NVIDIA Omniverse

What it does

Omniverse takes in 3D scene data, especially OpenUSD files, along with CAD and other 3D content converted into OpenUSD, and produces simulation-ready scenes. Its libraries add RTX rendering, sensor outputs such as camera, lidar, and radar, GPU-accelerated physics, and asset validation against SimReady requirements. Outputs can include rendered visualizations, sensor data, and synthetic training data.

Developers typically use Omniverse by integrating its libraries into existing applications rather than using a standalone end-user app. Documented uses include AI factory digital twins, industrial facility digital twins, robot simulation and learning, synthetic data generation, and autonomous vehicle simulation. Its distinct feature is that the libraries are modular and designed so AI agents can prepare content and run simulation steps while developers keep control of their applications.

Omniverse suits developers and engineering teams building robotics, digital twin, or sensor-simulation workflows that already work in OpenUSD. Access is through SDKs, APIs, and libraries on GPU hardware; some libraries are pre-release and not enterprise-supported. Community support is the default without an NVIDIA AI Enterprise subscription, and enterprise support is priced separately per GPU with multi-year options. Use is governed by NVIDIA's EULA and product-specific terms.

What it can do

• Map native application data into structured OpenUSD scenes for exchange across tools and simulation pipelines
• Check OpenUSD assets against SimReady requirements and report which checks pass or fail
• Render camera, lidar, radar and other sensor outputs from OpenUSD scenes using NVIDIA RTX rendering
• Simulate motion, contacts and collisions with GPU-accelerated physics for robotics and digital twins
• Generate physically based synthetic data from simulation-ready worlds with configurable scene variation
• Build custom applications and tools with the Kit SDK and Python or C++ extensions

Works with

• OpenUSD
• NVIDIA Isaac

Example workflows

Validate an OpenUSD asset before robot training

Input: an OpenUSD asset with materials and physics properties. Steps: run SimReady validation to see which requirements pass or fail, fix flagged properties, then load the scene into the physics library for motion and contact testing. Output: a validation report and a physics-tested scene ready for downstream training.

Generate sensor data from a facility scene

Input: an OpenUSD facility scene with sensor layouts and lighting. Steps: configure scene variation and sensor placement, render camera and lidar outputs with the RTX rendering library, and review the generated data for issues. Output: rendered sensor datasets and flagged problems for developer review.

Before you choose

• Several libraries, including ovrtx, are pre-release software and not enterprise-supported.
• Community support is the only support included without an NVIDIA AI Enterprise subscription; enterprise support is priced separately.
• Use of all Omniverse software is subject to the NVIDIA EULA and product-specific terms.

Listed under

• Coding
• 3D rendering
• Developers
• Digital twins
• Openusd
• Physical AI
• Robotics

Get started

Developer page links to SDKs and libraries; docs cover installation and samples.

1. Review the NVIDIA Omniverse technical requirements and drivers in the documentation before installing libraries.
2. Get started with the Kit SDK or OpenUSD Exchange SDK from the documentation to build a first application.
3. Run a sample such as the ovrtx example or the ovphysx quickstart to confirm the environment works.

Enterprise NVIDIA AI Enterprise subscriptions list at $4,500 per GPU for one year on self-managed systems. Cloud Marketplace consumption is $1 per hour per GPU plus CSP instance costs for production use. Development use may be free or bring-your-own-license plus CSP instance costs. Education and Inception pricing applies to qualified organizations only.

Source: docs.nvidia.com/ai-enterprise/planning-resource/licensing-guide/latest/pricing.html, checked 28 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 28 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Develop Physical AI Applications | NVIDIA Omniverse ↗www.nvidia.com
• Pricing — NVIDIA Enterprise Licensing Guide ↗docs.nvidia.com
• NVIDIA Omniverse - NVIDIA Docs ↗docs.nvidia.com

Company website

nvidia.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Web search

Confirms the official site

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 28 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.