As AI reshapes the productivity outlook, the Federal Reserve faces a critical question: can technology deliver the disinflation that monetary policy has struggled to achieve?

Key Highlights

  • AI-driven productivity gains are increasingly framed as a structural disinflationary force by institutional asset managers.
  • The Federal Reserve remains divided on whether AI justifies easing monetary policy in the near term.
  • Massive capital investment in AI infrastructure is generating immediate inflationary pressure, offsetting anticipated supply-side gains.
  • Fed Chair nominee Kevin Warsh has drawn parallels to the 1990s productivity cycle to argue for rate reductions.
  • Policymakers face a timing problem: the productivity payoff from AI may lag significantly behind its demand-side costs.

The Productivity Argument Gains Institutional Weight

A quiet but consequential debate is reshaping how institutional investors and policymakers think about the inflation outlook. The proposition is straightforward: if artificial intelligence delivers even a fraction of the efficiency gains now being reported by major corporations, the macroeconomic consequences could rival the most significant positive supply shocks in modern economic history.

Asset managers overseeing trillions in capital are beginning to treat this not as speculative optimism but as a legitimate variable in monetary policy analysis. The logic is grounded in basic economics. When technology raises output per worker at scale, it expands productive capacity without proportionally increasing wage costs or input demand. The result, in theory, is downward pressure on prices across goods and services, without the recessionary pain that typically accompanies disinflation.

Fed Chair nominee Kevin Warsh has characterised the AI boom as the most productivity-enhancing wave of our lifetimes, drawing comparisons to the internet's disinflationary impact when it first began reshaping the economy roughly three decades ago. His argument is that technology-led productivity gains create room for the Federal Reserve to reduce borrowing costs without reigniting inflation, much as Alan Greenspan navigated the late 1990s expansion.

The Fed's Internal Divide

Not all policymakers share that confidence. The Federal Reserve's institutional response to the AI productivity thesis has been cautious, and for reasons that are analytically sound.

Fed Chair Jerome Powell acknowledged that the Fed's long-run growth estimates have been revised upward, perhaps reflecting growing optimism around AI, but characterised near-term productivity gains as theoretical rather than empirically confirmed. In his assessment, AI-driven improvements do not yet constitute a basis for lower interest rates.

The structural concern is one of timing asymmetry. AI technology requires data centres that compete with other production processes for land, energy, and broader inputs, placing upward pressure on specific price categories even as the promise of efficiency gains remains on the horizon. Fed Vice Chair Philip Jefferson has cautioned that AI's effects on inflation are not solely disinflationary and that monetary policy must be calibrated to the broad economy, not a single sector.

This creates a genuine policy dilemma. Capital expenditure on AI infrastructure is a present-tense demand stimulus. The productivity dividend, if and when it materialises, belongs to a future that central banks cannot yet price into their models with confidence.

Adam Posen of the Peterson Institute for International Economics has argued that the current AI cycle is generating positive real income effects and rising asset prices, but very little actual disinflation, and that those framing AI as a near-term disinflationary force may have the causality inverted.

A Supply Shock Without a Timeline

The historical precedent most frequently cited, the 1990s technology boom, is instructive but imperfect. Productivity growth in that era did allow the economy to expand at an above-trend pace without triggering inflation. However, that outcome unfolded over a decade, not a quarter. The Federal Reserve under Greenspan benefited from favourable geopolitical conditions, a structurally declining oil price environment, and the gradual rather than disruptive diffusion of technology into the labour market.

Today's environment is materially different. AI-driven displacement is already visible in the labour market, with firms announcing significant workforce reductions tied explicitly to changes in how they deploy human capital. Rising unemployment would conventionally signal the need for looser monetary policy. Yet officials have acknowledged that AI-related job displacement may sit outside the traditional cyclical framework, complicating their dual mandate calculus.

Federal Reserve Bank of Boston President Susan Collins has noted that the US economy remains in relatively early days of AI adoption, and that a well-informed understanding of the technology's labour market and productivity implications is essential before the Fed can responsibly adjust its policy stance.

What Markets Are Pricing In

Equity markets have already begun discounting a version of the AI productivity thesis, with capital flows concentrating heavily in technology infrastructure, semiconductor supply chains, and enterprise software. Valuation multiples in these segments reflect expectations of durable revenue growth tied to AI adoption. Whether those expectations are anchored in realistic productivity timelines or in the early-stage enthusiasm that tends to overprice transformational technologies remains an open question for institutional investors to assess.

The risk embedded in current valuations is that the disinflationary payoff is real but delayed, while the demand-side costs of building the infrastructure are immediate. That gap between cause and consequence is precisely where monetary policy errors tend to occur.

The Structural Outlook

The macroeconomic case for AI as a disinflationary force is analytically coherent. History suggests that sufficiently broad technology adoption does raise productive capacity, moderate input costs over time, and expand the economy's non-inflationary growth ceiling. The debate is not about whether this outcome is possible, but whether it is proximate enough to influence monetary policy today.

The San Francisco Fed has noted that AI adoption continues to evolve rapidly, and that what is currently understood about its impact on productivity and the broader economy remains materially uncertain. Early indicators in business data and macroeconomic statistics will be essential signals before any policy recalibration can be justified.

For now, the Federal Reserve appears content to observe. That posture is defensible. A supply shock that has not yet shown up consistently in productivity data is a hypothesis, not a policy foundation. Central banks are not in the business of pricing in potential; they respond to conditions as they are, while attempting to anticipate what they are becoming.

The AI disinflation thesis may ultimately prove correct. If it does, it will represent one of the more significant structural shifts in the post-pandemic macroeconomic landscape. But the timeline, and the policy implications that follow, remain genuinely uncertain.