On September 30, 2025, Crusoe announced that the first phase of the Abilene, Texas data center campus - the flagship site of the Stargate Project - was operational, running on Oracle Cloud Infrastructure with NVIDIA GB200 GPUs. Construction had begun in June 2024, the first two buildings were energized within about a year, and Oracle delivered the initial GB200 racks in June 2025, after which the campus began running early training and inference workloads.
The campus is being built out at the Lancium Clean Campus to eight buildings totaling roughly 4 million square feet and a planned 1.2 gigawatts of power, supporting hundreds of thousands of GPUs on a single network fabric. Each building was designed to hold up to 50,000 of NVIDIA’s GB200 NVL72 units, and Crusoe described a construction workforce scaling from around 2,000 to nearly 5,000 workers on site.
Abilene was the concrete first realization of Stargate, the $500 billion infrastructure program announced by OpenAI, Oracle, and SoftBank in January 2025. Seeing the flagship campus actually energized and running production AI workloads turned the program’s headline number into a physical fact, and demonstrated the scale - gigawatts of power and millions of square feet - that a single frontier-AI training site had come to require.