Overview#
NVIDIA® AI Enterprise is a software platform that accelerates every stage of the AI lifecycle - from prototyping to production - across cloud, data center, and edge environments. A single license covers all software components, enterprise support with SLAs, security patches, and maintenance updates.
Audience#
This documentation is for IT administrators, platform engineers, DevOps teams, data scientists, and AI infrastructure architects who deploy, manage, or consume NVIDIA AI Enterprise software.
Two-Layer Architecture#
The platform is organized into two layers with independent release cadences, allowing organizations to adopt new application capabilities without disrupting validated infrastructure:
Application Development - NIM™ microservices, NeMo™ tools for agentic AI, Omniverse™ libraries for physical AI, and AI frameworks and ML libraries, all built on CUDA®-X and CUDA.
Infrastructure Management - GPU drivers, Run:ai for workload orchestration, vGPU™ and MIG for GPU partitioning, Kubernetes operators, and Base Command Manager for deploying and managing GPU infrastructure at scale.
The stack is composable - foundation components are common to all deployments, while the remaining components are assembled based on your use case.
Where Documentation Lives#
NVIDIA AI Enterprise documentation is published across several sites. This Software catalog and its sibling Lifecycle Policy are the two main Docs Hub spokes:
Resource |
Description |
|---|---|
Landing page with tabs for Getting Started, Software, Infrastructure, Planning, and Support |
|
NVIDIA AI Enterprise Software (this document) |
Application and Infrastructure Layer component catalogs with NGC™ links |
Branch types, support timelines, the Interactive Explorer, and EOL tracking |
|
Per-release documentation for drivers, operators, vGPU, and deployment guides |
|
Container images, Helm charts, models, and collections for download |