Nvidia has introduced Vera, its first processor specifically designed for the emerging era of AI agents, marking a significant expansion of the company’s ambitions beyond GPUs and AI accelerators. Announced at GTC Taipei during Computex, the new CPU is now in full production and is expected to become a foundational component of next-generation AI infrastructure.
According to Nvidia, Vera delivers up to 1.8 times faster task completion than traditional x86 processors across workloads including agentic AI, reinforcement learning, software development, data processing, and orchestration tasks. The company argues that CPUs are becoming increasingly important as AI systems evolve from generating responses to running code, using tools, coordinating workflows, and making decisions autonomously.
The processor is powered by Nvidia’s custom Olympus CPU architecture and features 88 cores, support for Spatial Multithreading, and up to 1.2TB/s of memory bandwidth through LPDDR5X memory. Vera also serves as the host processor for Nvidia’s upcoming Vera Rubin platform and supports up to 1.8TB/s of coherent bandwidth between CPUs and GPUs through NVLink-C2C interconnect technology.
Several leading AI organizations are already evaluating or planning deployments of Vera. Nvidia said customers include Anthropic, OpenAI, SpaceXAI, ByteDance, CoreWeave, Oracle Cloud Infrastructure, Lambda, Nebius, and Nscale. The processor will also be integrated into systems from Dell, HPE, Lenovo, Supermicro, ASUS, Foxconn, Gigabyte, and other major hardware manufacturers.
The announcement was part of a broader Computex presentation in which CEO Jensen Huang outlined Nvidia’s vision for the next generation of AI computing. Alongside Vera, Nvidia introduced updates to its Nemotron 3 Ultra model family, new robotics and autonomous vehicle platforms, AI development tools, and DGX Station systems for Windows. Huang also argued that future PCs will increasingly ship with dedicated AI processors and dismissed claims that AI will inevitably lead to widespread job losses.
The Infrastructure Behind AI Agents
Nvidia believes AI agents will become one of the largest consumers of computing resources in the coming years. Unlike traditional chatbots, agentic systems continuously process information, execute code, interact with software tools, and coordinate complex workflows, placing significantly greater demands on CPUs.
As a result, Nvidia is increasingly positioning CPUs as a critical component of AI factories alongside GPUs. The company argues that improving CPU performance directly increases agent throughput, reduces latency, and ultimately generates more AI output from existing infrastructure investments.
Industry benchmarks cited by Nvidia suggest Vera performs particularly well in workloads involving Python execution, code compilation, databases, and orchestration systems, all of which play a growing role in autonomous AI environments.
Expanding Beyond GPUs
The launch reflects Nvidia’s broader strategy to control more layers of the AI technology stack. While the company built its dominance through graphics processors, it has steadily expanded into networking, software, CPUs, storage infrastructure, robotics, and edge computing.
At Computex, Nvidia also unveiled RTX Spark, its first Arm-based PC processor for Windows devices. Developed with Microsoft and MediaTek, the chip is designed to bring Nvidia’s AI capabilities directly to laptops and desktops, extending the company’s reach from data centers to personal computing.
Nvidia is also investing heavily in the infrastructure needed to support future AI workloads. The company committed at least $6.5 billion to photonics companies developing optical networking technologies that could improve data transfer efficiency and reduce energy consumption across large-scale AI systems.
