Borrowing costs for US investment-grade technology bonds have climbed to roughly 110 basis points in November, up from about 80 basis points in October. The spreads, the widest in at least two years, reflect growing investor concern over the potential returns on the massive capital expenditures required for artificial intelligence infrastructure. As a result, some banks are now telling market participants they are reluctant to finance new AI-related deals, regardless of whether the loans are guaranteed by the sector’s largest technology firms.
The tightening of credit conditions follows a surge in debt issuance from ‘hyperscalers’ like Microsoft, Meta, and Amazon, who are raising billions to fund data centers. In 2026, five such firms raised $182B in US bonds, while global investment-grade debt for AI-related projects and companies surpassed $415B. This volume of issuance has increased competition for capital, leaving data center builders to face steeper borrowing costs as billions in existing debt come due for repayment.
