Supermicro Rolls Out Nvidia Vera Rubin NVL72 Racks for AI Workloads

Supermicro has started shipping Nvidia's latest AI platform, Vera Rubin NVL72 racks, to customers. These racks combine 72 Rubin GPUs and 36 Vera CPUs, making them a powerful tool for artificial intelligence workloads.

The systems are part of Supermicro's broader push to supply complete AI data-center infrastructure, including servers, cooling, networking, and deployment infrastructure. The company aims to provide customers with production-ready AI systems spanning everything from the server rack to the cooling system.

Supermicro's larger configuration connects 16 NVL72 racks, creating a system with 1,152 Rubin GPUs and 331 TB of HBM4 memory. This scale is part of Supermicro's strategy to capture more of the spending around AI data centers through its infrastructure offerings.