US chipmaker Nvidia has unveiled Alpamayo, a new artificial intelligence system designed to improve the safety and performance of autonomous vehicles. The announcement was made at the Consumer Electronics Show (CES) 2026 in Las Vegas, where the company showcased its latest advances in automotive AI.
Alpamayo is a reasoning-based AI model that allows self-driving cars to better understand their surroundings and decide how to respond to real-world situations. Unlike traditional systems that mainly detect objects, Alpamayo is built to analyse complex traffic scenarios, predict risks and choose safer driving actions.
At the core of the system is Alpamayo-1, a large Vision-Language-Action model with 10 billion parameters. It processes data from vehicle cameras and sensors, interprets what it sees and determines the most appropriate response, such as slowing down, stopping or changing lanes. Nvidia says the model can also explain its reasoning, which is expected to help developers improve safety and transparency.
Nvidia CEO Jensen Huang described Alpamayo as a breakthrough for what he called “physical AI”, comparing it to how conversational AI transformed digital applications. He said the company’s goal is to make autonomous driving systems more reliable, especially in rare and unpredictable road situations, often referred to as edge cases.
The Alpamayo platform also includes AlpaSim, a simulation environment that allows developers to test self-driving software in virtual settings before deploying it on real roads. Nvidia has released the tools and models as open source, encouraging global researchers and automakers to use and improve them.
The company plans to begin deploying Alpamayo-powered systems in vehicles later this year, starting with select Mercedes-Benz models in the United States.
Reacting to the announcement, Tesla CEO Elon Musk commented that achieving most of autonomous driving is relatively easy, but solving the final, rare scenarios remains the biggest challenge. His remarks highlight the growing competition and debate among technology leaders racing to perfect self-driving technology.