Anthropic has agreed to a $45 billion AI computing lease agreement with Nscale for a new campus in West Virginia.
Anthropic has agreed to spend $45 billion over six years to lease AI cloud computing capacity from infrastructure provider Nscale, according to a Bloomberg News report cited by Reuters. The capacity will be sourced from Nscale’s developing data center campus in West Virginia, which is expected to run on Nvidia’s next-generation Vera Rubin technology.
The agreement encompasses approximately 460 megawatts of power capacity, as Anthropic moves to secure the computing resources required to support growing demand for its artificial intelligence products, including the Claude family of models and the Claude Code development tool. Neither Anthropic nor Nscale publicly commented on the reported deal following Reuters’ requests.
The arrangement illustrates a broader industry pattern: leading AI firms are increasingly securing computing capacity years before anticipated demand materializes. Training and running advanced AI models requires large numbers of specialized processors, data center space, and reliable electricity, making access to infrastructure a strategic constraint.
Nscale’s West Virginia initiative, branded the Monarch Compute Campus, is being developed as a large-scale AI data center and power facility. Earlier this year, the company stated the campus spans thousands of acres in Mason County and carries a potential power runway exceeding 8 gigawatts.
Nscale previously announced a letter of intent with Microsoft to deliver up to 1.35 gigawatts of AI computing capacity at the same site using Nvidia Vera Rubin NVL72 systems, with phased deployments slated to begin in late 2027. The reported Anthropic contract would thus add another major AI customer to a campus being positioned as a large-scale computing hub.
Nvidia unveiled the Vera Rubin platform in March, characterizing it as a new generation of integrated AI infrastructure engineered for large-scale training and inference. The company noted that Vera Rubin-based products would roll out through partner networks starting in the second half of 2026. Nscale has been identified by Nvidia as one of the cloud partners expected to deploy the architecture.
Designed to consolidate GPUs, networking, storage, and power systems into unified AI platforms, Vera Rubin offers distinct architectural advantages. For Anthropic, access to the newer hardware stack could provide additional computing capacity as its models become more complex and demand for AI applications increases.
The scale of the commitment highlights the extraordinary capital requirements fueling the current AI infrastructure buildout. At $45 billion, the six-year arrangement ranks among the largest disclosed allocations by an AI company toward dedicated computing capacity. It also reinforces the centrality of power logistics: the reported 460-megawatt requirement is equivalent to the continuous electricity demand of a large number of households, further intertwining electricity generation, transmission, and data center construction with AI expansion.
For West Virginia, the project could further establish the state as an emerging destination for energy-intensive AI infrastructure.
The agreement remains contingent on the completion of site development and equipment installation. Consequently, observers will closely monitor construction timelines and the rollout schedule for Vera Rubin systems at the Monarch Compute Campus.
Ultimately, the deal secures a substantial long-term compute pipeline for Anthropic while marking a pivotal milestone for Nscale and Nvidia in constructing the foundational infrastructure required to sustain the next phase of global AI expansion.