CoreWeave bond yield surpasses 16%, raising costs for AI data centers

Debt markets are tightening for AI data center builders, as CoreWeave's bond yield has surged above 16%. This shift signals increased costs for companies seeking financing in an environment where credit is becoming more expensive. The rise in yields reflects broader concerns about the political landscape, particularly with new Democratic governors in key states and a Democratic-controlled Congress that could impose more regulations on data center operations.

The current climate has led investors to seek greater compensation for the risk of default, even among well-established firms like Nvidia, Broadcom, and Amazon. The credit default swaps (CDS) for these companies have seen significant increases—Nvidia's up 76% in just a month, Broadcom's up 49%, and Amazon's climbing 44%. While these figures remain low in absolute terms, they indicate a growing apprehension in the market, suggesting that financing costs could rise for all companies as a result.

The situation is further complicated by the possibility of moratoriums on data center construction in states such as Texas, Ohio, and Georgia, which are crucial for future capacity. Texas governor-elect Gina Hinojosa has expressed her intention to support data centers, but she also advocates for a pause until protections for local communities are established. Goldman Sachs estimates that proposed regulations could affect 45% of planned capacity in the U.S., with Texas being a major contributor to that figure.

Despite these challenges, the overall market for AI-related debt financing remains robust. Since late October, Goldman Sachs has tracked five AI bond deals totaling $67 billion, bringing the year's total hyperscaler debt issuance to $296 billion. Looking ahead, Goldman projects an even larger issuance of $440 billion for next year, alongside an additional $300 billion in project-level financing. National Economic Council Director Kevin Hassett has voiced confidence that data center growth will continue as anticipated, suggesting that while the credit environment is changing, the demand for AI infrastructure remains strong.