Automakers can't wait to launch lucrative subscription services, such as autonomous driving assistance features – and semiconductor giants say they are providing the tools to make that happen.
Intel, Nvidia, and Arm, to name three, are designing chipsets, frameworks, and services so cars can be improved through software upgrades. That provides an avenue for chip and auto companies to get recurring revenue through over-the-air updates, subscription services, and robotaxi fares.
"We've been preaching for a long time: you should really build a software-defined computer in the car," said Ali Kani, vice president of general manager for automotive products at Nvidia during an Evercore ISI webcast this month.
"It's not just easier for OEMs to update, support and upgrade that experience over time, but if you think of it in terms of software and services ... it transforms the business model of OEMs."
Tesla just expanded an earlier beta of its Full-Self Driving (FSD) software, which for $199 a month (or $10,000 up front) gives access to self-driving-ish features like automatic lane changes, auto parking, and identification of stop signs and traffic lights. Early results are not exactly great, with missed lanes, slow driving, and a car that was foxed by an intersection, as you can see below.
While Tesla is mostly self-reliant on hardware, top car and equipment makers are partnering with chip companies on hardware and software. Nvidia's "modular" approach involves providing car manufacturers with autonomous hardware and software, or only the hardware on which automakers can put their own artificial-intelligence algorithms, applications, and sensors.
Betting on upgrades
Nvidia foresees ongoing sales for itself, let alone vehicle manufacturers, through software and services it provides over the lifetime of a car.
"There is recurring revenues in many ways. If you build an application and when a customer upgrades a feature or a function after they buy the car, then there is recurring revenues from that," Kani said.
"Even when we don't build an application, when you're building a production vehicle that has to be safe and secure, we need to maintain that release over the life of the car and so there's fees there."
Nvidia is working with companies like Volvo and Mercedes Benz to build these kinds of software-defined cars.
- If you're Intel, self-driving cars look an awful lot like PCs
- Nvidia gobbles up mapping startup to help automakers install its self-driving platform
- Intel's Mobileye unveils first 'production-grade fully electric self-driving vehicle,' partners with Sixt for Munich launch
- Poltergeist attack could leave autonomous vehicles blind to obstacles – or haunt them with new ones
To be sure, cars need expensive hardware to run autonomous driving applications. Nvidia's roadmap includes producing the 250-teraflop Orin and the 1000-teraflop Atlan systems for powering AI in vehicles in 2024. The computers will be form and power compatible, meaning car owners can upgrade vehicles by swapping out Orin with Atlan.
"We're designing it such that we're going to make it better. Customers are going buy the car, but what the car can do once after it goes off the lot will improve over the life," Kani said.
There will be a price to pay. Tesla is charging $1,000 to upgrade computers on older cars to support advanced FSD self-driving service.
Intel's Mobileye is the opposite of modular Nvidia: it tightly integrates hardware and software for co-pilot and fully autonomous systems. Ford last year signed a deal to include Mobileye electronics and code, on which the car maker is building paid-for driver-assistance services.
The concept of getting a cut on each robotaxi ride powered by one's technology was raised by Erez Dagan, executive vice president at Mobileye, during a chat at the Evercore ISI Autotech & AI Forum. Intel last month announced plans to launch autonomous taxis in Munich, Germany, next year with Sixt as an operational partner.
The x86 giant is focusing on economizing systems that can be subsequently upgraded to provide premium autonomous driving features, Dagan said.
"It is now becoming more critical that we put emphasis on consumer [autonomous vehicles] than in the world of robotaxis where the monetization through a service is a bit more forgiving as far as the economies of the system," Dagan said.
Arm, meanwhile, has thrown itself into the deep end of software-defined vehicles with software frameworks and support for hardware platforms.
We asked the processor core designer how it intends to prevent miscreants from hijacking over-the-air updates to compromise people's vehicles. A spokesperson told us "the adoption of zero-trust security architecture in the automotive space is critical" to thwarting these supply-chain attacks.
Also, security defense mechanisms present in today's system-on-chips that can be used to secure updates and services need to be accessible in a way that's easy for developers to use, or as Arm put it, a "standardized way across the cloud-edge automotive software stack, so that over-the-air solution providers can take advantage of these silicon features using the cloud-native infrastructure." ®