Abilene Data Center

Overview 

This confidential data center will power one of the world's largest graphics processing unit (GPU) clusters, driving advancements in artificial intelligence (AI). 

Purpose-built to support AI workloads, the campus features high-density data halls and integrates local renewable energy sources as part of its long-term sustainability strategy. At full capacity, each building will support up to 100,000 GPUs on a single integrated network fabric, setting a new benchmark for AI training and inference infrastructure. 

The design is optimized for direct-to-chip liquid cooling and rear-door heat exchangers while maintaining flexibility for air cooling. Initial scope includes two buildings totaling 200 megawatts (MW) of information technology (IT) capacity, with the site designed to scale up to 1 gigawatt (GW). Each single-story, 474,000-square-foot building houses four 106,000-square-foot data halls connected by a central utility hall, each capable of accommodating 500 racks and up to 25 MW of IT load. 

Our Approach 

As the design-build mechanical and plumbing contractor, we are leveraging our in-house engineering and VDC teams to rapidly produce shop and spool drawings. This integration was critical to meeting the project's aggressive schedule and speed-to-market goals. The repetitive building design allows for concurrent work in multiple halls, enabling each to progress from start to energizing in roughly 10 months. 

We are integrating prefabrication and innovative construction techniques to improve efficiency, maintain quality, and enhance safety. Recognizing the site's high water demand, our team designed a high-flow reverse osmosis (RO) flushing system projected to reduce chilled water pipe flushing water use by 75 percent. This benefits both the client and the city of Abilene by significantly reducing the need for municipal water supply. 

Workforce projections show that peak construction will involve more than 900 of our employees onsite, including about 200 field supervisors. 

Result 

This project broke ground in summer 2024, with the first 28 MW data hall wing and core delivered in just 12 months. Within 6 months of groundbreaking, the scope expanded from 2 buildings to 8, designed for a total of 1.2 GW of computing power and capable of operating up to 50,000 Nvidia GB200 NVL72 GPUs per building. 

The additional six buildings began construction in March 2025 and are scheduled for completion by mid-2026. The accelerated construction schedule and resource-driven approach have positioned the client to bring industry-leading AI infrastructure online faster while delivering lasting operational benefits.