Cutting rates based on a hypothesis is a weak position
As Kevin Warsh assumes leadership of the Federal Reserve, the very force he believed would tame inflation — the artificial intelligence investment boom — is instead feeding it, flooding capital markets with demand, driving chip prices skyward, and pushing long-term interest rates toward levels unseen in a generation. The tension between a theory of future productivity and the present reality of accelerating prices places Warsh at a crossroads that echoes one of history's most instructive monetary episodes: Greenspan's 1990s, when technological optimism did not spare the economy from eventual tightening. What unfolds now is less a question of who is right about AI's ultimate promise, and more a question of whether a central banker can govern by a thesis the market has not yet confirmed.
- Warsh arrives at the Fed with a deflationary AI thesis already under siege — inflation jumped 3.8% in April, the sharpest annual rise since 2023, while bond markets price in a possible rate hike by December.
- The $700 billion annual wave of tech infrastructure spending is doing the opposite of what Warsh predicted: it is crowding capital markets, spiking borrowing costs, and generating chip inflation so severe that DRAM memory prices have multiplied seventeen times in a single year.
- Fed colleagues Jefferson and Barr are openly warning that massive AI capital expenditure raises the neutral rate rather than lowering it, fracturing internal consensus before Warsh has even taken his seat.
- The 1990s Greenspan precedent looms large — productivity booms historically push real interest rates higher, not lower, and the Fed eventually tightened even as technology transformed the economy.
- Warsh now faces a defining gamble: resist rate hikes to protect AI investment and risk entrenching inflation, or tighten policy and risk being blamed for strangling the very revolution he championed.
Kevin Warsh takes the helm of the Federal Reserve with a theory the market is already stress-testing. For months, the incoming chair argued that AI would supercharge productivity and create a deflationary tailwind powerful enough to justify lower interest rates. The bond market disagrees. Long-term Treasury yields have climbed to near two-decade highs, and measures of the neutral rate suggest monetary policy is still accommodative even as inflation runs hotter than expected.
The difficulty is not that AI investment is illusory — it is enormous. Four major technology companies alone plan to spend over $700 billion this year on data centers, chips, and electrical infrastructure. But that spending is flooding capital markets with demand right now, driving up borrowing costs and igniting what analysts are calling chip inflation. Memory chip prices have multiplied seventeen times over the past year. Software and computing equipment costs in the U.S. jumped 14 percent in April year-over-year. Tech firms have already issued more than $300 billion in bonds to finance AI buildouts — a volume the Dallas Fed estimates has effectively expanded long-term Treasury supply by over 10 percent.
This places Warsh in a bind his predecessor did not face so acutely. Appointed by a president who accused Jerome Powell of strangling growth, Warsh had himself criticized the Fed for overestimating inflation's persistence. Yet he arrives with the consumer price index at 3.8 percent annually — its steepest climb since 2023 — and futures traders speculating about rate hikes in December, not cuts.
Fed colleagues are already pushing back. Vice Chair Jefferson and Governor Barr have both cautioned that the capital demand unleashed by AI deployment could itself raise the neutral rate. The 1990s offer a sobering parallel: Greenspan was right that productivity gains would help contain inflation, but the Fed still held rates steady and ultimately raised them as investment accelerated. Productivity booms, history suggests, can lift real interest rates rather than suppress them.
Some analysts believe Warsh may eventually be vindicated — Vanguard, for one, acknowledges AI could meaningfully boost productivity over time. But for now, supply shocks and investment-driven demand are winning. Warsh's choice is stark: resist tightening to protect the AI investment he believes will deliver deflation tomorrow, or accept that governing by an unconfirmed hypothesis is a fragile position. Until productivity gains actually appear in the broader data, the technology Warsh counted on to solve inflation is, at least for now, making it worse.
Kevin Warsh takes the helm of the Federal Reserve on Friday with a theory about artificial intelligence that the market is already testing—and failing. For months, the incoming Fed chair has argued that the central bank underestimated how AI would turbocharge productivity, creating a powerful deflationary force that would allow interest rates to fall. But the bond market is telling a different story. Long-term Treasury yields have climbed to near two-decade highs, and a closely watched measure of what economists call the neutral rate—the interest rate that neither stimulates nor restrains economic growth—suggests monetary policy remains accommodative even as inflation runs hotter than expected.
The problem is not that AI isn't real. It is. Four major technology companies alone plan to spend more than $700 billion this year on data centers, computing equipment, and electrical infrastructure to power the AI revolution. The problem is what that spending is doing right now. It is flooding capital markets with demand, pushing up borrowing costs. It is creating what analysts call chip inflation, as the global scramble for semiconductors needed to run AI systems drives prices skyward. Memory chips used in data storage have multiplied seventeen times in value over the past year. Software and computer equipment prices in the United States jumped 14 percent in April compared to a year earlier. Microsoft and Meta have begun raising prices on some of their products. Tech companies have already sold more than $300 billion in bonds to American investors to finance AI-related investments, a volume so large that the Federal Reserve Bank of Dallas estimates it has the same effect on bond markets as if the long-term Treasury supply had increased by more than 10 percent.
This creates a bind for Warsh that his predecessor Jerome Powell did not face in quite the same way. President Trump appointed Warsh after repeatedly saying Powell was strangling the economy by not cutting rates fast enough. In a Wall Street Journal editorial last November, Warsh himself criticized Fed officials for predicting that inflation would remain elevated, arguing they had missed the productivity gains that AI would deliver. Yet here he is, about to take office, with inflation accelerating—the consumer price index jumped 3.8 percent in April, the largest annual increase since 2023, partly because oil prices have surged amid the U.S. conflict with Iran. Futures traders are now speculating that the Fed might be forced to raise rates in December, not cut them.
Warsh's colleagues at the Fed are already pushing back on his theory. Vice Chair Philip Jefferson and Governor Michael Barr have both warned that the massive business investment required to deploy AI could itself raise the neutral rate by increasing demand for capital. The parallel to the 1990s is instructive. Alan Greenspan, then Fed chair, correctly predicted that rapid productivity gains would help contain inflation. But the Fed did not slash rates. It held them steady at first, then raised them sharply as investment accelerated and the economy picked up speed. The lesson: productivity booms can actually push real interest rates higher, not lower.
Some investors believe Warsh may ultimately be vindicated. Analysts at Vanguard, one of the world's largest asset managers, acknowledge that AI spending could eventually make the economy more productive and help ease inflation. But they are watching carefully for signs that productivity gains are actually showing up in the broader economy. For now, supply shocks and investment-driven demand are pushing prices up. Christoph Rieger, head of rates and credit research at Commerzbank, put it plainly: the base case is that AI will contribute to higher inflation in the years ahead.
Warsh faces a choice that will define his tenure. He could argue, as some analysts expect, that raising rates would be foolish because it might dampen the very AI investment that could deliver deflation tomorrow. But the market is not buying that argument. Blake Gwinn, head of U.S. rates strategy at RBC Capital Markets, said it plainly: cutting rates based on a hypothesis is a weak position. The markets are not convinced. What happens next depends on whether the productivity gains Warsh predicted actually materialize—and how quickly. Until they do, the Fed's new leader will be managing an economy where the technology he thought would solve inflation is, for now, making it worse.
Citações Notáveis
Our base case hypothesis is that AI will contribute to higher inflation in the coming years— Christoph Rieger, head of rates and credit research, Commerzbank
Cutting rates based on a hypothesis is a weak position. Markets are not buying that idea— Blake Gwinn, head of U.S. rates strategy, RBC Capital Markets
A Conversa do Hearth Outra perspectiva sobre a história
So Warsh argued that AI would be deflationary, but the market is seeing inflation instead. What changed?
Nothing changed about AI itself. The technology is real and powerful. What changed is the timing. Right now, companies are spending hundreds of billions on infrastructure—data centers, chips, power systems. That spending creates immediate demand for capital and materials, which pushes prices up. The deflation Warsh expects might come later, when all that infrastructure is built and productivity actually increases. But "later" is not now.
The Fed's neutral rate is rising. Why does that matter so much?
Because it's the speed limit for monetary policy. If the neutral rate is rising, the Fed has less room to cut rates without overstimulating the economy. Warsh came in saying the Fed should cut rates because AI would make the economy more productive. But if the neutral rate is rising because of AI investment, then cutting rates might actually be dangerous. It's a trap.
His colleagues at the Fed seem skeptical of his theory.
They are. Jefferson and Barr are saying the same thing: heavy business investment in AI raises the neutral rate by itself, just by demanding more capital. They're pointing to the 1990s, when Greenspan was right about productivity but the Fed still had to raise rates eventually because the economy was overheating. History suggests that productivity booms don't automatically mean lower rates.
Could Warsh still be right in the long run?
Possibly. Vanguard thinks so. If AI actually delivers the productivity gains Warsh expects, inflation could come down and rates could fall. But that's a bet on the future. Right now, the market is pricing in the present: higher inflation, higher rates, higher borrowing costs. Warsh has to manage the present while hoping the future proves him right.
What's the real risk here?
That Warsh becomes so committed to protecting AI investment that he keeps rates too low for too long, and inflation gets away from him. Or that he's simply wrong about how quickly productivity will show up in the data. Either way, he's betting his credibility on a theory that the market is actively testing—and failing.