Anthropic expects to commit at least $518 billion over the next decade to cloud services, computing capacity, and AI infrastructure, according to details from its confidential IPO prospectus reported by Reuters and the Financial Times. About 80% of that amount cannot be canceled or must be paid even if Anthropic does not use the contracted capacity.
The commitments show how strongly the maker of Claude believes computing power will shape competition among leading AI companies. Anthropic tells prospective investors that demand for advanced AI systems could exceed the available supply of compute. It describes access to computing capacity as a central constraint on future growth.
The disclosures also illustrate the financial risk behind the AI industry’s infrastructure race. Anthropic reported a net loss of roughly $42 billion for 2025, Reuters reports. That figure includes a large accounting charge related to financing instruments that could convert into company shares. On an operating basis, excluding such writedowns, the company lost more than $8 billion while revenue rose twelvefold to nearly $4.6 billion.
Long-term contracts with major technology companies
Several of Anthropic’s largest agreements require payments regardless of actual usage. The company has committed at least $111.1 billion to Google, $110 billion to Amazon, and $31.4 billion to Microsoft through long-term infrastructure service obligations.
- Google’s agreement runs from April 2026 to July 2033.
- Amazon’s agreement runs from May 2026 to April 2036.
- Microsoft’s $31.4 billion commitment runs from November 2026 to May 2033 and is largely non-cancelable.
- Broadcom-related equipment leases account for about $161.2 billion and are also mostly non-cancelable.
Anthropic has additional arrangements with xAI that could lead to up to $84.5 billion in spending on Nvidia-based computing capacity through 2029. Reuters reports that these agreements are more flexible, with cancellation generally possible on 90 days’ notice. AMD has also agreed to supply more than $20 billion in AI computing capacity and may buy up to $5 billion in Anthropic stock.
The company is moving beyond a cloud-only strategy by developing dedicated data centers and leasing chips directly. That approach could give Anthropic more control over its capacity, but it also locks the company into major financial obligations before future demand is certain.
Anthropic identifies another complication: several of its infrastructure providers are also investors, customers, distributors, and competitors. Amazon, Google, and Microsoft all have their own AI products and models. Anthropic warns that their incentives may not always match its own, and that a reduction, repricing, or termination of third-party compute access could harm its business.
Safety warnings and founder control
The prospectus also devotes substantial space to AI safety risks. The Financial Times reports that Anthropic warns its systems could behave unpredictably, including by manipulating people, enabling fraud, or creating more severe harms. The filing states that advanced AI may pose “existential risks to humanity.”
Anthropic plans to preserve unusually strong founder influence after the IPO. Its seven co-founders would initially control 50.1% of voting power through a new Founder LLC and a special Class F share. The company remains a Delaware Public Benefit Corporation, a structure that allows leaders to balance investor returns with a stated public-benefit mission.
For public-market investors, the filing presents two connected bets: that Anthropic can sustain rapid revenue growth, and that demand for its AI models will justify one of the largest long-term infrastructure commitments ever disclosed by an AI company.
Sources
- IPO prospectus: Anthropic expects to spend $518B+ over 10 years with six partners on AI infrastructure; ~80% is non-cancelable or payable regardless of usage – Reuters
- IPO prospectus: Anthropic reports a net loss of $42B in 2025 and plans to spend $518B on cloud, computing, and infrastructure obligations in the coming year – Reuters
- IPO filing: Anthropic’s seven co-founders will initially have 50.1% of the total voting power through a new “Founder LLC” aimed at serving the common good – Reuters
- Sources: Anthropic dedicated nearly a third of its IPO prospectus to detailing “risk factors”, including that its AI may pose “existential risks to humanity” – Financial Times
Stay up to date
AI for content creation: the latest tools, tips and trends. Every two weeks in your inbox:
