Key Notes
- Kevin Warsh outlined scenarios in which AI lifts productivity, disrupts jobs or disappoints investors, without presenting them as forecasts.
- The Fed’s stability report documents more than $680 billion in advanced computing debt issuance through October 2026.
- Financing links among technology firms, developers and lenders could transmit an investment downturn beyond stock prices.
Federal Reserve Chair Kevin Warsh said Monday that disappointing returns from the artificial intelligence investment boom could push stock valuations lower, slow the economy and drive unemployment higher, even as he outlined the technology’s potential to lift long-term growth.
Speaking at the Fed’s Productivity and Jobs Task Force Conference in San Francisco, Warsh described several paths for the economy. He stressed that the staff’s exercises were “illustrative scenarios, not forecasts,” rather than predictions of an impending crash.
The remarks put investment returns and employment at the center of the Fed’s AI discussion. They also arrive as the central bank documents a growing network of borrowing, leases and guarantees behind the construction of data centers and computing infrastructure.
Three Paths for Growth and Employment
Warsh’s most favorable scenario combines faster productivity with gains shared by workers and investors. Economic growth would remain above its long-run average while employment stays close to its full potential.
A second possibility is more complicated: output rises, but automation displaces workers faster than the labor market can absorb them. Stronger growth could then coexist with higher unemployment.
In the downside scenario, investors reassess expected returns after substantial spending and stock-market gains. Falling valuations would accompany weaker activity and a larger increase in joblessness, without the hoped-for productivity improvement.
These are different risks. A technology can improve the economy’s ability to produce goods and services while still disrupting particular jobs. An investment can also prove disappointing even if the underlying technology remains useful.
A $680 Billion Borrowing Surge
The Fed’s November stability report, released Friday, shows why the investment cycle matters beyond technology shares. Debt issuance connected to advanced computing infrastructure exceeded $680 billion through October 2026, compared with $130 billion in all of 2023.
That measure covers public issuance and trackable private securities, rather than every private financing arrangement. The report separately estimated outstanding public debt tied to advanced computing at more than $1.1 trillion as of October. Issuance during a year and the stock of debt outstanding measure different things.
Other commitments are substantial too. Across seven major technology firms, the report identified more than $1 trillion in undiscounted leases that had not yet started and more than $2 trillion in contracts and residual-value guarantees. Those nominal commitments cannot be compared directly with discounted balance-sheet debt.
How an AI Setback Could Reach Lenders
Long-term leases can help developers obtain construction financing, while guarantees on the future value of chips can support borrowing against that equipment. Such arrangements link the finances of large technology companies with smaller businesses that depend more heavily on outside funding.
The report estimates that the largest US bank holding companies have close to $1 trillion in exposure to large technology firms, data-center developers, related financing vehicles, private-credit vehicles and utilities. That is a measure of exposure, not a forecast that those banks will lose $1 trillion.
The issue overlaps with the IMF’s warning about interconnected AI financing, as we previously reported. When a supplier helps finance its customer, weak demand can affect both the customer’s ability to pay and the supplier’s sales.
Barr Warns That Useful Technology Can Still Be Overpriced
In a separate speech Monday, Fed Governor Michael Barr distinguished AI’s long-term promise from the returns investors receive. Even a transformative technology could disappoint if its earnings arrive later, or in smaller amounts, than expected.
Barr identified two ways an investment reversal could weaken growth: businesses could spend less on new capacity, while losses in wealth could reduce household and business spending. He also warned that leverage could amplify a reversal through rapid asset sales.
His assessment was not uniformly negative. Barr said strong capital at large banks provides protection against spillovers, while emphasizing the need to monitor counterparties with large AI positions. The concern is how an adjustment travels through financing relationships, not simply whether share prices move down.
The Focus Is Returns After the Buildout
For investors, the financing figures make cash generation an important counterpart to announced spending. A signed lease or equipment guarantee can support construction today, but the project still needs customers and revenue over time.
The practical questions are when capacity becomes productive, how much customers pay to use it, and which company bears the cost if demand falls short. Our earlier coverage examined how cloud-capacity commitments and guarantees can extend a supplier’s exposure beyond hardware sales.
Warsh’s remarks do not settle those questions or announce a new interest-rate decision. They show the Fed preparing for outcomes in which productivity, investment returns and employment move in different directions.