---
title: "Nvidia opens its Vera Rubin rack designs to the Open Compute Project at OCP Global Summit"
url: https://anawords.com/story/nvidia-vera-rubin-rack-designs-ocp-mgx-1-2
published: 2026-10-13T11:07:02+05:30
updated: 2026-10-13T11:07:02+05:30
section: Chips & Compute
author: Mayank Sahu
checked: sources checked by AI, approved by editor
version: 1
---

# Nvidia opens its Vera Rubin rack designs to the Open Compute Project at OCP Global Summit

The updated MGX 1.2 specification supports racks up to 250kW. Nvidia also announced Meta’s adoption of Spectrum-6 switches and a new optics format with Cisco.

**In short:** Nvidia said on 12 October 2026 at the OCP Global Summit in San Jose that it has contributed its Vera Rubin rack and compute-tray designs to the Open Compute Project through an updated MGX 1.2 specification supporting racks of up to 250kW.

## What Nvidia shared

Nvidia said on 12 October 2026, at the Open Compute Project (OCP) Global Summit in San Jose, that it has contributed its Vera Rubin rack and compute-tray designs to OCP through an updated MGX 1.2 specification.[^1][^2] OCP is an industry group that publishes open hardware designs for data centres.

Nvidia says MGX 1.2 supports racks of up to 250kW and adds rack-scale blind-mate liquid cooling, a liquid-cooled 5,000A busbar and support for standard 19-inch equipment.[^1]

## Shared specs for AI data centres

- **Open Data Center Specification v0.7:** published by Nvidia, AMD, Google and Meta as a common blueprint for power, cooling and layout of AI data centres.[^1]
- **OCP Grid Resilience Specification:** published by Google, Meta, Microsoft and Nvidia. It sets out how AI data centres should behave during grid disturbances, including returning to near-full load in under two seconds.[^1][^2]

## Networking and optics

Nvidia says Meta is adopting its Spectrum-6 102.4Tb/s Ethernet switches, which have twice the switching capacity of Spectrum-4.[^1][^2] With Cisco, Nvidia unveiled USFP, a pluggable optics format that it says offers a path to 400Gb/s per lane.[^1][^2]

## Open-source software

Nvidia released open-source components for managing AI clusters, covering switch configuration, cluster stress testing, rack firmware and bare-metal provisioning.[^1]

## Why it matters

Open designs let server makers and data centre operators build compatible hardware without starting from scratch. The performance and efficiency figures here are Nvidia’s own.

## Same event

Intel launched its GPU D270 inference card at the same summit, which we [cover separately](/story/intel-gpu-d270-crescent-island-ai-inference-card).

## Fact ledger

- [Confirmed] Nvidia contributed Vera Rubin rack and compute-tray designs to OCP through MGX 1.2. (source 1,2)
- [Confirmed] MGX 1.2 supports racks of up to 250kW. (source 1,2)
- [Confirmed] Meta is adopting Nvidia Spectrum-6 Ethernet switches. (source 1,2)
- [Company claim] USFP optics offer a path to 400Gb/s per lane. (source 1,2)
- [Not known] Which server makers will build to the new designs, and when.

## Sources

1. [Nvidia: OCP Global Summit, Vera Rubin and AI factories](https://blogs.nvidia.com/blog/ocp-global-summit-vera-rubin-scale-ai-factories/) — official
2. [AI Weekly: Nvidia opens Vera Rubin rack blueprints via MGX 1.2](https://aiweekly.co/alerts/nvidia-opens-vera-rubin-rack-blueprints-via-mgx-12-at-ocp)