Morgan Stanley-Led Banks Plan to Sell $15 Billion Debt for Anthropic Data Center

A group of banks led by Morgan Stanley is preparing to offload a $15 billion debt package tied to a new data center campus in Texas leased to artificial intelligence firm Anthropic. The banks intend to refinance the debt they committed to the project by selling it to investors in the bond market, the Financial Times reported, citing several people familiar with the matter.

The Refinancing Plan

The debt is associated with a 2,000-acre data center campus currently under construction in Hubbard, Texas. The project developer is Nexus Data Centers, and the primary tenant will be the Google-backed AI company, Anthropic.

According to the report, the refinancing is expected to proceed as soon as the initial loans are drawn. The plan involves several key components:

• The $15 billion debt package will likely be divided into multiple bond sales over time.
• The original bank loan includes a delay-draw feature, allowing the developer to access funds as specific construction milestones are met.
• Some portions of the debt could also be refinanced through the leveraged loan market.

Speculative Grade and Key Risks

Ad

Despite backing from Google for the project's tenant, the bonds are expected to receive a speculative-grade (or "high-yield") credit rating. This reflects the risks associated with the construction phase of the project.

Crucially, Google's financial backing only becomes effective after the data center is fully completed and operational. This contingency means that until construction is finished, investors in the debt bear the project risk, which is typical for infrastructure developments at this stage.

Market Context

This large-scale financing highlights the immense capital required to build the infrastructure powering the artificial intelligence boom. The plan to syndicate the debt to public bond and loan markets demonstrates how Wall Street is structuring deals to fund these capital-intensive projects while managing its own balance sheet exposure.

The deal will serve as a significant test of investor appetite for debt linked to AI infrastructure, particularly for projects that are still under construction and carry a non-investment-grade rating.