SAN JOSE: NVIDIA on March 16 expanded its physical AI push at its GTC conference, unveiling new Cosmos world models, Isaac simulation frameworks and Isaac GR00T N robot models while naming a broad group of robotics partners that are moving the technology into factories, warehouses and healthcare settings. The company said industrial and humanoid robot developers including ABB Robotics, AGIBOT, Agility, FANUC, Figure, Hexagon Robotics, KUKA, Skild AI, Universal Robots, World Labs and YASKAWA are building on its platform, alongside surgical robotics groups including CMR Surgical and Medtronic.
NVIDIA said the new software stack is designed to help robotics companies design, train, test and deploy machines with more autonomy in real-world environments. It introduced Cosmos 3, a world foundation model that combines synthetic world generation, vision reasoning and action simulation, and launched Isaac Lab 3.0 in early access for large-scale robot learning. The company also said GR00T N1.7 is available in early access with commercial licensing, while GR00T N2 was previewed during the keynote as a next-generation foundation model aimed at broader robot tasks.
The announcement also broadened NVIDIA’s industrial validation effort. The company said FANUC, ABB Robotics, YASKAWA and KUKA, whose combined installed base exceeds 2 million robots, are integrating Omniverse libraries and Isaac simulation frameworks into virtual commissioning systems used to model production lines before deployment. NVIDIA said the manufacturers are also integrating Jetson modules into controllers for real-time AI inference at the edge, linking digital twins and robot control in live operating environments.
Factory deployments gain traction
Among the clearest real-world deployments, NVIDIA said Skild AI is partnering with ABB Robotics and Universal Robots to extend generalized robot intelligence across industrial and collaborative systems. It also said Skild AI is working with Foxconn on high-precision assembly for NVIDIA Blackwell production lines. Foxconn’s Houston lines producing Blackwell servers are deploying Skild AI software in robots, marking an early commercial use of generalized physical AI in semiconductor and electronics manufacturing.
Warehouse automation was another focal point of the March 16 announcements. NVIDIA said KION and Accenture are using Omniverse-based digital twins to train and test fleets of Jetson-powered autonomous forklifts for GXO. KION said the work has moved into live operations at a GXO warehouse in Épinoy, France, where an AI-supported autonomous industrial truck is carrying out end-to-end transport missions in a facility that also operates manual trucks. The company said the site is serving as a pilot environment for physical AI in supply-chain operations.
Healthcare and data infrastructure expand
Healthcare robotics formed a third pillar of the rollout. NVIDIA said CMR Surgical is using Cosmos-H simulation to train and validate robotic intelligence for its Versius surgical system before clinical deployment. Johnson & Johnson MedTech is using Isaac Sim and Cosmos-based post-training workflows for the Monarch platform for urology, while Medtronic is exploring NVIDIA IGX Thor for surgical robotic systems that require high precision and functional safety. The company framed the sector as one where autonomy must meet strict safety and regulatory requirements.
NVIDIA also paired the robotics announcements with a Physical AI Data Factory Blueprint that Microsoft Azure and Nebius are integrating into cloud services for generating, augmenting and evaluating training data at scale. The broader package presented in San Jose ties together model development, simulation, digital twins and deployment across manufacturing, logistics and healthcare. Taken together, the March 16 announcements show NVIDIA moving physical AI from research and simulation into named commercial pilots and operational environments across the global robotics sector – By Content Syndication Services.