Science & Tech
Anthropic Locks In $518 Billion AI Buildout
Most of the spending is fixed even if demand changes.
Most of the spending is fixed even if demand changes.
is racing to secure the computing power it needs to keep its AI models growing, and the cost is huge. Its filing shows at least $518 billion in long-term spending commitments for , , and other equipment, with much of it locked in even if demand changes. That makes the company’s growth look both bold and risky.
The story shows that AI is now a battle over , electricity, and long contracts, not just better software. and chip companies benefit from this buildout, while and its investors take on large fixed costs that only make sense if usage keeps rising fast. For people, that could mean better AI tools, but also more strain on power grids, , and the companies trying to turn excitement into profit.
Reuters said Anthropic reported a of $42 billion in 2025, even as it prepared for a possible . The filing turned Anthropic’s rapid growth into a of whether huge infrastructure commitments can be converted into durable revenue. It also sharpened investor attention on how much AI progress now depends on long-term , power, and chip supply.
Reuters and other reporting said the filing also included $161.2 billion of largely non-cancelable tied to . The disclosure showed that Anthropic’s infrastructure strategy is not only a cloud story, but also a . It underscored how capital-intensive has become, with suppliers and lenders sharing the risk.
The prospectus said Anthropic plans to spend at least $111.1 billion with Google, $110 billion with Amazon, and $31.4 billion with Microsoft under . Reuters reported that the commitments run for seven to 10 years and are payable regardless of usage. The structure showed how deeply Anthropic’s and needs are tied to major .
Reuters reported that about 80% of the $518 billion in obligations is non-cancelable or requires payment regardless of usage. That made the unusually rigid, with future costs locked in even if demand shifts. The disclosure highlighted the embedded in Anthropic’s growth strategy.
Reuters disclosed that Anthropic expected to spend at least $518 billion over a decade on with six partners, according to its . The filing put the company’s , computing, and infrastructure obligations in the same category as the largest technology buildouts ever described. Reuters said the total reflected long-term commitments rather than a single contract.
Reporting based on The Information said Anthropic had agreed to about $517 billion in covering 14.8 of capacity over 11 months through August 2026. The report also said this dwarfed an earlier investor disclosure of roughly $180 billion through 2029. It placed Anthropic’s spending surge in the context of a rapidly intensifying AI infrastructure race.
Reuters said Anthropic’s revenue for Claude had surged this year, and the company responded with a major compute shopping spree. The reported deals included a $30 billion commitment tied to Microsoft and Nvidia, a roughly $45 billion infrastructure deal with Nscale, and a $1.25 billion monthly agreement with SpaceX for compute access. The sequence showed how quickly usage growth was turning into .
Reuters reported that Anthropic would spend $45 billion to from ’s data center campus in West Virginia. The six-year agreement was tied to about of power capacity and came as Anthropic was preparing for a major IPO. The report said the company was aggressively expanding compute to meet demand for products such as .
Reuters reported that Anthropic for a in June 2026, moving the company toward what could be the first public listing by a . The prospectus also framed Anthropic as a company racing to scale in a market where compute and infrastructure had become central strategic assets. Reuters said the filing came as Anthropic’s public-market ambitions drew fresh attention.
Anthropic said it expected to bring roughly of capacity through Amazon’s chips by year-end, a sign that its compute needs were scaling quickly. Reuters reported the deal as part of a broader push to secure specialized infrastructure ahead of demand. The arrangement reinforced how closely growth was becoming tied to cloud and chip suppliers.