Nscale and Microsoft launch an AI campus in West Virginia with 430,000 NVIDIA GPUs

Nscale announced a letter of intent with Microsoft for 1.35 GW of AI capacity for the new Monarch campus in West Virginia. The project was presented on March 16, 2026 at GTC: it centers on approximately 430,000 NVIDIA Vera Rubin GPUs, with initial deliveries scheduled for the end of 2027.

Scale of the Deal

This is not simply a purchase of chips, but rather a long-term scheme in which Nscale will build and operate infrastructure under Microsoft's load. The company speaks of a multi-year agreement for compute services and long-term data center lease, with deployment occurring in multiple phases. Formally, this is still a letter of intent, not a final contract. But even at this stage, the declared parameters look like one of the largest announcements on the AI infrastructure market in 2026. 1.35 GW of computing power in an AI context is a scale already comparable not to a single data center, but to an entire industrial complex.

At the heart of the project will be NVIDIA Vera Rubin NVL72 architecture and the NVIDIA Vera Rubin DSX AI Factory reference design. Nscale calls this the first major commercial deployment of such a configuration. If the project reaches its stated volume, Microsoft will get a dedicated site for training and inference of large models, and Nscale will gain the status of one of the key operators of the new wave of GPU factories.

Site and Energy

The campus is being built in Mason County, West Virginia, on the Monarch Compute Campus site. Nscale simultaneously announced the acquisition of AIPCorp, which includes this parcel. The site size is up to 2,250 acres, and its main distinction is that the project is described as the first state-certified AI microgrid in the USA. In other words, it is a data center designed from the ground up around its own power generation, rather than waiting for connection to an external grid over several years.

• approximately 430,000 NVIDIA Vera Rubin GPUs
• 1.35 GW of power for Microsoft's needs
• 2 GW of own generation capacity by the first half of 2028
• potential to scale beyond 8 GW on a single site

Power generation is handled by a partnership with Caterpillar: Nscale plans to deploy G3500 series gas generators and reach 2 GW of on-site generation in the first half of 2028. The campus is planned to launch independently of the local power grid to avoid waiting for scarce grid connections and not shifting load to existing consumers. At the same time, the project includes the possibility of future grid connection and even reverse power supply. Nscale also speaks of plans for carbon capture and claims that the site will use less water with no impact on municipal water supply, though such claims still need to be verified in practice.

Why This for Nscale

For Nscale itself, this is not an isolated deal, but part of a much larger bet on vertical integration. A week before the announcement, the company closed a Series C round at 2 billion dollars with a valuation of 14.6 billion dollars, and on March 16, 2026, announced not only a letter with Microsoft, but also the acquisition of AIPCorp. In parallel, Nscale is creating an Nscale Energy & Power division in Houston. The logic is simple: in AI infrastructure, it's no longer enough to simply order GPUs—you need to control land, energy, construction, capacity deployment, and operational management of the entire stack.

"Our approach is to combine our own data centers, leasing, and strategic partnerships,"

Microsoft stated. The companies already have a shared history: Microsoft was already a client of Nscale at the Narvik data center in Norway. The new campus in the USA brings the partnership to another level—closer to Ashburn in Virginia and Chicago, that is, to major hubs of cloud and network infrastructure. For Microsoft, this is a way to quickly increase AI capacity through a mix of owned facilities, leasing, and external partners. For Nscale, it's a chance to establish itself in the world's largest AI market and demonstrate that it can sell not only GPU time, but also megawatts.

What This Means

The Nscale and Microsoft deal shows that the race in AI is increasingly shifting from models and chips to land, electricity, and infrastructure deployment timelines. The winner is no longer the one who simply found more GPUs, but the one who managed to assemble a complete package in advance: a site, energy, network, financing, and a customer. That said, it's important to remember that we're currently talking about a letter of intent. If deliveries indeed start at the end of 2027, and power generation reaches the first 2 GW in the first half of 2028, Monarch could become one of the key AI sites in the USA.