Nvidia (NVDA) unveiled new plans for humanoid robots and self-driving vehicle technology at CES 2026, positioning the company as a central supplier for the next phase of robotics and automation development.

Speaking during the company’s keynote address on Monday, Chief Executive Jensen Huang said a growing number of companies are using Nvidia’s robotics platforms to design and operate advanced machines. He cited partners including Boston Dynamics, Caterpillar, LG Electronics, and NEURA Robotics.

Nvidia said interest in humanoid robots is expanding as manufacturers and logistics firms look for new ways to automate physical tasks. The company estimates those industries represent a combined market worth $50 trillion.

At CES, Nvidia introduced new artificial intelligence models designed to train robots to perceive and respond to real-world environments. The company also detailed new hardware systems intended to serve as the core computing platforms for robotic systems.

Nvidia said its tools are being used to improve how robots move, interact with objects, and perform complex actions. The company described these efforts as part of a broader push toward what it calls physical AI.

Alongside robotics, Nvidia presented a new set of models for autonomous vehicles known as Alpamayo. The company said the models rely on vision, language, and action capabilities combined with chain-of-thought reasoning.

Nvidia said the technology allows vehicles to recognize unusual or unexpected driving scenarios and determine appropriate responses. One example involved identifying a malfunctioning traffic signal at an intersection and deciding how to proceed safely.

According to Nvidia, Alpamayo is intended to act as a large-scale teacher model. Developers can adapt and refine it for use within their own self-driving software systems.

The company said several organizations, including Lucid, Uber, and Berkeley DeepDrive, have expressed interest in the technology.

Self-driving vehicles are already operating in parts of the world, but challenges remain. Some systems have struggled with rare edge cases and complex traffic situations.

Nvidia said virtual training environments can help address those issues. By simulating difficult scenarios, developers can improve autonomous systems without relying solely on real-world driving.