Anthropic has signed compute contracts valued at up to $517 billion over an eleven-month period, according to a report by The Information. Since October 2025, the artificial intelligence company has secured at least 14.8 gigawatts of computing power in addition to its existing capacity and is planning to construct its own dedicated data centers.
The aggressive infrastructure expansion follows public warnings from Anthropic CEO Dario Amodei regarding competitors taking uncalculated financial risks through rapid investment. Despite Anthropic’s annualized revenue topping $65 billion according to Bloomberg, the company’s long-term compute commitments extend past 2030 and exceed its current direct revenue capacity, mirroring similar multi-gigawatt buildout targets pursued by OpenAI.
The rapid capacity expansion highlights an intense infrastructure deployment race among frontier model developers. Meanwhile, OpenAI CEO Sam Altman has urged caution regarding speculative infrastructure investments, warning that rapid technological progress could render current long-term compute commitments financially unsustainable.
Why it matters
Capital requirements for frontier AI models are reaching unprecedented scales, forcing heavy reliance on multi-year infrastructure commitments.
Rapid hardware innovation risks devaluing massive long-term compute contracts before the underlying infrastructure is fully deployed.
Source: the-decoder.com
