Anthropic Secures $517 Billion in Compute Contracts

AI safety and research firm Anthropic has locked in massive computing power agreements valued at up to $517 billion over an eleven-month span starting in October 2025. According to reports from The Information, the startup has secured at least 14.8 gigawatts of capacity to supplement its existing one to two gigawatts. In addition to these third-party contracts, the company is actively planning to build its own dedicated data centers to support its long-term scaling goals.

This aggressive infrastructure expansion places Anthropic in direct competition with OpenAI, which has targeted a capacity of 30 gigawatts by the year 2030. While Anthropic's total planned capacity currently falls short of that milestone, many of its newly signed contracts extend well past 2030, complicating direct comparisons. Crucially, neither firm can fund these massive infrastructure commitments solely through current operations. Anthropic's annualized revenue recently topped $65 billion, according to Bloomberg, while OpenAI reported annualized revenue above $40 billion as of July.

The massive spending spree represents a swift pivot for Anthropic. In early 2026, CEO Dario Amodei publicly cautioned the industry against rapid capital deployment, warning that competitors "don't really understand the risks they're taking." Now, Anthropic finds itself in a high-stakes race to secure hardware. Meanwhile, OpenAI CEO Sam Altman has raised his own concerns, warning of "unsustainable silliness" among emerging cloud providers and noting that rapid technical progress could quickly render today's expensive infrastructure investments obsolete.

For AI practitioners and developers, this massive capital commitment guarantees that Anthropic will have the raw computational horsepower needed to train and deploy next-generation frontier models. It signals that the industry-wide push for scale is not slowing down, ensuring a steady pipeline of increasingly capable models. However, the sheer financial risk involved suggests that the cost of accessing these frontier APIs may remain high, as providers struggle to amortize their historic infrastructure debts.