Nvidia's Vera Rubin platform reaches partners as chief executive touts performance per watt gains

9:20 AM · July 22, 2026

Nvidia detailed how its Vera Rubin platform is delivering improved performance per watt and lower per token costs for cloud partners worldwide, according to the company's own blog, as production continues ramping toward a planned fall shipping window across eight cloud partners including Microsoft Azure, CoreWeave, Google Cloud, and Oracle Cloud Infrastructure. The platform combines Nvidia's Vera CPU with its Rubin generation GPUs into a unified rack scale system, continuing Nvidia's shift toward selling complete integrated systems rather than individual chips as the primary unit of competition in AI infrastructure. Nvidia frames the performance per watt improvements as directly responsive to the power constraints increasingly shaping AI infrastructure decisions industry wide, as electricity availability rather than capital alone increasingly limits how much AI compute companies can practically deploy regardless of budget. The efficiency gains matter most for inference workloads run continuously at scale, where even modest per token cost reductions compound significantly over the billions of queries large AI services now handle daily, making the platform's efficiency claims as commercially relevant to Nvidia's cloud partners as its raw performance numbers. The update continues a pattern of incremental technical disclosures from Nvidia building toward the platform's formal fall launch, each aimed at reinforcing customer confidence in Vera Rubin's readiness ahead of full scale deployment.