German-British ADAS partnership
Mercedes-Benz to bring Wayve AI Driver into production models
Mercedes-Benz is testing advanced driver-assistance systems in locations including San Francisco. Future models are set to integrate Wayve AI Driver into the carmaker’s vehicle architecture to support automated driving in a wider range of traffic situations.
Mercedes-Benz
Mercedes-Benz is taking its collaboration with Wayve towards production. The Wayve AI Driver is due to enter future models within two years, combining the UK company’s data-driven driving intelligence with MB.OS, Mercedes hardware and existing map interfaces for series deployment.
Mercedes-Benz and British AI company Wayve have signed a production agreement covering the use of Wayve AI Driver in future vehicles. The software is expected to reach customers within the next two years. The planned system is designed to support point-to-point journeys in both urban traffic and on motorways, combining Mercedes-Benz vehicle technology with AI driving software from Wayve.
The partners have not yet named the model lines that will receive the technology. Nor have they specified an SAE automation level. That leaves the exact scope of the functions available at launch open, including the level of driver supervision that will be required.
The production agreement builds on several years of technical collaboration. Mercedes-Benz also participated in Wayve’s Series D funding round in early 2026. During the partnership, the AI Driver has already been connected to Mercedes-Benz’s production architecture, using the manufacturer’s hardware together with the MB.OS operating system and map interfaces. Testing of the joint technical basis has taken place in Stuttgart, London and San Francisco.
How does Wayve’s learning-based driving approach work?
Wayve follows a data-driven development model for automated driving. The company calls its architecture AV2.0 and places it within the broader concept of embodied AI. Rather than modelling driving behaviour mainly through fixed rules and high-definition maps, the system is designed to interpret traffic situations using trained AI models.
According to Wayve, AI Driver does not require HD maps and is intended to transfer to new cities and different traffic environments without being retrained separately for every location. The underlying approach is also designed to avoid fixed geographical operating boundaries. In technical terms, this is part of the concept of embodied AI, in which an AI system perceives and responds to the physical world instead of operating purely in a digital environment.
Wayve says its models are trained on NVIDIA AI infrastructure provided through Microsoft Azure. These statements describe the intended technical capabilities of the platform. Mercedes-Benz has not yet specified which of those functions will be enabled when the first production vehicles launch, or under which regulatory conditions they will be offered.
What automation levels can the common platform support?
Wayve is developing AI Driver as a shared software foundation for different levels of vehicle automation. According to the company, the same technological basis is intended to span advanced driver-assistance functions and eventually fully driverless use cases.
That means the architecture could in principle serve both privately owned passenger cars and robotaxi applications. The agreement with Mercedes-Benz, however, is not an announcement of a driverless production car. What has been confirmed is an advanced assistance function for point-to-point driving in urban areas and on motorways, with the technology ultimately intended for multiple markets.
This distinction is important because software architecture and product functionality are not the same thing. A platform may technically support a broad range of automation capabilities, while the functions activated in a production vehicle still depend on sensor configuration, validation, homologation and the regulatory framework in each market.
