Federal Reserve Chair Kevin Warsh champions parsimony in central bank communications and loathes so-called forward guidance. But he’s said plenty about the future of artificial intelligence: He’s an arch optimist on the technology who thinks it will help him vanquish inflation. In a November Wall Street Journal column published shortly before getting picked to lead the central bank, Warsh wrote that “AI will be a significant disinflationary force, increasing productivity and bolstering American competitiveness.”
So how will he react if the AI boom stalls, as more now predict as evidence emerges that the sector may be facing a budding funding crunch? It’s concerning to think that Warsh may have boxed himself into overreacting to such a wobble and abandoning the possibility of further interest-rate increases. Fed-watchers may have gotten a taste of his views Wednesday at what markets broadly perceived as a dovish press conference following the Federal Open Market Committee’s decision to keep rates steady at 3.75%-4%.
The Bloomberg Artificial Intelligence Index of 50 notable AI developers, facilitators and users dropped as much 10% this month, amid concern that revenue growth might slow from spectacular to simply great, and that data center companies might struggle to keep financing their expansions as borrowing costs rise. Debt yields surged for CoreWeave Inc., the AI cloud-computing provider, and data center peers with growth models that are dependent on borrowing at a reasonable price. As I’ve noted earlier, the recent turbulence calls for a cautious approach to monetary policy that guards against the possibility that this trickle of credit stress becomes a flood, and the Fed was right to take a break from increasing rates this month. But caution doesn’t mean capitulating to inflation that’s been above the Fed’s 2% target for almost six years. Warsh should still aim to shepherd policymakers toward a quarter-point hike in December.
In some ways, an AI wobble might have a silver lining for monetary policy. Warsh’s disinflationary thesis on AI is fundamentally a long-run bet even though AI is clearly adding to inflation at the moment. The high bandwidth memory that AI companies have gorged on soaks up a massive amount of the wafer space that the memory industry produces, much of which would typically go to conventional memory for personal devices. Personal computers and peripheral equipment prices are up 9.7% in the past year. To quote Federal Reserve Bank of Philadelphia President Anna Paulson: “While the impact of AI on productivity remains unclear, its effect on prices is not.”
If you get AI companies to pump the brakes in a meaningful way, wafer availability will improve and inflation could cool in related categories. And most Americans would be none the worse for it. A slowdown would hurt the famously boom-and-bust semiconductor industry, but new specialized factories, known as fabs, are already underway and poised to arrive in 2027 and 2028, including those from Micron Technology Inc. and SK hynix Inc.
Warsh seemed to at least entertain the possibility on Wednesday that the Fed’s quarter-point rate increase in September is already bringing supply and demand into better balance. Asked at his press conference whether the Fed might already be at the “right” policy setting, bond markets read his hemming and hawing as a maybe, with the yield on the two-year Treasury note dropping as much as 11 basis points in intraday trading. “Well, we’ve got to determine the appropriate protocol,” Warsh said. “If you were listening next to me in that room the last couple of days, the overwhelming consensus, the overwhelming judgment, is that what we’ve done in this monetary medicine — it’s starting to work some, but there’s no firm conclusions.”
Later, Warsh said that the September boost to rates already had an effect on the tech sector. “You see market participants that are discriminating between the kinds investments they want to make,” he said. Swaps now imply roughly 83% odds of a rate hike in December, down from around 95% prior to the Fed decision and Warsh’s remarks.
No one’s rooting for the Fed to keep pushing rates until credit markets buckle, and that’s why moving slowly makes sense. But the risk of a crunch is still extremely modest relative to the threat of entrenched inflation. A government report Thursday may show that gross domestic product expanded at a strong 3.2% annualized pace in the third quarter, according to a Bloomberg survey of economists, and the S&P 500 Index is less than 1% from its all-time high. If Fed hikes managed to, say, take 10% to 20% off the stock market, the reverse wealth effect would primarily hit the wealthiest 10% of American households and have little direct effect on the bottom 50%, which suffers the most from inflation.
All of this supports the Fed’s case for pausing now, but a quarter-point rate increase in December is prudent at this point — a hike that’s already mostly priced into markets and would demonstrate Warsh’s commitment to restoring price stability. Credit spreads would have to widen to the point that companies start to meaningfully revise capital expenditures expectations downward before markets think about pricing out a rate boost at year-end. It’s not clear where Warsh stands on that matter, due to his general aversion to forward guidance. But if you believe the professional speculators in bond and derivatives markets, the subtle shakeout in the AI trade might already have caused Warsh to blink.
