Nscale Targets 1.35GW AI Deployment with Microsoft at Monarch AI Campus

Key Takeaways

• Nscale to deliver 1.35 GW AI compute for Microsoft.
• Campus can scale power to 8 GW via on‑site microgrid.
• NVIDIA Vera Rubin GPUs power the deployment.
• Caterpillar to supply 2 GW natural‑gas generators by 2028.
• Project aims to offset emissions with carbon sequestration.

Summary

Nscale has signed a letter of intent with Microsoft to install 1.35 GW of AI compute at the Monarch AI campus in West Virginia, using NVIDIA’s next‑generation Vera Rubin NVL72 GPUs. The campus, acquired through AIPCorp, already hosts a state‑certified AI microgrid that can scale to over 8 GW of power. A partnership with Caterpillar will add 2 GW of natural‑gas generation by mid‑2028, accelerating the rollout. Deployment will begin in late 2027 and become one of the world’s largest dedicated AI compute facilities.

Pulse Analysis

The AI boom is reshaping data‑center economics, with analysts projecting more than 150 GW of AI‑specific capacity needed by 2030. Nscale’s agreement with Microsoft positions the Monarch campus as a strategic counterweight to overseas hyperscalers, offering a U.S.‑based, low‑latency environment for training massive models. By anchoring the deployment on NVIDIA’s Vera Rubin architecture, the facility promises unprecedented throughput while maintaining energy efficiency, a critical factor as model sizes continue to expand.

Beyond raw compute, the campus’s engineering focuses on power resilience. The on‑site microgrid, already certified for AI workloads, can be expanded to eight gigawatts, while Caterpillar’s G3500 natural‑gas generators will provide two gigawatts of rapid‑deployment capacity by 2028. This hybrid approach shortens construction timelines and reduces reliance on the regional grid, ensuring consistent performance for mission‑critical AI inference and training tasks. Proximity to major fiber corridors linking Ashburn and Chicago further reduces latency for enterprise customers.

Environmentally, Nscale is leveraging carbon‑sequestration opportunities in West Virginia and designing the facility to operate independently of municipal utilities, mitigating grid strain and rate‑payer impact. The project illustrates a broader industry shift toward modular, sustainable data‑center models that can be scaled quickly to meet surging AI demand. As competitors scramble to secure power‑intensive sites, Monarch’s blend of high‑density GPU clusters, on‑site generation, and green initiatives could become a template for future AI infrastructure investments.