NVIDIA and Mercedes-Benz are extending their automated-driving partnership into the next decade. In an announcement published by NVIDIA's newsroom on October 16, 2026, the two companies said they will jointly develop the next generation of driver assistance and automated driving systems, combining Mercedes-Benz's automotive engineering, safety expertise and MB.OS operating system with NVIDIA's Hyperion compute and sensor architecture, the Alpamayo software stack and the Halos safety system.
A full stack, not a single chip
Automaker-chipmaker deals often stop at buying compute. What was announced here is a full stack. Inside the vehicle, MB.OS carries new driving functions and over-the-air updates, while Hyperion provides the in-vehicle computing platform and sensor architecture. On the software side, the Alpamayo stack supplies full-stack driving software with AI reasoning, supporting point-to-point urban assistance, active safety and automated parking. On the safety side, Halos adds a parallel safety stack on top of the Mercedes-Benz safety foundation, bringing redundancy and guardrails. The stated ambition goes beyond assistance features to future SAE Level 4 applications, including robolimousines.
Jörg Burzer, the Mercedes-Benz board member responsible for development and procurement, called NVIDIA the company's most important strategic partner in automated driving, noting that the pair had already built the first Mercedes-Benz urban point-to-point assistance system for the western world. NVIDIA automotive vice president Ali Kani framed the contribution as accelerated computing that lets vehicles understand the world and reason through complex driving situations. Put together, the two statements describe the logic of the deal: the carmaker guards production safety standards, the platform supplies the reasoning.
The factory is on the same table
One line in the announcement is easy to skip: the partnership is not confined to the car. Mercedes-Benz also uses NVIDIA technology for vehicle engineering simulation and inside MO360, its digital production ecosystem, from virtual planning to AI-based production applications, and the two plan to keep improving efficiency, flexibility and quality across the global production network. The same compute and simulation capability thus connects product features at one end and manufacturing at the other, which makes the relationship stickier than a single model program.
For the industry, the signal is that a leading carmaker is placing key pieces of its automated-driving roadmap on an external platform for the long term rather than building everything in-house. Urban point-to-point driving, active safety and automated parking will be contested on reasoning quality and safety redundancy, not on sensor counts alone. NVIDIA's platform moves outside the car are just as busy; its open Indic-language corpus is covered in NVIDIA's 2.8-trillion-token Indic corpus release. What the announcement does not give is a timetable: which models, and which markets first, remains the open question to watch.
