Nvidia Unveils Alpamayo for Reasoning Self-Driving Cars
At CES 2026, Nvidia introduced Alpamayo open models to help AVs reason through rare scenarios, debuting in the new Mercedes-Benz CLA in the US soon.
TL;DR
At CES 2026, Nvidia CEO Jensen Huang introduced Alpamayo, an open-source platform designed to help self-driving cars reason through rare road situations and explain their choices. Nvidia says it will debut in the Mercedes-Benz CLA (US first, then Europe/Asia), with models shared via Hugging Face, as it pushes deeper into physical AI despite Tesla’s long-tail warnings.
At CES in Las Vegas, Nvidia CEO Jensen Huang unveiled Alpamayo, a new platform aimed at giving autonomous vehicles stronger “reasoning” so they can handle unusual edge cases and explain why they make certain driving choices. Huang said the initiative is designed to help cars operate more safely in complex real‑world conditions, especially when the situation doesn’t look like the data they see every day.
Huang also announced that Nvidia and Mercedes‑Benz have started building a vehicle powered by Nvidia’s autonomous-driving stack: the new Mercedes‑Benz CLA. The CLA is expected to launch in the US first, with Europe and Asia to follow later. Nvidia showcased a demo of the Mercedes driving through San Francisco while the passenger kept hands off the wheel, highlighting that the system was trained from human driving demonstrations and can narrate its next move.
Nvidia says Alpamayo is open to researchers, with models and tools available through Hugging Face so teams can access and retrain them. Analysts framed the move as Nvidia expanding from a chip supplier into a full platform provider for “physical AI,” and Nvidia’s shares edged up slightly in after-hours trading after the keynote. Elon Musk responded by arguing that getting close to perfect autonomy is easier than solving the “long tail” of rare events, as Tesla continues developing its Autopilot/FSD approach. Nvidia also reiterated that its next-generation Rubin AI chips are in manufacturing and are expected to launch later this year with improved energy efficiency versus current chips.
