Anthropic Signs $19 Billion Data Center Lease With TeraWulf in Kentucky
Anthropic has signed a 20-year, $19 billion lease for a large AI data center campus in Kentucky with energy company TeraWulf. The deal is one of the largest data center commitments in history and reflects the massive infrastructure investment required to train and run frontier AI models.

Anthropic has signed a 20-year, $19 billion lease for an AI data center campus in Kentucky with TeraWulf, an energy and infrastructure company. The deal is one of the largest data center commitments in history and signals the scale of investment that AI companies now require to compete at the frontier.
The Kentucky campus will provide Anthropic with dedicated computing capacity to train and run its Claude family of AI models. TeraWulf will build and operate the facility, which will draw on a mix of power sources including nuclear and renewable energy. Anthropic has made clean energy a priority in its infrastructure decisions, citing the environmental impact of large-scale AI computing.
The deal comes as AI companies race to secure long-term access to computing power. Training the most capable AI models requires enormous amounts of electricity and specialized hardware, and demand for both is outpacing supply. Companies that lock in capacity now are betting that the competitive advantage will justify the cost.
Anthropic is not alone in making large infrastructure commitments. Amazon, which has invested heavily in Anthropic, is pursuing a $25 billion bond sale to fund its own AI infrastructure expansion. Google reported a 37 percent increase in electricity consumption in 2025, driven almost entirely by AI data center growth.
The Kentucky deal also reflects a broader trend of AI companies moving infrastructure out of coastal tech hubs and into states with lower energy costs and more available land. Kentucky has been actively courting data center investment, offering tax incentives and working to expand its power grid capacity.
Anthropic is also in talks with Samsung about manufacturing custom AI chips, a move that would reduce its dependence on Nvidia, which currently supplies the graphics processing units that power most AI training workloads. Custom chips could give Anthropic more control over its computing costs and performance over time.


