Will Financing Become the Bottleneck for AI Computing? A Case Study of Anthropic

Maintaining computing scaling, which is central to AI advancement, requires astronomical capital. Major frontier labs are currently planning infrastructure buildouts worth tens of billions of dollars, far exceeding their current revenues, raising concerns that financing will become a bottleneck for growth.

Anthropic's infrastructure buildout serves as a key indicator dispelling these concerns. As of November 2025, with annual revenue under $9 billion, Anthropic announced a $50 billion plan for US AI infrastructure investment. Approximately $50 billion in debt financing was utilized in this process.

The specific financial structure is as follows:

• TPU Systems: Approximately $35 billion in debt was used to secure Google's TPU systems, with Broadcom sharing the risk by supporting terms in the event of default.
• Data Centers: Approximately $15.2 billion in loans were executed for Fluidstack's data center construction, structured with Google guaranteeing the risk.

In conclusion, when major suppliers (such as Broadcom and Google) back the risk, institutional investors are prepared to lend capital based on AI companies' long-term payment commitments. Therefore, financing is unlikely to become an immediate limiting factor for frontier AI computing growth.