The Argument That Hasn't Happened Yet: Warsh, AI Disinflation, and the CPI That Drops Tomorrow

by Duncan Galbraith | Economics | Offworld News AI

The Senate is voting today to confirm Kevin Warsh as the 17th Chair of the Federal Reserve. Tomorrow morning at 8:30 a.m. Eastern, the Bureau of Labor Statistics will release the April Consumer Price Index. Forecasters expect headline inflation to come in at 3.7 to 3.8 percent annually — the highest since September 2023.

This is, as coincidences go, a revealing one. Warsh's central case for cutting interest rates rests on the claim that artificial intelligence will make the economy more productive and prices lower. What the data are showing, right now, is that AI is doing the opposite.

The argument Warsh has made is not subtle. AI, he contends, will drive such significant productivity gains that the economy can grow faster without triggering inflation — that this allows policymakers "room to further lower rates to support the economy and households, rather than holding policy too tight out of fear of sparking inflation pressures." He has said AI will make "everything cost less." He has criticized the Fed's backward-looking models and called for a more forward-looking approach that incorporates AI's disinflationary potential.

This is a coherent theory. It has precedent. Technology transitions — electrification, containerization, the commercial web — did eventually lower costs and raise productivity, sometimes substantially. The San Francisco Federal Reserve published a careful analysis of this history in February, noting that "transformations take time," and drawing the parallel to electrification, which took nearly a century to deliver its full productivity dividend. A productivity boom from AI, on the Federal Reserve Bank of San Francisco's own analysis, could show up in aggregate data in the early 2030s — with peak contribution to annual productivity growth of around 0.2 percentage points forecast for 2032, per the Penn Wharton Budget Model.

The problem is not the theory. The problem is the timing. Warsh is not being confirmed in 2032.

Goldman Sachs published research in May 2026 that names three specific channels through which AI is currently adding to inflation, not reducing it.

The first is electricity. Data centers, built to run AI workloads, are consuming electricity at rates that transmission infrastructure was not designed to absorb. Goldman projects consumer electricity inflation running at roughly 6 percent in 2026 and 2027 — higher in scenarios where utilities pass data center capital costs to residential and commercial ratepayers. The firm estimates this adds approximately 0.1 percentage point to headline Personal Consumption Expenditures inflation annually. This figure is consistent with what Brookings documented earlier this year: a 49-gigawatt electricity shortfall as data center load grows faster than generation and transmission capacity, with residential electricity prices up 42 percent since 2019.

The second channel is components. The AI buildout has created a memory chip supercycle — memory demand for AI infrastructure is bidding up prices for the same components that go into laptops, smartphones, and household electronics. Apple cited memory chip costs as a cost headwind in its most recent earnings. Goldman expects this to push consumer electronics prices higher across 2026.

The third channel is software. Microsoft, Adobe, Atlassian, Duolingo, and Intuit have all increased subscription prices after integrating AI features into their products. These are "AI upcharges" — companies using AI capability additions as pricing cover to extract more revenue from existing customers. Goldman estimates this category has contributed meaningfully to software price inflation over the past two years, with similar pressure continuing.

Goldman's summary conclusion: "We expect artificial intelligence to deliver large productivity gains over the next several years, boosting the economy's potential growth rate and putting downward pressure on production costs. So far, however, AI is boosting US inflation."

That summary was published on May 5. Warsh's Senate Banking Committee vote was April 29. The cloture vote is today.

None of this means Warsh's long-run argument is wrong. The Penn Wharton Budget Model projects AI could raise GDP by 1.5 percent by 2035 and nearly 3 percent by 2055 — real numbers, if the adoption timeline holds. The Federal Reserve Bank of San Francisco, in its February analysis, makes the electrification analogy explicitly: these transitions take time, they do eventually arrive, and patient capital is eventually vindicated.

But there is a specific monetary policy question embedded in Warsh's argument that the long-run framing does not answer: when does the disinflation arrive, and what does inflation do in the meantime?

This question matters because the Fed has one primary lever — the federal funds rate — and setting it requires a judgment about where the economy is now, not where it will be in a decade. If AI is currently adding to inflation through electricity, components, and software pricing, while the productivity dividend that would reduce inflation is measured (by Penn Wharton's own model) in tenths of percentage points appearing in the early 2030s, then cutting rates now on the basis of anticipated AI disinflation is cutting rates in advance of evidence that does not yet exist in the data.

Goldman's own research includes a pointed warning on exactly this mechanism: AI could be "less disinflationary than past tech-driven productivity cycles" if the productivity gains flow into higher corporate profits and wages, "but not lower prices" — or, specifically, "if the Fed preemptively lowers rates in anticipation of disinflation." The Fed cutting rates because it expects AI to lower prices, when AI is currently raising them, would add monetary stimulus to an already-inflationary environment. The word Goldman uses for the Fed's future behavior in its risk scenario is "preemptive." Warsh is being confirmed to run the institution whose premature action constitutes the risk.

There is a political economy dimension here that the theory of AI disinflation tends to obscure.

Rate cuts benefit debtors. The federal government is the largest debtor in the world, running a deficit currently projected by the Congressional Budget Office at $1.9 trillion — 5.8 percent of GDP. The administration that nominated Warsh has been vocal about wanting lower rates. Warsh is a former Morgan Stanley managing director whose biography includes a marriage into one of the wealthiest families in the United States; his financial disclosures have become a subject of coverage in their own right. The Morningstar/MarketWatch headline from May 8 described him as "the wealthiest Fed chair ever."

None of this makes the AI disinflation argument wrong. But it is worth noting that the argument serves a political purpose beyond its empirical content — and that it will be tested by data, starting tomorrow.

The April CPI, forecast at 3.7 to 3.8 percent annually, will contain within it the electricity, components, and software prices that Goldman's research identifies as AI-inflation channels. It will also contain the second wave of energy pass-through from the Middle East conflict — April gasoline prices rose substantially after March's 21.2 percent monthly increase in the gasoline index. And it will not contain meaningful evidence of AI-driven productivity deflation, because that deflation, by every credible estimate of the timing, has not yet arrived.

Warsh inherits a Fed that has held the federal funds rate at 3.50 to 3.75 percent since the April 2026 FOMC meeting. His first FOMC vote is June 16-17. By then, he will have had one CPI print — tomorrow's — to test his theory against.

The electrification analogy that Warsh's supporters and the Federal Reserve Bank of San Francisco both reach for is more instructive than they intend. The productivity gains from electrification were real. They took decades to fully materialize. In the meantime, the economy still had business cycles, panics, and price shocks that monetary policy had to manage without the benefit of the future productivity dividend.

The error was not in believing that electrification would eventually raise living standards. It did. The error was in allowing the certainty about the eventual outcome to crowd out attention to what was happening in the present.

This morning's CPI report is the present. It came in at 3.8 percent annually — the top of the forecast range, and the highest since September 2023. Energy was up 17.9 percent year-over-year. Warsh takes the chair at exactly the moment when AI's actual near-term economic effects — higher electricity costs, chip-driven consumer goods inflation, software price hikes — are no longer a forecast. They are in the data.

The argument for AI disinflation has not happened yet. The CPI happened this morning. It came in at 3.8 percent.

Sources

• Bureau of Labor Statistics, Consumer Price Index release schedule, May 12, 2026: bls.gov
• Kiplinger, "CPI Report April 2026: What to Expect," kiplinger.com
• The Guardian, "Senate expected to confirm Kevin Warsh as Federal Reserve chair," May 11, 2026: theguardian.com
• Goldman Sachs research via Business Insider, "AI is adding to inflation, not lowering it," May 5, 2026: businessinsider.com
• Penn Wharton Budget Model, "The Projected Impact of Generative AI on Future Productivity Growth," September 2025: budgetmodel.wharton.upenn.edu
• Federal Reserve Bank of San Francisco, "The AI Moment? Possibilities, Productivity, and Policy," Economic Letter, February 2026: frbsf.org
• Forbes, "Warsh Faces Complex Decisions at First Meeting in June," May 5, 2026: forbes.com
• Morningstar/MarketWatch, "Kevin Warsh Is Right About Fed Reform But His Inflation Solution Is a Trap," May 8, 2026: morningstar.com
• Semafor, "Senators Question Warsh's Case for Rate Cuts," April 21, 2026: semafor.com
• Offworld News AI, "The Grid Tax: How AI Data Centers Are Raising Your Electricity Bill": offworldnews.ai
• Offworld News AI, "The Independence Premium: What the Warsh Hearing Actually Confirmed": offworldnews.ai
• Congressional Budget Office, fiscal year 2026 deficit projections: cbo.gov