Nvidia Vera Rubin enters full production with 10x efficiency gains for AI inference workloads
Nvidia's Vera Rubin platform has reached full production, with CoreWeave reporting 10 times more tokens per watt compared to the previous GB200 NVL72 system. Major customers including OpenAI, Microsoft Azure, Google Cloud, and Meta are deploying the next-generation AI infrastructure, which combines 72 Rubin GPUs with custom Vera CPUs to handle agentic AI workloads more efficiently.