Vera Rubin Platform

Active
Nvidia
Nvidia

Overview

Nvidia's next-generation seven-chip platform - Vera CPU, Rubin GPU, NVLink 6 switch, ConnectX-9, BlueField-4, Spectrum-6, and the Groq 3 LPU integrated after a December 2025 licensing deal. Rubin NVL72 racks train mixture-of-experts models with a quarter of the GPUs Blackwell needs and deliver up to 10x inference throughput per watt at a tenth the cost per token.

Reached full production and began shipping to customers (OpenAI, CoreWeave, Google Cloud, Microsoft Azure, Meta, Dell) on July 21, 2026; OpenAI plans to deploy it at scale in Q3 2026

Key Specs

Training efficiency

MoE training with 1/4 the GPUs of Blackwell

Inference

Up to 10x throughput/watt at 1/10 cost per token

Partner availability

Shipping since July 21, 2026; CoreWeave's early NVL72 racks delivering ~10x the token throughput of the prior generation