The Fed Chair Says AI Is Moving Faster Than Even Its Believers Predicted

Federal Reserve Chair Kevin Warsh said at Jackson Hole on August 28, 2026, that artificial intelligence may become a new factor of production, alongside labor, capital, and land, with annualized token sales for the two leading AI labs exceeding $100 billion, a 500% increase from 12 months earlier. Warsh noted the Fed has established an internal task force on productivity and jobs to study AI's implications, though current policy remains unchanged with the fed funds target upper bound at 3.75%.

The Fed Chair Says AI Is Moving Faster Than Even Its Believers Predicted Federal Reserve Chair Kevin Warsh just rewrote the economic textbook at Jackson Hole, and the implications for every interest rate decision going forward are ones most investors have not yet priced in. Federal Reserve Chair Kevin Warsh used his Jackson Hole platform on August 28, 2026, to say something the central bank has resisted saying out loud. Artificial intelligence may belong alongside labor, capital, and land in the models economists use to think about growth. That reframing matters because the Fed sets interest rates based on assumptions about how much the economy can produce before wages and prices overheat. If those assumptions are wrong, current policy could be too tight or too loose in ways officials cannot yet see. A Fourth Factor of Production Changes the Math Warsh described the moment plainly. “Times sure have changed. We’ve come to a hinge point in history,” he said. The claim underneath that framing is the one worth marking. He said the Fed now recognizes “AI is a new variable, potentially a new factor of production that will have consequences both for the economy and for the conduct of monetary policy.” Factors of production is a phrase from introductory economics, where labor, capital, and land are the classical three. Adding a fourth item is more than rhetorical, because Fed forecasts of potential output rest on how those inputs combine. If AI genuinely joins that list, the level of activity the economy can sustain without generating inflation rises with it. A stance that looks restrictive today would look neutral tomorrow, and the framing of every future rate decision would shift. Warsh said these considerations will not immediately shape current policy decisions. The fed funds target upper bound https://fred.stlouisfed.org/series/DFEDTARU has held at 3.75% since the easing cycle that ran through late 2025. What Token Sales Actually Measure Warsh anchored his case in a specific figure. “Reports put annualised token sales for the two leading labs alone at more than $100 billion, an increase of 500% from just 12 months ago,” he said. A token is the unit AI companies bill by, roughly a fragment of a word, counted whenever a model reads a prompt or writes a response. Token revenue is therefore a rough proxy for actual usage of AI products, different from data-center capital expenditure, which tracks capacity built rather than capacity consumed by paying customers. Warsh did not name the labs, and neither will this article, because the two firms most readers would guess are private companies without listed shares. A revenue jump of that magnitude in a single product line, if the reports Warsh cited are accurate, is the kind of adoption curve that would eventually show up in productivity statistics. It has not yet shown up in official Fed data, and that gap is part of his point. Why Warsh’s Framing Breaks With the Post-2008 Consensus For most of the past 15 years, macro debate centered on secular stagnation. The worry was that the economy had permanently run out of profitable investments, leaving rates and growth stuck at low levels. Warsh contrasted the current moment with that older concern. Real GDP growth came in at 1.5% for the quarter ending April 1, 2026, below the 2 to 3% range the Fed’s own guide identifies as healthy. Yet Core PCE, the Fed’s preferred inflation gauge, stood at 130.658 in July 2026, the 91.7th percentile of its trailing year. Inflation has not fallen the way stagnation theorists predicted. The labor market shows the same in-between quality. Total nonfarm payrolls stood at 158,858 in July 2026, and average hourly earnings at $37.62, both drifting up rather than accelerating. The Fed has established an internal task force on productivity and jobs to study the implications of AI. That is a meaningful institutional signal that the framing is changing inside the building, not just at the podium. Which Question Long-Term Investors Should Track Warsh posed the central question directly. “Will the application of AI cause a significant, sustained rise in productivity across the economy? If so, when will token usage be complementary or competitive to labor?” A sustained rise in productivity would allow the economy to grow faster without forcing the Fed to raise rates, a combination that has historically supported equity valuations broadly. The complement-or-compete question is harder because both can be true in different jobs at once. Real average hourly earnings in July 2026 came in at $11.30, close to the $11.32 recorded a year earlier, which argues AI has not yet moved the aggregate wage picture. Bond markets are pricing something between hope and doubt. The 10-year Treasury yield sat at 4.67% on August 27, 2026, near the top of its trailing-year range. The useful takeaway from Warsh’s remarks is that the Fed has begun treating AI as a monetary policy problem to solve rather than a technology story to observe. Long-term investors in duration-heavy assets, whether growth equities or long bonds, will find the answer in every discount rate they use, and the buildout enabling all that token growth power, cooling, and networking suppliers we profiled in a free report on AI infrastructure https://247wallst.com/pages/ai-power-seven-offer-d905ec99.html is where spending shows up first. Contact email protected for any questions or corrections.