Fed's Bowman: AI Now Drives About a Quarter of US Excess Inflation

Artificial intelligence accounted for roughly one-fourth of America's above-target core PCE inflation over the 12 months ending in September, according to Fed Vice Chair for Supervision Michelle Bowman, who delivered the estimate at the Council on Foreign Relations in Washington on Tuesday, November 10, 2026.

Bowman attributed a similar share of the excess to tariffs and a similar share to energy. Core PCE is the Federal Reserve's preferred inflation gauge, and it excludes food and energy prices. The figure is her own estimate; the Fed has published no such statistic.

The stakes for markets are direct. The same AI infrastructure spending that powers NVIDIA and Microsoft revenue growth now sits, in Bowman's framing, on the Fed's list of inflation drivers — a shift that ties the fate of AI equities to monetary policy more tightly than at any point since the buildout began.

What did Bowman say about current inflation data?

Alongside the attribution estimate, Bowman said 12-month core PCE likely rose to 3.3% in October. The October consumer price index rose 0.4%, above the consensus forecast.

The market reaction was mixed and cannot be clearly linked to her speech, as 24/7 Wall St. noted. NVIDIA (NASDAQ:NVDA) closed Tuesday at $168.89, down 5.47%. Around 1:32 p.m. ET on November 11, it traded at $168.89's rebound level of $177.46, up 5.07%, as a CoreWeave (NASDAQ:CRWV)-led rally lifted AI cloud stocks. Microsoft (NASDAQ:MSFT) traded at $521.28, down 2.23%.

Why can rate hikes barely touch AI inflation?

Bowman said the AI expansion has pushed investment demand toward high-tech equipment whose supply cannot expand quickly. Prices for memory chips, semiconductors and computers rose by double digits, and she said monetary policy has limited influence over them.

That supply constraint helps NVIDIA. On its August earnings call, the company's chief executive said: "Even though our demand is much greater than 70%, our supply allows us to confidently deliver 70%."

Microsoft sits on the paying side of the same shortage. Its quarterly capital spending reached $39.367 billion, and roughly three-quarters of that went to short-lived assets such as GPUs and costlier memory.

What financing risks did she flag?

Bowman said AI-related debt is on track to exceed $800 billion in 2026. She estimated that leases, contract commitments and residual value guarantees could add more than three times that amount — figures that, by 24/7 Wall St.'s arithmetic, imply more than $2.4 trillion in total obligations.

She also raised concerns about hyperscaler equity stakes in smaller customers, saying they create "the appearance of circular financing." NVIDIA fits that description, according to the report: its guarantee obligations are limited to $108.5 billion, and it has invested nearly $50 billion in frontier AI labs.

Her downside scenario was explicitly conditional. If AI revenue falls short of rising costs, AI stocks could face a sharp correction and tighter credit.

What did Bowman leave out about December?

Bowman backed holding the target range at 3.75% to 4% in October and said policy is on no fixed course. She did not endorse a December hike, even though Governor Christopher Waller and Dallas Fed President Lorie Logan argued for more increases on the same day.

Futures markets were less restrained. Early on November 11, they priced an 88.5% chance of a quarter-point hike and an 11.5% chance of a half-point move.

Despite the elevated readings, Bowman said she still expects core PCE to move closer to 2% next year.

How do AI stocks stack up before a possible December hike?

The bull case rests on demand still exceeding supply. NVIDIA's guidance points to $108.0 billion in third-quarter revenue, and Microsoft's commercial backlog stands at $677 billion.

The bear case is mechanical. A hike raises borrowing costs for data center developers, and it lowers the present value of profits expected years from now — a discount that matters most for growth stocks.

The 24/7 Wall St. analysis judged NVIDIA better placed than Microsoft under the pressures Bowman described. NVIDIA collects the price increases she described and trades at 34 times earnings after a -22.6% month. Microsoft trades at 30 times earnings and pays those higher prices.

The near-term calendar sets the test. NVIDIA's earnings report is expected on November 17, though the company has not confirmed the date. If revenue misses the $108.0 billion guide by more than its 2% band, that analysis concedes its view is wrong. The FOMC then meets on December 8-9.