Nvidia has revealed new details about its next-generation Vera central processing unit (CPU), marking a bigger push into the AI server market traditionally dominated by AMD and Intel. The company said Vera chips were delivered in June to customers including OpenAI, Anthropic, and SpaceX. Unlike Nvidia’s graphics processing units (GPUs), which power AI training and deployment, the Vera CPU is designed to handle tasks that support AI systems, particularly the growing use of autonomous AI agents.
Nvidia says Vera was built from the ground up rather than using an existing Arm design, with a focus on faster single-core performance, lower latency, and higher memory bandwidth. According to the company, the chip delivers up to 50% better performance for AI agent workloads than traditional x86 processors. Vera will be sold both as a standalone CPU and as part of Nvidia’s integrated AI server systems, including its upcoming Vera Rubin platform.
The launch signals Nvidia’s ambition to compete more directly with AMD and Intel in the server CPU market while expanding its strategy of offering complete AI computing systems. Analysts say adoption may take time because cloud providers already have strong relationships with existing CPU suppliers. However, Nvidia believes demand for AI infrastructure will continue to grow, with OpenAI expected to begin deploying Vera chips at scale this quarter.
Related Readings:
