Nvidia plans to test a robotaxi service in 2027 in self-driving push

Nvidia has announced its ambitious plans to launch a robotaxi service by 2027, marking a significant step in its quest to dominate the self-driving vehicle market. At a recent demonstration in San Francisco, company officials revealed that the service will utilize cars equipped with 'Level 4' autonomous driving capabilities, allowing them to operate without human intervention in designated areas. However, details regarding the specific location and the partner involved in this venture remain undisclosed. Xinzhou Wu, Nvidia's vice president of automotive, indicated that the initial rollout would be limited as they collaborate with their partner to establish a solid foundation. Nvidia has been a player in the automotive sector since 2015 with its Drive brand, although sales from automotive and robotics chips accounted for only $592 million in the last quarter, making up a mere 1% of the company's overall revenue. In a noteworthy partnership announced in October, Nvidia joined forces with Uber to further enhance its robotaxi ambitions. The company also recently developed software that will be incorporated into Mercedes-Benz models set to debut in late 2026, enabling them to navigate complex urban settings like San Francisco. CEO Jensen Huang emphasized the strategic importance of robotics, particularly self-driving cars, as a key growth area for Nvidia, second only to artificial intelligence. "We envision a future where a billion cars on the roads are autonomous," he stated at the recent CES conference in Las Vegas, suggesting that users could opt for robotaxis or own autonomous vehicles. Nvidia's offerings extend beyond chips for self-driving cars; the company also provides access to its advanced AI chips and simulation software for automotive manufacturers. These tools enable car makers to accelerate the development of self-driving technologies while optimizing costs. Nvidia's Drive AGX Thor automotive computer, priced at approximately $3,500 per chip, is designed to aid manufacturers in bringing self-driving features to market more efficiently. As interest in robotaxis surges, with companies like Alphabet's Waymo already operating commercial driverless services in multiple U.S. cities, Nvidia's announcement signals a strategic pivot towards self-driving fleets alongside personal vehicle sales. In a recent demonstration, reporters experienced a ride in a 2026 Mercedes-Benz CLA sedan, which showcased the car's self-driving capabilities—albeit with a safety driver present. The vehicle managed to navigate autonomously for the majority of the journey, though there were moments requiring human intervention during complex traffic scenarios. Nvidia's technology is currently classified as 'Level 2 Plus Plus,' indicating features akin to Tesla's Full Self-Driving mode. While the self-driving capabilities of Nvidia-powered vehicles will evolve, the responsibility for safety remains with the driver, who must remain attentive at all times. Future updates promise to enhance these vehicles with additional features, such as park-to-park driving. At an Nvidia event, Mercedes-Benz Group CEO Ola Källenius stated that the company is committed to providing solutions for challenging parking situations. The Mercedes-Benz models equipped with Nvidia's technology are set to launch in the U.S. this year, featuring enhanced lane-keeping and driver assistance systems that will continue to evolve through software updates. Nvidia is employing two distinct AI systems in its Drive-powered vehicles to bolster safety. The primary system relies on an 'end-to-end' vision-language model that interprets visual data to navigate, while a secondary safety stack enforces strict rules to manage uncertainties. Looking ahead, Nvidia aims to leverage advances in generative AI to refine self-driving algorithms, with an eye on achieving point-to-point autonomous driving by 2028. Their vision includes creating a vehicle that interacts with users as if it were a human driver, responding to voice commands seamlessly.