Federal Reserve: Warsh Confirms Fed Removed Accommodation in September, Frames AI as a Demand Shock Before a Supply Windfall
Fed Chair Kevin Warsh in conversation with IMF Managing Director Kristalina Georgieva at the IMF/World Bank Annual Meetings in Bangkok, October 2026
A tightening acknowledged, with no guidance on what comes next
Fed Chair Kevin Warsh offered the clearest recent description of how he views the Fed's most recent policy move, though he gave no hint of what the October and December FOMC meetings will bring. He said the Fed "would deliver price stability, and we removed a dose of accommodation about a month ago." He then declined to prejudge the next two meetings. Georgieva noted that he had "disappointed half of the audience or more," and Warsh replied that "it's a new chapter at the Fed, and we're not in that business anymore."
The refusal to pre-signal is itself informative. Warsh described his approach as looking "out the window" rather than leaning on lagging data or fallible forecasts. He said the world "has broadly conformed to our expectations" since the September meeting, which suggests no urgency to change course. Investors looking for forward guidance will not find it from this chair. The reaction function appears to rest on realized output, employment and inflation readings at each meeting, not on a pre-announced path.
AI is a demand shock first and a supply shock later
The most substantive part of the discussion was Warsh's framing of AI for monetary policy. He called it "probably the most consequential change to the US and the global economy in my lifetime," but he approached it as a central banker. His argument is that the technology "is obviously hitting the demand side of the economy before it's gonna hit the supply side." That makes it an inflationary impulse today, even though he "fundamentally" believes it becomes "a positive supply shock in the medium term." He explicitly distinguished it from the war and energy shocks that Georgieva cited as sources of price pressure, saying this one carries "a huge benefit to the global economy" over the horizon.
The implication is that the Fed is willing to treat AI-related investment as a source of near-term demand strength without assuming it is disinflationary now. Warsh made no productivity-based argument for easing. He conceded that the labor market effect is unknown: "We don't know in the next 12 or 18 months its effect on employment."
He was candid that the transition will be uneven, and he raised a point that applies directly to equity investors. The path will be bumpy "because we really don't know what the market structure is going to be of these new technologies. Who's gonna get to make the money? What are the consequences for those that bet that they were gonna make money, but the market structure doesn't allow it?" He said the Fed does not have to pick winners, because "that's what financial markets are for doing." Its task is to ask what various scenarios would mean for its mandate. He added that he does not think central banks should try to suppress variance around a modal outcome. They should instead ask whether there are negative tail risks they can actually address, "and we have to be pretty humble about that question."
Investment surge is broadening beyond AI
On the real economy, Warsh was constructive on the US. He said America is in "a massive investment surge," which he described as the result of "good policy," both macroeconomic and microeconomic, and not an accident. He contrasted it with the secular stagnation debates of two decades ago. He added that the surge "isn't just AI" and "is broadening," which he said gives the economy "strength and durability." He acknowledged that this momentum "is certainly not found everywhere in the world."
The comments point to a capital spending cycle that the Fed sees as wider than the data center build-out. That reduces the risk that a reassessment of AI returns would take down the whole investment story, though Warsh did not quantify the breadth.
Global resilience and synchronized tightening
Warsh said he was surprised by how well G20 economies have held up. "If you had made me give some IMF forecast... I'm not sure I would have found that so many members of the G20 were hanging in there," he said, adding that he might have said so of the US, but "not so many" others. The common problem is inflation. He characterized recent months as "a synchronous tightening in monetary policy in most places around the world," which he said means "there's no free riders." He called that a positive development and said he heard "resolve" from foreign counterparts to deliver price stability. Georgieva endorsed the point, calling the common commitment to price stability a central message of the meetings.
Synchronized tightening against an energy shock and an investment boom is a more restrictive global backdrop than the headline resilience suggests. Warsh did not address how long it would last.
Task forces, better data and a de-emphasis of models
Warsh described a structural effort to bring in outside thinking. He has set up five external task forces, including ones on productivity and jobs and on data collection and inflation. The first field hearing of the productivity and jobs task force will be held November 16 in the Bay Area. He said he is "pretty sure I've got the right questions. Not sure I have the right answers," and that he wants the groups to "produce new intellectual capital" and present it to the FOMC. He called the work "global public goods" that other central banks might adopt.
He also said AI will change how central banks read the economy. Today, "we torture data with old fashioned models" to find underlying inflation trends. Soon, he said, central banks will have "access at the fingertips... to all the information in the world," with new analytical tools to distill it. That would give policymakers more confidence in separating real signals from lagged or mismeasured data. "I think it's gonna fundamentally change the conduct of monetary policy."
He closed with an unusually direct rebuke of model-driven policymaking: "I prefer frameworks to mean regressing models." He said the Fed should not hear "this is how we've always done it, or this is what the model says." Combined with the data and AI comments, this suggests a Fed that will lean more on judgment and real-time information than on traditional model outputs. That could make its reaction function harder for markets to forecast.
Leadership: credibility over the printing press
Warsh, now 148 days into the chairmanship, said his main shortfall has been giving too little attention to the 75% of Fed employees who do not work on monetary policy, including those in supervision and regulation, payment systems, and consumer affairs. He pledged to give those functions "equal focus" over the next 100 days. His rationale was institutional: "the power really comes from our institutional credibility," and a stumble in supervision or payments "reflects on our institutional credibility."
He framed his own tenure in terms of reform, not continuity. He said the Volcker, Greenspan and Bernanke eras each required leaders who "love the institution enough to be unafraid to challenge it, to change it." He also said "there's no status quo ante to return to" and that the world is "changing faster than at any point in my adult life." He said he is "not in the legacy business" and hopes the remaining "three years and eight months or so" of his term are benign. Investors should read this as a signal that institutional change at the Fed, not just rate decisions, is a live agenda item under his leadership.