S1 EP31 - Automotive Megatrends

much more. To stay up to date on future episodes, please be sure to subscribe to the Marvell essential technology podcast. Hey Willard, it's great to have you on today's episode, really looking forward to our discussion today. One of my first questions is what are some of the future megatrends in automotive?

ones too. There's autonomous, there's connected, and there's software defined vehicles and connected as when, you know, we're using, you know, Wifi, Bluetooth, the cloud, you know, connectivity to do things like vehicle to infrastructure, you know, leveraging the cloud to do voice recognition, or over the air update. So there's a lot of activities. And then again, like I said, there's software defined vehicles, which leverages both the concept of connectivity, but enables things like autonomous.

So, you know, like I said, three major mega trends out there autonomous, connected software defined vehicles.

software defined vehicles. And the reason why they love it is that they want to get reoccurring revenues the same way mobile phone, you know, game, game application guys will get, you know, a monthly fee, perhaps to get a game or even to have like, you buy things in the game and things like that. So, for the OEMs, they want to get this reoccurring revenue for things like self driving mode.

And, you know, and be able to get this reoccurring revenue, a subscription fee on top of selling you a vehicle and want this reoccurring revenue. And then the consumers kind of like it, because you think about it, the potential of getting new features over the life of the vehicle that may not even existed at the time they

purchased the vehicle, right? So maybe this feature didn't exist, but you know, you bought your car, and now it does, and you now want it but you don't want to go buy a new vehicle, you just kind of want to upgrade your car to be able to support that new feature. Well, you just download it, so to speak, and, and then you have it. And then the consumer also kind of likes it. Because you can imagine that if you had a feature and it improved over time, you could then say, hey, I want to upgrade

my feature. Yeah, I might have to pay a little bit more this might be like, let's say your engine control or your EV car had a better algorithm to get better mileage, well, who wouldn't want better mileage, and if it only cost you, you know, $1 a month, you know, hey, you probably be for it. Or maybe there's a new graphical cluster, graphics display to make your car look more cooler than it did when you first bought it. You know, those things people may pay, you know, various amounts

of money. You know, in some cases, maybe you're talking $1,000 a month for self driving. But maybe for something like a little bit better mileage or a cooler graphic, maybe you're paying a couple of dollars a month. So and then, of course, the consumer could like it for another case where let's say you're buying a used car from somebody. And now, you know, the previous owner didn't have the feature that you want it but you can just add. So there's a lot

of things for the consumer. And then the third party, you know, service providers. I mean, think of it as an ecosystem of app developers, you know, who wouldn't want to be able to leverage an existing car base or deployment out there like the the Fords and the GMs and all these car makers who have this great deployment of vehicles out there, and then be able to sell new features and capabilities into that market?

concurrently. And, and those are things that you have to start thinking about now, just like on your mobile phone, you can be running in a lot of different apps. At the same time, I could imagine that there could be this mode for all those creators out there that create content on the internet. And, you know, as a creator, he might be driving on down the road, and he's, let's say, in the National Park, and

he wants to take a selfie. And, and that would be great that he could leverage that what we call a driver monitoring camera, take a selfie, but then he could use the surround view camera, to look at the surroundings around his vehicle. So he could then have this overlay where he has a selfie of himself, and the surround view picture and say, hey, yeah, I'm in the National Park. And this is where I'm at today. And here's what I'm seeing today, isn't it awesome. It's not a feature I would want.

But the new generation, certainly, it's probably something they would want. But then you can have more practical applications that could also add to the compute requirements, like student driver mode, you've got a young teenager, and, you know, he's learning how to drive. And if he had an abrupt braking event, you'd want to capture the 10 seconds before

the 10 seconds after. So you could sit there and review all the cameras and say, Hey, I know why you had a near accident, you were on your cell phone, here's the camera picture of you looking at your cell phone while you were driving. So you know, there's a lot of compute, that needs to support all these different applications. And you got to aggregate all these things, because it could be all running concurrently or instantaneously. And, you know, it's complicated, it's pretty

complicated. There are other compute elements that could come in, in terms of, you know, concurrent workloads. And it could be anything from other applications like pet mode, where you're, you're monitoring, whether your pets are in distress when you leave the vehicle, or it could be sentry mode, where you have the surround view and the interior camera act as a security system for your vehicle. So all these different types of applications

can kind of come into play. But you know, you can have other things just as logical as park assist, biometric identification, or even video conferencing. So a variety of things that could be out there in terms of that would demand more compute power, and they all could be running at the same time in theory.

picture. As you add more, you're going to need five nanometer technology because five nanometer allows us to integrate more CPUs, more hardware accelerators, more security, more Ethernet switches, more AI, you know, five nanometer allows us to do the integration and do it at 30% more density than seven nanometer. Five nanometer allows us to consume less power 40%, more than than seven

nanometers. So, five nanometer is a huge enabler to allow the OEMs to put more compute capability in the vehicle so they can run more software defined workloads.