Silicon Valley-based Nvidia, a global leader in artificial intelligence hardware and software, is set to release a powerful new processor named Drive Thor in 2024. Nvidia thinks it could revolutionize the electric vehicle industry and lead the way into fully autonomous self-driving vehicles in time for the 2025 EV lineup.
“It absolutely will scale up to full autonomy,” Nvidia’s vice president of automotive, Danny Shapiro, reportedly said at a press briefing on September 19 at the Nvidia GTC conference. Full autonomy could refer to a level 4 on the self-driving scale where the driver is present, but not paying any attention to the road or level 5 where both steering wheels and drivers are optional, hence unnecessary, since the car is fully autonomous.
Shapiro explained that the Thor processor is based on Nvidia’s new Hopper graphics processing unit which accelerates a powerful AI technique referred to as transformers. Thor’s utilization of that Hopper feature is to greatly enhance the handling of the artificial intelligence software used for self-driving cars.
The first carmaker to sign up to receive Thor is Chinese premium electric vehicle manufacturer Zeekr, a company started just last year.
Combining and Streamlining Systems
In addition to its autonomous self-driving artificial intelligence, Thor is being designed to integrate into one system many of the smaller processors that handle the operation of things such as navigation, self-parking, braking, door locks, engine control and entertainment. Currently, many electric vehicle manufacturers use different systems from different vendors.
“If we look at a car today, advanced driver assistance systems, parking, driver monitoring, camera mirrors, digital instrument cluster and infotainment are all different computers distributed throughout the vehicle,” Shapiro said. “In 2025, these functions will no longer be separate computers. Rather, Drive Thor will enable manufacturers to efficiently consolidate these functions into a single system, reducing overall system cost.”
Nvidia had initially been planning a chip called Atlan for a 2024 release, but canceled when Thor arrived on the scene handling 2 quadrillion operations per second, or twice what Atlan and eight times what Atlan’s predecessor chip, Orin, could.
Once fully produced, Thor is expected to be sold by Nvidia in various configurations, not all with top-of-the-line self-driving abilities. Simpler, lower-end versions of Thor could be sold to handle more run-of-the-mill driver-assist technologies, such as lane keeping, etc.
The automotive processor chip producer community is still feeling pressure from the strained global supply chain caused by the Covid-19 pandemic. Not to mention the strong demand from its carmaker customers looking for more ways to use semiconductor chips to lower production costs and/or offer car buyers more options.
The long, drawn-out supply chain debacle has also produced fundamental changes in the relationship between auto manufacturers and their chip suppliers. EV and other vehicle makers are now looking to make large capital investments in their chip producers in exchange for ready, long-term supply.
Prior to the pandemic, chip producers found their car manufacturing customers had little stomach for making investments in chips to secure supply. But like many other things, the chip shortages have reshaped the industry by slowing down manufacturers’ abilities to elicit revenue from hungry car buyers.