Nvidia has introduced generative AI models and blueprints to expand its Omniverse platform, targeting physical AI applications like robotics and autonomous vehicles. Partners like Accenture, Microsoft, and Siemens are integrating Omniverse into their products and services. Nvidia’s offerings include AI models that accelerate 3D world creation, synthetic data generation, and digital twin capabilities. The company’s AI Enterprise suite is priced at $4,500 annually per GPU.
Nvidia has introduced generative AI models and blueprints to expand its Omniverse universe, including physical AI applications such as robotics, autonomous vehicles, and vision AI.
Software development partners and professional services are, or will be, using Omniverse to develop new products and services that will accelerate industrial AI. Whatever will they run such apps on, one wonders.
Companies such as Accenture, Altair, Ansys, Cadence, Foretellix, Hyundai, Microsoft, and Neural Concept have integrated, or are integrating, Omniverse into their software products and services. Siemens announced the availability of Teamcenter Digital Reality Viewer, a Siemens Xcelerator application powered by Nvidia Omniverse libraries.
Creating 3D worlds for physical AI simulation involves worldbuilding, labeling the world with physical attributes, and making it photoreal. Nvidia is offering generative AI models that it claims accelerate each step. The USD code and USD Search NIM microservices are now available, allowing developers to use text prompts to generate or search for open assets.
Nvidia AI Blueprints, built on NIM microservices, are reference implementations for complex AI workflows. They help developers connect several components, including libraries, software development kits, and AI models, together in a single application. Nvidia NIM costs about $4,500 annually per GPU. The price is based on the cost of the Nvidia AI Enterprise suite for the number of GPUs being using.
Generative AI model and repositories provider Hugging Face launched an alternative to Nvidia’s NIM. Hugging Face Generative AI Services, or HUGS, is currently the only available alternative to NIM. Nvidia is both a partner and an investor in Hugging Face.
A new Nvidia Edify sim-ready generative AI model can automatically label existing 3D assets with attributes like physics or materials, enabling developers to process 3D objects more efficiently. Edify is a multi-modal architecture for developing visual generative AI models for image, 3D, 360 HDRI, PBR materials, and video. Using Nvidia AI Foundry, service providers can train and customize Edify models to build commercially viable visual services built on top of Nvidia NIM.
Nvidia Omniverse, paired with new Nvidia Cosmos world foundation models, can be used to create a synthetic data multiplication engine. This lets developers generate controllable, photoreal synthetic data. Developers can compose 3D scenarios in Omniverse and render images or videos as outputs. These can then be used with text prompts to condition Cosmos models and generate synthetic virtual environments for physical AI training.
One of the significant aspects of Omniverse is its capability to use digital twins. Altair, Cadence, Hyundai, Siemens, and others use that capability in their products and operations.
Nvidia has taken Pixar’s Universal Scene Description (USD) to new heights and applications that the developers never imagined, and they probably now wish they hadn’t made it open source. Nvidia has added a vast library of tools, examples, and applications, plus a large population of partners with complementary software and applications. Like all other universes, Nvidia’s Omniverse is continuing to expand—let’s hope entropy isn’t its final destination.
LIKE WHAT YOU’RE READING? INTRODUCE US TO YOUR FRIENDS AND COLLEAGUES.