344
Productivity & Workflow355
Automation & Workflow225
Software Development251
Marketing & Growth192
AI Infrastructure & MLOps175
Writing & Content Creation203
Data & Analytics142
Photography & Imaging156
Design & Creative170
Customer Support132
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
Anthropic will rent data center capacity from Nscale in West Virginia in a deal worth about $45bn over six years, as it seeks more computing power.
In short: Anthropic has agreed a roughly $45bn, six year deal to rent data center capacity from UK start-up Nscale in West Virginia.
Anthropic, the company behind the Claude chatbot, has struck a major agreement with Nscale, a UK cloud computing start-up. Anthropic will rent capacity from Nscale’s flagship data center site in Mason County, West Virginia. The deal is worth about $45bn over six years, according to the Financial Times.
A data center is a building filled with computers that run online services. For AI companies, these buildings act like giant “engine rooms” that power both training and everyday use of AI tools.
People familiar with the matter told the FT that Anthropic expects to access 460 megawatts of Nvidia’s next-generation “Vera Rubin” processors when those chips become available late next year. Processors are the parts of computers that do the work, similar to the engine in a car.
The report says Anthropic has also been signing other computing deals recently, including with Amazon and SpaceX. It has separately committed to leasing more than $150bn of Google’s AI chips in data centers built specifically for them. The company is also reported to have held early talks with Meta about renting computing power.
Nscale is building a 1.35 gigawatt data center in West Virginia, with a 2 gigawatt natural gas power plant on site to supply it. Nscale has estimated the overall project could cost $69bn, and it has faced local pushback over noise and other impacts.
This deal shows how much computing capacity AI companies need, and how hard it can be to secure enough of it. More capacity can mean faster responses and the ability to build larger, more capable models, but it also brings big costs and local concerns about power use and construction.
Source: Financial Times