Corporate credit priced for heavy AI capex by hyperscalers

Investment-grade spreads sit near mid-1990s lows as hyperscalers drive record bond issuance to finance large AI infrastructure spending.

Investment-grade corporate credit markets are pricing in heavy capital spending by large cloud and online platforms, with benchmark spreads trading near levels last seen in the mid-1990s.

Meta, Microsoft, Alphabet and Amazon are scheduled to increase AI-related capital expenditures sharply in 2026. Market estimates put their combined planned spending at about $700 billion for 2026, roughly 80% higher than their 2025 total and equal to about 2.2% of U.S. gross domestic product. A large portion of that spending is expected to be financed through bond issuance.

Investment-grade corporate issuance has reached about $976 billion through May, ahead of the record pace set in recent years. Alphabet, Amazon, Meta, Microsoft and Oracle have issued roughly $110 billion of U.S. investment-grade paper year-to-date, representing nearly 16% of total issuance compared with roughly 3% a year earlier.

Despite the surge in supply, the benchmark investment-grade spread sits around 80 basis points. Pension funds, insurance companies and other liability-driven investors have taken much of the new issuance, attracted by all-in yields that remain above 5% for many investment-grade corporates even as spread compensation narrows.

Market participants note several potential triggers that could alter sentiment: a pickup in merger-and-acquisition activity that increases borrowing across sectors; lower-than-expected returns on hyperscalers’ AI investments; or persistent high issuance that outpaces demand. Debt from the largest cloud and online platforms currently trades more than 25 basis points wider than the broader investment-grade index, a gap not seen in about a decade, indicating more issuer-specific pricing.

Proceeds from the deals are being used to build AI infrastructure, including data centers, custom chips and related power and cooling systems needed to run large-scale models. The concentration of issuance among a handful of companies has made those firms the dominant marginal borrowers in the corporate bond market this year.

Articles by this author