Uber and Israeli autonomous driving company Autobrains have announced a strategic partnership to launch a robotaxi program in Munich, Germany, marking a major step toward scaling commercial autonomous ride-hailing in Europe.
The initiative, unveiled at NVIDIA’s GTC Taipei conference, will integrate three core components: Uber’s global ride-hailing platform, Autobrains’ Agentic AI autonomous driving system, and NVIDIA’s DRIVE Hyperion autonomous vehicle computing architecture.
Pending regulatory approval, Munich is set to become the first deployment city for the program. The companies said the pilot is intended to serve as a foundation for an OEM-agnostic model, meaning it could enable autonomous ride-hailing across multiple vehicle manufacturers and urban markets, rather than relying on proprietary fleets or single-platform deployments.
Munich was selected due to its status as one of Europe’s key automotive hubs, its combination of dense urban streets and higher-speed road networks, and Germany’s established regulatory framework for autonomous vehicle testing and innovation.
A central feature of the collaboration is Autobrains’ Agentic AI approach to autonomous driving. Unlike conventional systems that depend on a single end-to-end model to manage all driving tasks, Autobrains structures decision-making into multiple specialized AI agents, each responsible for a specific aspect of driving behavior.
These agents continuously process road conditions, interpret context, evaluate possible actions and select responses in real time. According to the company, this modular approach is designed to improve robustness in complex and unpredictable driving environments while maintaining computational efficiency suitable for large-scale fleet deployment.
Autobrains said the system is built to operate on standard automotive sensor configurations and automotive-grade computing hardware, reducing the need for costly custom sensor stacks and highly specialized vehicle platforms that have limited scalability in earlier autonomous driving programs.
“Autonomous driving will not scale by relying on a single model to solve every driving scenario,” said Autobrains founder and CEO Igal Raichelgauz. “It requires systems that can reason, adapt and make decisions under uncertainty. With Uber and NVIDIA, we are bringing this approach into autonomous ride-hailing.”
Sarfraz Maredia, global head of autonomous mobility and delivery at Uber, said the goal is to create a pathway for automakers and autonomy developers to participate in ride-hailing networks without needing to build separate commercial transportation systems.
NVIDIA’s DRIVE Hyperion platform will provide the high-performance computing architecture required for Level 4 autonomous driving capabilities. Level 4 autonomous driving means a vehicle can handle all driving on its own without a human driver, but only in specific areas or conditions, such as certain city zones or mapped routes. Within those limits, it can steer, brake and navigate traffic independently and does not require human intervention. However, outside approved areas, Level 4 systems may not function or may safely stop.
Autobrains, which holds more than 300 patents, is based in Tel Aviv.
