Physical AI Takes the Wheel: How the World’s Robotaxi Leaders Are Building With NVIDIA Technologies
nvidia ·
NVIDIA outlines its modular, full-stack robotaxi platform as a three-computer solution spanning AI training, simulation and safety validation, and in-vehicle computing. The article states the global robotaxi market is projected to reach $400 billion by 2035, with more than 6 million commercial vehicles in operation. For training, NVIDIA points to its DGX systems and Alpamayo open reasoning vision-language-action models, and says adding meta-action and chain-of-thought reasoning data improved one VLA model’s trajectory prediction accuracy by reducing minimum average displacement error 43%, from 2.08 to 1.18. Simulation and validation rely on Omniverse NuRec models and Cosmos world foundation models running on RTX PRO Servers, with the AlpaSim framework supporting closed-loop testing. For in-vehicle use, DRIVE Hyperion 10 combines dual DRIVE AGX Thor systems-on-a-chip with 14 high-definition cameras, nine radars, three lidars and 12 ultrasonics, while Halos supplies a production-ready safety foundation. NVIDIA says every major robotaxi program operating at commercial scale today uses its modular stack or parts of it across Asia, Europe, the Middle East and North America. The article lists numerous adopters, including Uber, which plans to reach 28 cities by 2028, as well as Waymo, May Mobility, Bolt, Lyft, WeRide with Grab, Wayve with Nissan and Uber, Autobrains, Zoox, Momenta, Pony.ai, Tensor, Waabi, TIER IV with Isuzu, Lenovo with SWM, and DeepRoute.ai. It also notes automakers bringing robotaxis into production, such as Tesla, Mercedes-Benz, Stellantis, Lucid with Nuro, Hyundai and Kia, Geely and Zeekr.
nvidia