SF Fed's Daly warns AI chip and energy squeeze could outlast the Fed's usual patience

The Federal Reserve can usually wait out a supply shock. Mary Daly, president of the Federal Reserve Bank of San Francisco, is not sure it can wait out this one.

In an interview with Axios, Daly said that surging AI demand for chips and related technology is beginning to look like a prolonged inflation pressure rather than a temporary disruption — and that it could compound with tariffs and rising energy costs to keep inflation elevated well beyond the window the Fed would normally tolerate.

I see it less as a one-off.— Mary Daly, President, Federal Reserve Bank of San Francisco

The Fed's standard framework treats supply shocks as temporary — typically expecting relief within one to three years, after which policymakers can look past the price pressure without raising rates further. Daly told Axios that AI demand does not fit that template.

This is probably further out before we get relief.— Mary Daly, President, Federal Reserve Bank of San Francisco

It doesn't seem like the demand for AI is going down. If anything, it seems like it's going up.— Mary Daly, President, Federal Reserve Bank of San Francisco

Daly's district covers Silicon Valley and many of the companies at the center of the AI buildout, giving her an unusually direct line to what businesses are actually doing. In conversations with those contacts, she said she is starting to hear that the AI boom is reshaping purchasing and product-design decisions well beyond the data center sector — a sign the chip squeeze is spreading.

One concrete signal: some firms are seeking forward contracts for memory chips, according to Daly, to lock in a guaranteed supplier before supplies tighten further. That kind of hedging behavior suggests companies themselves do not expect the crunch to resolve quickly.

Daly's warning aligns with remarks from other senior Fed officials. New York Fed President John Williams, in late September 2026 remarks reported by Crypto Briefing, described AI-driven demand as an increasingly salient inflation force, framing it as 'a race between available supply and surging demand' for chips, power equipment, and data-center inputs. Williams said AI capital expenditures are running in the hundreds of billions of dollars each year, and that without that investment, U.S. growth would have been approximately one-third weaker during certain periods of 2026.

Williams also flagged a complication that limits how much the AI spending boom actually boosts the U.S. economy: much of the investment is offset by imports. When companies buy chips and equipment from abroad, the spending flows out of the country, reducing its contribution to GDP even as it adds to demand-side price pressure.

There is a further wrinkle. Williams noted that AI's expected productivity gains — the eventual payoff that could help offset higher costs — have not yet appeared in the economic data. Until they do, the Fed faces a situation where AI is pushing prices up without yet delivering the growth dividend that would make that inflation easier to absorb.

For the Fed, supply shocks present a structural problem: the central bank's main tool is interest rates, which work by cooling or heating demand. Rate hikes cannot build a chip factory or expand a power grid. If AI demand keeps outrunning supply for years rather than months, the Fed may face sustained pressure to keep rates higher than it otherwise would — or risk letting inflation become entrenched.

Williams laid out a gradual inflation path in his September remarks, according to Crypto Briefing: 3.5% for 2026, falling to just above 2% in 2027, and reaching the Fed's 2% long-run target only by 2028. That forecast assumes energy prices stabilize and AI supply and demand move into better balance — two conditions that, as Daly's comments suggest, are far from guaranteed.

Energy and AI are not separate stories, either. Data centers are power-hungry, meaning that AI demand amplifies energy costs at the same time that Middle East tensions are already pressuring commodity prices. Daly's concern is precisely that these forces do not stay separate — that AI, tariffs, and energy costs could compound each other in ways that keep inflation elevated and require more policy tightening than markets currently expect.

Why it matters — If senior Fed officials conclude that AI demand is a lasting inflation driver rather than a passing shock, the central bank may keep interest rates higher for longer than markets expect — affecting borrowing costs for consumers, businesses, and governments.

Reported by axios.com, reuters.com, cryptobriefing.com