CoreWeave’s New $3.1B Loan Tests Wider Market for AI Infrastructure Debt

CoreWeave (NASDAQ: CRWV) Inc. closed a $3.1 billion delayed draw term loan facility, adding another layer of debt financing to fund its rapid buildout of AI cloud infrastructure tied to customer contracts.

The Livingston, New Jersey-based company said the new facility, known as DDTL 5.0, will support the purchase and deployment of infrastructure dedicated to contracts with two large non-investment-grade customers. The loan matures in about 5.5 years and is priced at SOFR plus 4.50%, after investor demand allowed the company to tighten pricing by 50 basis points from initial discussions.

CoreWeave said the transaction was “meaningfully oversubscribed” and described it as the first publicly syndicated financing vehicle backed by high-performance computing infrastructure. The structure is designed to broaden the investor base for AI infrastructure debt by allowing secondary market trading, a step that could make GPU-backed financing more accessible to institutional investors beyond private credit and bank-led arrangements.

The facility received a Ba2 rating from Moody’s and BB+ from Fitch, placing it below investment grade. Morgan Stanley and Mitsubishi UFJ Financial Group served as joint lead arrangers and bookrunners.

The new loan follows CoreWeave’s earlier $8.5 billion delayed draw term loan facility, DDTL 4.0, which the company completed earlier this year to finance AI infrastructure backed by customer contracts. That facility was notable because it received investment-grade ratings — including A3 from Moody’s and A (low) from DBRS Morningstar — reflecting the perceived strength of the underlying collateral and contracted cash flows. Borrowers under that structure could initially draw up to about $7.5 billion, with an accordion feature expanding total capacity to $8.5 billion.

By contrast, the new DDTL 5.0 facility is tied to deployments for two large customers that are not investment grade, which helps explain the lower credit ratings and higher spread. Still, the deal underscores how lenders and institutional investors are increasingly treating AI infrastructure as a financeable asset class, with GPUs, data center capacity and long-term customer commitments forming the basis for large debt packages.

CoreWeave said the delayed draw structure aligns funding with deployment schedules and the useful life of the underlying GPU infrastructure. Such structures have become a key part of the company’s capital strategy as it races to meet demand for AI compute from enterprise and hyperscale customers.

The latest facility brings CoreWeave’s year-to-date debt and equity financing to more than $20 billion, according to the company. That capital raise comes as AI cloud providers and data center operators face heavy upfront costs to secure GPUs, power, networking equipment and data center capacity before revenue from contracted deployments is fully realized.

CoreWeave, which went public in 2025, has become one of the most aggressive users of asset-backed and infrastructure-backed financing in the AI cloud market. Its financing strategy mirrors a broader shift in the AI infrastructure sector, where companies are seeking ways to turn customer contracts and expensive compute equipment into collateral for large-scale debt funding.

The new loan also highlights the bifurcation emerging in AI infrastructure credit. Investment-grade customer contracts can support lower-cost debt, as seen in CoreWeave’s earlier DDTL 4.0. Deals tied to weaker-rated or non-investment-grade customers may still attract demand, but at higher yields and lower ratings, reflecting greater perceived repayment risk.