Microsoft's first production NVIDIA Vera Rubin systems have arrived — and NVIDIA says the next-gen AI computing platform is now in full production.
CEO Satya Nadella shared a photo from a Microsoft data center showing the racks on site, just as NVIDIA confirmed Rubin is ramping up volume shipments.
Vera Rubin is NVIDIA's rack-scale AI platform after Blackwell. Each NVL72 cabinet packs 72 Rubin GPUs and 36 Vera CPUs. NVIDIA claims that compared to the GB200 NVL72, it cuts inference cost per million tokens to roughly one-tenth — a major step for training and running mixture-of-experts models.