Uber to Build 100,000 AI-Driven Robotaxis with Nvidia Chips

Uber Technologies Inc. has announced plans to build a fleet of 100,000 autonomous vehicles powered by Nvidia Corp.’s artificial intelligence technology, marking one of the most ambitious efforts yet to scale self-driving ride-hailing services globally.

The collaboration, unveiled during Nvidia’s GTC conference in Washington, D.C., aims to begin large-scale deployment by 2027, according to a statement from Nvidia on Tuesday. The expansion builds on a partnership established earlier this year, under which Uber agreed to share its vast driving data with Nvidia to enhance the company’s AI models and automotive chips used by carmakers developing autonomous systems.

Nvidia also introduced its latest vehicle technology platform, called Nvidia Drive AGX Hyperion 10, which will enable manufacturers to integrate compatible hardware and sensors designed to work with advanced autonomous driving software.

As part of the initiative, Stellantis NV will be among the first automakers to supply Uber with at least 5,000 Nvidia-powered robotaxis for use in both U.S. and international markets. Uber will manage complete fleet operations for these vehicles, including remote assistance, charging, cleaning, maintenance, and customer service. Stellantis said it would collaborate with Foxconn for hardware and systems integration, with production expected to start in 2028 and pilot operations launching in the United States before expanding globally.

The agreement positions Uber to accelerate the availability of robotaxis on its platform over the next several years, a move that could significantly reduce operating costs and improve profitability. Developing and commercializing autonomous ride-hailing remains an expensive undertaking, but scaling through Nvidia’s technology could make it more sustainable.

Uber has long signaled its intention to blend autonomous and human-driven services, partnering with more than a dozen self-driving technology developers. It has also invested in several of them as part of its strategy to eventually operate a mixed fleet of human drivers and AI-powered vehicles.

Currently, Uber offers limited autonomous rides in Austin and Atlanta through Alphabet Inc.’s Waymo, as well as with WeRide Inc. in Abu Dhabi and Saudi Arabia. These programs, however, remain small compared to Uber’s massive human driver network, which includes millions of active rideshare drivers and couriers worldwide.

The limited size of its autonomous fleet has made it difficult for Uber to achieve meaningful profit margins from robotaxi services. Much of the operational work — including daily charging, cleaning, and maintenance — is outsourced to third-party fleet operators, adding to costs.

The Nvidia partnership aims to change that dynamic. By leveraging Nvidia’s AI and hardware solutions, Uber expects to significantly expand its robotaxi availability while gradually lowering expenses related to maintenance and technology development.

In addition to Stellantis, Uber’s future and existing partners — including Avride, May Mobility Inc., Momenta, Nuro Inc., Pony.ai, Wayve Technologies Ltd., and WeRide — will also be able to use Nvidia’s systems to contribute vehicles to the 100,000-car fleet. The target figure includes approximately 20,000 Lucid Gravity and Nuro vehicles that Uber previously committed to acquiring over the next six years under separate agreements.

To support this ambitious rollout, Uber and Nvidia are jointly developing a “robotaxi data factory” — a large-scale data infrastructure designed to accelerate the training and validation of autonomous driving models. Uber will collect more than three million hours of robotaxi driving data, which will be used to refine AI performance and safety. Nvidia will supply the processors, neural models, and supporting tools required for data curation, search, simulation, and large-scale training.

According to Uber, this integrated data and AI system will form a “powerful data engine” capable of processing everything from raw driving footage to scenario simulation and synthetic data generation. The goal, the company said, is to dramatically shorten the time between pilot testing and profitable autonomy deployment.

While Uber’s stock initially rose by as much as 1.4% following the announcement, those gains were later erased amid broader market volatility. Still, analysts view the partnership as a major step toward Uber’s long-term goal of achieving profitability through automation.

By combining Nvidia’s computing power and AI software with Uber’s operational scale and transportation network, the project could redefine the economics of ride-hailing — and potentially usher in a new era of large-scale, commercially viable robotaxi services.