In a move that screams “the future is finally getting a business plan,” NVIDIA and Uber announced a partnership at GTC Washington D.C. to deploy one of the world’s largest Level 4 autonomous vehicle networks. The ambitious plan involves scaling up to 100,000 robotaxis and autonomous delivery vehicles, with the rollout starting in 2027. Kicking things off, automotive giant Stellantis is on deck to supply the first 5,000 L4 vehicles for Uber’s robotaxi operations.
The brains of this massive operation will be NVIDIA’s DRIVE AGX Hyperion 10 platform, a production-ready reference architecture designed to make just about any vehicle L4-ready. This isn’t just a beefed-up cruise control system; the platform is built around two DRIVE AGX Thor SoCs based on the Blackwell architecture, processing a firehose of data from a sensor suite including 14 cameras, nine radars, and one lidar. This gives vehicles the computational grunt to handle complex urban driving without a human babysitter, fusing sensor data to navigate the unpredictable chaos of city streets. Uber’s grand strategy is to create a single, unified ride-hailing network where your car might be driven by a human or a silicon brain with more teraflops than you can shake a stick at.
This isn’t just an exclusive club. The partnership extends across a growing ecosystem, with Lucid and Mercedes-Benz also leveraging the DRIVE Hyperion platform for their own L4 ambitions. The autonomous trucking sector is also getting a piece of the action, with companies like Aurora, Volvo Autonomous Solutions, and Waabi developing L4 trucks on NVIDIA’s hardware. To feed the AI models required for this scale, NVIDIA and Uber are also developing a joint “data factory” using the NVIDIA Cosmos platform to process the millions of hours of driving data needed for training and validation.
Why is this important?
This partnership signals a critical shift from scattered, localized autonomous vehicle pilots to a concerted effort at mass-scale industrialization. By combining a standardized, high-performance computing platform (NVIDIA) with a massive global mobility network (Uber) and a legacy automotive manufacturer (Stellantis), the pieces are being set for the widespread commercialization of robotaxis. It suggests the industry is moving past the “is it possible?” phase and into the far more complicated “how do we make this profitable and scalable?” phase. For cities, commuters, and a legion of human drivers, the robotaxi apocalypse—or utopia, depending on your point of view—just got a firm date on the calendar.
