Nvidia Ships Hundreds of Thousands of Grace CPUs, Shifts Focus to Agentic AI
Nvidia has announced the shipment of hundreds of thousands of its Grace standalone servers, primarily utilizing its CPU technology. This development comes as the company increasingly emphasizes its role in agentic data centers, where AI agents autonomously perform tasks. The company is rolling out Vera, its first custom CPU specifically designed for agentic AI applications. While the new Vera CPU targets advanced agentic workloads, Nvidia revealed that its previous generation Grace CPU design has already achieved significant mass deployment. These earlier Grace CPUs have been utilized for non-agentic tasks, demonstrating broad applicability of Nvidia's CPU architecture beyond the cutting-edge AI frontier. This strategic pivot in messaging highlights Nvidia's expanding ambitions in the data center market, moving beyond its traditional GPU dominance to encompass a more comprehensive CPU strategy.
Nvidia's strategic emphasis on its CPU offerings, particularly for agentic AI, signals a maturation of the AI infrastructure market. As AI systems become more autonomous, the demand for efficient and specialized processing units, including CPUs, will likely grow. This diversification beyond GPUs suggests a recognition of the complex computational needs of future AI deployments, where different workloads may benefit from tailored hardware. The success of the Grace CPU in standalone deployments indicates a strong market appetite for Nvidia's silicon across various applications, potentially challenging established CPU vendors. This move positions Nvidia to capture a larger share of the data center hardware spend, leveraging its deep expertise in high-performance computing.
AI-generated to prompt reflection — not editorial opinion, not advice, not a statement of fact. How this works.