Nvidia launched the Vera CPU, the world's first processor purpose-built for agentic AI and reinforcement learning. The chip delivers twice the efficiency and 50% faster performance than traditional rack-scale CPUs, addressing the growing infrastructure demands of AI systems that plan tasks, run tools, and validate results.

Major cloud providers and system makers are already adopting Vera, including Alibaba, Meta, Oracle Cloud Infrastructure, Dell Technologies, HPE, and Lenovo. The CPU features 88 custom Olympus cores with Spatial Multithreading, enabling up to 1.2 TB/s of memory bandwidth while consuming half the power of general-purpose CPUs.

Vera integrates with Nvidia's MGX modular architecture and pairs with GPUs via NVLink-C2C interconnect for 7x the bandwidth of PCIe Gen 6. The platform supports both dual and single-socket server configurations optimized for workloads like reinforcement learning, agentic inference, and high-performance computing. With full production underway, Vera will be available from partners in the second half of this year.