^{Photo: Forbes}

Nvidia Pushes Deeper Into Autonomous Mobility

Nvidia is accelerating its ambitions in autonomous driving, aiming to power commercial robotaxi fleets with its AI chips and Drive AV software stack as early as 2027. The company says it is actively working with robotaxi operators to deploy vehicles capable of operating without human drivers in specific, mapped regions.

The target capability is Level 4 autonomy, where vehicles can handle all driving tasks independently under defined conditions. According to Nvidia, initial rollouts are expected to be limited in scale, expanding gradually as partners validate safety, reliability, and regulatory compliance.

From Gaming Chips to Self-Driving Systems

Nvidia has been supplying automotive technology under its Drive brand since 2015, but the segment has historically represented a small slice of its business. In the quarter ending in October, automotive and robotics revenue totaled approximately $592 million, accounting for roughly 1 percent of Nvidia’s overall revenue.

That figure stands in sharp contrast to Nvidia’s data center business, which has surged on the back of generative AI demand. However, executives increasingly describe robotics and autonomous vehicles as the company’s second most important long-term growth category after AI infrastructure.

CEO Jensen Huang has repeatedly framed self-driving technology as a massive future market, envisioning a world where autonomous systems power everything from personal vehicles to shared robotaxi networks.

Targeting Robotaxis and Consumer Vehicles

Nvidia’s strategy spans both fleet-based robotaxis and consumer-owned vehicles. The robotaxi push signals a direct challenge to incumbents like Alphabet’s Waymo, which already operates fully driverless commercial services in multiple U.S. cities, including San Francisco.

In parallel, Nvidia continues to deepen relationships with traditional automakers. In December, the company said Mercedes-Benz models scheduled for release in late 2026 will integrate Nvidia’s autonomous driving technology, enabling advanced city navigation in dense urban environments.

Mercedes-Benz plans to customize Nvidia’s software stack to align with its brand experience, selling advanced autonomy features either bundled with vehicles or as optional upgrades.

Inside Nvidia’s Drive Platform

At the core of Nvidia’s autonomous offering is its Drive AGX Thor automotive computer, priced at approximately $3,500 per chip. The company argues that using a standardized, high-performance AI platform can significantly reduce research and development costs for automakers while accelerating time to market.

Beyond in-car hardware, Nvidia also generates revenue by selling access to its powerful AI chips and simulation software. These tools allow carmakers and mobility startups to train self-driving models, test edge cases, and refine vehicle behavior in virtual environments before deployment.

Automakers can choose how deeply they integrate with Nvidia’s ecosystem, ranging from full-stack autonomy to selective use of training or optimization tools.

Real-World Testing in San Francisco

In December, Nvidia offered journalists and analysts a demonstration ride through San Francisco in a 2026 Mercedes-Benz CLA sedan equipped with its autonomous technology. The vehicle operated with a safety driver present, as required for testing.

According to the driver, the system handled approximately 90 percent of the route autonomously. The drive covered steep hills, dense traffic, and frequent stops, with smooth performance throughout most of the journey.

One notable exception occurred during a complex traffic bottleneck involving buses, parked trucks, and a self-driving Waymo vehicle. In that scenario, the safety driver intervened, reversing the car and waiting for the congestion to clear.

Nvidia classified the system as “Level 2 Plus Plus,” similar in responsibility to advanced driver assistance systems such as Tesla’s Full Self-Driving, where the driver must remain attentive at all times.

How Nvidia Approaches Safety

Nvidia says its Drive-powered vehicles rely on two complementary AI systems. The primary system uses an end-to-end vision-language model that processes sensor data and plans driving actions in real time.

A secondary safety-focused system operates in parallel, applying rule-based logic to enforce strict behaviors such as stopping at stop signs or yielding in uncertain situations. This redundancy is designed to reduce risk when the AI encounters ambiguous or unfamiliar scenarios.

Over time, Nvidia expects advances in generative AI and transformer models to significantly improve autonomous decision-making, enabling smoother navigation and greater adaptability.

The Road Ahead for Autonomous Driving

Nvidia is targeting 2028 for point-to-point self-driving capabilities in consumer vehicles, where cars can drive from one destination to another with minimal human input. Features such as park-to-park autonomy are expected to follow, allowing vehicles to navigate complex parking environments independently.

Ultimately, Nvidia envisions autonomous cars functioning like intelligent digital chauffeurs that passengers can interact with using natural language.

As robotaxi adoption grows and automakers race to differentiate their software-defined vehicles, Nvidia is positioning itself not just as a chip supplier, but as a foundational platform provider for the future of transportation. Whether robotaxis or privately owned autonomous cars dominate the market, Nvidia intends to power both.