Roughly $37 billion of the $80 billion raised for US data centers in 2026 sits below investment grade, and investors are demanding nearly 10% to hold it
The AI buildout has a financing problem, and it is showing up in the bond market. Nearly half of the bonds issued to fund US data centers this year carry junk ratings.
The figures come from a Forbes analysis of FactSet data, with input from Columbia Business School professor Stijn Van Nieuwerburgh. Of the $80 billion in bonds issued for US data centers in 2026, approximately $37 billion is rated below BBB, the line where investment grade ends and junk begins.
The year-over-year jump is the striking part. In all of 2025, junk-rated data center bond issuance came to approximately $8.6 billion. This year’s approximately $37 billion is more than four times that total.
The money is going into 16 projects tied to some of the biggest names in AI, including Amazon , Meta , and Nvidia . The bonds themselves, however, are not necessarily backed by those balance sheets. The junk ratings reflect the creditworthiness of the developers and structures doing the borrowing, which is a different thing from the tenants they hope to sign.
By mid-October 2026, average yields on junk-rated data center bonds had climbed to nearly 10%, with the move reportedly happening within the span of a single week.
Stress is also visible in corporate debt tied to so-called neoclouds, the newer cloud providers renting out GPU capacity to AI customers. CoreWeave’s corporate debt was reportedly trading below 80 cents on the dollar.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Heavily indebted developers Firmus and AirTrunk have postponed or canceled initial public offerings amid the turbulence.
The underlying tension is a split in who is actually renting these facilities. On one end sit the hyperscalers, the giant cloud companies with stable cash flows and strong credit. On the other end are riskier developers and younger operators whose revenue depends on AI demand holding up for years.
Layered on top are soaring construction costs and energy supply constraints. The research also notes that broader AI-related debt issuance has reached hundreds of billions of dollars so far this year, with data centers accounting for a significant portion.
A near-10% average yield means lenders are no longer treating data center debt as a safe proxy for the AI boom. Higher borrowing costs eat into project economics, and with IPO windows narrowing, lower-rated developers may find capital harder to access at the very moment they need it to finish construction.
Watch three signals in the coming months. First, whether junk-rated data center yields stabilize or keep climbing. Second, whether distressed pricing spreads beyond CoreWeave to other neocloud borrowers. Third, whether developers like Firmus and AirTrunk revive their listing plans or keep them on ice.