Jamie Dimon is starting to sound a bit like Alan Greenspan—and that should make investors nervous.
The JPMorgan Chase CEO warned in a Bloomberg TV interview this week that markets may be showing “too much exuberance,” pointing to frothy valuations around artificial intelligence and the Big Tech giants building out the infrastructure behind it. His choice of words evokes Greenspan’s infamous “irrational exuberance” line from 1996, when the then–Federal Reserve chair cautioned that animal spirits can push asset prices far beyond what fundamentals justify, leaving them vulnerable to painful reversals.
At the time, Greenspan was notably not predicting a bubble and kept monetary policy loose even as a massive one formed around the dawning internet age, bursting painfully at the turn of the millennium. In financial terms, “irrational exuberance” has since come to mean investor optimism that pushes asset prices far beyond what underlying earnings, cash flows, or economic fundamentals justify, fueled by herd behavior and narratives more than data. Nobel laureate Robert Shiller’s book of the same name, first published in 2000, reinforced the idea, over 100 years old by that point but still often forgotten, that bubbles are less about spreadsheets and more about contagious stories that drown out skepticism.
About a week before Dimon’s interview, a bearish voice emerged to warn exactly about “irrational exuberance” and pointing out that from a macro perspective, the AI boom is already 60% larger than the technology-media-telecom (TMT) bubble that Greenspan was talking about.
“There are increasing signs of ‘irrational exuberance’ in the AI boom,” Panmure Liberum strategist Joachim Klement wrote, concluding that “AI is in a bubble,” although it can last another one to two years and investors may not want to sell just yet.
That makes Dimon’s warning more than just a curmudgeonly aside—and makes an understanding of just what Greenspan said, and how he said it, important for understanding this moment.
From Greenspan’s warning shot to today’s AI mania
Greenspan coined the term “irrational exuberance” in a December 1996 speech at the American Enterprise Institute, asking how policymakers could know when speculative fever had driven asset prices to unsustainable levels. Markets around the world sold off on the remark, taking it as a hint that the Fed might eventually lean against the boom. Yet U.S. stocks, and especially tech names, continued to surge for several more years before the dot-com bust, turning the phrase into a kind of shorthand for late-stage bubble psychology rather than a precise top-tick.
“How do we know when irrational exuberance has unduly escalated asset values, which then become subject to unexpected and prolonged contractions?” Greenspan asked the crowd, before vowing that “we as central bankers need not be concerned if a collapsing financial asset bubble does not threaten to impair the real economy, its production, jobs, and price stability.” He was unable to prevent that happening just a few years later himself, from too much exuberance.
Today’s AI surge has clear echoes of that era—but the numbers are even starker. Panmure’s team calculated that from a macro perspective, the current AI boom is already about 60% larger than the late-1990s TMT episode when measured by the contribution of tech capex to U.S. GDP growth. They estimate that almost all of U.S. real GDP growth right now is being driven by technology investment, a concentration that leaves the broader economy heavily exposed if the trend slows or reverses.
According to Panmure, hyperscalers have embarked on the largest capex boom in history, pouring money into data centers and chips at a pace that only makes sense if they can eventually find an extra $2 trillion to $5 trillion in annual revenue to justify it. At the same time, the authors pointed out that’s not even clear whether flagship model providers like OpenAI and Anthropic have viable business models, suggesting some highly valued players may effectively be trying to ride the hype to IPOs that transfer risk from founders to public shareholders.
Why Dimon’s ‘exuberance’ line matters in that context
Dimon has been broadly optimistic about AI as a long-term productivity engine, comparing the technology’s potential to the internet’s and arguing that it will reshape how companies operate. He told Fortune in October 2025 that “AI itself is real” as a technology and everyone “should be using it,” and yet it was also true at that point that “some asset prices are high, in some form of bubble territory.”
Dimon is not a central banker, but he occupies a unique perch: he runs the largest U.S. bank by assets, sits at the center of global capital markets, and has successfully navigated multiple crises from the 2008 financial meltdown to the pandemic. When he starts echoing Greenspan’s concerns about exuberance—even without using the exact phrase—markets listen.
Three things make his latest warning especially resonant in 2026:
- He’s pushing against the narrative that AI alone can bail out everything else. Fresh research from Deutsche Bank argued that the world is entering a period where most structural “megatrends”—from sovereign deficits to domestic political discontent and demographic drag—are working against growth, with technology and AI as the primary offsetting positive. Their model shows the combined impact of these megatrends is currently deeply negative, a pattern previously seen only around the 1970s oil shocks and the run-up to the 2008 crisis. In that context, exuberant pricing in risk assets looks less like rational discounting of an AI-driven productivity boom and more like investors underpricing the macro headwinds.
- He’s cautioning just as “status quo” assumptions look least safe. Deutsche’s AI-driven megatrend framework concludes that betting on a “muddle-through” world of 1% to 3% growth, stable debt markets, and average stock returns is “almost certainly wrong.” The model suggests developed economies are more likely to see either a powerful productivity boom—if AI adoption accelerates as expected—or a severe, prolonged downturn if technology fails to outrun the drag from deficits and demographics. Dimon’s concern about exuberance effectively says markets are pricing something close to the best-case AI scenario long before the macro risks are resolved.
- He’s flagging exuberance in an era when traditional shock absorbers work less well. The Deutsche analysis finds that classic “haven” assets—Treasuries, the dollar, gold, the yen, and bunds—have failed to consistently protect portfolios during major risk-off episodes in the 2020s, including the pandemic, the 2022 rate shock, tariff scares, and the Iran conflict. That breakdown of hedges means that if exuberance does give way to a sharp repricing, investors may have fewer reliable places to hide than they did in Greenspan’s day.
Echoes of 1996—with worse fundamentals
In some ways, today does rhyme with the mid-1990s. Then, as now, a transformative technology (the commercial internet) was colliding with a new policy and macro regime. But Deutsche’s megatrend history underscores a crucial difference: in the 1990s, the tech boom was reinforced by a rare alignment of other forces—falling sovereign debt ratios, relatively benign domestic politics, supportive demographics, and a positive energy backdrop. That broad tailwind amplified the upside and cushioned the eventual bust.
Today’s AI boom is unfolding against almost the mirror image: worsening sovereign debt trajectories, aging workforces in most rich countries, elevated social discontent, and rising geopolitical fragmentation. Deutsche’s model suggests that makes this rally more dependent than past episodes on AI actually delivering sustained productivity gains that exceed even the 1990s, and doing so quickly enough to outrun the structural drags. Against that backdrop, Dimon’s “too much exuberance” warning is less about technophobia than about the mismatch between long-term promise and near-term pricing.
So, do you feel exuberant?
For this story, Fortune journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing.