That creates a one-way risk. If expectations for earnings, adoption or the cost of capital change, investors may demand a higher risk premium and reprice the same companies abruptly. The ECB’s historical comparison is therefore not a prediction that the current market will repeat the exact events of 2000. It is a warning that genuine technological revolutions can coexist with excessive valuations and subsequent corrections.

What the Federal Reserve is watching

The Fed’s May 2026 Financial Stability Report said asset-valuation pressures were elevated. The forward price-to-earnings ratio for S&P 500 companies remained in the upper range of its historical distribution, while the estimated equity premium—the compensation investors receive for taking equity risk—remained well below its historical average. Corporate bond and loan spreads were also low by historical standards.

Together, those conditions can leave markets sensitive to disappointing earnings, higher interest rates or a decline in risk appetite. But the Fed’s assessment is more cautious than the ECB’s: a senior Fed official said the situation did not look like a bubble, pointing to the stronger earnings of the companies leading the investment cycle.

So the two messages are compatible. The Fed is identifying vulnerability in prices and risk premia without forecasting a crash; the ECB is making a stronger historical argument that a correction should be expected.

Europe is exposed even without its own AI mega-cap bubble

Europe’s smaller technology sector may reduce the risk of a home-grown, dot-com-style collapse. It does not insulate the region from a fall in U.S. technology stocks, however. Euro-area households, insurers, pension funds and other investors have exposure to the Magnificent Seven through direct holdings, index funds and other investment products.

That exposure creates a wealth channel. A sharp fall in companies such as Apple, Alphabet, Microsoft, Amazon, Meta, Nvidia and Tesla would reduce the value of portfolios held by European investors. Correlated selling and weaker market sentiment could also weigh on European equities, even if European companies were not responsible for the original sell-off.

The ECB has cited roughly €440 billion in euro-area household exposure to U.S. technology stocks, much of it held through funds tracking major indexes. The figure should be read as an exposure estimate, not as a forecast of losses: the eventual impact would depend on the size, speed and composition of any market decline.

AI debt adds a credit-market transmission channel

The AI investment cycle is no longer financed only from the internal cash flows of large technology companies. Technology firms are increasingly tapping bond and equity markets to fund data centres, cloud capacity and related infrastructure.

The scale of issuance has become significant. U.S. corporate issuance totaled $1.68 trillion from January through mid-August 2026, up nearly 27% from the same period a year earlier, according to data cited by Reuters. AI hyperscalers had issued about $220 billion in debt in 2026 as of August 10, compared with $12.5 billion in the comparable period the previous year.

This does not mean the borrowing is automatically destabilizing. Large issuers generally have stronger earnings and balance sheets than many companies involved in the dot-com boom, and AI-related debt remains only one part of the wider credit market. But borrowing creates a second route for disappointment to spread. If expected AI returns weaken, technology shares could fall while bond investors demand higher yields, refinancing becomes more expensive and the value of existing corporate debt declines.

Why record-high U.S. equity levels matter

The Fed’s July meeting minutes said broad equity indexes remained close to all-time highs, while credit spreads for hyperscaler firms had widened relative to investment-grade issuers. High prices are not, by themselves, evidence that a crash is imminent. They do mean that markets have less room for a disappointment without changing the assumptions embedded in valuations.

The key questions are whether AI revenues and productivity gains will arrive quickly enough to justify current prices, whether infrastructure spending will generate adequate returns, and whether interest rates will remain compatible with those valuations. A change in any of these variables could affect both equity and credit markets.

Limited policy room could make a correction harder to absorb

The ECB argues that policymakers may have less room than in earlier downturns to cushion a major shock through aggressive rate cuts or fiscal stimulus. Its concern is that existing policy-rate and public-finance constraints could make it harder to offset a sudden loss of market wealth and confidence.

That does not imply that central banks would be unable to respond. It means the response could be less powerful or less immediate than during periods when rates were much higher and fiscal capacity was more flexible.

The starting point in Europe is already not entirely comfortable. The ECB’s May Financial Stability Review reported tighter bank credit standards, weaker reported loan availability and some softening in new lending to euro-area non-financial companies. A market shock arriving alongside those conditions could intensify the pressure on businesses seeking finance.

The possible feedback loop from markets to jobs

The broader risk is not simply that expensive technology shares lose value. A potential chain of events would look like this:

1. AI earnings or productivity expectations disappoint, or interest rates rise.
2. Investors reprice technology equities and the bonds financing AI infrastructure.
3. Lower asset values weaken household and institutional confidence.
4. Higher risk premia and tighter lending conditions raise the cost of finance.
5. Companies defer investment, expansion and hiring.
6. Weaker demand then reinforces the original pessimism about AI-related returns.

This sequence is a risk scenario, not a forecast. The ECB describes the potential consequences for the euro area as severe, while the Fed’s published assessment establishes elevated valuation vulnerability without endorsing the ECB’s conclusion that a correction is likely.

Is this another dot-com crash?

The comparison is useful when applied to market structure rather than treated as a timetable. Both episodes involve powerful technology narratives, concentrated leadership and uncertainty about which companies will ultimately capture the economic gains. The main lesson is that a technology can transform the economy while investors in highly valued companies still experience disappointing returns.

The available evidence does not establish that a 2000-style collapse is inevitable. The leading firms in today’s AI cycle have, in general, more substantial earnings, cash flow and balance sheets than many dot-com-era businesses. The warning is narrower but still consequential: even rational enthusiasm can be followed by a valuation correction if future gains are already priced in.

For investors and policymakers, the most important indicators are therefore not just headline share prices. They include the concentration of portfolios in the Magnificent Seven, the gap between AI spending and realized returns, the pace and terms of corporate borrowing, bond-market absorption capacity, credit standards and the willingness of firms to keep investing and hiring when financing conditions tighten.

That creates a one-way risk. If expectations for earnings, adoption or the cost of capital change, investors may demand a higher risk premium and reprice the same companies abruptly. The ECB’s historical comparison is therefore not a prediction that the current market will repeat the exact events of 2000. It is a warning that genuine technological revolutions can coexist with excessive valuations and subsequent corrections.

What the Federal Reserve is watching

The Fed’s May 2026 Financial Stability Report said asset-valuation pressures were elevated. The forward price-to-earnings ratio for S&P 500 companies remained in the upper range of its historical distribution, while the estimated equity premium—the compensation investors receive for taking equity risk—remained well below its historical average. Corporate bond and loan spreads were also low by historical standards.

Together, those conditions can leave markets sensitive to disappointing earnings, higher interest rates or a decline in risk appetite. But the Fed’s assessment is more cautious than the ECB’s: a senior Fed official said the situation did not look like a bubble, pointing to the stronger earnings of the companies leading the investment cycle.

So the two messages are compatible. The Fed is identifying vulnerability in prices and risk premia without forecasting a crash; the ECB is making a stronger historical argument that a correction should be expected.

Europe is exposed even without its own AI mega-cap bubble

Europe’s smaller technology sector may reduce the risk of a home-grown, dot-com-style collapse. It does not insulate the region from a fall in U.S. technology stocks, however. Euro-area households, insurers, pension funds and other investors have exposure to the Magnificent Seven through direct holdings, index funds and other investment products.

That exposure creates a wealth channel. A sharp fall in companies such as Apple, Alphabet, Microsoft, Amazon, Meta, Nvidia and Tesla would reduce the value of portfolios held by European investors. Correlated selling and weaker market sentiment could also weigh on European equities, even if European companies were not responsible for the original sell-off.

The ECB has cited roughly €440 billion in euro-area household exposure to U.S. technology stocks, much of it held through funds tracking major indexes. The figure should be read as an exposure estimate, not as a forecast of losses: the eventual impact would depend on the size, speed and composition of any market decline.

AI debt adds a credit-market transmission channel

The AI investment cycle is no longer financed only from the internal cash flows of large technology companies. Technology firms are increasingly tapping bond and equity markets to fund data centres, cloud capacity and related infrastructure.

The scale of issuance has become significant. U.S. corporate issuance totaled $1.68 trillion from January through mid-August 2026, up nearly 27% from the same period a year earlier, according to data cited by Reuters. AI hyperscalers had issued about $220 billion in debt in 2026 as of August 10, compared with $12.5 billion in the comparable period the previous year.

This does not mean the borrowing is automatically destabilizing. Large issuers generally have stronger earnings and balance sheets than many companies involved in the dot-com boom, and AI-related debt remains only one part of the wider credit market. But borrowing creates a second route for disappointment to spread. If expected AI returns weaken, technology shares could fall while bond investors demand higher yields, refinancing becomes more expensive and the value of existing corporate debt declines.

Why record-high U.S. equity levels matter

The Fed’s July meeting minutes said broad equity indexes remained close to all-time highs, while credit spreads for hyperscaler firms had widened relative to investment-grade issuers. High prices are not, by themselves, evidence that a crash is imminent. They do mean that markets have less room for a disappointment without changing the assumptions embedded in valuations.

The key questions are whether AI revenues and productivity gains will arrive quickly enough to justify current prices, whether infrastructure spending will generate adequate returns, and whether interest rates will remain compatible with those valuations. A change in any of these variables could affect both equity and credit markets.

Limited policy room could make a correction harder to absorb

The ECB argues that policymakers may have less room than in earlier downturns to cushion a major shock through aggressive rate cuts or fiscal stimulus. Its concern is that existing policy-rate and public-finance constraints could make it harder to offset a sudden loss of market wealth and confidence.

That does not imply that central banks would be unable to respond. It means the response could be less powerful or less immediate than during periods when rates were much higher and fiscal capacity was more flexible.

The starting point in Europe is already not entirely comfortable. The ECB’s May Financial Stability Review reported tighter bank credit standards, weaker reported loan availability and some softening in new lending to euro-area non-financial companies. A market shock arriving alongside those conditions could intensify the pressure on businesses seeking finance.

The possible feedback loop from markets to jobs

The broader risk is not simply that expensive technology shares lose value. A potential chain of events would look like this:

1. AI earnings or productivity expectations disappoint, or interest rates rise.
2. Investors reprice technology equities and the bonds financing AI infrastructure.
3. Lower asset values weaken household and institutional confidence.
4. Higher risk premia and tighter lending conditions raise the cost of finance.
5. Companies defer investment, expansion and hiring.
6. Weaker demand then reinforces the original pessimism about AI-related returns.

This sequence is a risk scenario, not a forecast. The ECB describes the potential consequences for the euro area as severe, while the Fed’s published assessment establishes elevated valuation vulnerability without endorsing the ECB’s conclusion that a correction is likely.

Is this another dot-com crash?

The comparison is useful when applied to market structure rather than treated as a timetable. Both episodes involve powerful technology narratives, concentrated leadership and uncertainty about which companies will ultimately capture the economic gains. The main lesson is that a technology can transform the economy while investors in highly valued companies still experience disappointing returns.

The available evidence does not establish that a 2000-style collapse is inevitable. The leading firms in today’s AI cycle have, in general, more substantial earnings, cash flow and balance sheets than many dot-com-era businesses. The warning is narrower but still consequential: even rational enthusiasm can be followed by a valuation correction if future gains are already priced in.

For investors and policymakers, the most important indicators are therefore not just headline share prices. They include the concentration of portfolios in the Magnificent Seven, the gap between AI spending and realized returns, the pace and terms of corporate borrowing, bond-market absorption capacity, credit standards and the willingness of firms to keep investing and hiring when financing conditions tighten.