Super Micro Computer is highlighting an expanded collaboration with Nvidia at the Nvidia GTC event in Washington, D.C., focusing on Artificial Intelligence infrastructure designed for federal customers. The company plans to deliver next-generation Nvidia Artificial Intelligence platforms in 2026, specifically the Nvidia Vera Rubin NVL144 and Vera Rubin NVL144 CPX, signaling a roadmap aimed at meeting stringent government requirements.
In addition to its forward-looking platform plans, Supermicro introduced U.S.-manufactured, Trade Agreements Act compliant systems tailored for public sector procurement. The portfolio features a high-density 2OU Nvidia HGX B300 8-GPU system that can scale to as many as 144 GPUs per rack. Supermicro is also expanding with a Super AI Station based on the Nvidia GB300 and new rack-scale solutions built on the Nvidia GB200 NVL4 for high performance computing, underscoring a range that spans from workstation-class development to large-scale deployment.
The company framed its U.S.-based manufacturing as a key differentiator for federal Artificial Intelligence deployments. According to president and CEO Charles Liang, with headquarters, manufacturing, and R&D co-located in San Jose, California, Supermicro is positioned to deliver first-to-market systems that are developed, constructed, validated, and manufactured domestically for American federal customers. Liang characterized the long-running partnership with Nvidia, also located in Silicon Valley, as central to establishing Supermicro’s role in American Artificial Intelligence infrastructure.
Presented as part of its presence at Nvidia GTC, the announcements emphasize compliance, density, and a broadening product set aligned to government needs. By combining TAA-compliant, U.S.-manufactured systems with a planned path to next-generation Vera Rubin platforms and GB300 and GB200-based offerings, Supermicro is pitching an end-to-end approach intended to satisfy the performance and procurement requirements of federal agencies.
