Everything About AI
NextFin News - Michelle W. Bowman, the Federal Reserve's vice chair for supervision, delivered one of the most candid inflation assessments of the 2026 policy year on Monday, telling the Council on Foreign Relations that "our progress on lowering inflation has disappointed" even as the labor market "improved more than I expected this year." The divergence is the story: with core PCE inflation likely at 3.3 percent in October and the federal funds rate held at 3-3/4 to 4 percent, Bowman framed the path ahead as data-dependent rather than preset, while identifying the artificial-intelligence investment boom and Middle East energy shocks as the two forces most likely to decide whether inflation returns to 2 percent next year. The Fed is not done with inflation — but it is not about to break the labor market to finish a fight that supply shocks started.
The Setup: A Split Mandate
The core tension in Bowman's remarks is that the two halves of the Fed's dual mandate are moving in opposite directions, and the side that is supposed to justify patience — the labor market — has healed faster than the side that demands action. Growth has stepped up this year. Real GDP likely expanded at more than 3 percent in the third quarter, well above last year's 2 percent pace, after rising a solid 2.4 percent during the first half. She attached one important caveat: currently reported first-half growth may be overstated by about 0.3 percentage point, because downward revisions in the retail-sales annual update were not folded into the Bureau of Economic Analysis' 2026 revision of the national accounts. Even net of that adjustment, the economy is running well above its 2025 speed.
The labor market has stabilized on similar terms. The unemployment rate edged down to 4.1 percent in October, close to Bowman's estimate of its natural rate. Job gains have averaged about 80,000 per month through October — above last year's "anemic" pace — and have broadened across industries beyond health care and social assistance. Consumer spending has strengthened despite softness in real incomes and declining sentiment, both partly driven by higher energy prices. Residential investment has declined, and the recent rise in mortgage rates is dragging home construction and sales down further, but the aggregate picture is one of resilience.
Inflation, by contrast, has not cooperated. Progress on lowering it "has stalled over the past two years or so," and the latest consumer price data suggest 12-month core PCE inflation "likely moved up further to 3.3 percent in October," as upward pressures from higher energy prices and AI-related demand continue to flow through to core goods and services prices. Bowman's bottom line was unambiguous: "Therefore, bringing inflation down remains a priority." That sentence, more than any other in the speech, defines the December meeting. With inflation moving the wrong way and the labor market no longer weak, the burden of proof has shifted decisively toward hawks.
"While the labor market has improved more than I expected this year, our progress on lowering inflation has disappointed. Therefore, bringing inflation down remains a priority."
The Mechanism: Three Shocks, Not a Labor-Market Story
The most analytically important claim in the speech is Bowman's decomposition of above-target inflation. Tariffs, energy, and AI-related price pressures each accounted for roughly one-fourth of above-target core PCE inflation over the 12 months ending in September. The remaining one-fourth reflects the persistent echo of the inflationary shocks of recent years. Put differently, about three-quarters of the inflation problem sits outside the labor market — which is precisely why the wage data look so benign, and why this cycle does not look like 1970s cost-push inflation rolling into a wage-price spiral.
The wages and salaries component of the employment cost index rose 3.1 percent over 12 months in September, the same pace as the change in average hourly earnings for October. Both measures have slowed noticeably from a year ago. Given strong productivity gains in recent years, Bowman noted, wage growth at this pace is "in line with or even below the pace we would expect if inflation was running at 2 percent." That is the crux of the policy dilemma. Monetary policy cannot offset the direct effects of tariffs and energy costs on the price level — no interest rate can make oil cheaper or a tariff disappear. What policy can do is contain the second-round effects: stop a relative-price shock from becoming embedded in consumer and business expectations. Bowman's forecast is that as these shocks wane or unwind, core PCE inflation will move closer to 2 percent next year. Tariff effects should further fade. Energy-related pressures should unwind when the Middle East conflict resolves. But inflation pressures related to the AI buildout "seem to persist despite some early signs of slowing," because the boom has shifted investment demand toward specific types of high-tech capital with relatively inelastic supply in the near term.
This decomposition is also a defense. By attributing three-quarters of above-target inflation to non-labor factors, Bowman is pre-emptively answering the question of why the Fed has not tightened more aggressively into a strong labor market. If the labor market is not generating inflation — and the ECI and AHE data say it is not — then hiking harder risks breaking the wrong side of the mandate without fixing the price level. The transmission mechanism she describes runs through expectations and second-round pass-through, not through excess demand for labor. That is a materially different model of inflation than the one that governed the 2022-2024 tightening cycle, and it has direct implications for how much restraint is actually needed.
The AI Buildout: Growth Engine and Financial-Stability Risk
Bowman's treatment of artificial intelligence is the most distinctive part of the speech, and it cuts both ways. On the growth side, she credits "increased investment in and use of artificial intelligence (AI), or super intelligence" with boosting economic growth and productivity. AI-related capital expenditures have been robust enough to help lift third-quarter GDP above 3 percent, and over time she expects increased AI adoption to raise productivity and put downward pressure on inflation as lower production costs pass through to consumer prices. This is the optimistic structural case: a general-purpose technology expanding the economy's productive capacity, the same way electrification and the internet did.
But the financial-stability warning that follows is detailed enough to read like a pre-emptive stress test. AI-related debt is rising rapidly across corporate bonds, private credit, and project finance, and is on track to exceed $800 billion this year. Other obligations — long-term lease and contract commitments and residual value guarantees — have surged to more than three times the amount of debt. Hyperscalers and large tech firms have assumed off-balance-sheet liabilities to allow smaller and riskier companies to use them as collateral in asset-based financing for data-center construction and the purchase of accelerated computing equipment. These connections among AI firms, Bowman said, create "the appearance of circular financing."
"AI-related debt has been rising rapidly, with notable gains in corporate bonds, private credit, and project finance, and is on track to exceed $800 billion this year. Other obligations, in the form of long-term lease and contract commitments and residual value guarantees, have also surged and their potential payments are estimated at more than three times the amount of debt."
The price signals are already flashing. Prices of memory chips, semiconductors, computers, and other electronic products have increased sharply this year — in a few categories by double digits — as demand from data centers has outpaced supply. Bowman expects production capacity to catch up gradually, but the near-term effect is inflationary. And financing is getting more expensive: as investors reassess expected returns on AI projects, equity valuations in the tech sector have dropped, corporate bond and mortgage spreads have widened, and credit default swaps have risen for some companies involved in the buildout. Financial conditions have tightened a bit recently because of lower equity prices, higher private interest rates, and a stronger dollar.
The risk scenario is explicit and internally coherent. Risks are concentrated in a few large firms that provide AI services, including AI labs and hyperscalers. If AI-related revenues fall short of sharply rising costs — especially as public sentiment and local opposition could lead to increased regulation and delays — the result could be a sharp correction in equity prices for AI firms, higher financing costs, and reduced access to credit. That shock would slow the AI buildout and soften both economic activity and inflation. In Bowman's framing, the same boom that is lifting GDP today is the single biggest source of downside risk to the outlook. She may have already seen the first act: equity valuations dropped, spreads widened, and CDS widened "in recent weeks" as investors reassessed expected returns.
Energy: A Shock the Fed Tries to Look Through
The second force Bowman flagged is energy. The Middle East conflict has disrupted energy markets, impairing the flow of oil and other commodities through the Strait of Hormuz, which accounted for about 20 percent of global oil supply before the conflict — roughly equivalent to total U.S. oil consumption. Brent crude moved to and above $110 per barrel in the second half of October before dropping sharply last week on optimism that an agreement between the U.S. and Iran could be reached. As of Tuesday, Brent had fallen below $90 a barrel for the first time since August, with West Texas Intermediate near the mid-$80s, as crude flows in the Persian Gulf recovered to near pre-conflict levels.
Bowman's stated approach is to look through temporary energy-driven inflation when possible, while remaining alert to signs that it is spreading into broader prices or expectations. That is textbook optimal policy for a temporary supply shock: reacting aggressively would add unwarranted restraint and weigh unnecessarily on employment. But she drew a clear line. If geopolitical tensions persist, energy prices may stay elevated in the near term, increasing the risk that core inflation stays elevated through pass-through to production and transportation costs. Freight rates and insurance premiums have already surged because of more complex logistics and ongoing geopolitical risk, raising the cost of delivered oil even where crude itself has retraced. And even with a quick resolution to the conflict, she noted, futures markets expect oil prices to decline only gradually, as global oil consumption is projected to exceed supply.
The asymmetry matters for policy. An energy shock that resolves quickly is disinflationary in 2027 — exactly the scenario in which Bowman's "core PCE moves closer to 2 percent next year" forecast works. An energy shock that persists or escalates is the scenario in which core inflation stays above target and the Fed faces a genuine trade-off. Bowman said she would be "especially concerned about effects on both sides of our mandate if there were further significant disruptions to energy markets," noting that escalation would have more meaningful effects on real activity because of higher uncertainty and the larger impact of energy prices on lower- and middle-income consumers.
The Path Forward: Not on a Preset Course
On policy, Bowman was careful not to pre-commit. At the October FOMC meeting she supported holding the federal funds rate target range at 3-3/4 to 4 percent, consistent with her view that further changes should wait until there is better understanding of how conditions would evolve and how the recent rate increase would influence the underlying trends. Maintaining the stance allowed the committee to assess incoming data — and the data since have been mixed, with inflation persisting and the labor market firming.
Before the December meeting, policymakers will receive the October PCE inflation report and the November employment report. Bowman's read of the current moment is pointed: "Today, I am seeing signs that inflation pressures are persisting, while labor market conditions appear to be firming. This indicates that there is still work to do to bring inflation closer to 2 percent." But she paired that with a warning about the cost of tightening into already-tightening financial conditions. Against the backdrop of lower equity prices, wider spreads, and a stronger dollar, "the benefit of additional policy tightening has to be carefully weighed against its costs for interest-rate-sensitive sectors" — housing, small-business investment, and the newer, medium-sized AI firms where higher rates are already holding back investment and purchasing decisions.
"It is important to remember that monetary policy is not on a preset course. At each FOMC meeting, my colleagues and I will evaluate incoming data, the evolving outlook, and the implications for the risks to our dual-mandate goals of maximum employment and price stability."
Two features of that statement deserve attention. First, "not on a preset course" is a deliberate rejection of forward guidance — after the rate cuts toward the end of last year and the recent rate increase, the committee is refusing to telegraph the next move. Second, the reference to "the recent rate increase" confirms that policy has moved toward restraint since the late-2025 cuts that Bowman says helped address labor-market weakness. The stance is restrictive; the question is whether it is restrictive enough.
What This Means: Cyclical Shocks on a Structural Platform
The judgment that matters for markets is whether Bowman's inflation story is cyclical or structural, because the answer determines the policy endpoint. Her answer is layered, and it is the right one. The inflation problem is cyclical in origin — three identifiable supply shocks, each contributing roughly a quarter of above-target core PCE, that should abate or unwind next year. None of them reflects an overheated labor market, and none of them requires a recession to cure. But these cyclical shocks are playing out on top of a structural shift in how the economy invests. The AI buildout is not a one-off demand spike; it is a reorientation of capital formation toward a new general-purpose technology, with financing structures, off-balance-sheet exposures, and cross-firm linkages that did not exist a decade ago.
That distinction drives the policy conclusion. If inflation is mostly cyclical supply shocks with a benign labor market, the Fed can afford to wait. Cutting too soon risks re-anchoring expectations above 2 percent; hiking too hard breaks the labor market to solve a problem the labor market did not create. Bowman's wait-and-see stance at 3-3/4 to 4 percent is the logical position: policy is restrictive enough to contain second-round effects, but not so aggressive that it manufactures a downturn. Time is an ally — "Time will help us learn more about the durability of momentum in AI-related demand and the evolution of geopolitical risks," she said, adding that greater clarity on AI-related financing conditions and the sustainability of AI investments "should increase my confidence in future policy decisions."
The counter-thesis is that inflation expectations are already de-anchoring, or at least that waiting is itself a policy error. A hawk would argue that three-quarters of above-target inflation coming from supply shocks is exactly when credibility matters most: if households and firms see 3.3 percent core PCE and conclude the Fed will tolerate it because "it's just tariffs and energy," then second-round effects embed themselves in wages and contracts before policy can react. The strongest version of this argument does not attack Bowman's decomposition — it accepts it and says the decomposition is the point. The specific falsifying signal is broadening combined with persistence. Bowman herself noted that price pressures "had been concentrated in a few categories, but the latest data suggest that these pressures may be broadening." If the October PCE report confirms 3.3 percent year-over-year with a monthly core reading at or above 0.3 percent for two consecutive months, and if the breadth of rising prices continues to expand beyond energy and tech hardware, then the "shocks will abate next year" forecast is wrong — and the case for additional tightening becomes compelling despite the labor-market drag.
Scenarios and What to Watch
The base case is that energy tensions ease, AI financing conditions stabilize without a disorderly repricing, and core PCE drifts back toward 2 percent through 2027 with the Fed holding at 3-3/4 to 4 percent into year-end. The upside case for growth is that AI productivity gains arrive faster than expected, offsetting the inflationary impulse from the buildout and allowing the Fed to cut sooner. The downside case is a simultaneous energy escalation and AI valuation correction: Brent back above $110, tech spreads widening further, and the $800 billion AI debt overhang showing stress. In that scenario, the Fed faces stagflationary pressure — higher inflation from energy, weaker activity from the AI shock — and "not on a preset course" becomes the polite way of saying the next move could go either way.
What to watch before December:
- The October PCE inflation report and the November employment report — the two data releases Bowman explicitly named as decision-relevant.
- Whether core PCE confirms 3.3 percent year-over-year and whether monthly prints show continued broadening beyond energy and tech hardware.
- Brent crude, freight rates, and insurance premiums — a renewed Middle East escalation would rekindle the energy shock and invalidate the "abate next year" forecast.
- AI financing conditions: corporate bond spreads, credit default swaps on AI-exposed names, and any sign of stress in the $800 billion-plus AI debt complex.
- Equity valuations in the tech sector — a large correction would tighten financial conditions further and, in Bowman's own chain, undermine the resilience in consumer spending.
The December meeting is a data gate, not a direction. If the AI boom cools and energy prices normalize, a 2027 return toward 2 percent is plausible with policy on hold. If neither happens, the Fed's patience will be tested — and Bowman's candid admission that inflation progress has disappointed suggests she would rather hold too long than cut too soon. The central bank that spent 2022 learning it was behind the curve on inflation has no intention of making the same mistake in reverse.
Explore more exclusive insights at nextfin.ai.
