What the record says
The Federal Reserve released the minutes of the 15–16 September meeting on 7 October. They are notable less for the decision than for how often one phrase appears.
In the staff’s account of why inflation was higher than a year earlier, the causes listed are past tariff increases, higher energy and input costs from geopolitical developments, and “an increase in technology-related consumer goods prices associated with the AI buildout”. That last clause puts data-centre demand in the price index for things households buy.
The sentence that matters
In the participants’ discussion: “Several participants observed that the rate of price increases in the core goods category also remained elevated, as effects of the AI buildout appeared to increase while the effects of tariff increases waned.”
Two forces, moving in opposite directions, in one clause. For two years the standard explanation for stubborn goods prices in the United States has been tariffs. The Committee’s own record now says that explanation is weakening and another one is strengthening.
Where else it turns up
The build-out appears across the document. It “continued to support robust increases in business investment spending” and “also fueled strong imports of high-tech capital goods”. Several participants said its scale and pace “had continued to surprise to the upside”. Some noted that “strong demand for skilled workers in sectors related to the ongoing AI buildout had been driving strong wage gains for these workers”. Equity indexes rose on earnings “bolstered by the AI buildout”, and spreads on bonds issued by AI hyperscalers narrowed slightly after widening earlier in the year.
On risks: “Some participants commented that the AI buildout could cause aggregate demand to outpace aggregate supply over the medium term, putting upward pressure on inflation.” A few participants also flagged repercussions from rapid AI adoption, including cybersecurity risks, that could act as a drag on productivity in some cases.
The decision
All members agreed to raise the target range by a quarter point to 3.75–4%. Voting against: none. The staff put total inflation at 3.6% in August and core at 3.2% under the BEA’s new methodology, with unemployment at 4.1% in July and August.
The staff’s forecast has inflation stepping down over the next two years “as the effects of tariffs, geopolitical developments, and the AI buildout were expected to wane”, reaching 2% in 2029. That forecast is somewhat higher for 2026 through 2028 than the one prepared in July.
What this is and is not
It is not a finding that AI is inflationary. It is a central bank’s committee recording, in its own minutes, that an investment boom in one sector is now large enough to show up in consumer goods prices, business investment, import volumes, skilled wages and corporate credit spreads at the same time.
The word “data centre” does not appear anywhere in the minutes. “AI buildout” does, repeatedly, and that is the Fed’s chosen term for the same thing.
What to watch
Whether the next set of minutes still has the two forces moving in opposite directions, or whether the tariff clause drops out altogether. The staff is forecasting the AI effect to fade; the participants are describing it as surprising to the upside. Those two cannot both keep being true.
