Nvidia Vera CPU Aims to Dominate AI Workloads

Nvidia has unveiled new details about its Vera CPU, positioning the chip as a strong contender in the growing AI CPU market. The Vera CPU features custom Olympus cores designed to maximize single-thread performance for branch-heavy, latency-sensitive workloads common in agentic AI applications. Nvidia claims Vera can deliver up to 1.8x higher performance on these workloads compared to traditional x86 CPUs, thanks to improvements in Python execution, code compilation, static analysis, and tool-driven software workflows.

The company emphasized that agentic AI shifts more execution work to CPUs, as AI agents increasingly execute code, invoke tools, retrieve data, interact with databases, and analyze results. This requires stronger single-thread performance, predictable latency under heavy concurrency, and higher memory bandwidth to keep AI agents running efficiently. Nvidia’s Olympus cores, Scalable Coherency Fabric, and LPDDR5X memory architecture are designed to sustain high performance even under full socket load.

Nvidia’s announcement comes as the server CPU market is expected to expand significantly. BofA recently raised its forecast for the global server CPU market to over $170 billion by 2030, up from a prior estimate of $125 billion. AMD has also raised its outlook, expecting the total addressable market to exceed $120 billion by 2030 due to the rise of agentic AI. This growth presents a major opportunity for both traditional x86 chipmakers and Arm-based CPU vendors, benefiting companies like Nvidia, AMD, Intel, and Arm Holdings.

Nvidia’s shares were up nearly 2% in Tuesday morning’s trade, reflecting investor confidence in the company’s expansion into the CPU market. The stock has seen a 10% year-to-date increase and a 20% rise over the past 12 months. The broader semiconductor and technology sectors have also performed well, with the iShares Semiconductor ETF (SOXX) up 123% and the iShares U.S. Technology ETF (IYW) up 36% over the same period.