ECB Vice President Warns of Extreme AI Valuations, Rising Risk of Stock Market Correction
nashnova research
ECB Vice-Governor Boris Vujčić warned that AI-driven equity valuations have reached levels not seen in years, with a clear risk of correction. This means → the ECB is formally treating the AI valuation bubble as a financial-stability issue, not just an investor problem.
How extreme are valuations, exactly?
The S&P 500's Shiller CAPE — a price-to-earnings ratio smoothed over ten years of average earnings — sits just above 40, closing in on the all-time record of 44.2 and roughly 2.3× its long-run average of 17.
The only other time it stayed at this level for months on end was just before the dot-com peak in March 2000.
The "Buffett indicator" (total U.S. market cap ÷ GDP) has climbed past 237%, well above the 200% danger line Buffett himself has flagged.
In plain terms = almost every classic valuation yardstick points to the same conclusion: the optimism baked into today's prices has historically appeared only on the eve of a bubble bursting.
Why is this rally especially fragile?
Since the current bull market began in October 2022, the S&P 500 has gained 127% and the Nasdaq 161%, driven largely by AI infrastructure spending.
Apple, Microsoft, Nvidia, Google, Amazon, Meta, and Tesla account for over 35% of the S&P 500's market cap; the top ten make up about 38% of cap but only 31% of earnings.
This means → there is a 7-percentage-point gap between weight and profit. A stumble by a handful of stocks is not a sector rotation — it is a market-wide event.
What does this have to do with the euro area?
Five ECB economists concluded on August 17 that a correction would be "a financial-stability issue, not merely a private one" for the euro area.
Euro-area households hold roughly €440 billion in the top-seven U.S. stocks; pension and insurance funds carry exposure on a similar scale.
Much of that exposure was built passively through index products — "involuntary" concentration. In plain terms = many European families never chose to bet on AI. They bought index funds, and the money automatically piled into those seven stocks.
What signals is the market already sending?
The Philadelphia Semiconductor Index plunged 5.86% on Monday, its steepest drop since July 1.
The U.S. 10-year Treasury yield breached 5% intraday — the first time since October 2023.
U.S. equity margin debt surged 77% in roughly 14 months, topping $1.5 trillion.
This means → the market faces a triple squeeze: historic valuations + rising risk-free rates + fading AI momentum expectations — all hitting at once.
How is this different from 2000?
When the dot-com bubble burst, the Fed had ample room to cut rates and governments had fiscal headroom to intervene.
Today, rates are already relatively low and public debt is already elevated — the policy buffer is visibly thinner.
Brent crude topped $107 a barrel on Monday; Middle East conflict is adding inflation and bond-market pressure on top of the repricing risk.
In plain terms = if a correction does arrive, central banks and governments have far fewer tools this time than last.
What is the ECB actually saying — and not saying?
Vujčić said on the ECB podcast that these P/E levels "may eventually be justified by fundamentals — but they may not."
The ECB analysis explicitly states it is not predicting a crash; the timing of a correction "cannot be known in advance" and can only be identified after the fact.
It also notes that "this does not mean current prices are the ceiling" — if AI proves truly transformative, valuations could still go "much higher" even after a reset.
This reflects the ECB's real concern: two questions the long rally has wrongly bundled together. Whether AI delivers transformative value is one question; whether the price investors are paying for it is reasonable is quite another.
市场有风险，内容仅供研究参考，不构成投资建议。