TL;DR

A European Central Bank blog post argues that a correction in US technology valuations is likely, and that the fallout would be harder to contain than in previous episodes. Euro zone households hold roughly €440bn of Magnificent Seven stock, with pension and insurance exposure of about the same size. The post does not necessarily represent the ECB’s own view.

The argument has two halves

The first is historical. Studies of earlier technological revolutions point, in the authors’ words, to a worrisome conclusion — that today’s valuations are likely to correct. The mechanism does not require AI to fail. Even where the technology delivers and profits climb, prices can still fall, because markets have priced in profit growth that is very hard to actually deliver.

The second is behavioural. Optimistic investors push prices above what fundamentals justify; when that optimism fades, the fall tends to overshoot what a purely rational repricing would produce. Timing, the post concedes, “is unknowable in advance”, and these patterns are “only identifiable with hindsight”.

Why the cushion is thinner

The sharper claim is not about equities alone. The dangerous case is a correction landing alongside broader instability that policymakers cannot quickly calm — and unlike the dot-com period, there is markedly less scope now to cut rates or spend the problem away.

European valuations look more rational on their own terms, but that offers little shelter. Local markets track US moves closely enough that a correction there arrives here regardless.

Looking forward

For UK readers the transmission is familiar. Resultsense reported on 14 August that AI-driven borrowing was pushing up real yields, gilts included, and the exposure a UK saver carries to these seven US companies runs through pension and insurance allocations rather than direct holdings. The awkwardness of the timing is worth noting: the warning landed the same day Anthropic’s revenue run rate was reported above $65bn. A central bank forecasting a correction and a lab reporting sevenfold growth are not contradictory — both can be describing the same bubble.