NVIDIA Launches Vera CPU to Target $200 Billion Market

NVIDIA has launched the Vera CPU, the first processor purpose-built for agentic AI, as part of a strategic shift to become a full-stack semiconductor platform company. The company is targeting a total addressable market estimated at $200 billion and expects nearly $20 billion in standalone CPU revenue this year. To meet demand, NVIDIA is collaborating with over 300 global partners and a dozen manufacturing partners to produce up to 1,000 Vera Rubin racks per day.

This expansion accompanies the volume ramp of the Vera Rubin NVL72 AI system, a rack-scale platform integrating compute, networking, and cooling. The system is currently deployed at Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and CoreWeave Inc. CoreWeave Inc. benchmarks show the platform delivers 10 times more tokens per second per megawatt than the Grace Blackwell NVL72. The system utilizes a liquid-cooling method designed to save water in AI factories and reduces overall assembly time.

In Europe, the platform supports a multibillion-dollar infrastructure partnership between Microsoft and Mistral AI. Additionally, the London-based startup Ineffable Intelligence is utilizing Google Cloud's A5X instances, which are based on the Vera Rubin architecture, to test infrastructure for its superlearners.