Anthropic's $518 billion AI buildout hinges largely on deals that cannot be canceled, filing shows

Anthropic, the AI lab, plans to spend at least $518 billion over a decade on AI infrastructure with six partners, according to a confidential IPO prospectus. This significant commitment ranks among the largest in the field. Over 80% of this amount is non-cancelable or requires payment regardless of usage. The company attributes this necessity to the scarcity of computing power, as future demand for advanced AI systems may exceed available supply.

Anthropic commits $111.1 billion with Google, $110 billion with Amazon, and $31.4 billion with Microsoft to long-term infrastructure service obligations. Additionally, the company carries about $161.2 billion in Broadcom-related equipment lease obligations that are largely non-cancelable. Anthropic disclosed agreements with xAI and AMD, with the latter committing to buying up to $5 billion of Anthropic stock and supplying AI computing capacity expected to exceed $20 billion.

As Anthropic increasingly builds its own AI infrastructure, shifting from a cloud-only model, it highlights the risk of dependence on Amazon, Google, and Microsoft, which provide computing infrastructure and develop rival AI models.