Photo: Konstantin Lanzet / Wikimedia Commons / CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0) The GPU-as-a-Service company is staking its claim in Pennsylvania's increasingly crowded AI infrastructure corridor with a major land and energy acquisition.
Alpha Compute Corp is spending $55 million to acquire land and gas rights in Pennsylvania, where it plans to build a 200-megawatt data center campus. The deal marks one of the more aggressive moves by the Nasdaq-listed company as it races to expand its physical footprint in the AI infrastructure space.
Alpha Compute, which trades on Nasdaq under the ticker ALP, positions itself as a GPU-as-a-Service provider specializing in confidential computing. The company is headquartered in the British Virgin Islands but operates offices across New York, Los Angeles, Miami, Amsterdam, and Toronto.
The Pennsylvania acquisition comes shortly after the company signed a $32.2 million compute off-take agreement with a frontier AI lab. That two-year deal is expected to generate annual recurring revenue of $16.1 million, powered by 504 NVIDIA B200 GPUs.
No specific location within Pennsylvania, construction timeline, or detailed power sourcing plans have been disclosed yet. But the combination of land and gas rights in a single transaction points toward a vertically integrated approach where the company controls both the facility and at least some of its energy supply.
Pennsylvania’s AI infrastructure gold rush #
Pennsylvania has become one of the hottest markets in the US for AI and hyperscale data center development. CoreWeave and Amazon are among the companies that have made substantial commitments to develop facilities in the state. Total investment pledges in Pennsylvania’s data center sector are projected to reach tens of billions of dollars.
The GPU-as-a-Service business model #
Rather than leasing rack space or selling colocation services, Alpha Compute focuses on delivering GPU compute power as a managed service. Customers rent access to high-performance NVIDIA hardware without needing to own, maintain, or house the physical equipment themselves.
The $32.2 million off-take agreement with a frontier AI lab demonstrates that this model can attract serious customers. The contract deploys 504 NVIDIA B200 GPUs, one of NVIDIA’s latest data center GPUs designed specifically for large-scale AI training and inference.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our