What is NVIDIA Cloud Accelerator?
NVIDIA Cloud Accelerator (NCX) is a portfolio of open, modular infrastructure software components that cloud partners can use to build and operate NVIDIA-powered AI clouds. Built from NVIDIA’s own AI factory learnings, it consists of composable building blocks across infrastructure and platform layers, including hardware lifecycle management, health monitoring, and operational automation.
NVIDIA Cloud Functions (NVCF) is a unified API layer for scaling inference and simulation workloads across one or more Kubernetes clusters.
KAI Scheduler is a scalable Kubernetes scheduler optimized for GPU resource allocation across large-scale AI and machine learning systems.
NVIDIA Fleet Intelligence is a powerful agent-based managed service that offers continuous GPU health monitoring and predictive failure signals for maximum uptime and stability. Read about data center fleet management software.
Bare-metal provisioning and secure lifecycle management for multi-tenant GPU infrastructure.
AI Cluster Runtime provides a canonical, continuously validated definition of the NVIDIA-accelerated Kubernetes runtime for reproducible AI infrastructure. Read more.
Open-source, Kubernetes-native GPU monitoring and fault remediation. NVSentinel helps detect issues early and automate recovery to keep GPU fleets productive. Read about automating Kubernetes AI cluster health with NVSentinel.
The NVIDIA DOCA Platform Framework (DPF) is an orchestration system to build, deploy, and operate BlueField‑accelerated infrastructure services enabling partners to build secure, multi‑tenant cloud infrastructure for AI and other modern applications.
Project GPUd is a lightweight, production-proven GPU telemetry agent. It integrates with Docker, containers, Kubernetes, and NVIDIA ecosystems, while providing a unified view of critical metrics.
The NVIDIA AI Cloud-Ready ISV Validation Initiative qualifies and validates AI infrastructure and platform software from ISVs for deployment on NVIDIA Cloud Partners (NCPs).
This guide is intended to provide NVIDIA Cloud Partners (NCPs), Cloud Service Providers (CSPs), and Independent Software Vendors (ISVs) with an infrastructure-native Northstar reference for building AI cloud services on NCP hardware platforms that can be operated with multi-tenancy and elastic resource allocations.
This document outlines a software architecture intended to help NVIDIA Cloud Partners (NCPs)—sometimes called operators—build a performant, cost-effective, solution for large-scale AI inference workloads. It is intended to provide NCPs and ISVs with a Northstar definition that will best serve AI practitioners and cloud operators alike.
These are the standards and expectations for NVIDIA Cloud Partners (NCPs) operating NVIDIA GPU-accelerated AI cloud infrastructure. They cover the full operational stack, from compute and Kubernetes to storage, networking, security, telemetry, and fleet management, and expand on the NVIDIA hardware reference design and NCP Software Reference Guide.