI, Nvidia and others; a similar pattern appeared in the late dot-com years.

The bears include heavyweights: GMO's Jeremy Grantham expects a deflation that would bring "a major stumble for the economy, a plunge in profits, and a severe decline in valuations."

What regulators say

Official bodies have grown more vocal. The IMF warned in October 2025 that an AI investment bust could resemble the dot-com crash. The Federal Reserve's May 2026 Financial Stability Report found a sharp rise in the share of market contacts naming AI as a possible source of systemic risk, and flagged stretched tech valuations as an amplifier.

The market's own hedge

Investors are already pricing in some doubt. Boursel has reported on the rotation out of megacap tech in 2026 — money spreading from the dominant few into the broader market. The equal-weight and small-cap parts of the market have outpaced the cap-weighted index, a shift that can read either as healthy broadening or as quiet hedging against the AI trade.

How to think about it

The honest answer is that no one knows. Evercore's Julian Emanuel, who likens today to 1999 more than 1997, told Fortune that "FOMO has proven a stronger incentive than poor stock performance." Morningstar offers a middle path: AI is probably neither the clean boom nor the certain bust, with real gains arriving more slowly than current prices assume. The outcome hinges on things that cannot be known today — how fast businesses adopt AI, whether the spending earns a return, and how long the circular financing holds.

What is not in dispute is the stakes: the AI trade is among the most concentrated and capital-intensive bets in market history. Whether it turns out to be 1997 or 1999, the resolution — in either direction — will matter for almost every portfolio.