AI in action · Markets
Lenders Turn Wary of AI Debt After the Tech Sell-Off
Banks and credit investors are scrutinising AI debt more closely after the sell-off, Reuters reported on November 9, 2026. Hyperscaler bond spreads have widened about 50 basis points this year.
The money behind the AI build-out is getting more careful. Banks and credit investors have stepped up their scrutiny of AI-related debt deals after the stock market sell-off, Reuters reported on Monday, November 9. Some lenders that were eager to fund data-centre projects a few weeks earlier are now reluctant to add more, the report said, even when the deals carry guarantees from large technology companies.
The exposure is already sizeable. At JPMorgan, loans backed by income-producing properties including data centres reached $29 billion at the end of the third quarter, 12% of its commercial real estate book and second only to apartments; three years earlier the share was 8%. Goldman Sachs Research expects global hyperscaler debt issuance of $420 billion in 2027, more than 60% above its $250 billion estimate for this year.
The bond market is charging for it. Bank of America strategist Yuri Seliger noted that the spread on 10-year bonds of the big cloud companies has widened by about 50 basis points this year, against about 20 for high-quality industrial companies with no AI exposure. Smaller players feel it more: Nebius took a $3.5 billion senior secured loan from Citigroup in October after dropping a planned high-yield bond.
"Over the past year there was unbridled enthusiasm to provide debt to neoclouds," Chris Poggi, a finance partner at the law firm Ropes & Gray, told Reuters. Now, he said, "debt investors' books have become full." OCBC chief executive Tan Teck Long said at the bank's results briefing on November 6 that "we are very focused on the risk aspect" in the deals it looks at, and Jay Bala, chief executive of AIP, expected most banks and credit investors to "take a wait and see approach".
This matters for stock investors as much as for lenders. The AI build-out has relied on borrowed money for a growing share of its spending, so tighter credit slows it even if demand holds up. The Federal Reserve's Michelle Bowman flagged the same channel in a speech on November 10, and CoreWeave's results the same evening showed what it looks like in practice: quarterly interest costs of $846 million.
Sources: Reuters via Yahoo Finance — Banks, investors grow wary of AI debt, spooked by market selloff (Nov 9, 2026)