NEW YORK — Kevin Warsh's proposal to factor AI's disinflationary potential into Federal Reserve forecasts faces delays as policymakers grapple with immediate cost pressures from AI adoption.
Computer software and accessory prices have risen since the start of 2026, marking the first sustained increase in nearly three decades and breaking a deflationary trend dating to 1998. The shift has forced the Fed to prioritize short-term inflation risks over Warsh's longer-term productivity argument.
Warsh advocated for projecting AI's potential to drive down costs and efficiency gains into the Fed's long-term economic outlook. But current Fed assessments emphasize the near-term inflationary drag: specialized hardware purchases, software licenses and infrastructure upgrades.
Fed Governor Lisa Cook previously said long-term productivity gains could generate disinflation within years. Yet for now, AI appears a net inflationary force—one that could pressure bond yields if markets price in sustained price acceleration.
If inflation expectations broaden beyond software, the yield curve could steepen. Longer-dated Treasuries face increased duration risk as fixed-income investors demand higher compensation for holding bonds in a higher-for-longer rate environment.
The Fed's current focus on short-term AI inflation reduces near-term rate-cut probability, keeping policy restrictive against price stability threats. The debate over AI's dual inflation-deflation arc will shape the next Summary of Economic Projections and future policy moves.
