NVIDIA has announced a partnership with Uber to expand the deployment of level 4 autonomous vehicles globally. The collaboration will use NVIDIA's DRIVE AGX Hyperion 10 platform and DRIVE AV software, designed specifically for advanced autonomy in robotaxi and delivery fleets.
Starting in 2027, Uber plans to scale its autonomous fleet to 100,000 vehicles using this technology. The companies are also developing a joint AI data factory based on the NVIDIA Cosmos platform to support data processing for vehicle development.
The DRIVE AGX Hyperion 10 is a reference architecture that enables any vehicle to be level 4-ready. It provides automakers with validated hardware and sensors capable of hosting compatible autonomous-driving software, aiming for safety and scalability across different types of vehicles.
Uber's approach will integrate both human drivers and autonomous vehicles into one ride-hailing network. This system will allow for seamless operation between current human-driven mobility services and future autonomous fleets.
“Robotaxis mark the beginning of a global transformation in mobility — making transportation safer, cleaner and more efficient,” said Jensen Huang, founder and CEO of NVIDIA. “Together with Uber, we’re creating a framework for the entire industry to deploy autonomous fleets at scale, powered by NVIDIA AI infrastructure. What was once science fiction is fast becoming an everyday reality.”
“NVIDIA is the backbone of the AI era, and is now fully harnessing that innovation to unleash L4 autonomy at enormous scale, while making it easier for NVIDIA-empowered AVs to be deployed on Uber,” said Dara Khosrowshahi, CEO of Uber. “Autonomous mobility will transform our cities for the better, and we’re thrilled to partner with NVIDIA to help make that vision a reality.”
Several automakers are collaborating on level 4-ready vehicles compatible with DRIVE AGX Hyperion 10. Stellantis is working on platforms optimized for robotaxi requirements using NVIDIA’s full-stack AI technology. Lucid is advancing autonomous capabilities in its next-generation passenger vehicles with this platform as well. Mercedes-Benz is testing collaborations involving its MB.OS operating system alongside DRIVE AGX Hyperion.
In addition to passenger transport, companies such as Aurora, Volvo Autonomous Solutions, and Waabi are extending level 4 autonomy into long-haul freight trucking using NVIDIA's technology.
The modularity of DRIVE AGX Hyperion 10 allows manufacturers flexibility in customization while providing access to prequalified sensor suites and rigorous safety standards. At its core are two DRIVE AGX Thor systems based on Blackwell architecture, each delivering over 2,000 FP4 teraflops of real-time compute power.
NVIDIA’s approach leverages foundation models trained on extensive real-world and synthetic driving data. New reasoning models combine visual understanding with natural language processing to enable adaptive decision-making in complex urban environments. Foretellix is partnering with NVIDIA to validate these models through integrated toolchains.
To further support development in this area, NVIDIA has released what it describes as the world’s largest multimodal AV dataset—comprising thousands of hours of camera, radar, and lidar data from multiple countries—to aid in training foundation models for self-driving systems.
NVIDIA has also introduced Halos Certified Program through its Halos AI Systems Inspection Lab—the first lab accredited by ANSI Accreditation Board—to evaluate physical AI safety standards for automotive applications. Initial members include AUMOVIO, Bosch, Nuro, and Wayve.