Citadel's Griffin calls AI "almost certainly" a bubble despite strong demand

One of finance's most prominent voices has now made a bubble call, a comment that could sharpen the debate about AI valuations even though Griffin did not go as far as urging investors out of the market. Because he sees sweeping shifts as normally arriving with bubbles, his argument points to churn and losses on the way rather than a breakdown of the larger trend, so any immediate market response may be muted.

The warning arrived at a notable moment: SpaceX's AI revenue beat, Super Micro's surprise upgrade and Musk's contention that compute remains scarce all coincided, leaving investors to weigh solid near-term demand against concerns about how much is already reflected in prices. That friction may feed the stock-by-stock volatility analysts have been forecasting for this earnings period.

For Griffin, AI can be simultaneously a speculative excess and a genuine development, no more contradictory than the fact that Amazon survived the dot-com crash.

Key points:

• Ken Griffin, founder of Citadel, says investors are almost certainly in an AI bubble
• He argues that radical economic transformations generally come with a bubble
• Griffin draws a parallel with the dot-com era and its implausible business models, saying he wished he had held on to his Amazon shares
• He was speaking on Tuesday at the Future of Global Markets Conference in New York
• His AI views have shifted during 2026, from dismissive at Davos in January to unsettled by the technology's abilities in May
• The remarks landed on a day when SpaceX, Super Micro and Elon Musk all signaled strong AI demand

Billionaire Ken Griffin, who founded hedge fund Citadel and market maker Citadel Securities, warned that an AI bubble is almost certainly where investors now find themselves, delivering a sceptical note on a day full of strong AI-demand indicators.

At the Future of Global Markets Conference in New York on Tuesday, Griffin said sweeping economic shifts usually arrive hand in hand with bubbles. He recalled the run-up to the dot-com bust, an era in which many business plans turned out to be fantasy, yet some companies came through, and he added that he wished he had held on to his Amazon shares back then.

The distinction is significant. Griffin is not brushing AI aside as noise. Rather, he argues that a genuine technological leap and a speculative bubble are compatible, with inflated prices and failed businesses appearing next to durable winners.

His comments are the newest twist in a position that has moved repeatedly over 2026. At Davos, early in the year, he was bluntly skeptical about AI. By May, during a talk at Stanford's business school, he described himself as unnerved by how far the technology could go. The latest remarks fall between those poles: they acknowledge that AI can change the economy while cautioning that stock prices have moved beyond that reality.

His warning carried extra force because of when it came. Earlier in the day, SpaceX posted roughly $7.6 billion in AI revenue, over half of its quarterly figure, while CEO Elon Musk said processing capacity is so constrained that older Nvidia chips now carry rental prices above what they commanded when new. In a separate release, Super Micro Computer offered preliminary results that put September-quarter revenue well above the company's own forecast.

The numbers point to genuine demand that is still gaining speed for now. Griffin's concern lies with valuation, not demand: can the prices assigned to AI-related businesses all prove realistic in the end?

With big-tech earnings season now underway and Meta leading off, investors are searching for answers, especially on whether massive AI outlays are starting to produce returns that justify today's valuations.