NVIDIA Opens Vera Rubin Architecture to Scale AI Factories Worldwide

NVIDIA contributed Vera Rubin rack and compute-tray designs to the Open Compute Project through the open NVIDIA MGX architecture. The announcement was made at the OCP Global Summit at the San Jose McEnery Convention Center.

The updated MGX 1.2 specification includes blind-mate liquid cooling, interoperable rack-scale interconnects and a 5,000A liquid-cooled busbar. The architecture supports rack deployments of up to 250kW.

Open standards for AI factories

NVIDIA, AMD, Google and Meta published Open Data Center Specification v0.7, covering AI-factory design from power architecture and cooling to physical layout. A separate OCP Grid Resilience Specification from Google, Meta, Microsoft and NVIDIA defines how infrastructure should respond to utility disturbances and return to near-full load in less than two seconds.

NVIDIA also released open-source DSX AI Factory components for network configuration, cluster validation, rack management and bare-metal provisioning. They include NVIDIA Config Manager, Cluster Readiness Engine, Rack Management Service and Cluster API Provider for NVIDIA Infrastructure Controller.

Partner designs for Vera Rubin

Meta and NVIDIA announced open networking advances involving Spectrum-X Ethernet SuperNIC and Spectrum-6. NVIDIA and Cisco announced the Ultra-Dense Small Form Factor Pluggable standard for 400Gb/s networking fabrics, while NVIDIA and Google announced Falcon Transport support for future ConnectX-10 SuperNICs and BlueField DPUs.

Visionbay introduced K-1, an AI-factory platform built around 800 VDC distribution and designed for future Vera Rubin deployments. The design combines NVIDIA’s 800 VDC reference architecture with power and thermal technologies from Vertiv.

The Vera CPU rack specification extends MGX beyond accelerated computing. It is designed for tool calling, orchestration and sandboxed execution, while Vera Rubin NVL72 is positioned for large-scale training and inference.