The ECB Says an AI Correction Is Coming Whether or Not the Bubble Is Real

Central banks rarely tell you a crash is coming. They hedge, they qualify, they publish charts and let you draw the line yourself. So when five economists at the European Central Bank published a blog post on August 17 arguing that the AI-driven surge in US equity valuations points to what they called "a worrisome conclusion," the notable thing was not the caution. It was the absence of it.

"We argue that economic research on past technological revolutions points to a worrisome conclusion: a correction of current stock market valuations is likely," wrote Malin Andersson, Johannes Breckenfelder, Stefano Corradin, Kalin Nikolov and Maria Antonietta Viola in a post titled "The AI boom: rational enthusiasm or the next dot-com bubble?" The analysis, which carries the standard disclaimer that the views are the authors' own rather than the ECB's, has circulated widely across European financial press through the week and was still drawing coverage on Friday.

The argument's power is that it does not depend on calling the bubble. The authors lay out two competing explanations for the AI rally and show that both land in the same place. The first is what they call the rational view: extreme uncertainty about a new technology's eventual productivity creates something like an option value, which legitimately inflates the price-to-earnings ratios of early adopters. "Why has Nvidia's share price risen 20-fold since 2022?" they ask. "Because investors rationally perceived that the company would become the next Google — with a highly uncertain and potentially large upside."

But even if that bet pays off, prices still fall. As adoption spreads, the authors write, uncertainty migrates from a "single sector" to the "entire" economy, and risk that could once be diversified away cannot be. Investors demand a higher risk premium, and historically that premium wins out over the cash-flow boost unless profit growth is exceptional. The second explanation, the behavioural view, is the familiar one: overconfident investors bid prices past fundamentals, and when the mood turns, "prices can fall even more sharply than in the rational scenario." Rational or irrational, the conclusion holds. "The case for expecting a correction is not dependent on whether today's prices are rational or irrational," the authors write. "We should be aware of that and prepare."

Europe's Passive Exposure Problem

The second half of the analysis is the part that should unsettle European readers, because it quantifies something most households have never consciously chosen. Euro-area investors hold the Magnificent Seven overwhelmingly through funds rather than directly — mutual funds and ETFs tracking global indices where those seven names now dominate. Euro-area households, the ECB team found using look-through data on underlying fund investors, carry "around €440 billion of exposures to US technology equities without necessarily being aware of the associated concentration risk." Insurance companies and pension funds hold significant Mag7 positions too, based on holdings data through the third quarter of 2025.

That fund structure is not just a measurement detail; it is an amplifier. A sharp selloff triggers redemptions, redemptions force funds to sell — liquid assets first, then distressed ones if the drawdown persists — and the forced selling pushes valuations down further, triggering more redemptions. "This is why a Mag7 correction is a question of financial stability for the euro area, rather than just a private one," the authors write.

The authors are careful to note that Europe's own market looks less frothy. Euro-area price-to-earnings ratios remain well below US levels, the region's exchanges are dominated by "old economy" stocks, and the ICT sector's fundamentals — rising productivity, rising markups, no exuberance in digital-services sentiment — bear little resemblance to 2000. Digital investment across the euro area rose by more than three times the cumulative growth in GDP over the past decade. None of it helps much. US and euro-area equity markets have historically been very highly correlated, and the ECB team flags a worse starting point than the dot-com era: today there is "markedly less room to cut interest rates or use fiscal policy to cushion the fallout."

Why It Matters

The warning lands against roughly $700 billion in combined 2026 capital-expenditure guidance from Amazon, Microsoft, Alphabet and Meta, and a US index in which the Magnificent Seven account for roughly a third of S&P 500 market capitalisation — up from around an eighth a decade ago. Every European saver who bought a low-cost global tracker bought into that concentration, whether or not they read the fact sheet.

It also sits awkwardly beside the ECB's own policy posture. Two days after the blog post, in Geneva, President Christine Lagarde told the World Economic Forum's International Business Council that "Europe largely missed out on the first digital revolution, as the commercial gains from the spread of information and communication technologies were captured disproportionately elsewhere," adding: "We cannot afford to repeat that experience with artificial intelligence, the second digital revolution." Frankfurt is simultaneously telling Europe to run faster at AI and warning that its households are dangerously long the trade.

What to Watch

Three things. First, whether the analysis graduates from a blog post to the ECB's Financial Stability Review, which would move it from staff opinion toward institutional position and could invite supervisory attention to insurer and pension-fund concentration limits. Second, the next hyperscaler earnings cycle — the ECB's mechanism turns on the gap between capex and realised profit growth, and that gap is now the market's central question. Third, any move on retail fund disclosure: if €440 billion of household exposure is genuinely unrecognised, the policy fix starts with telling people what they own.