Anthropic Signs $45 Billion Nscale Compute Deal
Anthropic has agreed to spend $45 billion on AI computing capacity from Nscale’s West Virginia data center campus. The six-year agreement covers about 460 megawatts of power capacity.
The deal comes as Anthropic rapidly expands computing capacity for Claude and its AI coding tool, Claude Code. Meanwhile, the company is preparing for a potential initial public offering, according to reports.
Nscale plans to deploy Nvidia’s next-generation Vera Rubin systems for the Anthropic workload. The hardware is expected to begin coming online late next year.
Nscale Campus Will Supply Frontier AI Capacity
The computing agreement will use Nscale’s planned data center development in Mason County, West Virginia. The campus covers up to 2,250 acres and has a planned power runway exceeding 8 gigawatts.
Furthermore, Nscale is developing the site as a large-scale AI infrastructure campus. Its first phase targets 1.35 gigawatts of AI computing capacity. Therefore, Anthropic’s 460-megawatt commitment represents a significant portion of the initial planned capacity.
The arrangement also highlights the growing importance of power availability in frontier AI. As demand for model training and inference rises, AI companies increasingly need dedicated access to electricity, GPUs, and data center infrastructure.
Nscale acquired the Monarch Compute Campus earlier this year through its acquisition of American Intelligence & Power Corporation. The company said the site includes the United States’ first state-certified AI microgrid.
Moreover, Nscale is building infrastructure to support long-term power availability. That approach could help Anthropic secure computing resources as demand for its models continues to increase.
Anthropic Accelerates Its Compute Expansion
The Nscale agreement adds to Anthropic’s growing list of major infrastructure commitments. Previously, the company announced an agreement with SpaceX covering more than 300 megawatts and over 220,000 Nvidia GPUs at the Colossus 1 data center.
In addition, Anthropic has announced major arrangements involving Amazon, Google, Broadcom and Fluidstack. The company says these agreements will expand its access to GPUs, custom AI chips and other computing resources.
Anthropic’s infrastructure spending reflects strong demand for Claude. The company reported that its revenue run rate had exceeded $47 billion earlier this month. Furthermore, its latest $65 billion funding round valued Anthropic at $965 billion.
The company has also projected substantial future revenue growth. According to reporting, Anthropic expects 2028 revenue of roughly $190 billion to $200 billion.
Consequently, securing reliable compute has become central to Anthropic’s growth strategy. The company must support model training, inference, and growing enterprise usage while competing with other major AI developers.
AI Infrastructure Becomes Strategic Asset
For Nscale, the agreement strengthens its position as a major AI infrastructure provider. The company is also preparing for a possible public listing while expanding its data center operations in the United States and Europe.
However, the scale of the West Virginia project creates significant execution requirements. Nscale must coordinate power generation, data center construction, networking and advanced chip deployment before the planned capacity becomes operational.
At the same time, Anthropic gains greater visibility into future availability of computing resources. This matters because frontier AI development increasingly depends on long-term infrastructure planning rather than short-term cloud capacity.
Ultimately, the agreement shows how AI companies are securing computing resources years before demand peaks. As AI models grow larger and enterprise adoption expands, access to power and specialized hardware is becoming a strategic asset alongside capital and talent.
