The new NVIDIA Vera Rubin NVL72 platform delivers 10x more tokens per megawatt than Blackwell, aiming for the lowest token cost in the agentic era. Cloud providers including CoreWeave, Google Cloud, Microsoft, and Oracle Cloud are deploying the Vera Rubin NVL72. Benchmark results from DeepInfra also show the NVIDIA Vera CPU is over 2x faster and supports more concurrent AI agents than other CPUs.
NVIDIA's Vera Rubin platform significantly lowers inference costs for agentic AI, driving hyperscaler compute demand and increasing the performance gap against competitors.