NVIDIA Opens Vera Rubin Rack Designs at OCP Summit

At the Open Compute Project Global Summit, NVIDIA contributed Vera Rubin rack and compute tray designs to OCP through an updated MGX 1.2 specification. The specification adds rack-scale blind-mate liquid cooling, an interoperable rack-scale interconnect, a liquid-cooled 5,000A busbar, and supports rack deployments up to 250kW with native 19-inch equipment support. A new Vera CPU rack design extends the architecture for agentic AI workloads such as tool calling and orchestration. Alongside the hardware contributions, NVIDIA released four open source software tools under its DSX AI Factory platform: NVIDIA Config Manager for switch configuration, a Cluster Readiness Engine for GPU cluster stress-testing, an NVIDIA Rack Management Service for firmware management, and a Cluster API Provider that reduces bare-metal-to-production cluster setup from days to hours. NVIDIA and Cisco also unveiled the Ultra-Dense Small Form Factor Pluggable standard, a transceiver module delivering 400Gb/s-per-lane networking with 16 full-duplex lanes and plans to scale to 32, in roughly the same footprint as the existing OSFP form factor. Meta announced adoption of NVIDIA Spectrum-6 switches in its N-IceCube switch running FBOSS, with Spectrum-6 offering 102.4Tb/s switching capacity and Spectrum-X Multiplane technology enabling AI factories with more than 100,000 GPUs. Startup Visionbay.ai introduced K-1, an 800VDC AI factory platform built for Vera Rubin deployments in partnership with Vertiv. AMD also announced its Helion rack at the same summit, an open design for agentic AI workloads supporting CPU-only or mixed configurations with up to 16 Instinct MI350P GPUs, with Dell, HPE, Cisco, and Supermicro planning availability from November.