Nvidia's Vera CPU starts shipping, with AWS among the first hyperscalers taking delivery
Anthropic, OpenAI, Oracle Cloud and SpaceX AI received the earliest systems as AWS and Nvidia expand a 16-year partnership to include Vera CPU-based infrastructure.
Original cover art, generated for this story. THE VISSION does not republish third-party press imagery.
- Nvidia's Vera CPU, its first processor built specifically for agentic AI workloads, has moved into production shipping.
- AWS received its first Vera CPU server and Vera Rubin GPU, hand-delivered in Seattle, as part of an expanded partnership that includes plans for 2 million additional GPUs.
- Anthropic, OpenAI, Oracle Cloud Infrastructure and SpaceX AI were named as the first organizations to receive early Vera systems.
- Oracle Cloud Infrastructure plans to deploy hundreds of thousands of Vera CPUs beginning this year, positioning it as the first cloud provider running Vera at hyperscale.
Nvidia says its Vera CPU — the company's first processor built specifically for agentic AI workloads rather than general compute — has moved into production shipping at scale. The 88-core chip is designed around the demands of AI agents that act rather than just answer, and Nvidia claims up to 1.8 times faster per-core performance on agentic workloads than prior-generation designs, alongside 1.2TB/s of memory bandwidth.
AWS is among the first hyperscalers taking delivery. Nvidia and AWS announced an expansion of their 16-year partnership that includes plans for 2 million additional GPUs and Vera CPU-based infrastructure on AWS; the first Vera CPU server and Vera Rubin GPU were hand-delivered to AWS in Seattle by Nvidia Vice President of Hyperscale and HPC Ian Buck.
Nvidia named four early recipients of initial Vera systems: Anthropic and OpenAI in San Francisco, Oracle Cloud Infrastructure, and SpaceX AI in Palo Alto. Oracle has committed to the most aggressive rollout, planning to deploy hundreds of thousands of Vera CPUs beginning this year — making it, on Nvidia's account, the first cloud provider to run Vera at hyperscale.
Buck framed the chip as a response to a shift already underway in how AI workloads run: "Agentic AI is creating a new CPU moment in the AI factory — as models move from answering to acting, Vera is purpose-built to keep that work moving at scale."
A CPU built around agent workloads rather than general-purpose compute is Nvidia betting that the shift from chatbots to autonomous agents is durable enough to justify purpose-built silicon, not just software tuning on existing chips. The list of first recipients — two frontier labs, a major cloud provider and a rocket company's AI division — signals which customers Nvidia expects to push agentic workloads hardest first.