Supermicro announces an end-to-end systems blueprint for NVIDIA’s Vera Rubin platform, described as an NVL4 design intended for converged high-performance computing (HPC) and AI workloads. The plan is presented as a “total solution” that scales in liquid-cooled racks and supports both AI processing and double-precision (FP64) simulation performance. According to the available reports, configurations can include up to 1,152 NVIDIA Rubin GPUs and 576 NVIDIA Vera CPUs within liquid-cooled racks, packaged for high-density deployment. The PR release also describes a 3.2MW scalable unit and notes that the design uses DLC-2 Direct Liquid Cooling, with a stated cooling capacity of 362 kW per rack. The offerings are framed as research-cluster-ready, with scalability aimed at systems of different sizes. While the details provided focus on the integration approach, the communication does not add independent performance results beyond the claim of native FP64 capability. Other sources provided repeat the same announcement without additional substantiating specifics.