The Federal Reserve meets this week facing a paradox. Artificial intelligence is supposed to make everything cheaper. For now, it is making the Fed’s inflation problem harder.
I expect the Fed to pause, as I argued before September’s meeting. That is the right decision, and it reflects strength rather than weakness. Last month the question was credibility: would the Fed act against renewed inflation? It did. The question now is how much more it needs to do. My answer is not much.
September CPI rose 3.7 percent from a year earlier, core inflation has been running at a 3.4 percent annualised pace over three months, and oil remains above US$100 a barrel. Energy is only part of the story. AI investment is creating immediate demand for electricity, construction, equipment and scarce computing capacity. Computers, which for decades got cheaper every year, are now going up in price at double-digit rates.
We are paying for the infrastructure before collecting the productivity dividend.
But that dividend is coming, and the Fed should not lose sight of it. Central banks cannot set today’s interest rates against tomorrow’s efficiencies, but nor should they ignore them. Higher rates cannot manufacture turbines or expand electricity grids. They work only by persuading somebody to cancel or delay spending, a crude fix for a physical shortage. Pushing rates up until the most dynamic part of the economy slows would do disproportionate harm everywhere else, starting with housing.