# Nvidia Ships Vera Rubin NVL72 With 72 GPUs and 36 CPUs

By Simon Yoon

Canonical URL: https://www.tokenpost.com/news/technology/60609
Published: 2026-11-18T15:58:59.000Z
Updated: 2026-11-18T15:58:59.000Z
Section: Technology

> The platform delivered up to twice its predecessors’ per-GPU performance as of Nov. 16. Earlier this fall, AI executives raised concerns about cyberattacks and model control.

Nvidia is shipping its Vera Rubin NVL72, a liquid-cooled AI computing rack with 72 Rubin GPUs and 36 Vera CPUs, expanding access to a system built for large-scale AI training.

The NVL72 is a rack-scale system, not a standalone processor. Its GPUs and CPUs connect within one NVLink domain, Nvidia’s high-speed interconnect. Nvidia called the rollout “the fastest product ramp in the company’s history.”

As of Nov. 16, the Vera Rubin platform delivered up to twice the per-GPU performance of its predecessors in MLPerf Training v6.1, an AI training benchmark suite. It led in six of seven categories. The performance comparison is per GPU, while the category tally counts results across the suite.

Deployment has also advanced through Nvidia’s partners. Supermicro began shipping integrated liquid-cooled racks Sept. 23, CoreWeave announced early availability Sept. 30, and Oracle Cloud Infrastructure did so Oct. 27. CoreWeave has put NVL72 systems into production and opened customer access through its cloud platform. [CoreWeave’s production rollout](<https://www.tokenpost.com/news/technology/25883>) made Cognition the first customer to run production workloads on the platform.

That deployment produced a separate performance comparison. Cognition tested software-engineering AI inference tasks and measured total token throughput, recording up to 4.8 times the throughput of Nvidia’s GB200 NVL72. This was a different metric and comparison from MLPerf’s per-GPU training results.

AI safety concerns had been voiced earlier this fall. Anthropic CEO Dario Amodei called for slower advances in the most powerful AI models in an essay published Sept. 12, warning that the systems could enable cyberattacks and become harder for their creators to control. OpenAI CEO Sam Altman voiced support for Amodei’s position that day. OpenAI paused its largest planned frontier training run Sept. 20 and resumed it Oct. 1 after adding security measures.



## Links in this article

- [CoreWeave’s production rollout](https://www.tokenpost.com/news/technology/25883)
