This article is more than 1 year old
Car makers lock in long-term deals with chip giants for future autonomous vehicles
Not going to get stuck in first gear again with supply-chain crises
The battle between chip giants in the autonomous vehicle space isn't a two-horse race like it was for desktop PCs. Car makers have a somewhat wider choice of silicon vendors.
At CES, Intel, Qualcomm, and Nvidia emerged as leaders in that race, with top automakers locking in long-term commitments on chipsets, components, and software for the brains of autonomous vehicles. And by autonomous we mean technologies ranging from next-generation cruise control and driving assistance to potentially driverless solutions.
Vehicle manufacturers are said to be getting comfortable with these chip suppliers after years of stringent testing and validation of self-driving systems for stability, safety and operation in harsh conditions.
Cars are turning into moving computers, with sensors and AI systems guiding autonomous driving. The technology advances are coming with the rapid electrification of vehicles. Hyundai Motor Company this week announced it wouldn't develop new internal combustion engines, speeding up its transformation to an all-electric vehicle outfit.
Automakers moved quickly to secure technology assets amid the chip shortage and rapid digitization of cars, said Qualcomm CEO Cristiano Amon, during a CES keynote.
"Car companies started to develop direct relationships with technology companies and chipset providers. That got taken to the whole next level as we looked at the supply chain crisis," he said.
Renault announced a partnership with Qualcomm for advanced driver-assistance systems (ADAS) in its cars. Renault CEO Luca de Meo said the partnership will involve the development and implementation of Qualcomm's Digital Chassis, which includes 5G connectivity, telematics, driver assistance, autonomous driving, and a digitized cockpit.
Volvo Cars announced it would use Qualcomm's technology for infotainment systems in its first fully electric SUV launching later this year, said Henrik Green, the company's chief product officer.
Automakers are using conventional engineering approaches to integrate autonomous systems across car models, Amon said. It's understood manufacturers are keen to use computer systems that can have common elements across their vehicles, whether those systems are in the cloud and always connected, or in the car.
"They would develop a drivetrain that they would use across consumer cars," he added as an example.
- Power management IC shortage holding cars, laptops, hostage
- Ford taking control of chip supply in Globalfoundries deal
- Self-driving towards an IPO? Intel unveils plans for Mobileye offering
- Whenever automakers get their hands on chip supplies, the more expensive vehicles are first in line – NXP
Nvidia's Hyperion 8 car computing platform platform, which includes the company's Drive Orin system-on-chip, has been adopted by hundreds of automakers, truck makers, and tier-1 OEMs, said Nvidia's vice president and general manager Ali Kani during a prerecorded CES press conference.
Some car makers buy chips and software from Nvidia and put their own software on top of it, Kani said. Others like Mercedes have relied on Nvidia for the entire stack, including self-driving software, hardware and simulation.
Intel's Mobileye unit, which is being spun off into a public company, announced three new EyeQ chips to power autonomous vehicles. Mobileye late last year passed the threshold of selling 100 million chips.
The forthcoming EyeQ Ultra has 12 RISC-V CPU cores with 24 threads and is made using a 5nm process node. It has 64 accelerator cores, and a GPU for visualization that delivers 256GFlops of performance. The chip delivers 4.2 TFLOPS of performance, and 176 TOPS for AI operations, it is claimed.
The full ECU will be available for under $1,000, and is targeted at cars with high levels of autonomy reaching the market in 2025, said Mobileye CEO Amnon Shashua, during a taped keynote.
Shashua said the EyeQ Ultra chip was the "crowning jewel" of its offerings, and roughly equal to five previous generation EyeQ5 chips, which is based on MIPS CPU cores. The company also announced new EyeQ6L and EyeQ6H chips, which are not as fast as the Ultra chip.
Shashua said Mobileye was working with Ford on new ADAS technologies, and that could include the EyeQ Ultra chip in the future.
"Further out, Ford and Mobileye are working on ... new generation of autonomous technologies that are going to change customers' lives, moving from safety to a doing all sorts of things in a vehicle," Ford CEO Jim Farley said in a broadcast.
Ford also announced it would use Mobileye's cloud-based mapping technology called Road Experience Management in its cars. With REM, Mobileye uses harvested information from existing sensors and cameras in millions of cars to create maps with road information. Ford will feed that information to its autonomous cars for features like lane detection. ®