NVIDIA Vera CPU Rack
Purpose-built for reinforcement learning and agentic AI.
Overview
Rack-Scale Infrastructure for AI Factories
Built on NVIDIA MGX™, the Vera CPU Rack delivers rack-scale CPU infrastructure for modern AI factories. As reinforcement learning (RL) and agentic AI systems scale, CPUs execute the code, tools, and data workflows that drive results. Vera combines leading per-core performance, massive memory bandwidth, and exceptional energy efficiency, completing workloads up to 50% faster with twice the efficiency of traditional CPU infrastructure.
CPU Sandboxes:
Executing AI Actions at Scale
Environments Powering RL and Agentic AI
RL and agentic AI systems rely on software environments outside the model that run model-generated code, tools, and data operations, evaluate results, and return feedback before the next step proceeds. Each environment typically runs on a dedicated CPU core and, at scale, thousands to millions operate in parallel. Their fully loaded per-core performance directly determines how quickly accelerators can continue working.
The NVIDIA Vera CPU Rack is purpose-built to support these environments at scale. A single liquid-cooled rack integrates up to 256 Vera CPUs, supporting more than 22,500 concurrent CPU environments. It’s designed for sustained per-core performance and up to 2x higher energy efficiency, keeping evaluation cycles short, and AI factories operating at peak throughput.
Specifications1
NVIDIA Vera
| NVIDIA Vera CPU Rack | NVIDIA Vera CPU | |
|---|---|---|
| Configuration | 256 Vera CPUs | 1 Vera CPU |
| Cores | Threads | 22,528 NVIDIA Olympus cores 45,056 threads |
88 NVIDIA Olympus cores 176 threads |
| Memory Capacity | Up to 400 TB | Up to 1.5 TB |
| Aggregate Bandwidth | Up to 315 TB/s | Up to 1.2 TB/s |
| N/S Networking | NVIDIA BlueField-4 DPU | N/A |
| Cooling | Liquid Cooled | N/A |
1. Preliminary information. All values are up to and subject to change.
Partners
Meet Our Partners
Get Started
Stay Up to Date on NVIDIA News
Sign up for the latest news, updates, and more from NVIDIA.