Anthropic’s $518 Billion AI Buildout Is Mostly Locked In, Filing Shows

Anthropic expects to spend at least $518 billion over ten years building AI infrastructure with six partners, according to a confidential IPO prospectus. About 80% of that total is non-cancelable or must be paid whatever the usage. The company filed confidentially with the SEC in June, and the paperwork has not been made public. It says computing power is becoming the main limit on AI, and that demand for advanced systems will likely outrun supply.

Google, Amazon and Microsoft Cloud Commitments

Anthropic plans to spend at least $111.1 billion with Google, $110 billion with Amazon and $31.4 billion with Microsoft over the next seven to 10 years. If actual spending falls short, it must pay Google the difference, and Amazon has similar terms. The Microsoft deal, running from November 2026 to May 2033, can only be ended if Microsoft commits an uncured material breach.

Broadcom, xAI and AMD Deals

Anthropic also carries about $161.2 billion in Broadcom-related equipment lease obligations that are largely non-cancelable. Its agreements with Elon Musk’s xAI could bring up to $84.5 billion of Nvidia-based computing spend through 2029, but those can be cancelled with 90 days’ notice. AMD has committed to buy up to $5 billion of Anthropic stock and supply more than $20 billion of computing capacity.

Anthropic’s Shift to Its Own Data Centers

The company says it is moving from a cloud-only model toward dedicated data centers and directly leased chips. It also flagged a risk in relying on Amazon, Google and Microsoft, which act as investors, customers, cloud providers and rivals at once, so their interests may not fully match its own. The scale is close to OpenAI’s $500 billion Stargate plan, shared by OpenAI, SoftBank, Oracle and MGX.