At the Consumer Electronics Show 2025 (CES), Uber and Nvidia announced a major collaboration to promote the development of the autonomous driving field through advanced technologies. The core of this collaboration is to use Nvidia's newly launched simulation tool for generating world models and DGX Cloud, a cloud-based artificial intelligence supercomputing platform, to provide strong technical support for Uber's autonomous driving technology. This cooperation not only demonstrates Uber’s ambitions in the field of autonomous driving, but also marks its continued investment in technological innovation.
Cosmos is a tool designed for robotics and autonomous driving companies, with the ability to generate physics-based high-definition video from a variety of inputs, and can simulate real industrial and driving environments. Behind this tool is a massive amount of data support, including 900 trillion data tokens generated from 20 million hours of video. Through this technology, Uber can test and optimize its autonomous driving algorithms in a variety of complex environments, thereby improving the safety and reliability of autonomous driving systems.
Uber has adopted a "light asset" model in the research and development of autonomous driving technology, and has accelerated the implementation of technology by establishing partnerships with several autonomous driving companies. Over the past year, Uber has signed 14 cooperation agreements with companies including Waymo and Aurora Innovation, covering multiple fields such as robot taxis, truck transportation and sidewalk delivery. This cooperation model allows Uber to flexibly respond to market demands, and at the same time, with the help of its partners' technical strength, further promote the commercial application of autonomous driving technology.
Looking back on Uber's development history in the field of autonomous driving, Uber launched the self-driving project through cooperation with Carnegie Mellon University as early as 2015. Although it has acquired Otto, a self-driving truck startup, Uber's self-driving business has not been able to proceed smoothly due to various reasons. In 2018, Uber's self-driving test car had a fatal accident in Arizona, which also puts the company under great pressure on its development of autonomous driving. Uber eventually sold its autonomous driving division to Aurora Innovation in 2020.
Now, Uber has more clarified its role in connecting passengers and drivers, whether humans or robots. Uber CEO Dara Khosrowshahi mentioned in the company's third-quarter earnings report that they and Waymo launched services in only two cities this year in Austin and Atlanta, highlighting the deployment of automatic deployment in new cities. Investment and preparation required for driving services. This strategy shows that Uber's layout in the field of autonomous driving is more cautious and pragmatic.
Despite the challenges facing the advancement of autonomous driving, Uber believes that through cooperation with Nvidia, safe and scalable autonomous driving solutions can be achieved more quickly. “Working with Nvidia, we are confident that we can accelerate the timeline of safe and scalable autonomous driving solutions within the industry,” Kosroshahi said in a statement. This cooperation not only brought technological breakthroughs to Uber, but also injected new impetus into the development of the entire autonomous driving industry.