Every summer the world’s most powerful central bankers decamp to a hillside town outside Lisbon to argue about the economy in relative calm. This year the argument had a single organising subject, and it was not inflation in the usual sense.
It was artificial intelligence, and specifically the awkward fact that nobody in the room could say with confidence whether it will make their job easier or a great deal harder.
The occasion was the European Central Bank’s annual Forum on Central Banking, held in Sintra from 29 June to 1 July under the theme “Shaping Europe’s future: innovation, growth and stability”.
On the marquee policy panel, Fed Chair Kevin Warsh, ECB President Christine Lagarde, Bank of England Governor Andrew Bailey and Bank of Canada Governor Tiff Macklem sat together to work through what AI actually means for growth, for prices and for financial stability. The tone was less triumphant than searching.
The problem they kept circling is a genuinely hard one. AI promises a productivity boom that could, in theory, let economies grow faster without pushing prices up. Getting there, though, runs through an investment surge so large it is inflationary in the near term.
The major AI firms committed roughly $300bn to capital spending in 2025 alone, pouring money into chips, power and data centres, and that spending lands as demand on the economy long before any productivity gains show up in the figures.
So far the gains are real but modest. US output per hour rose about 2.2% last year, which looks more like a recovery from a weak patch than the step change the technology’s boosters describe. Warsh said inflation remains too elevated even as Fed officials have grown more open-minded about AI eventually proving deflationary.
Easing policy today on the strength of a productivity leap that has not yet arrived, most policymakers agreed, would be a risky bet to make with demand already running hot.
Lagarde used her time to make a point that is uncomfortable for her own continent. Europe is lagging on AI investment and on the frontier companies driving the breakthroughs, she acknowledged, before adding that Europe and the United States are, in her phrase, “sort of hostage to each other” when it comes to making progress.
It was a rare admission of dependence from a central bank chief who has spent years arguing for European strategic autonomy.
The labour-market thread ran underneath all of it. A recent survey by the Federal Reserve Bank of New York found firms are not planning mass layoffs so much as quietly scaling back hiring, a shift that may already be feeding the unusually low rate of job creation in the US.
That is a subtler kind of disruption than the wave of redundancies people tend to fear, and a harder one for a central bank to read in real time.
None of this is abstract worry. The Bank for International Settlements has warned that a bust in AI investment could hit credit markets with a force comparable to 2008, and Lagarde herself has gone further, arguing that AI could trigger financial crises and calling for governance modelled on Cold War arms control.
Others are already reaching for the technology as a fix rather than a threat. The Bank of Italy has opened talks with the big developers, pitching AI as a cure for chronic low productivity, while Morgan Stanley now expects European banks to shed a fifth of their jobs to it by 2030.
What Sintra produced, in the end, was not a decision but a shared unease. The people who set the price of money left having agreed on the size of the question and very little about the answer.
The data they need to resolve it does not yet exist, and by the time it does, the rates that hinge on it will already have been set.
