NVIDIA DGX Cloud

Type: Platform Tags: NVIDIA, DGX Cloud, cloud AI, AI supercomputing, enterprise AI Related: NVIDIA-DGX, NVIDIA-DGX-SuperPOD, NVIDIA-AI-Enterprise, NVIDIA-Enterprise-AI-Factory, NVIDIA-AI-Data-Platform, NVIDIA-Certified-Storage, NVIDIA-Base-Command, NVIDIA-AI-Grid, NVIDIA-NIM, NVIDIA-Quantum-Cloud, CUDA-Q, NVIDIA-Cloud-Accelerator-NCX, NVIDIA-Fleet-Intelligence, NVIDIA-Project-GPUd, NVIDIA-Isaac-GR00T Sources: https://docs.nvidia.com/dgx-cloud/index.html, https://www.nvidia.com/en-us/data-center/dgx-cloud/ Last Updated: 2026-04-29

Summary

NVIDIA DGX Cloud is NVIDIA’s cloud AI supercomputing platform for accessing NVIDIA AI infrastructure through cloud service environments. In the wiki it should be treated as a deployment platform adjacent to NVIDIA-DGX, NVIDIA-DGX-SuperPOD, and NVIDIA-AI-Enterprise.

Detail

Purpose

Not every organization can deploy a full AI supercomputer on premises, and many teams need elastic access to NVIDIA AI infrastructure. DGX Cloud provides a cloud-hosted path for training, customization, and inference workloads that still align with the NVIDIA software stack.

Key capabilities

• Cloud access to NVIDIA AI supercomputing infrastructure.
• Enterprise AI workflow alignment with NVIDIA-AI-Enterprise and NVIDIA software containers/models.
• Fits hybrid AI factory planning across on-prem, hosted, and cloud environments.
• Relevant to large training, model customization, inference, and evaluation workflows.

NVIDIA context

DGX Cloud is part of the NVIDIA-AI-Grid story: workloads can run where infrastructure, data locality, cost, and latency make the most sense. It is also a cloud-accessible AI factory path alongside on-premises NVIDIA-DGX-SuperPOD, with NVIDIA-AI-Enterprise, NVIDIA-AI-Data-Platform, and storage/networking choices determining how training, inference, retrieval, and agent workflows are placed. NVIDIA-Quantum-Cloud is the quantum-facing cloud access path for CUDA-Q projects and is adjacent to DGX Cloud capacity when quantum workloads need GPU-backed cloud resources.

Connections

• NVIDIA-DGX - on-prem DGX systems are the hardware/platform counterpart.
• NVIDIA-DGX-SuperPOD - large on-prem AI supercomputing reference architecture.
• NVIDIA-AI-Enterprise - enterprise software stack used across DGX Cloud environments.
• NVIDIA-Enterprise-AI-Factory - DGX Cloud is a hosted capacity option in the broader AI factory pattern.
• NVIDIA-AI-Data-Platform - data access and retrieval design influence how DGX Cloud workloads connect to enterprise data.
• NVIDIA-Certified-Storage - cloud partner and AI factory storage choices affect large-scale AI workload throughput.
• NVIDIA-Base-Command - AI development and job management layer historically tied to DGX Cloud.
• NVIDIA-NIM - inference microservices can be deployed on NVIDIA cloud or self-hosted GPU infrastructure.
• NVIDIA-Quantum-Cloud - cloud/API access path for CUDA-Q quantum workloads on NVIDIA GPU systems.
• CUDA-Q - quantum workloads can use CUDA-Q in NVIDIA cloud contexts.
• NVIDIA-Cloud-Accelerator-NCX - cloud partner software components support NVIDIA-powered AI cloud operations.
• NVIDIA-Fleet-Intelligence - fleet health and predictive failure signals are relevant to DGX Cloud-scale operations.
• NVIDIA-Project-GPUd - GPUd is described as used in DGX Cloud Lepton production infrastructure.
• NVIDIA-Isaac-GR00T - humanoid robotics foundation model training can use DGX/DGX Cloud-style AI infrastructure.

Source Excerpts

• NVIDIA documentation and product pages position DGX Cloud as cloud-accessible NVIDIA AI supercomputing infrastructure.