NVIDIA has officially launched the NVIDIA Vera CPU, a new processor designed specifically to run artificial intelligence agents. This hardware represents a shift in data center architecture. Consequently, it focuses on token generation rather than traditional core metrics.
Introduction to the NVIDIA Vera CPU
The new processor completes tasks 1.8 times faster than standard x86 processors. Meanwhile, it builds on the previous Grace processor line, which has shipped nearly 2.5 million units. This performance boost helps modern data centers run complex agentic workloads.
“AI agents will be the largest users of computing. Vera is the first CPU designed for that future, built to run agentic AI at hyperscale with extraordinary performance, efficiency and programmability.”
Jensen Huang, Founder and CEO of NVIDIA
Technical Specifications and Architecture
The NVIDIA Vera CPU features 88 custom Olympus cores and a high-bandwidth memory subsystem. Specifically, this memory subsystem delivers up to 1.2 terabytes per second of bandwidth using LPDDR5X technology. As a result, the processor handles concurrent environments and data pipelines efficiently.
In addition, the processor integrates with other hardware platforms. For example, it serves as the host processor for Rubin platforms using NVLink-C2C technology. This connection provides up to 1.8 terabytes per second of coherent bandwidth.
Industry Adoption and Partnerships
Several major technology firms plan to adopt the NVIDIA Vera CPU. Notably, companies like Anthropic, OpenAI, and SpaceXAI are evaluating the processor for their workloads. Furthermore, hyperscale cloud providers such as Oracle Cloud Infrastructure plan to deploy these systems.
“By deploying NVIDIA Vera CPUs, OCI will support high-throughput reasoning and data processing workloads across next-generation AI environments.”
Mahesh Thiagarajan, Executive Vice President of Oracle Cloud Infrastructure
Market Availability and Future Outlook
Hardware manufacturers are preparing to build systems using the new processor. Specifically, companies like Dell Technologies, HPE, Lenovo, and Supermicro will offer standalone configurations. These systems will provide a new alternative to traditional x86 server setups. Meanwhile, Taiwan-based system builders are also developing compatible hardware.
The new systems will be available starting this fall. Customers can choose between liquid-cooled racks for large installations or air-cooled systems for standard enterprise setups. This rollout will impact the global economy of data centers.
According to independent benchmarks, the processor delivered fast performance across database processing and code compilation. These tasks are critical for modern digital factories. Consequently, the hardware helps reduce the time agents spend waiting on CPU-bound steps.
