Key facts
- AI financing needs now extend beyond data-center construction to land, power, semiconductors, GPUs and TPUs.
- Capital spending by five largest US hyperscalers and CoreWeave is expected to reach nearly $1 trillion in 2027.
- A wave of loans maturing between 2027 and 2029 will test weaker borrowers' refinancing ability.
- 17% of 607 tracked borrowers use payment-in-kind interest, adding to debt burden.
- Moody's flagged "circular" financing risks due to correlated exposures among tech companies.
High levels of borrowing to fund AI infrastructure is expanding private credit beyond traditional corporate lending, but a wave of loans coming due in the next few years could leave weaker borrowers struggling to refinance, Moody's Ratings said on Tuesday. The ratings agency expects capital spending by the five largest US hyperscalers and CoreWeave to reach nearly $1 trillion in 2027, financed through cash flows, debt and structures such as joint ventures and special-purpose vehicles. About 17% of the 607 borrowers tracked by Moody's used payment-in-kind interest, which allows companies to defer cash interest payments but adds to their debt burden. Moody's also flagged "circular" financing risks as technology companies, chipmakers and AI developers increasingly invest in, lend to and buy from one another, raising fears that correlated exposures could weaken together during a downturn.