Nvidia says Vera Rubin enters deployment as Nebius completes early Finland test
Nvidia said Vera Rubin systems are in full production and were delivered in June to OpenAI, Anthropic and SpaceX, while Nebius tested an early NVL72 rack in Finland ahead of a planned 2026 rollout.
Nvidia said computing systems based on its Vera Rubin architecture are in full production and entering deployment, with Vera server CPUs delivered in June to OpenAI, Anthropic and SpaceX. Separately, Nebius said it completed and tested its first Vera Rubin NVL72 rack at its Mäntsälä, Finland data center on July 23, describing it as one of the earliest deployments of Nvidia’s next-generation AI supercomputer platform globally. The rack includes 72 Rubin GPUs, 36 Vera CPUs and Nvidia Spectrum-6 switches, and Nebius plans commercial availability through its AI Cloud and Token Factory services in the second half of 2026.
- Vera Rubin: Nvidia’s latest-generation AI computing architecture for data center systems, including Rubin GPUs and Vera CPUs.
- Vera Rubin NVL72: An Nvidia AI supercomputer rack configuration that includes 72 Rubin GPUs and 36 Vera CPUs.
- Spectrum-6 switches: Nvidia networking hardware used in high-performance AI data center deployments.