Anthropic has entered into agreements for at least 14.8 gigawatts of compute capacity since October, according to an analysis by Valida Pau at The Information summarized by Techmeme. The same analysis says Anthropic may spend as much as $517 billion over the next decade. The wording matters: the figure is presented as a possible upper bound, not a confirmed budget or company forecast. Techmeme’s summary says Anthropic has spent the last year trying to line up cloud computing deals. The available feed item does not specify the counterparties, contract structures, deployment schedule, or how much of the 14.8 GW is firm versus conditional. For now, the story should be read as a signal of scale rather than a fully validated capital plan. It points to the size of the infrastructure race around frontier AI models, but the provided material does not establish how Anthropic will finance the capacity, when it will come online, or how much will ultimately be used. Who benefits: Cloud and infrastructure providers tied to large AI compute contracts stand to benefit if the reported capacity agreements convert into deployed workloads. Anthropic would benefit if the capacity improves its ability to train and serve models reliably. Who's exposed: Anthropic is exposed to execution and financing risk if long-term compute commitments grow faster than revenue or deployment capacity. Customers and partners may also face uncertainty until more details emerge on timing and availability.