
Fed Chair Warsh to Grill AI's Economic Claims in San Francisco
Fed Chair Kevin Warsh convenes AI panels in San Francisco as the 10-year yield hits 5.15%, forcing the question of whether the AI buildout is productivity or overheating.
- By
- Sophie Lindqvist
- Filed
- Channel
- AI & Compute
- Read
- 4 min read
Federal Reserve Chair Kevin Warsh will open a conference at the San Francisco Fed on Monday at 9 a.m. Pacific with 15 minutes of remarks. He will then spend most of the morning leading panels with Stripe co-founder Patrick Collison, Stanford economist Nick Bloom, and venture capitalist Marc Andreessen, per the published agenda. The topic sounds academic: how artificial intelligence, and what the Fed calls superintelligence, will change productivity and jobs. The stakes are not academic at all.
The interest-rate question is concrete. If AI raises the economy's speed limit, the Fed can let growth run faster without tightening. If it does not, today's AI building boom starts to look like old-fashioned overheating. The answer determines whether capital keeps flowing into the supply chain behind that boom — power generation, data center construction, and the silicon itself.
Why now: long-term rates just jumped
The timing is awkward. The Fed lifted the upper bound of its target range to 4.00% in September, from 3.75%. Real GDP grew at a 3.4% annualized pace in the third quarter, above the 2% to 3% range considered healthy trend growth. In the week ending November 13, the 10-year Treasury yield rose from 4.86% to 5.15%, and the 30-year reached 5.47%. When bond investors demand more to lend long-term, they are signaling doubt that faster growth comes without inflation.
That doubt matters for the hardware industry. AI infrastructure is financed on long-duration assumptions. Yields above 5% raise the discount rate on every data center, fab expansion, and power project on the drawing board.
What has Warsh actually said?
Warsh has framed the issue himself. At Jackson Hole in August, he said "We've come to a hinge point in history." He called AI "potentially a new factor of production" and asked whether it would "cause a significant, sustained rise in productivity across the economy." He also asked who ends up with the gains: chipmakers, cloud providers, AI developers, other businesses, or consumers.
At his October 28 press conference, he said "Productivity growth is strong and capital investment robust," while noting that inflation has run above target "for close to six years." He asked whether AI would bring a rapid productivity miracle or gradual gains. He also asked whether it would make price stability easier or harder to achieve. These are open questions. He has not answered them.
The bills arrive before the payoff
The timing gap is the core problem. The efficiency gains come later. The construction bills are due now.
Warsh said capital spending sits "at a historically high share of GDP," led by high-tech investment. Data centers could use up to 12% of U.S. electricity demand by 2028, according to Energy Department projections — a projection, not a measured figure. The independent market monitor for PJM Interconnection, which runs the largest U.S. power grid, concluded that "data center load growth is the primary reason" for its tight supply and high prices. Data centers compete with everyone else for power, chips, and electricians. That competition pushes prices up well before any productivity return.
Households are already absorbing the costs. Inflation-adjusted average hourly earnings were $11.26 in October, a preliminary figure, down from $11.38 a year earlier. The University of Michigan's consumer sentiment index fell to 48.1 in September, a level associated with recessions. Unemployment sits at 4.1%. AI has not caused mass layoffs. It also has yet to show up in paychecks.
What does each outcome mean for rates and stocks?
If productivity rises fast enough to lower the cost of producing each unit, the Fed could hold or cut rates while growth stays strong. That would bring down mortgage rates and business borrowing costs, and support stock valuations well beyond NVIDIA (NASDAQ:NVDA). If spending keeps outpacing the efficiency gains, the Fed is dealing with a classic overheating economy. Long-term yields above 5% could then stick, putting pressure on housing and on stocks whose prices depend on low interest rates.
Right now, the bond market is pricing in the second scenario. It is up to AI to prove otherwise. Note what that implies: investors currently discount the productivity story that much of the industry's capital spending assumes.
What to watch on Monday
Warsh has said the task force's findings will inform future policy judgments, not dictate immediate decisions. Monday is unlikely to bring a rate signal. Two things are worth tracking instead.
- Whether Warsh presents strong productivity as a reason for patience on rates.
- The next reading of core Personal Consumption Expenditures prices, the Fed's preferred inflation measure, which rose 0.3% in September.
Cooler readings alongside 3% growth would be the first real evidence of an AI productivity boom. Absent that, the market's overheating diagnosis stands, and the cost of financing AI's physical buildout keeps rising.
Source: 247wallst.com
More from Sophie Lindqvist
Show full bio
News editor covering money, funding and market shifts around Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,004 articles
Related articles
warsh-says-ai-investment-cannot-escape-interest-rates
Warsh Says AI Investment Cannot Escape Interest Rates
87-of-fed-survey-contacts-now-call-ai-a-top-financial-risk
87% of Fed Survey Contacts Now Call AI a Top Financial Risk
ai-semiconductor-trade-remarkably-resilient-despite-rising-rates-2c3687f8
AI Semiconductor Trade 'Remarkably Resilient' Despite Rising Rates
ai-infrastructure-funding-demand-holds-despite-credit-stress
AI infrastructure funding demand holds despite credit stress



