Nvidia is accelerating its CPU product roadmap to further solidify its market position as a full-stack systems provider.Ian Buck, vice president of high-performance computing and ultra-large-scale systems business, revealed,The company has shipped "hundreds of thousands" of Grace standalone servers, with cumulative shipments exceeding 2.5 million Grace chips.
Grace is Nvidia's first independent architecture CPU for data centers. It is designed based on Arm architecture and is mainly used in high-performance computing and AI infrastructure fields, and provides independent server solutions.
The market expectations of the new generation of Vera CPU are significantly higher than those of the previous generation. In response to the growth in demand brought by Agentic AI, NVIDIA is collaborating with more than 300 partners around the world to ensure adequate supply of Vera chips.
Andrew Bell, senior vice president of hardware engineering, told the media that Nvidia has about 12 manufacturing partners and has a production capacity goal of producing 1,000 Vera racks per day.Based on an estimated daily production capacity of 1,000 racks, the revenue potential of this business in a single quarter can reach at least US$630 billion.
