NVIDIA DGX BasePOD

Type: Platform Tags: NVIDIA, DGX BasePOD, DGX, reference architecture, AI factory, enterprise AI, Mission Control, Base Command Manager Related: NVIDIA-DGX, NVIDIA-DGX-BasePOD-B200-H200-H100-RA, NVIDIA-DGX-B200, NVIDIA-DGX-SuperPOD, NVIDIA-DGX-SuperPOD-B200-RA, NVIDIA-DGX-B300, NVIDIA-GB300-NVL72, NVIDIA-AI-Enterprise, NVIDIA-Mission-Control, NVIDIA-Base-Command-Manager, NVIDIA-BaseOS, NVIDIA-Run-ai, NVIDIA-Certified-Storage, NVIDIA-AI-Data-Platform, NVIDIA-Certified-Systems, NVIDIA-DGX-Enterprise-Support Sources: https://docs.nvidia.com/dgx-basepod/index.html, https://www.nvidia.com/en-us/data-center/dgx-basepod/, https://docs.nvidia.com/dgx-basepod/reference-architecture-infrastructure-foundation-enterprise-ai/latest/index.html Last Updated: 2026-05-09

Summary

NVIDIA DGX BasePOD is NVIDIA’s prescriptive enterprise AI infrastructure reference architecture for building and scaling DGX-based AI infrastructure. It sits between individual DGX systems and full NVIDIA-DGX-SuperPOD deployments by combining DGX systems, NVIDIA networking, storage, software, MLOps components, and deployment guidance into a validated base architecture for enterprise AI factories.

Detail

Purpose

Enterprise teams often need a validated path to deploy AI infrastructure without designing every compute, network, storage, software, and operations component from scratch. DGX BasePOD provides that prescriptive base so organizations can build an AI Center of Excellence or enterprise AI factory with lower deployment complexity and more repeatable operations.

Platform role

  • Reference architecture for DGX infrastructure using DGX B200, H200, H100, and A100 era systems in current documentation.
  • The current NVIDIA-DGX-BasePOD-B200-H200-H100-RA document covers a two-to-eight-node NDR400 InfiniBand design with DGX B200, H200, and H100 options.
  • Connects compute, networking switches, software, MLOps solutions, and third-party storage into validated configurations.
  • Provides deployment guides for DGX B200 with NVIDIA-Mission-Control, DGX H200 and DGX A100 with NVIDIA-Base-Command-Manager, Run:ai on DGX BasePOD, AWS connectivity, RHEL deployment, and update workflows.
  • Works with NVIDIA-AI-Enterprise for supported enterprise AI software.
  • Connects to NVIDIA-Certified-Storage and partner storage reference architectures for high-throughput training, inference, and data pipeline workloads.

BasePOD vs SuperPOD

DGX BasePOD is the enterprise foundation pattern: validated, repeatable infrastructure for organizations building production AI capacity. NVIDIA-DGX-SuperPOD is the larger leadership-class AI supercomputing architecture for the most demanding training, simulation, and AI factory workloads. BasePOD can be a stepping stone toward SuperPOD-scale operations.

NVIDIA context

BasePOD makes several wiki areas queryable together: NVIDIA-DGX systems, NVIDIA-Mission-Control operations, NVIDIA-Base-Command-Manager cluster management, NVIDIA-Run-ai workload orchestration, NVIDIA-AI-Enterprise software, and certified storage/data platform guidance.

Connections

Source Excerpts

  • NVIDIA’s DGX BasePOD docs describe a prescriptive enterprise AI infrastructure that reduces design, deployment, and management complexity.
  • NVIDIA’s DGX BasePOD product page positions it as the infrastructure foundation for enterprise AI factories and connects it to AI Enterprise and partner storage reference architectures.

Resources