Economy

Add as a preferred source on Google

Fed's Barr says AI financing obligations exceed $3 trillion

AI companies' obligations outside their balance sheets exceed $3 trillion, the Federal Reserve's Barr said, citing risks from increasingly complex financing arrangements.

Updated 2 min read
Data centre cooling plant.
Data centre cooling plant.

AI companies’ financial obligations outside their balance sheets now exceed $3 trillion, potentially magnifying losses if investment values fall, Barr said in a speech published by the Federal Reserve Board on Nov. 16.

Those commitments have grown faster than their recorded debt, Barr said. He made the remarks in “Some Potential Financial Stability Implications of Artificial Intelligence.” He said increasingly complex financing could make it harder for investors to identify who bears losses.

Barr cited the Federal Reserve’s latest Financial Stability Report, published the preceding Friday, which assessed borrowing and other arrangements financing data centers and related investments. Its comparisons included spending and debt issuance since 2020.

The central bank estimated the potential maximum nominal value of obligations outside company balance sheets at more than $3 trillion, Barr said. He said these include energy contracts, long-term data center leases and guarantees covering assets’ resale values.

Debt recorded on the balance sheets of the seven largest AI firms is expected to exceed $800 billion before year-end, the central bank said. Barr said borrowing grew substantially in 2026, drawing comparisons with the earlier telecommunications investment boom.

The central bank compared large cloud-computing companies’ cumulative capital spending and debt issuance with those of telecommunications companies during the broadband investment cycle that began in 1995. Barr said the cloud companies had not yet surpassed those earlier totals.

Large cloud companies and graphics-chip producers also hold more than $100 billion in cumulative equity investments in language-model developers and cloud-computing start-ups, Barr said. Recipients use that money to buy or rent computing services and chips from their investors, he said.

Barr said those investments account for a relatively small share of the suppliers’ balance sheets but provide essential funding for smaller recipients. Those businesses eventually need paying customers to replace financing supplied by their vendors, he said.

Some larger companies also guarantee the resale value of data centers or chips pledged to secure loans, Barr said. These guarantees could leave them absorbing losses when borrowers default and lenders sell the assets, he said.

Barr said he was less concerned about direct losses because debt backing the investment boom remains limited. He said strong capital levels at the largest banks also help protect the financial system from losses spreading.

However, high hedge-fund borrowing could force rapid asset sales after a sharp reversal, pushing prices lower and prompting further selling, Barr said. He said banks should keep monitoring exposures to trading partners with large AI positions.

Barr said AI could improve productivity and living standards over time. Even so, he said disappointing investment returns could reduce economic growth through weaker investment and lower consumer and business spending.

Related articles

View all