Nvidia Unveils Vera Processor Featuring 88 Olympus Arm Cores and High-Bandwidth LPDDR5X Memory

• July 22, 2026

Nvidia has revealed extensive details about its upcoming Vera processor designed for data center applications. The new chip features a robust architecture centered around specialized cores codenamed Olympus, utilizing the Arm instruction set.

Powerful Multi-Core Setup and Advanced Memory Architecture

The Vera processor integrates 88 Olympus cores on a single monolithic computing die, supporting a total of 176 hardware threads. This multi-threaded configuration aims to deliver significant processing capability across highly parallel workloads commonly encountered in modern data centers.

A notable feature of Vera is its large unified L3 cache, sized at 164 megabytes. This cache architecture is intended to improve data locality and reduce latency, boosting overall efficiency and throughput of the processor.

In terms of memory technology, Vera employs LPDDR5X memory with an impressive bandwidth capacity of 1.2 terabytes per second. Such high throughput supports the processor’s ability to manage extensive streams of data, which is critical for performance in demanding data center environments.

By combining Arm-based Olympus cores with advanced memory subsystems and a substantial cache, Nvidia’s Vera processor aims to meet the growing compute demands of large-scale server deployments and cloud infrastructure solutions. Further details regarding availability, pricing, and specific use cases have not yet been disclosed.

Nvidia details its Vera processor with 88 Olympus Arm cores, 176 threads, and LPDDR5X memory delivering 1.2 TB/s bandwidth for data centers.