AI shifts from “reason for rate cuts” to “argument for rate hikes”: Cook says only a small portion of the 2 trillion dollars capital expenditure is spent, and electricity and water prices have increased by about 5%.

Federal Reserve Governor Lisa Cook on Monday delivered the most specific quantitative warning yet from a Fed official about AI’s inflationary effects: productivity gains driven by artificial intelligence cannot offset short-term inflationary pressure in time. The role of AI within the Fed's policy framework is shifting—from a market-anticipated “reason for rate cuts” to a “justification for hikes” supporting further tightening.

According to Cook’s speech in Oakland, California, large-scale data center investment has already increased electricity and water costs by around 5%; of the $2 trillion in AI capital expenditure pledged by companies, only a small portion has been spent so far, and the release of remaining funds will intensify price pressure. She also noted that the AI-driven stock market wealth effect is boosting consumption, further offsetting the mild disinflation brought by productivity gains.

“At present, I expect that productivity gains will bring mild inflation moderation in the coming years; however, I do not believe these positive effects will appear in time to offset the mounting inflation pressures later this year.”

Cook stated that the Fed’s interest rate hike earlier this month was necessary to address high inflation, and that the future policy path will be guided by economic data. She also commented that the labor market can withstand further rate hikes and that economic growth has “remained notably resilient over the past year.”

On the market level, federal funds futures currently show a roughly 70% probability of another rate hike in October. Fed officials have spoken intensively recently, generally citing ongoing economic momentum and a strong labor market as justification for further tightening.

5% rise in electricity and water prices is just the beginning

Cook pointed out that large-scale data center investment is intensifying competition for shared resources such as power and construction labor, which is already reflected in the roughly 5% rise in electricity and water costs over the past year.

More importantly, of the $2 trillion in AI-related capital expenditure pledged by companies, only “a small portion” has been spent so far. As more funds are deployed, demand for energy, land, equipment, and labor will increase further, and price pressures may spread from localized sectors to the broader economy.

This assessment contrasts with previous market perceptions about AI. Over the past two years, investors have generally viewed AI as a supply-side positive, making operations more efficient and lowering costs. Cook’s remarks, however, reveal an overlooked transmission chain: before productivity dividends are realized, large-scale AI deployment first creates a demand-side shock, with capital expenditure itself pushing up inflation.

Wealth effect becomes the second inflationary pathway

Cook identified a second AI inflation transmission channel—the wealth effect from the stock market. AI-driven rallies in tech stocks have pushed up household net worth, spurring consumption. Against a backdrop of a still-strong labor market, this wealth-driven increase in demand could counteract the mild disinflation provided by productivity gains.

Cook acknowledged that AI “could be the most significant technological transformation of our lifetimes,” but expressed clear uncertainty about the speed and magnitude of the productivity boost.

She said that “any estimate about how and when this mechanism will work involves uncertainty and is worth further study and discussion”; her current expectation is that productivity gains will provide “mild disinflation in the coming years,” but “I do not believe these effects will arrive in time to offset the mounting inflation pressures later this year.”