Nvidia details its next-generation Vera CPU for AI, setting up challenge to AMD and Intel - CNBC

What the report says

CNBC reported that Nvidia has released more technical details about Vera, its data center central processing unit designed for artificial intelligence servers. The company said the new disclosures include specifications, benchmarks and architecture information needed by potential buyers assessing the chip. Nvidia representatives said Vera chips were delivered in June to clients including OpenAI, Anthropic and SpaceX.

The product marks a deeper move by Nvidia into CPUs, a market long dominated by Intel and AMD. Nvidia rose to the top of the chip industry largely on demand for its GPUs, but CNBC said the company is increasingly trying to sell complete AI computing systems rather than individual chips. Vera will be offered on its own, in systems paired with Nvidia GPUs under the Vera Rubin lineup, and in liquid-cooled rack configurations.

According to Nvidia, Vera is its first server CPU designed from the core rather than built around an off-the-shelf Arm design. The company says the chip is tuned for AI agents, emphasizing single-core speed, memory bandwidth and lower latency so GPUs can stay more fully used. Nvidia said Vera can outperform x86 chips in AI-agent workloads by 50%, while CNBC noted that Intel and AMD use x86 architecture.

The stakes are significant because CPUs are again drawing attention as agentic AI workloads require fast coordination and data handling alongside GPUs. CNBC cited market data showing AMD and Intel shares had outperformed Nvidia so far in 2026, and reported analyst views that AMD has strong hyperscaler relationships while Intel remains a major server CPU supplier. Analysts also cautioned that winning broad cloud-provider adoption may still be difficult, even for Nvidia.