AI borrowing risks drive demand for credit default swaps

Weekly trading in credit default swaps, or CDS, tied to debt issued by Oracle, SpaceX, Alphabet, Amazon, Meta, Broadcom and Nvidia exceeded $9 billion in September, compared with less than $2 billion in July, according to Morgan Stanley Research.

The increase came as technology companies raised more debt to fund artificial intelligence infrastructure and other capital spending. Companies have raised an estimated $466 billion through AI-related borrowing so far in 2026, compared with $101 billion during the same period last year.

Investors have raised concerns about higher debt levels, concentrated exposure to technology companies and whether AI infrastructure spending will generate enough profit. Multi-asset and equity investors are also using CDS as a lower-cost way to hedge technology exposure than buying put options on shares.

CDS are contracts that protect investors against a borrower failing to repay debt or experiencing another defined credit event. Investors pay a premium for the protection and receive compensation if the event occurs.

The cost of default protection has risen for several large technology companies. Five-year CDS spreads for Meta and Nvidia reached record highs of 108 and 101 basis points, respectively, in October. Oracle’s spread rose to 322 basis points on Monday, more than twice its level at the start of the year. SpaceX’s spread reached a record 226 basis points.

A wider spread means investors are demanding more money for protection. The increase reflects concern that borrowers could have difficulty servicing debt if AI spending slows, expected returns fail to materialize or financing conditions worsen.

The sharpest deterioration has occurred at CoreWeave, an AI cloud infrastructure provider. Some of its bonds have fallen into distressed territory, while its five-year CDS spread has reached about 1,090 basis points.

At that level, protection on $10 million of debt would cost about $1.82 million upfront, plus quarterly payments of $125,000. The pricing reflects a higher perceived credit risk than that of the larger technology companies.

Robert Dafforn, chief investment officer at distressed-debt specialist Polus Capital, warned that high debt levels and linked financing arrangements across the AI sector could increase losses if weaker borrowers face difficulties.

Investors are assessing the sustainability of AI-related capital spending and the effect of new borrowing on technology companies’ balance sheets. Credit derivatives are being used to measure concerns about debt risks in the sector.