NVIDIA and Uber Partner to Launch Global Robotaxi Network

At its GTC Washington, D.C. conference, NVIDIA announced a landmark partnership with Uber to power the world’s largest Level 4 autonomous mobility network, signaling a major step toward robotaxi-ready cities. Using NVIDIA DRIVE AGX Hyperion™ 10 and DRIVE AV software, the companies will jointly develop next-generation autonomous fleets and data infrastructure.

Under the collaboration, Uber plans to scale its robotaxi and autonomous delivery network to 100,000 vehicles by 2027, built on NVIDIA’s L4-ready platform and AI systems. The alliance also includes development of a data factoryaccelerated by the NVIDIA Cosmos™ foundation model platform, enabling real-time curation of driving data for training and validation.

The DRIVE AGX Hyperion 10 architecture provides validated hardware, sensor suites, and compute capability for safe Level 4 autonomy. Automakers including Stellantis, Lucid, and Mercedes-Benz have joined the ecosystem, integrating NVIDIA’s full-stack AI to accelerate autonomous fleet deployment.

NVIDIA’s DRIVE AGX Thor chip, based on its Blackwell architecture, delivers over 2,000 FP4 teraflops of compute power, optimized for transformer and Vision-Language-Action (VLA) models that give AVs humanlike reasoning in unpredictable driving environments.

Through generative AI and foundation models, NVIDIA’s approach enables AVs to handle complex traffic scenarios safely. Its new Halos AI System Inspection Lab sets safety and certification standards for AI-driven vehicles, ensuring trusted large-scale deployment.

“NVIDIA is the backbone of the AI era,” said Uber CEO Dara Khosrowshahi. “This partnership transforms mobility, merging human and autonomous fleets into one intelligent network.”

With NVIDIA’s expanding ecosystem—including Aurora, Volvo, Waabi, and Wayve—the collaboration marks a pivotal moment in the evolution of AI-powered mobility, positioning both companies at the forefront of the autonomous transport revolution.