Analysis
Austan Goolsbee, president of the Federal Reserve Bank of Chicago and an FOMC member, used a [Fortune interview published Sept. 2](https://fortune.com/2026/09/02/austan-goolsbee-old-school-economics-consumers-health/) to make an argument that few people in the technology industry have had to answer: the AI data center boom may be reallocating economic activity rather than creating it.
His phrasing was direct. Data center expansion is "very hot, but largely shoving other parts of the economy down." The mechanism he cites is labor and input cost -- construction workers, electricians and HVAC technicians pulled toward campus projects, raising costs for every other structure that needs them. That is a crowding-out claim, and it is testable in a way most AI macro commentary is not.
On productivity he was skeptical of the industry's own forecasts, noting that technologists "have declared numerous technologies were going to totally change the world" and "we're still waiting." His framing for policy is precise: what matters to the Fed is not whether AI raises productivity but whether the gain is expected or unexpected, because an anticipated productivity increase is already in the price path while a surprise one is disinflationary.
“That is a crowding-out claim, and it is testable in a way most AI macro commentary is not.”
On rates, Goolsbee said he agrees with chair Kevin Warsh that inflation is the primary concern, that progress stalled and is now getting worse, and that price increases above the 2% target trace to supply shocks including tariffs and Middle East disruption. He supported holding steady at the July FOMC meeting, citing improved recent CPI prints as grounds for a wait-and-see stance. He names reduced consumer spending as the single biggest risk to continued growth.
Set this against what the administration argued the same week -- that AI compute investment reaching roughly 1.4% of GDP is the economy's "Golden Goose" -- and the disagreement is not about the magnitude of the spending. Both sides agree it is large. The disagreement is whether large spending in one sector, funded substantially by debt and depreciating on a three-to-five-year schedule, adds to output or borrows from it.
For anyone raising a fund or planning a 2027 financing, the practical translation is that the Fed's most vocal dove is now talking about sectoral overheating rather than about cuts. Terminal rate assumptions built into growth-stage models over the past year deserve a second look.
