NVIDIA's recent GTC 2025 announcement has sent shockwaves through the tech industry, unveiling a trio of groundbreaking open-source releases aimed at accelerating physical AI development. The company's new suite of world foundation models (WFMs), including Cosmos Transfer and NVIDIA Isaac GR00T N1, is set to revolutionize robotics systems and autonomous vehicle technology.

The announcement marks a significant leap forward in physical AI technology, offering developers powerful tools and resources to advance their work. At the heart of this innovation are three key releases: Cosmos Transfer, an open model for general humanoid reasoning; Physical AI Dataset, a commercial-grade dataset for developing physical AI; and NVIDIA Isaac GR00T N1, the world's first open foundation model for generalized humanoid robot reasoning and skills.

What Happened

Cosmos Transfer is a new suite of WFMs with multicontrols that introduces a new level of control and accuracy in generating virtual world scenes. Available in 7 billion parameter size, the model utilizes multicontrols to guide the generation of high-fidelity world scenes from structural inputs, ensuring precise spatial alignment and scene composition.

The model is built by training individual ControlNets separately for each sensor modality used to capture the simulated world. Input types include 3D bounding box map, Trajectory map, Depth map, Segmentation map. At inference time, developers can use various input types, including structured visual or geometric data such as segmentation maps, depth maps, edge maps, human motion keypoints, LiDAR scans, trajectories, HD maps, and 3D bounding boxes to guide the output.

Cosmos Transfer coupled with the NVIDIA Omniverse platform is driving controllable synthetic data generation for robotics and autonomous vehicle development at scale. Find more Cosmos Transfer Examples on GitHub. Cosmos Transfer samples built using post-training base model are also available for autonomous vehicles.

Background and Context

NVIDIA's GTC conference has become a premier event for showcasing the latest breakthroughs in AI technology. This year's announcement marked a significant milestone in the company's efforts to accelerate physical AI development. The Physical AI Data Factory Blueprint, announced earlier this month, is an open reference architecture that unifies and automates how training data is generated, augmented, and evaluated.

The blueprint enables developers to use NVIDIA Cosmos open world foundation models and leading coding agents to transform limited training data into large, diverse datasets — including rare edge cases and long-tail scenarios that are expensive, time-consuming, and often impractical to capture in the real world. NVIDIA is collaborating with Microsoft Azure and Nebius to integrate the open blueprint with their cloud infrastructure and services.

Why it Matters

The significance of these announcements cannot be overstated. Physical AI has the potential to transform industries such as manufacturing, logistics, healthcare, and more. The ability to generate massive amounts of data is crucial for developing autonomous systems and robots that can operate in complex environments.

NVIDIA's open-source releases are designed to accelerate physical AI development by providing developers with powerful tools and resources. Cosmos Transfer and NVIDIA Isaac GR00T N1 offer a new level of control and accuracy in generating virtual world scenes, while the Physical AI Dataset provides a commercial-grade dataset for developing physical AI.

What Comes Next

The path forward is clear: post-training is the key to advancing autonomous systems. Developers can use various input types to guide the output of Cosmos Transfer and NVIDIA Isaac GR00T N1. The Physical AI Dataset provides a commercial-grade dataset for developing physical AI.

NVIDIA's open-source releases are designed to accelerate physical AI development by providing developers with powerful tools and resources. The company is committed to advancing AI technology, as evident from its plans to open a dedicated research lab in Boston to advance quantum computing.

Key Facts

  • NVIDIA has announced three groundbreaking open-source releases aimed at accelerating physical AI development: Cosmos Transfer, Physical AI Dataset, and NVIDIA Isaac GR00T N1.
  • Cosmos Transfer is a new suite of WFMs with multicontrols that introduces a new level of control and accuracy in generating virtual world scenes.
  • The Physical AI Data Factory Blueprint is an open reference architecture that unifies and automates how training data is generated, augmented, and evaluated.
  • NVIDIA is collaborating with Microsoft Azure and Nebius to integrate the open blueprint with their cloud infrastructure and services.
  • Cosmos Transfer coupled with the NVIDIA Omniverse platform is driving controllable synthetic data generation for robotics and autonomous vehicle development at scale.
  • The Physical AI Dataset provides a commercial-grade dataset for developing physical AI, consisting of 15 terabytes of data representing more than 320,000 trajectories for robotics training.