Like it or not, the development of self-driving cars is going full-swing ahead. After ending its Cruise robotaxi operations last year, General Motors is now planning to develop its next-generation vehicles using Nvidia’s AI chips, the two companies announced today.
GM will leverage Nvidia’s expertise in high-tech chips to build new cars, optimize manufacturing and build robots using “AI and accelerated computing.” Nvidia already partners with dozens of global automakers to supply its cutting-edge chips that support advanced driver assistance systems (ADAS) and assisted driving, including several Chinese brands.
Now the race is on to procure the latest and greatest chips from the Silicon Valley giant to solve self-driving. Thanks to its AI expertise, Nvidia has become one of the world’s most valuable companies, worth nearly $3 trillion. That’s far more than what Tesla is worth.
Photo by: General Motors
Self-driving cars (which are not yet available to the public) and ADAS need powerful chips to process data from cameras, radars and sensors, helping the car “see” and react in real time. These chips run AI models that detect objects, plan routes and make active driving decisions. But initial efforts from automakers have been marred by safety investigations and controversies.
Nonetheless, Nvidia makes several chips that can fulfill such high processing needs. Instead of going solo, automakers are increasingly leveraging Nvidia’s expertise. Future GM vehicles will use Nvidia’s Drive AGX system-on-a-chip (SoC) that supports Level 2 and above ADAS. They will also use Nvidia’s Drive operation system for in-vehicle screen interfaces.
GM didn’t mention which AGX chip it will be using, as Nvidia offers several versions, including the AGX Orin, AGX Thor and AGX Hyperion. AGX Orin is capable of 254 trillion operations per second (TOPs). AGX Thor uses Nvidia’s latest processors and can deliver 1,000 TOPs, thanks to the Blackwell GPU that’s developed using Taiwanese semiconductor manufacturer TSMC’s fabrication process.
It’s unclear what type of self-driving applications—ride-hailing or consumer use—GM is aiming for with the Nvidia partnership.
Photo by: General Motors
The automaker closed its Cruise robotaxi division last year after a major accident led to the service being suspended. But GM said during its Q4 and full year 2024 earnings call in January that a growing number of GM customers were signing up for the Super Cruise advanced driver assistance system. The automaker said some 360,000 GM vehicles with Super Cruise are now on the road.
GM sunk billions of dollars in its initial effort to develop self-driving cars, but now is taking a more cautious approach as the technology hasn’t really matured. Most consumer applications are eyes-on systems, meaning drivers are required to supervise at all times. Some are hands-off, but still require full driver attention.
And except for Google-parent Alphabet’s Waymo, there’s no active robotaxi service in the U.S operating at scale. Tesla has said it will begin its first robotaxi service in Austin starting June—something CEO Elon Musk believes can rocket the company’s valuation into the trillions. But his autonomy timelines have long been optimistic.
Partnering with Nvidia may be a smart move for GM, as the chipmaker is a leader in this space with over a dozen clients already, including BYD, Li Auto, Geely Group, Volvo, Toyota, Mercedes-Benz and many others. Now we’ll see how fast the fruits of this partnership gets to the market here in the U.S.
Have a tip? Contact the author: [email protected]
