What are the European Central Bank and Federal Reserve warning about the sustainability of the AI-driven stock-market rally, including the ECB’s assessment that a correction is likely, Europe’s roughly €880 billion exposure to the “Magnificent Seven,” the Fed’s concerns about dot

The core warning is that AI optimism may be supporting valuations and investment spending that require exceptionally strong future earnings to justify. A correction need not mean AI fails—but a repricing could still be large and economically damaging, particularly because the rally, capital spending, and credit issuance are highly concentrated.

• ECB assessment: ECB economists argue that history of major technological revolutions makes a correction in present valuations “likely,” even if AI ultimately delivers genuine productivity gains. They identify direct euro-area exposure to the Magnificent Seven and contagion through weaker confidence and tighter financing conditions as the main transmission channels.

• Exposure figure clarification: The ECB material cited in reporting refers to about €440 billion of euro-area household exposure to the Magnificent Seven—not €880 billion. The larger number may reflect a different scope or a doubling of that estimate; the evidence available does not substantiate €880 billion.

• Valuation and investment concern: A Shiller CAPE around 41 is near the highest level since the 2000 dot-com peak, while U.S. technology investment as a share of GDP has surpassed its 1990s peak. These measures do not predict the timing of a fall, but indicate that investors are pricing in unusually optimistic long-run outcomes.

• Debt-financed buildout: AI-related debt issuance was estimated at nearly $500 billion in 2026 by early August, while Morgan Stanley had projected nearly $570 billion for the year. AI-linked borrowing had already approached 15% of investment-grade issuance, encouraging increasingly complex financing structures for data centres and supply chains.

• Why off-balance-sheet financing matters: Leasing, special-purpose vehicles, private credit, and similar arrangements can spread the financing beyond the headline balance sheets of large technology firms. That can make leverage, refinancing needs, and the ultimate holder of risk harder to see—an echo of past episodes in which apparently contained risks became broader financial-system risks. This is a risk comparison, not proof that today’s structures are identical to the dot-com era.

• “Deflate rather than burst” case: Today’s largest technology companies generally have substantial revenues, profits, cash flows, and stronger balance sheets than many dot-com firms. That makes a prolonged valuation compression—returns catching up with earnings—more plausible than an immediate wholesale collapse. But strong current earnings do not eliminate the risk that capital expenditure and debt service outrun eventual AI monetisation.

• Macro spillover: A sharp U.S. technology sell-off could reduce household wealth, dampen business and consumer confidence, tighten credit conditions, and lead companies to slow hiring and investment. The ECB’s concern is that the euro area could be hit despite being the secondary market, owing to its investor exposure and dependence on global financial conditions.

• Less policy room: The ECB argues that monetary and fiscal authorities have less capacity than around the dot-com episode to cushion a downturn with aggressive rate cuts or fiscal stimulus. That raises the potential economic cost of a disorderly correction.

One attribution caveat: the ECB explicitly made the “correction likely” argument. The CAPE, investment-to-GDP, debt, and record-index observations are market and research indicators; the available evidence does not show that the Federal Reserve itself issued all of those precise numerical warnings.