NVIDIA pushes Vera Rubin deeper into AI infrastructure

NVIDIA said its Vera Rubin NVL72 platform has entered a mass production ramp-up, with a supply chain spanning more than 350 factories across 30 countries and over 300 partners involved. The company said Vera chips had already been delivered to customers including OpenAI, Anthropic and SpaceX, while CoreWeave, Google Cloud, Microsoft Azure and Oracle Cloud Infrastructure are deploying related racks.

The Vera CPU is aimed at agentic AI workloads, where systems plan tasks, invoke tools, run code, retrieve data and orchestrate actions in real time. NVIDIA argues that these workloads make CPU latency, memory access and single-threaded performance more important, and says Vera includes 88 self-designed Olympus cores and 1.2TB/s of memory bandwidth.

NVIDIA is also framing Vera as part of a broader vertical integration strategy rather than a standalone CPU launch. By combining Vera CPUs, Rubin GPUs, networking chips and software into rack-scale systems, the company is trying to capture more of the AI server value chain. That approach challenges AMD and Intel by targeting high-growth AI infrastructure workloads before expanding into broader server CPU use cases.