Generative AI company Anthropic plans to invest $50 billion on data centers in the US.
The company said that it will build data centers with Fluidstack in Texas and New York, with more sites to come.
The first facilities will come online next year, and are expected to lead to 800 permanent jobs and 2,400 construction jobs.
“We’re getting closer to AI that can accelerate scientific discovery and help solve complex problems in ways that weren’t possible before. Realizing that potential requires infrastructure that can support continued development at the frontier,” said Dario Amodei, CEO and co-founder of Anthropic.
“These sites will help us build more capable AI systems that can drive those breakthroughs, while creating American jobs."
The Texas data center could be Fluidstack's previously announced 168MW Abernathy facility, co-developed with TeraWulf, or its 244MW project with Cipher Mining. The New York site is likely a 360MW Fluidstack-TeraWulf project in Lake Mariner.
Google, which has a 14 percent stake in Anthropic, has backed both TeraWulf projects. Earlier this year, it backstopped Fluidstack's loans and took a 14 percent stake in cryptominer TeraWulf.
It has also backstopped $1.4bn of Fluidstack’s lease obligations for the Cipher deal, and taken a 5.4 percent stake in the latter business.
Last month, Anthropic announced a 1GW+ Google Cloud deal with up to one million TPUs worth "tens of billions of dollars". DCD has contacted Anthropic to clarify if today's announcement is the same contract, with Fluidstack serving as the intermediary, or if this is a net-new deal.
Gary Wu, co-founder and CEO of Fluidstack, added: “Fluidstack was built for this moment. We're proud to partner with frontier AI leaders like Anthropic to accelerate and deploy the infrastructure necessary to realize their vision."
Rival cloud provider Amazon Web Services also owns a stake in Anthropic, and in October launched the Project Rainier cluster featuring its own Trainium2 chips, for use by the AI firm. It plans to spend as much as $11 billion on the site at full buildout.
While Anthropic's compute costs are rising, they pale in comparison to rival OpenAI's $1.4 trillion in commitments.
While the ChatGPT maker expects to lose $74bn this year and turn a profit in 2030, Anthropic expects to break even in 2028.
Internal documents seen by the Wall Street Journal estimate that OpenAI expects to burn through around 14 times as much cash as Anthropic before reaching profitability.
Documents seen by The Information show that Anthropic hopes to use more efficient models requiring less compute per user, reducing costs and helping it reach profit faster.
Comments