Bank of France Warns an AI Stock Crash Would Cross the Atlantic

Why the Bank of France Is Worried About AI Valuations

Speaking at the World In Progress conference in Barcelona on 2-3 November, Bank of France governor Emmanuel Moulin argued that the artificial intelligence boom is no longer just a technology story. It has become a macroeconomic force. AI-related investment, he said, will jump from roughly $400bn to $1,000bn next year, and a large share of that spending is financed with debt. That borrowing competes directly with governments for lenders' money, pushing up interest rates for everyone, including France.

Moulin's second warning was about contagion. European investors hold between 15% and 20% of US equities, so a sharp fall in the valuations of American AI companies would not stay on the other side of the Atlantic. Former ECB vice-president Luis de Guindos, on the same panel, went further: productivity gains from AI are being overestimated, and a market correction 'might not be such a bad thing'.

The panel also exposed a split. Bank of Spain governor Jose Luis Escriva Belmonte said he remains optimistic about AI's productivity potential, though he admitted nobody knows when or where those gains will materialise. On inflation, he took a hawkish stance, describing the energy-price disruption from the war with Iran as more persistent than expected and flagging a possible ECB rate increase in December. Moulin, by contrast, described the indirect effects of higher energy prices as moderate so far.

The backdrop was France itself. With French 10-year yields near 5% against 3.8% for Spain, Moulin acknowledged investor concern over the country's budget and political situation, but argued a budget is very likely to pass via constitutional procedures such as Article 49.3 and ordinances, and that markets had overreacted.

Moulin vs. Escriva: A Split Inside the Eurosystem

Moulin's Crowding-Out Argument

The core of Moulin's case is a mechanism economists call crowding out. When AI companies borrow hundreds of billions of dollars, they absorb credit that would otherwise fund governments. With lenders increasingly stretched, rates rise, especially for borrowers whose credit quality is not optimal. De Guindos endorsed this reading, calling AI, alongside fiscal policy, the main factor behind current bond-market tension. For France, already paying close to 5% on 10-year debt, that is an uncomfortable message: the competition for capital is not only domestic.

Why Europe Cannot Insulate Itself

Moulin's contagion warning rests on a concrete number: European investors own 15% to 20% of US equities. If AI-linked valuations in the US correct sharply, European portfolios take the hit directly, and the resulting wealth and confidence effects feed back into European spending and markets. De Guindos framed the risk differently, pointing to market fragmentation rather than correlation, but the practical conclusion for European asset owners is similar.

A Genuine Split Inside the Eurosystem

The panel made clear that eurozone central bankers do not share one view. Escriva remains confident in AI-driven productivity gains, though he concedes uncertainty over timing and location in the value chain. He is also the most hawkish voice on inflation, citing second-round effects from the Iran war's energy shock and floating a December rate hike. Moulin, carrying the weight of French debt, is the more dovish on prices but the more alarmed on AI. For context, this divergence matters because it shapes how the ECB's December meeting is likely to be framed: as a debate between energy-driven inflation risk and financial-stability risk.

The Valuation Question Nobody Can Settle

Economist Andres Velasco of the London School of Economics summed up the dilemma: either Wall Street is blind, or the world is witnessing a gigantic economic revolution. His addendum is the operative one for risk managers: crises always take longer to arrive than expected, and when they come, they come faster than anticipated. That asymmetry, not a prediction of a crash, is the practical takeaway from the Barcelona panel.

What the AI-Debt Warning Means for Investors and Treasurers

The most important insight: the AI trade is now a macro trade, and its funding costs are already showing up in government bond yields.

For investors and corporate treasurers, the Barcelona panel points to three concrete implications.

• Reassess exposure to the AI-debt nexus. Moulin's $400bn-to-$1,000bn investment figure is largely debt-financed. Anyone holding AI-linked equities or the credit of heavy AI borrowers is exposed to the same refinancing cycle that is lifting yields.
• Quantify the US-equity channel. European portfolios hold 15-20% of US equities. Stress-testing for a 20-30% drawdown in US AI names is now a reasonable exercise, not a hypothetical one.
• Watch the ECB's December meeting. Escriva's hawkish comments on energy-driven second-round inflation make a December rate decision a live event; Moulin's more dovish read means the outcome is genuinely contested.

Metrics worth tracking: French 10-year yields against the 3.8% Spanish level, the ECB's December rate decision, and any change in the pace of AI-related debt issuance.