Opinion | Kevin Warsh is caught between Trump and the ghosts of 1979

The article argues that expectations for AI-driven productivity gains could influence the Federal Reserve to cut interest rates faster, as lower costs would help keep inflation near the 2% target. It frames the policy debate around whether rate cuts should accelerate in response to a potentially faster disinflationary backdrop. The piece is analytical rather than event-driven, with no new economic data or policy decision announced.

Analysis

The market is at risk of mispricing AI as a near-term disinflation shock rather than a medium-term supply shock. In the next 6-12 months, AI mostly changes narrative and capex allocation; it does not instantly compress shelter, services, or wage inflation enough to justify aggressive easing. That means front-end yields can stay stubbornly elevated even if the growth mix improves, creating a regime where rate cuts are delayed but equity investors still bid up long-duration AI beneficiaries.

The second-order effect is that the biggest winners may be the picks-and-shovels names with pricing power and utility-like demand, not the application layer. Hyperscale buildouts, data-center interconnect, power equipment, and semis tied to training/inference should see persistent demand regardless of the macro path, while labor-intensive software and broad cyclical sectors face margin pressure if firms substitute AI for headcount before aggregate demand has time to respond. On the loser side, anything that trades on a simple “AI makes everything cheaper, therefore rates fall” story is vulnerable if the Fed stays data-dependent and real rates remain restrictive.

The contrarian read is that the consensus is assuming productivity gains will be quickly visible in the CPI, when the first observable impact may instead be higher investment intensity and stickier capital costs. If AI accelerates growth before it disinflates, the bond market can sell off on better nominal growth, not rally on lower inflation expectations. That creates a window where duration underperforms even as AI equities outperform, a combination the market is often slow to price.

Catalyst timing matters: over the next few months, any upside surprise in wages, services inflation, or capex commentary from large tech could invalidate the dovish-rate narrative; over 12-24 months, measurable labor substitution is the real disinflation channel. Until then, the risk is that policymakers delay cuts, long rates remain anchored above what growth bulls expect, and the market rotates toward companies monetizing AI spend rather than companies assuming cheaper money.

Key Decisions for Investors

• Stay long AI infrastructure beneficiaries and fund it with shorts in rate-sensitive, AI-narrative-exposed duration trades: e.g., long SMCI/AMAT/ANET basket vs short IWM or XLY for the next 3-6 months. Risk/reward favors the long leg if capex stays firm; the short leg hedges a higher-for-longer Fed.
• Add duration protection: short 2Y Treasury futures or buy payer swaptions into the next inflation print/CPI cycle. If services inflation remains sticky, front-end yields can reprice 25-50 bps higher even with continued AI optimism.
• Pair trade: long semis and data-center power names (NVDA/AVGO/ETN) vs short broad software (CRM/INTU basket) for 6-12 months. The setup is that capex monetization arrives before generalized software labor savings.
• Avoid extrapolating lower rates into cyclical beta too early; hold off on adding to homebuilders/REITs until real-rate relief is visible. If the Fed delays cuts by even 1-2 meetings, these names can underperform by 10-15% relative to the market.
• For higher-conviction expression, consider long-call structures on AI infrastructure leaders and put spreads on long-duration indices. This captures upside from AI spend while limiting damage if the Fed keeps policy restrictive longer than consensus expects.