Cook on AI and inflation: near-term pressure, slower relief

The news

Federal Reserve Governor Lisa Cook used a September 28 keynote at Oakland Tech Week to set out how she thinks about AI and inflation, and what the technology means for the labor market and for small businesses. She concluded that in the short term AI appears to add inflationary pressure, postponing the return to 2 percent, while the longer-term picture is more optimistic.

Key takeaways:

• Cook said AI-related demand is pushing up prices for chips, computers and software, and that some pressure is spreading more broadly.
• She expects productivity gains to bring modest disinflation within the next few years, but not in time to offset broadening pressure later this year.
• Total inflation was an estimated 3.8 percent in the 12 months through August, and core inflation 3.4 percent.
• She voted with the rest of the FOMC to raise rates 25 basis points in September.

Our analysis

The central argument on prices is about where the demand lands. Cook said part of the surge in AI-related prices reflects a shift in demand toward one sector, not an increase in economy-wide demand. Pressure of that kind should resolve as supply chains adjust, and fighting it with monetary policy could be a mistake, because the Fed’s tools are too blunt to target narrow sectors.

She nonetheless sees signs of broadening. Data-center construction draws on labor and energy used across the economy. Companies have spent only a small fraction of roughly $2 trillion in announced plans. A large part of the rise in equity prices is attributable to enthusiasm about AI, and that wealth appears to be feeding household spending. In the price data, she cited electricity and water costs each up about 5 percent over the past year and core goods prices running above a 3 percent annual pace this year, after drifting down before the pandemic.

The relief channel is productivity. If supply capacity grows faster than demand, prices can ease. Cook expects the effect to be limited but real if gains are spread evenly, because higher productivity also lifts demand through expected wages, returns and wealth, some of which she thinks is already in today’s data. She sketched scenarios with more relief, if gains are concentrated among higher-income consumers, are not passed on to wages or if job reallocation raises precautionary saving, and with less, if market concentration lets firms keep higher markups.

On jobs, Cook said there is limited evidence so far of significant changes to the labor market’s structure: unemployment and layoffs have been low and relatively flat for two years. She pointed to evidence that AI may be reducing labor demand in coding and simultaneous translation, and to harder searches for recent graduates. She flagged a scenario in which AI temporarily raises unemployment because of skills mismatch, in which case the Fed would have limited tools: cutting rates to address unemployment could risk fueling inflation.

She spent time on small businesses. The Fed’s Small Business Credit Survey found nearly half of small employer firms using AI, with 71 percent reporting higher productivity. Firms with fewer than 500 employees make up more than 99 percent of U.S. businesses and have accounted for 61 percent of net new job creation since 1995.

On policy, she said the September increase was meant to address inflation that has been too high for too long. The unemployment rate was 4.1 percent in August, and she sees the labor market as able to handle higher rates. She will consider what policy rate is needed to bring inflation to target, with the number and size of future adjustments guided by the economy’s reaction and by incoming data.

Room for disagreement

Cook herself flags uncertainty over the breadth and timing of the productivity channel and says she will adjust her view based on data. Other policymakers could weigh the same facts differently. If AI-related price increases are concentrated in a few sectors, as she argues, a case exists for looking through them rather than tightening. Her answer is that the pressure is broadening and that wealth effects add to demand.

The labor market evidence is also two-sided. Entry-level job difficulty and sector declines coexist with low unemployment, and she notes new tasks and occupations that cannot yet be foreseen. She also said adoption is happening faster than PCs or the internet did at comparable points, which she reads as encouraging for smaller firms.

What to watch

• Electricity, water and core goods prices, the categories she cited as showing broadening pressure.
• Labor data for recent graduates and AI-exposed occupations such as software development.
• Whether announced AI investment plans of roughly $2 trillion are carried out on schedule.
• Further Committee commentary on how many additional rate adjustments may be needed.