Uber relies on Nvidia to accelerate autonomous technologies

Uber announced at CES 2025 that it will use Nvidia's new tools, including the generative world model simulation tool Cosmos and the cloud-based AI supercomputing platform DGX Cloud, to advance its autonomous vehicle technologies. This collaboration promises to significantly strengthen Uber's position in the field of vehicle autonomy. Cosmos is aimed at companies in the fields of robotics and autonomy. It generates physics-based videos from diverse inputs, including 9,000 trillion tokens from 20 million hours of potentially copyrighted video material. These simulations create realistic industrial and driving environments. With Nvidia DGX Cloud, Uber and other companies can access a powerful AI infrastructure to train, customize, and deploy self-driving models. Although Uber disclosed few details about its specific plans with Nvidia's tools, the company is heavily investing in collaborative approaches. In recent years, Uber has formed 14 partnerships with AV companies, including Waymo in the robotaxi sector and Aurora Innovation in trucking. Uber's strategy deliberately focuses on partnerships rather than developing its own AV technology, an approach shaped by the company's tumultuous history in this area. The former strategy of in-house development led to controversial incidents, such as the legal dispute with Waymo and a fatal accident involving a self-driving Uber vehicle in 2018. Subsequently, Uber sold its self-driving unit to focus on partnerships. Today, the company positions itself as an interface between passenger and driver, whether human or machine, and is ready to efficiently implement advanced technology solutions. The recent introduction of Waymo services in Austin and Atlanta exemplifies Uber's determined yet cautious push into the future of mobility.