NVIDIA AI Enterprise Software

Type: Software Catalog Tags: NVIDIA, AI Enterprise, software catalog, application layer, infrastructure layer, NGC, support matrix Related: NVIDIA-AI-Enterprise, NVIDIA-AI-Enterprise-Quick-Start-Guide, NVIDIA-AI-Enterprise-Lifecycle-Policy, NVIDIA-AI-Enterprise-Infrastructure-Support-Matrix, NGC, NVIDIA-NGC-Catalog, NVIDIA-NIM, NVIDIA-NeMo, NVIDIA-Omniverse, NVIDIA-Run-ai, NVIDIA-Run-ai-Support-and-Lifecycle, NVIDIA-RAPIDS, RAPIDS-Accelerator-for-Apache-Spark, NVIDIA-FLARE, TensorRT, Triton-Inference-Server, PyTorch, NVIDIA-Riva, NVIDIA-TAO, NVIDIA-DeepStream, NVIDIA-Holoscan, Morpheus, NVIDIA-Data-Center-GPU-Drivers, NVIDIA-DOCA, NVIDIA-DOCA-Platform-Framework, NVIDIA-vGPU, NVIDIA-vGPU-for-Compute, NVIDIA-MIG, NVIDIA-Container-Toolkit, NVIDIA-GPU-Operator, NVIDIA-Network-Operator, NVIDIA-NIM-Operator, NVIDIA-Base-Command-Manager, NVIDIA-Enterprise-Support-and-Services Sources: https://docs.nvidia.com/ai-enterprise/software/latest/overview.html, https://docs.nvidia.com/ai-enterprise/software/latest/application-software.html, https://docs.nvidia.com/ai-enterprise/software/latest/infrastructure-software.html, https://docs.nvidia.com/ai-enterprise/index.html Last Updated: 2026-04-30

Summary

NVIDIA AI Enterprise Software is the current NVIDIA catalog of AI Enterprise application-layer and infrastructure-layer components. It describes the composable AI Enterprise stack, links components to NGC catalog entries and product docs, explains independent application/infrastructure release cadences, and points administrators to infrastructure support matrices and release notes for deployment compatibility.

Detail

Purpose

Use this page to answer “what is included in NVIDIA AI Enterprise?” at the software-catalog level. Use NVIDIA-AI-Enterprise for the platform overview, NVIDIA-AI-Enterprise-Lifecycle-Policy for branch support and compatibility timelines, and NVIDIA-Enterprise-Support-and-Services for support entitlements and case workflows.

Layer model

  • Application Development includes NIM microservices, NeMo tooling, Omniverse libraries, AI frameworks, ML libraries, domain SDKs, and pre-trained models built on CUDA-X and CUDA.
  • Infrastructure Management includes GPU drivers, networking drivers, Kubernetes operators, vGPU, MIG, Run:ai self-hosted, Container Toolkit, DOCA components, and Base Command Manager.
  • The two layers have independent release cadences so organizations can adopt new application capabilities without replacing validated infrastructure.
  • Components include enterprise support with SLAs, security patches, and maintenance updates when used under the AI Enterprise support model.
  • NGC is the distribution and catalog surface for supported software components, while some infrastructure software is supported but obtained through support portals, licensing portals, partner channels, or vendor channels.

Application layer

The current application-layer table includes catalog and documentation entries for inference/deployment, AI frameworks and libraries, domain SDKs/toolkits, and pre-trained models. Durable wiki connections include NVIDIA-NIM, TensorRT, Triton-Inference-Server, NVIDIA-NeMo, PyTorch, NVIDIA-RAPIDS, RAPIDS-Accelerator-for-Apache-Spark, NVIDIA-FLARE, NVIDIA-Riva, NVIDIA-TAO, NVIDIA-DeepStream, NVIDIA-Holoscan, Morpheus, NVIDIA-Omniverse, and model/NIM pages such as NIM-for-Large-Language-Models.

Infrastructure layer

The current infrastructure-layer table includes NGC catalog resources, core GPU/networking drivers, DOCA microservices, virtualization, container platform tooling, GPU orchestration, and Kubernetes operators. Durable wiki connections include NVIDIA-Data-Center-GPU-Drivers, NVIDIA-DOCA, NVIDIA-DOCA-Platform-Framework, NVIDIA-vGPU, NVIDIA-MIG, NVIDIA-Container-Toolkit, NVIDIA-Run-ai, NVIDIA-GPU-Operator, NVIDIA-Network-Operator, NVIDIA-NIM-Operator, and NVIDIA-Base-Command-Manager.

NVIDIA context

The software catalog is a map, not an installation guide. Use deployment-specific pages for bare metal, VMware, cloud, and Red Hat OpenShift installation paths; use the lifecycle policy to decide branches and validate compatibility; use NGC pages for artifact discovery.

Connections

Source Excerpts

  • The software overview describes AI Enterprise as a two-layer stack with independent release cadences.
  • The application-layer page lists NIM, TensorRT, Triton, CUDA Deep Learning, NeMo, PyTorch, RAPIDS, domain SDKs, Omniverse, Riva, TAO, and pre-trained models.
  • The infrastructure-layer page lists drivers, DOCA, vGPU for Compute, Container Toolkit, Run:ai, GPU Operator, Network Operator, NIM Operator, support matrices, and infrastructure release notes.

Resources