Analysis
US and European indices are at records and ECB economists chose this week to publish the opposite view. "Economic research on past technological revolutions points to a worrisome conclusion: a correction of current stock market valuations is likely," they wrote in a Monday blog post, reported by CNBC.
The argument is built on two independent paths to the same outcome, and the second is the one worth reading twice. The first is familiar: "overconfident, overoptimistic investors" bid prices above fundamental value, and the correction arrives when sentiment turns. The second requires no irrationality at all. Suppose current valuations accurately reflect AI's capacity to reshape the economy and lift corporate profits. As adoption spreads, "uncertainty becomes economy-wide. If something then goes wrong with that technology, the whole economy suffers." A more concentrated systemic dependency justifies a higher risk premium, and a higher risk premium mechanically lowers prices even while profit growth stays strong.
The historical analogues are the 19th century railway boom, the spread of electricity and radio in the 1920s, and the 1990s internet. In each case anxiety about a technology-linked transition eventually leaked from equity markets into the wider economy. "Both views imply a boom followed by a correction, or a pullback from wherever valuations have risen, at some point in the future," the economists wrote -- adding that recovery and further climbing typically follow.
“The historical analogues are the 19th century railway boom, the spread of electricity and radio in the 1920s, and the 1990s internet.”
The important caveat is theirs, not a critic's: "The exact timing is unknowable in advance. These boom-bust patterns are only identifiable with hindsight." That is an honest limit, and it also means the note is not actionable as a trade. Central bank economists have warned about equity valuations through most of the past decade of gains.
What This Means for Private Markets
The private-market translation of a public-market risk premium is dilution. When public multiples compress, crossover investors mark down private positions, late-stage rounds reprice, and the companies that raised at 2026 peaks face structured terms rather than clean down rounds -- liquidation preferences above 1x, ratchets, and pay-to-play provisions that punish existing holders who cannot follow on. That sequence played out through 2022 and 2023, and the founders who came through it best were the ones who had already extended runway before the window narrowed.
It is worth stating what the ECB note is not. It is not a forecast with a date, not a policy signal, and not a statement from the Governing Council -- it is research published on the bank's blog by staff economists. Central bank economists have published versions of this warning through most of the last decade of equity gains, and anyone who acted on the 2017 or 2019 iterations underperformed badly. The reason this one is worth reading is the second mechanism, which does not require anyone to have been irrational, and which most bubble commentary never considers.
What makes this one land differently is corroboration from inside the trade. Nvidia is now financing its own customers' capacity, Reuters has flagged persistent investor anxiety about AI capex and debt loads, and the OpenAI-Nvidia Ohio guarantee shrank by roughly $145 billion between the July leak and Monday's signature. Private markets are already repricing capital intensity. The ECB note is a description of that, arriving from the outside.
