AdvertiseFee Deal Desk

Macro

Fed Chair Warsh Outlines Three AI 'Super Intelligence' Scenarios

Federal Reserve Chair Kevin Warsh said Monday that Fed staff have started modeling how artificial intelligence could reshape the US economy, testing three…

Be a creator
November 16, 2026, 07:16 PM UTC4 min read
AI SummaryAI
  • Warsh said the scenarios are illustrative assumptions for policy discussion, not economic forecasts.
  • Fed Governor Michael Barr flagged over $3 trillion in potential off-balance-sheet AI financing obligations.
  • The third scenario sees AI returns disappoint, equity valuations fall and unemployment rise.
  • Warsh said US research effort grew more than 20-fold since the 1930s while research productivity fell.

Three Scenarios, Not Forecasts

Federal Reserve Chair Kevin Warsh said Monday that Fed staff have started modeling how artificial intelligence could reshape the US economy, testing three paths for growth, employment and markets before any of them arrives. Speaking at the Productivity and Jobs conference in San Francisco, an event co-hosted by the Fed Board and the San Francisco Fed, Warsh framed the work as scenario analysis rather than prediction. “These are illustrative scenarios, not forecasts,” he said, adding that simplified economic models are being used to test what might unfold over the next several years. In the first path, AI lets companies produce more with the same resources, pushing output above its historical growth rate while workers share the gains and unemployment holds near the level economists tie to a healthy job market. The second keeps the faster growth but has businesses automate roles faster than displaced workers can find new ones, lifting unemployment above that level even as productivity rises. The third assumes heavy spending on AI infrastructure fails to deliver the returns markets expect: stock prices rise on the expectation of huge profits, and those expectations eventually prove too optimistic. “Equity valuations could fall, leading to an economic slowdown,” Warsh said, with unemployment rising more sharply before the economy recovers without the promised productivity boost. The phrase is not loose talk: “super intelligence,” or SI, is now official Fed terminology, adopted in October after a Trump administration executive order, according to the central bank’s own terminology page. Governor Michael Barr warned in a separate speech Monday that the AI construction boom is increasingly financed through complex arrangements, including long-term computing leases, energy contracts and financial guarantees. The Fed estimates the potential maximum value of those off-balance-sheet commitments now exceeds $3 trillion. They are not debt already owed, but they show why disappointing AI revenue could create problems beyond the companies building data centers. Warsh proposed no interest-rate change in the remarks.

Why SI Differs From Electricity

Warsh gave the exercise a deeper rationale: super intelligence may carry economic properties unlike earlier general-purpose technologies such as electricity or the internal combustion engine. Those innovations worked by accelerating the innovations that followed them, he argued, whereas SI could automate or speed up the production of new ideas themselves. That distinction matters for growth accounting, because the constraint on long-run expansion has been the pace at which the economy generates ideas. He cited estimates that the United States' research effort has grown more than 20-fold since the 1930s, even as research productivity, the output gained per dollar spent or per researcher, has fallen sharply over the same period. If SI can break through that constraint, growth could sit above the long-term average for an extended stretch, the first scenario in practice. The chair set limits on the claim. Machines cannot substitute for human judgment and responsibility, he said: people must design the systems, build them and make the final decisions. He also flagged a security dimension, noting that SI could hand malicious actors powerful tools and that the Fed is watching the vulnerability of critical infrastructure. Board staff, he said, are analyzing scenarios keyed to how fast super intelligence develops and how far its economic effects reach. He did not attach dates to any path, and he left open how quickly the technology itself would mature. The central bank's role, in his framing, is neither to endorse nor oppose the technology's adoption, but to understand its economic impact and prepare monetary policy for what follows. He repeated that the scenarios are assumptions for policy discussion, not a projection of which future will materialize.

The Downside Path for Markets

For markets, the third scenario is the operative one, because it runs through asset prices directly: companies overspend on infrastructure, the market cap of AI builders compresses, growth slows and unemployment rises before any recovery arrives. That is the path where crowded AI-linked trades take the hit alongside the broader market, and riskier assets levered to tech sentiment, crypto included, would share the drawdown; traders weighing exposure across asset classes can compare how cryptocurrency vs stocks behave when a single theme dominates both. Barr's $3 trillion estimate gives the downside case a funding dimension of its own: obligations kept off balance sheet mean distress can surface at counterparties rather than at the builders themselves. Warsh proposed no rate change on Monday; the policy question returns at the Fed's next meeting, scheduled for December 8 and 9.

Primary sources

COINOTAG's editorial and research desk.

AI-Assisted

AI-generated, AI-reviewed, under COINOTAG editorial oversight.

Price-Impacting News

Live
All breaking news
Breaking News