CoreWeave Powers Up Nvidia Vera Rubin NVL72 System for Record-Breaking AI Performance
Dell Technologies has delivered the world's first Nvidia Vera Rubin NVL72 racks to CoreWeave, marking a major milestone in the development of high-performance AI computing. The Vera Rubin platform is a significant upgrade over its predecessor, the Blackwell platform, offering up to 10x better token throughput per megawatt on complex workloads.
The NVL72 rack-scale system features 72 Rubin GPUs and 36 Vera CPUs connected by NVLink 6 fabric, capable of delivering 3.6 exaFLOPS of NVFP4 inference capability. This represents a substantial improvement over previous systems, with Dell's integration speed also noteworthy - the team completed the system from physical delivery to production readiness in under 6.5 hours.
The partnership between CoreWeave and Dell has been ongoing for prior NVL system generations, and the news has already had an impact on CoreWeave's stock price, which spiked by roughly 14% following the announcement. With Nvidia expecting the Vera Rubin platform to enter full production by July 2026, deployments are planned across major cloud providers including OpenAI, Google Cloud, and Microsoft Azure.