Federal Reserve officials expected further interest rate increases would likely be appropriate under their baseline economic outlook, even as they voted to hold rates steady at their October meeting. Minutes released November 18 show that the discussion included a growing source of uncertainty: the effects of artificial intelligence investment on prices, productivity and financial stability. The rate decision left the Fed’s benchmark range at 3.75% to 4% following a 10-2 vote.
Almost all participants saw further increases as likely under their baseline inflation and employment projections, but that was not a commitment to a particular move. Officials said decisions would depend on incoming data, the outlook and the balance of risks. Lisa Cook and Beth Hammack dissented from the October decision, favoring an immediate quarter-point increase. The next meeting was scheduled for December 8-9.
A pause amid inflation uncertainty
The minutes showed differing views on the urgency and potential size of another increase. Some participants wanted a quarter-point rise in October, arguing that policy was not restrictive enough and that waiting could make inflation harder to contain. A couple said an increase larger than 25 basis points could warrant consideration at the next meeting if inflation data were unfavorable. Those views sat alongside the majority’s preference to gather more information before changing rates.
For those who backed holding steady, the unknowns reached beyond the latest inflation reading. Participants cited uncertainty about how durable demand for AI investment would prove, as well as financing conditions and geopolitical developments. They also wanted more information about how the Fed’s September rate increase was working through the economy. The decision therefore left officials with time to assess both a previous policy move and the forces shaping demand.
At the meeting, Fed staff estimated that headline personal consumption expenditures inflation had risen to 3.6% over the 12 months through September, while core inflation had edged up to 3.1%. Staff linked the year-over-year increases largely to earlier tariffs, energy prices and AI-related pressure on consumer goods prices. Most participants nevertheless expected inflation to move down toward the Fed’s 2% goal over time, according to the minutes.
AI’s competing effects on the economy
Many participants said demand tied to the AI buildout was straining supplies and pushing up prices for technology-related consumer goods. Several judged that those pressures were spreading beyond industries directly affected by data-center construction. The labor picture was more localized: many reported substantial wage premiums for skilled data-center construction workers, but officials said those pressures did not appear to be materially affecting aggregate wage measures. They reached a similar assessment of reduced labor supply in some immigrant-dependent industries.
Officials also considered potential benefits from AI investment. Some judged that stronger productivity could allow output to grow without a corresponding tightening of the labor market. Others emphasized uncertainty over when productivity gains would arrive, how large they would be and who would receive them. The discussion placed potential future gains alongside price and labor pressures that participants were already observing in particular parts of the economy.
The broader economy presented a mixed backdrop for that debate. Officials described strong business investment and household spending alongside weakness in housing and other interest-sensitive sectors. September unemployment had edged up to 4.2%, while payroll growth averaged about 50,000 a month over the preceding three months. Those conditions formed part of the picture available to officials in October; the minutes describe the meeting’s information and views, not necessarily conditions on the November release date.
AI financing was another focus. Fed staff warned that rapidly growing AI-related debt, off-balance-sheet leases and circular financing among AI firms could magnify losses if investment returns disappointed. Almost all participants discussed the risks of disappointing returns, and most noted the growth of complex, opaque financing structures. The concern was distinct from the discussion of AI-related consumer prices: it centered on how investments were funded and what could happen if their returns fell short.
The minutes also recorded market developments before the October meeting. The Fed’s market manager said sentiment toward the AI sector weakened after two leading firms reported revenue growth below expectations and a major cloud provider disclosed unexpectedly demanding capital-financing plans. AI-infrastructure shares fell particularly sharply, while spreads widened notably on high-yield bonds financing data centers. Staff nevertheless judged bank and household balance sheets strong enough to buffer shocks. Several participants stressed that monetary policy was not the appropriate tool for directly addressing financial-stability risks.
