Federal Reserve Governor Lisa Cook on Monday delivered the most specific quantitative warning yet from a Fed official on AI's inflationary effects: the productivity gains from artificial intelligence cannot offset short-term inflation pressures in time. The role of AI in the Fed's policy framework is reversing—from what markets once hoped would be a reason for rate cuts into an argument supporting tighter policy.

According to Cook's remarks in Oakland, California, large-scale data center investment has already pushed electricity and water costs up by about 5%; of the $2 trillion in AI capital expenditure pledged by companies, only a small portion has been spent so far, and the deployment of the remaining funds will intensify price pressures. She also noted that AI-driven stock market wealth effects are stimulating consumption, further offsetting the mild disinflation that productivity gains might bring.

"At present, I expect productivity gains to bring modest disinflation over the next few years, however, I do not believe these positive effects will arrive in time to offset the broadening inflation pressures later this year."

Cook said the Fed's rate hike earlier this month was necessary to address elevated inflation, and that the future policy path will be guided by economic data. She also said the labor market can withstand further rate increases, and that economic growth "has remained remarkably resilient over the past year."

At the market level, federal funds futures currently show about a 70% probability of another rate hike in October. Fed officials have recently spoken intensively, broadly citing continued economic momentum and a strong labor market as grounds for further tightening.

The 5% rise in power and water costs is only the beginning

Cook pointed out that large-scale data center investment is intensifying competition for shared resources such as electricity and construction labor, and this competition has already shown up as roughly a 5% increase in electricity and water costs over the past year.

More critically, of the $2 trillion in AI-related capital expenditure pledged by companies, only "a small portion" has been spent so far. As the remaining funds are gradually deployed, demand for energy, land, equipment, and labor will rise further, and price pressures may spread from localized areas to broader parts of the economy.

This assessment contrasts with the market's previous understanding of AI. Over the past two years, investors generally viewed AI as a supply-side benefit that boosts efficiency and lowers costs. Cook's remarks reveal a neglected transmission chain: before the productivity dividend materializes, the large-scale deployment of AI first creates a demand-side shock, and capital expenditure itself is pushing inflation higher.

Wealth effects become a second inflation path

Cook identified a second path for AI's inflationary effects—the stock market wealth effect. AI-driven gains in technology stocks have boosted household net worth, thereby stimulating consumption. Against the backdrop of a still-strong labor market, this wealth-driven demand growth may offset the mild disinflation that productivity gains would bring.

Cook acknowledged that AI "has the potential to be the most significant technological change in our lifetimes," but expressed clear uncertainty about the speed and magnitude of productivity gains.

She said, "There is uncertainty around any estimate of how and when this mechanism will work, and it warrants further research and discussion"; she currently expects productivity gains to provide "modest disinflation" over the next few years, but "I do not think these effects will arrive in time to offset the broadening inflation pressures later this year."