NVIDIA CloudXR

NVIDIA CloudXR™ 6.0 is a GPU-accelerated streaming platform that delivers high-fidelity spatial experiences from powerful GPU resources to an open ecosystem of AR, VR, and spatial computing devices. CloudXR 6.0 provides full OpenXR compliance, allowing developers to build once and deploy to any supported headset or operating system. By removing local hardware constraints, the SDK enables the streaming of photorealistic digital twins and complex simulations to lightweight devices—including native integration with Apple visionOS and frictionless web-based access via CloudXR.js.

Documentation Forum


Spatial Workflows

icon representing neural network

OpenXR Development

Support OpenXR applications with global-level scaling, enabling compatibility across a diverse range of XR devices.

Explore SDK Components
icon representing cuda containers

Digital Twins with Omniverse

Stream NVIDIA Omniverse™ Enterprise apps to immerse users in complex industrial spaces and simulate them with accurate digital twins in real time.

Spatial Streaming for Omniverse
icon representing workflows

Robotics & Teleoperations

CloudXR enables robotics developers to collect high quality demonstrations in the real world and simulation to train and test robot policies.

Explore Isaac TeleOp

How CloudXR Works

NVIDIA CloudXR 6.0 functions as a universal OpenXR bridge that decouples heavy compute from the display, enabling photorealistic spatial experiences through three primary distributed components:

  • CloudXR Runtime (Server): As the backbone of every deployment, this OpenXR-compliant runtime handles high-performance rendering and GPU-accelerated encoding on Windows or Linux servers. It acts as the bridge between your RTX-powered application and the network.

  • CloudXR Frameworks (Client): For native ecosystem support for Apple platforms, there are two pathways to build apps to receive CloudXR streams. On visionOS, you can leverage the Foveated Streaming framework to stream high quality OpenXR applications. The endpoint streams high quality content only where necessary based on information about the approximate region where the person is looking, ensuring performance. On iOS or iPadOS, you can leverage StreamingSession.xcframework to stream an OpenXR experience to an iPhone or iPad. StreamingSession.xcframework has a similar API to FoveatedStreaming.framework on visionOS, allowing you to easily build cross-platform streaming applications.

  • CloudXR.js (Web Clients): This JavaScript framework enables frictionless, browser-based XR. It allows devices like Meta Quest 3 and Pico 4 Ultra to access high-end Robotics, Omniverse, and OpenXR content via WebRTC without requiring an app store installation.

Diagram: OpenXR streaming with CloudXR to local devices


Get Started with CloudXR

CloudXR Runtime Server SDK for OpenXR 

Deploy the essential server-side engine required to render, encode, and stream your NVIDIA RTX™-powered applications. CloudXR Runtime 6.0 functions as a standardized OpenXR bridge, enabling any compliant application—from NVIDIA Isaac™ Lab to custom engines—to stream photorealistic content to lightweight wireless clients with ultra-low latency.
Download CloudXR 6.0 Runtime

Client SDK: Apple Platforms

Allows developers to incorporate immersive spatial streaming into new and existing iOS, iPadOS, and visionOS projects with full version control.  Built into Xcode for every Swift developer "out of the box.” These frameworks enables photorealistic, 1:1-scale experiences on iPhone, iPad, and Apple Vision Pro without sacrifice. 

Client SDK: WebXR Platforms

CloudXR.js delivers high-performance immersive content to Meta Quest, Pico devices and desktop browsers through a single secure web link. CloudXR.js is a framework-agnostic JavaScript library to stream Robotics simulations, Omniverse digital twins, and any OpenXR content directly to a browser.


CloudXR Learning Library

Documentation

CloudXR SDK Documents

This hub provides a complete developer journey for CloudXR 6.0, including SDK overviews, quick-start guides, and detailed technical requirements for the Runtime, native Apple Framework, and CloudXR.js. Access in-depth documentation for the XR Opaque Data Channel, Stream Manager, and native visionOS Message Channel APIs, as well as Unreal Engine integration guides and reference samples to help you scale spatial intelligence workflows.

Code Sample

CloudXR Stream Manager

CloudXR Stream Manager is a Windows service that manages NVIDIA CloudXR Runtime instances, consisting of a server component for hosting the RPC service and a sample client for demonstrating interaction.

Tech Blog

Spatial Streaming for Omniverse Digital Twins

Learn more about streaming immersive OpenUSD-based NVIDIA Omniverse™ digital twins to the Apple Vision Pro with this reference architecture

Tech Blog

Humanoid Robot Learning With Apple Vision Pro

Capture teleoperated data from Apple Vision Pro to generate large synthetic trajectories data sets from just a few human demonstrations using NVIDIA Isaac GR00T.

Get started with CloudXR today

Download Now