Vera Rubin Delivers 30x Throughput Boost for Agentic AI Workloads

Nvidia's Vera Rubin platform has been touted as a game-changer for agentic AI workloads, delivering up to 30x more throughput than its predecessor, the GB300 NVL72.

The rack-scale system, which combines CPUs, GPUs, networking, and software, was showcased at the AI Infrastructure Summit 2026, where Nvidia's vice president of hyperscale and high-performance computing, Ian Buck, presented the results.

Buck described Vera Rubin as a 'fungible' full-stack platform that can support open and proprietary models across training, inference, and other accelerated workloads. He also noted that agentic AI processing can require up to 100 times more compute than conventional chatbot interactions.

Nvidia co-designed the system's CPUs, GPUs, networking, and key-value cache management to address these demands. The platform has already entered full production, with Nvidia announcing at the end of May 2026 that it was available for customers.