Nvidia on Monday unveiled new artificial intelligence infrastructure and models aimed at accelerating the development of what it calls “physical AI,” as the semiconductor company pushes deeper into technologies that power robots and autonomous vehicles.
At the NeurIPS AI conference in San Diego, Nvidia introduced Alpamayo-R1, an open reasoning vision-language-action model designed specifically for autonomous driving research. The company said the model allows vehicles to interpret visual data and text simultaneously, enabling more human-like decision-making in real-world traffic conditions.
See also: Hyundai Partners with Nvidia to Advance AI for Autonomous Vehicles and Smart Factories
The new model is built on Nvidia’s Cosmos-Reason architecture, part of the Cosmos family of AI reasoning models first released in January 2025 and expanded later in the year. Nvidia said the technology is designed to help developers reach Level 4 autonomous driving, which enables full self-driving operation in defined environments and conditions.
Nvidia is also releasing a set of developer tools called the Cosmos Cookbook, which includes step-by-step workflows for data preparation, synthetic data generation, post-training optimization and model evaluation. The tools are intended to make it easier for researchers and engineers to customize Cosmos-based models for real-world applications.
See also: Stellantis Partners with NVIDIA, Uber, and Foxconn to Develop Level 4 Robotaxi Fleet by 2028
Both Alpamayo-R1 and the Cosmos Cookbook have been made publicly available through developer platforms including GitHub and Hugging Face.
Nvidia executives said physical AI — systems that can reason, perceive and act in the physical world — represents the company’s next major technology frontier beyond cloud-based artificial intelligence.
“I think eventually robots are going to be a huge player in the world and we want to basically be making the brains of all the robots,” Nvidia Chief Scientist Bill Dally said earlier this year. “To do that, we need to start developing the key technologies.”
See also: Nvidia Reportedly Developing Robotaxi Platform Based on DRIVE AGX Thor System
Chief Executive Jensen Huang has repeatedly described physical AI as the next major phase of artificial intelligence development, as Nvidia expands its role from data center computing into robotics, autonomous machines and intelligent edge systems.