As federal agencies modernize IT systems and prepare for artificial intelligence (AI) compute demands, Supermicro is working with NVIDIA to expand efforts to help the government rapidly deploy purpose-built “AI factory” infrastructure.
At NVIDIA’s GTC conference in Washington, D.C., industry experts said demand for AI factory designs is accelerating across government and regulated industries, aimed at improving data control and mission-specific performance. The growth of enterprise AI adoption, open-source models, and sovereign AI initiatives fuels the trend.
During the event, NVIDIA introduced the AI Factory for Government reference design, developed with a coalition of software and services providers to help agencies and other regulated organizations build and operate secure, scalable AI platforms.
Randy Kreiser, senior storage architect at Supermicro, said the company “helps the United States government build AI factories and the infrastructure behind AI.” Supermicro announced an expanded collaboration with NVIDIA during the event, emphasizing that the company develops, constructs, and validates in the United States, ensuring the solutions are Trade Agreements Act-compliant and Buy American Act-capable.
Supermicro’s “data center building block” approach integrates compute, cooling, power, network, and storage to accelerate deployment of complete AI systems, Kreiser said. “For example, a customer wants to deploy a SuperPod,” he said. “We can build it in our factory in San Jose and test the whole thing under liquid cooling … and allow the customer to touch it – either remotely or on site.”
Supermicro plans to offer the NVIDIA Vera Rubin NVL144 and NVL144 CPX systems beginning in 2026. In addition, the company introduced a new compact/high-density 2OU NVIDIA HGX B300 8-GPU system with up to 144 GPUs per rack and an expanded portfolio featuring a Super AI Station based on NVIDIA GB300 and the new rack-scale NVIDIA GB200 NVL4 HPC solutions. The company said the systems align with NVIDIA’s full-stack AI Factory for Government reference design and expand Supermicro’s AI solution portfolio.
The Super AI Station can operate on premises for AI model training, fine-tuning, application and algorithm prototyping and development. Designed for agencies without access to traditional server infrastructure or large-scale cloud services, the platform offers a self-contained, secure compute environment for missions that face constraints in availability, cost, privacy, or latency requirements.
The growing focus on AI factory architectures aligns with federal priorities to establish sovereign, secure, and scalable AI infrastructure. The Office of Management and Budget’s AI guidance and the National Science Foundation’s National AI Research Resource (NAIRR) Pilot both underscore the need for domestic compute capacity and trusted data environments to advance responsible AI development.