Nvidia is gearing up to deliver Drive Thor, its next-generation automotive-grade chip that the company claims will be able to unify a wide range of in-car technology from automated driving features and driver monitoring systems to streaming Netflix in the back for the kiddos.
Thor, which goes into production in 2025, is notable not just because it's a step up from Nvidia's Drive Orin chip. It's also taking Drive Atlan's spot in the lineup.
Nvidia is scrapping the Drive Atlan system on chip ahead of schedule for Thor, founder and CEO Jensen Huang said Tuesday at the company's GTC event. Ever in a race to develop bigger and badder chips, Nvidia is opting for Thor, which, at 2,000 teraflops of performance, will deliver twice the compute and throughput, according to the company.
"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," said Nvidia's vice president of automotive, Danny Shapiro, at a press briefing Monday. "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."
One chip to rule them all. One chip to help automakers build software-defined autonomous vehicles. One chip to continuously upgrade over-the-air.
Nvidia already has several automotive customers that are building software-defined fleets using Drive chips. For example, Volvo announced in January at the annual CES tech conference that its new automated driving features would be powered by Drive Orin. The automaker also said it would power its infotainment system with Qualcomm's Snapdragon chip. It's precisely this space sharing with competitors that likely drove Nvidia to create a more robust chip.
Zeekr is the first to raise its hand for Thor. The Chinese luxury EV startup owned by Geely said it will use the advanced chip for its next-generation of vehicles starting in 2025, according to Shapiro.
XPeng is already using Drive Orin, the latest generation chip, for its G9 SUV, which will be able to support highly advanced driver assistance functions, "such as parking and driving on main and secondary streets, highways and private roads, and safely handling, entering and exiting highways, city byways and toll collection routes," said Shapiro. No doubt XPeng, which has recently launched its City Navigated Guided Pilot ADAS in its P5 sedan and is planning to roll it out in the G9, will sign on for the upgraded chip.
Shapiro also noted that autonomous solutions provider QCraft will begin robotaxi operations in China powered by Orin.
Other automakers that have previously announced use of Nvidia's Drive Orin include Baidu's EV company JiDU Auto, NIO, Li Auto, R Auto, IM Motors and Polestar. It's notable that a fair number of Nvidia's automotive customers are based in China. While the chipmaker is based in California, its chips are produced, along with pretty much everybody else's, in Taiwan.
Can Nvidia deliver Thor to Chinese customers?
Earlier this month, the U.S. government imposed export restrictions on advanced AI chips to China, including Hong Kong and Russia. Nvidia doesn't sell to Russia, but the sanctions on China could cost the company as much as $400 million in potential sales in the third quarter. The U.S. said the move would address the risk of chips being used in or diverted to a "military end use" or "military end user" in China and Russia, but it's also a move by the Biden administration to keep China from becoming a more dominant player in the essential and lucrative chip production industry.
The government has restricted access specifically to Nvidia's A100 and H100 graphic processing units. Fortunately for Nvidia, the company can keep manufacturing the H100 in China, though purchases by Chinese customers will be restricted.
Shapiro said automotive customers won't be affected by the restrictions posed to Nvidia's high-end data center products, and that the company is working with Chinese customers and the U.S. government to "come up with different alternatives that are not subject to the same license requirements."
Updates to Drive SIM and digital twin tech
Nvidia Drive SIM is built on Nvidia Omniverse, which provides the core simulation and rendering engines for autonomous vehicle development. Image Credits: Nvidia
At the GTC event, Nvidia also announced that its end-to-end simulation platform, Drive SIM, is getting a new suite of AI tools that it's calling the "neural reconstruction engine" in order to assist the testing and development of self-driving vehicles.
"Using a neural engine made up by multiple [deep neural networks], real-world rides can be recreated and replayed precisely in simulation with the ability to change sensor configurations and location, create new scenarios, and modify or add new aspects of behavior of other road users," said Shapiro.
The way it works is the AI can deconstruct a 3D scene from recorded sensor data, and then that scene can be augmented in Drive SIM with either human-created content or AI-generated content.
Nvidia said these upgrades will also allow car designers, software engineers and electronics engineers to collaborate in Drive SIM in order to simulate the software inside the car.
The company also announced the second generation of Nvidia OVX, which will deliver immersive digital twins of cities that can be operated in Nvidia's Omniverse.
BMW Group and Jaguar Land Rover are some of the first customers to use OVX, said Shapiro, noting that marketing group WPP is using the Omniverse cloud to create a suite of services for automotive customers, like personalized programmatic ads that feature "perfectly photo real content via virtual sets, which will save the automotive companies orders of magnitude in costs" for expensive photo and video shoots that are usually done in scenic locations around the world.