Nvidia Corp (NASDAQ:NVDA, XETRA:NVD) used CES 2026 to underline its ambition to become a central force in the development of humanoid robots, as technology groups step up efforts to bring machines that can operate in the physical world closer to reality.
In a keynote address, chief executive Jensen Huang said a wide range of industrial and technology companies, including Boston Dynamics, Caterpillar, LG Electronics and NEURA Robotics, are already using Nvidia’s robotics platforms to design and power their machines.
The company believes so-called physical artificial intelligence has the potential to transform manufacturing and logistics, industries it estimates to be worth $50 trillion globally.
To support that push, Nvidia unveiled new artificial intelligence models designed to train robots to understand and interact with real-world environments, alongside the specialised hardware required to run them.
The aim is to provide a common technological foundation that robotics developers can adapt to their own needs, accelerating progress across the sector.
Beyond robotics, Nvidia also introduced a new family of models for autonomous vehicles, known as Alpamayo. These systems use advanced reasoning techniques to help cars interpret complex or unusual driving scenarios, such as navigating an intersection when traffic lights have failed.
Rather than being deployed directly in vehicles, the models are intended as large-scale training tools that developers can refine and integrate into their own self-driving software.
Interest in Alpamayo has already come from groups including Lucid, Uber and Berkeley DeepDrive. While autonomous vehicles are increasingly visible on public roads, led by services such as Waymo, the technology remains imperfect and prone to errors in unexpected situations.
Nvidia argues that virtual training environments can play a crucial role in improving safety and reliability, allowing developers to expose their systems to rare or hazardous scenarios without putting vehicles or people at risk.