When the Fed Finally Says the Quiet Part About AI
The moment arrived without fanfare, tucked inside a sober FedViews update from the San Francisco Fed. “Activities related to artificial intelligence” are now “a source of uncertainty around the 2026 outlook,” wrote senior policy adviser Mark M. Spiegel, noting they “played an outsized role” in 2025 growth. In central-bank speak, that is not a footnote. It is a declaration that AI has graduated from a story about which startups are hot to a variable that could tilt employment, consumption, investment, and inflation. The Fed just moved AI from the margins of the board to the squares that actually matter.
From firm strategy to aggregate demand
The clearest signal that AI has crossed into macro territory comes from where the hiring is—and isn’t. Using Lightcast data, the Fed finds that AI- and information-related postings accelerated in 2025, but the surge clustered in four knowledge-heavy arenas: information, advanced manufacturing, finance, and professional and technical services. Those sectors accounted for just over a quarter of U.S. output—26.3%—yet they delivered half of real GDP growth in Q3 2025. Inside that quartile of the economy, AI-related postings averaged 6.5% of listings. Everywhere else? 1.2%.
This is not diffusion; it’s concentration. And concentration is precisely how new technologies first move the macro needle. A handful of sectors scale compute spend, rewire workflows, and pull forward productivity and profits. Suppliers and investors chase them. The rest of the economy watches. For policymakers, that pattern translates into lumpy demand—investment booms in some corridors, slack or hesitation in others—and the risk that headline aggregates hide widening gaps beneath.
The labor market is being rerouted, not erased
The Fed’s broader labor snapshot reinforces the point. January payrolls grew by 130,000, with private employment up 172,000 offset by a 34,000 drop in federal jobs. Unemployment edged down to 4.3%. Benchmark revisions quietly recast 2025 as cooler than it looked at the time, trimming average monthly gains by roughly 35,000 to about 15,000. The composition is the tell: health care kept adding bodies; manufacturing slipped. What emerges is not a mass layoff cycle pinned on AI, but a channel shift toward knowledge-intensive roles and services that can absorb and direct new tools. This is how automation shows up at first—in hiring maps and wage ladders—before it appears in productivity tables.
Why the Fed cares now
Two macro channels are flashing at once. First is the demand side: AI buildouts—models, data infrastructure, energy, specialized labor—act like a capex surge. They can goose growth and, if bunched, stoke local price pressures. Second is supply: if those investments produce step-changes in output per worker, the economy’s speed limit rises, easing inflation even as growth holds up. The Fed’s word—uncertainty—captures the hard part. The same technology that juices investment today could, as adoption matures, compress hiring needs in routine functions and reorder bargaining power. A burst of high-wage hiring in a few sectors can coexist with tepid aggregate job growth if incumbents elsewhere automate their way to flat headcounts.
The diffusion dilemma
The 6.5% versus 1.2% posting split is more than a statistic; it is a roadmap for dispersion risk. If AI remains bottled inside the four leading sectors, productivity gains may be spectacular yet narrow, with unequal wage effects and a fragile demand base anchored to corporate investment cycles. If adoption jumps the fence—into logistics, retail operations, public administration, and the gray middle of office work—the macro profile changes. Broader diffusion would spread both the gains and the adjustment costs, possibly lifting trend productivity while cooling job growth in mid-skill roles. That is a different inflation path, a different wage dynamic, and a different political economy.
What “outsized” really means for 2026
To say AI “played an outsized role” in last year’s growth is to admit that a relatively small slice of the economy is pulling more than its weight. That worked in 2025. The test in 2026 is whether those sectors can keep investing at that clip while the rest of the economy recalibrates. Watch for three signals. First, whether AI-related postings in the leading quartet keep rising or plateau as firms shift from experimentation to standardization. Second, whether late-adopting sectors close the posting gap, which would indicate diffusion and foreshadow broader productivity effects. Third, whether measured productivity accelerates without an inflation penalty—evidence that supply is catching up to demand.
The subtext of the Fed’s note is pragmatic: risk management. Policymakers do not need a full theory of AI to recognize that it is already shaping aggregate outcomes. For workers and firms, the guidance is just as clear. The jobs aren’t vanishing wholesale; they are migrating toward domains that can orchestrate software, data, and capital at scale. The winners are drafting policy now, not waiting for it. And the central bank, at last, is writing AI into the macro narrative where it belongs.
