Super Micro Computer expands the Data Center Building Block Solutions (DCBBS) portfolio with the new NVIDIA Vera CPU rack, a rack-scale computing platform fully liquid-cooled to speed up agentic AI, reinforcement learning, and big data workloads. With a density of 256 NVIDIA Vera CPUs (that provides 22,528 cores and 45,056 threads), 32 NVIDIA Blue Field-4 DPUs, and up to 22,500 agent execution sandboxes, this high-density system is designed to meet the requirements of agentic AI workloads while operating as a standalone system or as an adjunct to the NVIDIA Vera Rubin NVL72 platform. The new rack supports third-generation modular NVIDIA MGX systems and is built using Supermicro’s DLC-2 liquid-cooling technology, capable of functioning as a standalone system or as a companion system for the NVIDIA Vera Rubin platform, with support for power options including the 800 VDC sidecar rack with battery backup units (BBUs).
Also Read: Smarsh Enhances Compliance Suite to Mitigate AI, Cyber, and Third-Party Risks
“As enterprises scale agentic AI and reinforcement learning, supporting thousands of concurrent agent workflows requires a new class of high-density CPU infrastructure,” said Chris Marriott, vice president of enterprise platforms at NVIDIA. “NVIDIA Vera CPUs and BlueField-4 DPUs, integrated with Supermicro’s liquid-cooled DCBBS architecture, give customers the dedicated compute and hardware-based security to scale the execution of agentic workloads across their AI factories.” Powered by the native NVIDIA Open Shell Reference Architecture and hardware-telmetry with NVIDIA Sentry, the new platform enables policy enforcement and containment in real-time, serving as a guide for hyperscale and enterprise AI factories.