AI Infrastructure: From Zero to $100B and Beyond: How the Emergent Sector Is Reshaping the Non-IG Market

Credit Research: Anton Gorbounov, Rucha Amdekar Relevant Item: Excel Download: AI Infrastructure Financing Summary Key Takeaways High-yield and unrated AI infrastructure issuers have raised more than $107 billion of committed and funded debt as of early May 2026. We believe the seven largest non-investment-grade AI infrastructure issuers may need to raise over $400 billion of asset-level and corporate-level financing over the next four years to meet their publicly disclosed capacity targets. The market has organized around two major ownership models away from hyperscalers: AI-native data center owners, which own and lease out the “powered shell” but not the underlying equipment, and independent neoclouds, which, in turn, both own and rent GPU and related hardware capacity. The typical financing stack for companies in the sector includes both asset-based and corporate-level debt. Novel sector-specific asset-based structures include GPU-backed deferred-draw term loans pioneered by CoreWeave, and secured single-asset data center SPV bonds employed by data center owners. Corporate issuance, on the other hand, has been dominated by convertible bonds, with CoreWeave being the sole corporate high-yield issuer in the space to date. Spreads remain wide to rating despite some secondary tightening. The 11 data center SPV notes priced at an average OAS roughly double[...]