How Aptiv is Driving Innovation in Software-Defined Vehicle Era

Software-defined vehicles (SDVs) are increasingly playing a crucial role in automakers’ product evolution. From consolidating hardware to providing regular software updates and the necessary compute power for future applications such as autonomous driving and digital cockpits, the SDV will become a defining enabler.

In this latest episode of ‘The Counterpoint Podcast’, host Murtuza Ali is joined by Benjamin Lyon, Vice President and Chief Technology Officer at Aptiv, a global technology firm and automotive supplier. The conversation dives into Aptiv’s software platform and service offering, which are enabling the SDV for automotive OEMs.

Follow us on social media platforms –
Twitter – https://twitter.com/counterpointtr
LinkedIn – https://www.linkedin.com/company/counterpoint-technology-market-research/
YouTube - https://www.youtube.com/@counterpointresearch/
Instagram - https://www.instagram.com/counterpointresearch/

Transcript

[Murtuza] (0:00 - 0:57)

Hello, and welcome to The Counterpoint Podcast. I'm your host, Murtuza Ali, a Senior Analyst at Counterpoint Research with a focus on automotive research. Today, I have a really exciting topic that we'll be discussing. It's innovation in the software-defined era. We have Benjamin Loyn, Senior VP and CTO at the global technology firm Aptiv joining us for this conversation. He leads the technology leadership at Aptiv, has an amazing career where he's been doing rocket research or rocket engineering, and also has two decades of experience at Apple. Welcome to the podcast, Benjamin. It'd be great if you can introduce Aptiv and yourselves in a little brief for our listeners and viewers. 

[Benjamin] (0:57 - 3:42)

Thank you, Murtuza. Well, thank you a ton. It's really great to be here, and thanks for inviting me to chat with you a little bit about software-defined technology and about Aptiv and what Aptiv's doing in the space.

Look, just to introduce Aptiv, Aptiv's actually been in the world of mobility for a very, very long time. And so, you think about myself, I come from the world of consumer electronics and consumer mobility and fundamentally trying to bridge between humanity and the humanities and technology. How do you make technology simple, seamless, personalized, easy to use?

And if you look at the world of automotive, it's going through those transitions now as well. And so, in some ways, it's kind of the last undiscovered, less, what do you call it? A new frontier, right? Exactly. Exactly. It's the new frontier.

So, in some ways, it's the new frontier of user experience and technology. And so, after spending two decades trying to figure that out, a thousand Psalms in your pocket, put the internet in your pocket. How do you take all of this complex technology and create a really simple, easy user interface that even your grandparents can use?

But it's also appropriate for the digital native generation. Similarly, Aptiv has been focusing on these mega trends of safety critical, which is a totally different thing than consumer, right? It's got to work every time and all the time. So, safety-critical, green, right? These are vehicles. They take people all around. And so, they consume energy. And so, you have to make them really efficient. So, how do you get green? How do you electrify and connect it? What does it mean when everything, all vehicles are connected to the internet? How do you create that seamless experience?

And so, Aptiv is a company that, while it has a very deep tradition in supporting all of automotive technologies, specifically in this era of electrification and connectivity, Aptiv really is a global leader, providing literally sensors to cloud, high voltage and electrification solutions. So, it really covers the gamut. So, it's a neat place to be because you get to tackle all sorts of interesting problems that fundamentally are enabling our customers, who are the OEMs, to step into that digital age, to step into that electrification age.

[Murtuza] (3:44 - 4:17)

Yeah, great. That's why I called Aptiv the technology firm rather than a supplier, so to speak, in the automotive sphere because we've been hearing a lot and know about Aptiv, you know, improving its offering in the automotive sector, where it's being quite connected, autonomous, and electric green. So, great introduction to Aptiv.How about yourself, Benjamin? Give a brief background about yourself as well for our viewers. Sure.

[Benjamin] (4:18 - 6:37)

Well, I mean, just really briefly, I love hard problems, right? And mobility and connected mobility is a really challenging set of problems. And that's been true for me my entire life.

I grew up in the Bay Area. So, I grew up in kind of the heart of near Silicon Valley, up near San Francisco, and in the United States, and went to Carnegie Mellon University. And I was actually an extremely young employee when I came to Apple.

I was 19 years old. And so, I was at the dead center of the kind of the rebirth of Apple, starting with the iMac, and then rolling through the iPod, and the iPhone, and the watch, and the iPad, and the like. I've always had a tremendous amount of passion for that connection between technology and people, and robotics is a natural place for that.

And of course, autonomous vehicles are robots at the end of the day. And so, I spent roughly a decade between Apple and a company called Astra working on autonomous robotics, both things that drive around on the ground as well as go to space, put 23 satellites into orbit, and learned all through that. But at the end of the day, here's what drove me to Aptiv, which was through the building of that beautiful house.

I realized it was all missing this core foundation, that operating system, and middleware, and the core sensing, compute technologies, so both hardware and software, that could really work every time and all the time, and be cost effective, be something people could actually afford. And Aptiv is committed to building that foundation. And so, after kind of taking this loop through a decade's worth of robotics, I was really excited about the opportunity to go and work together with all of the people at Aptiv to really work on that foundation, that we could then build an incredible future on top of for the flexible, internet-connected, self-driving future.

[Murtuza] (6:37 - 7:41)

Yeah, it's fascinating, isn't it? I've been nearly 20 years now in the automotive industry, and this is one of the most exciting times to be in the automotive industry. It's just so much happening in every aspect of the automotives or car that the mind boggles, and there's a lot of ground to be covered by the automotive industry as well to get to this seamlessness between, as you say, human and humanity.

So in terms of software-defined vehicles, this is a key focus for all of OEMs. How is Aptiv enabling this transition between the embedded hardware that automotive industries are used to, to software-defined vehicles? What is Aptiv doing in that space that can help their products in the future?

[Benjamin] (7:43 - 10:47)

Yeah, first of all, that's a really great question, and it's one that we think about a lot, because the answer is not vanilla. It's actually very different depending on each of the OEMs. While they're all shifting towards software-defined vehicles, really to prioritize choice and flexibility and openness in their platforms, and of course, they're looking for efficient ways to build and deploy this complex software in order to maintain cost while delivering the capability, its every OEM is on a slightly different journey, and they're in different places along that journey. So, Aptiv has to meet them where they are. And so, the easiest way to simplify it down is Aptiv needs to provide the right hardware, software, and services, or the right software, hardware, and toolchain for wherever that particular OEM is on the journey.

So, we have something called smart vehicle architecture, which is really a modern, sustainable platform, overall hardware platform, that manages and optimizes to minimize the overall complexity. So, as you add sensors, you add compute, the old days, you would add another box for every new feature. And think about how quickly that complexity gets out of control, and the associated cost, and the associated reliability challenges that come with that, especially in a safety environment, right?

And so, Aptiv provides this SVA, the smart vehicle architecture, really in order to simplify that down and provide solutions for our OEMs that are looking for that. The second is cloud-native software. So, software that is designed from the ground up for over-the-air updates.

And one key aspect of that cloud-native software is containerization. And if you think about software, the old way of building software is what we call a monolith. So, everything's interconnected, every piece of software, every feature depends on each other.

And so, it makes it really hard to do updates because whenever you want to update just one feature, think about like just a font, for example, or an icon on the screen, you have to end up testing all the rest of the software to make sure it still works.

So, but if you take that and you do like containers, it's like a puzzle with lots of puzzle pieces with clear interfaces for how they connect. And then you can test each of those puzzles, just the puzzle piece that you're changing. And then do the update of just that piece and expect the system.

So, that's cloud-native software. And then lastly is Wind River Studio, which is a complete DevOps tool chain. It's really a DevSecOps tool chain.

We can talk about what that means. That really streamlines a software development lifecycle. And that really improves efficiency and reliability in addition to kind of the time to market, right? The time you can deliver that new software feature to your customer.

[Murtuza] (10:49 - 11:18)                                  

No, that is just, you've just touched the right tunes there on what's going on for the challenges that the automotive industry has and how you have divided into these three parts where logically you can serve the OEMs. And like you said, meet them where they are because the priorities on like a different journey. Some people are much more, some OEMs are much more advanced than others. And there is a lot of variety and mix.

[Benjamin] (11:18 - 12:48)

Yeah, there totally is. Let me give you maybe a specific example. So, Aptiv also has these advanced safety and intelligent systems, including perception, decision-making, fully integrated, including mapping, ADAS solutions that range all the way from just compliance or regulations, right?

Like autonomous emergency braking, AEB, all the way up through level two plus plus. As a matter of fact, Aptiv was the first to get there with that kind of an offering. Aptiv was the first to put a radar on a car.

We have customers that what they want is they want a perception system. So, think cameras and radars, fusing all that together and feeding their system. Hey, what's out there in the world? Where are the people? Where are the cars? That kind of stuff. What's the state of the streetlights? 

There are other OEMs that want a complete package from us where they actually want hardware, software, the end-to-end algorithms, the mapping solution. They want it all integrated, tested, and it's just got to work.

And so, we have to not only develop this package that works as a whole, but we also have to be able to, because it's modular, because it's containerized, we can also just provide the piece that our customer needs that they're missing, right? So, that's one way we can kind of make it work. Yeah. Yeah.

[Murtuza] (12:48 - 13:36)

No, that's fantastic. That's a very well thought out way of dealing with this own complexity in the automotive industry where OEMs are different levels and you've got to, as a technology provider, meet them in their needs. Great, great, fascinating work at Aptiv.

When we talk about this specific subject where OEMs, like I said, are different paces, there's a lot of talk about OEMs getting into vertical integration and developing this in-house software. How do you see these make versus buy decisions, buy these OEMs for SDVs being made, and how does it affect or impact a company like yours in those decisions that they're doing?

[Benjamin] (13:37 - 15:50)

Yeah, well, this is definitely an evolving story, right? And it changes both regionally and in various different markets, as well as kind of where OEMs believe they can differentiate. And so, as we were just talking about, a key piece is we emphasize meeting customers where they are, right? We've got to do that. And that flexible approach is reflected in our ability to offer complete systems or individual components, really just like we were talking about. 

I think the other piece, though, is in China, for example, where local OEMs are really rapidly adopting EVs and software-defined technologies, we're seeing way more openness to full system solutions. And so, there's a real focus from them on, hey, we can differentiate in specific areas of the customer experience, but the underpinning technology, end-to-end technology solution, for example, for ADAS, or for driver or cockpit monitoring systems, or even for the core cockpit software, they're really open to sourcing that. And they actually want to source that because they see their differentiation as being really the layer above that in the direct user experience. The other thing we're doing is we're really partnering regionally with local technology, best-in-class technology suppliers, so that we make sure that the solutions that we offer really use the best of what we do, the best of what our partners do, in order to make one plus one equals three, right?

At the end of the day, you want the solution to be more than the sum of its parts, and that can be really compelling for our customers. So, it's not just about, especially with the regional approaches, it's not just about make versus buy, vertical integration versus not. It's also about making sure that we have regionalized solutions and we work with local suppliers, especially startups who are doing some great work.

[Murtuza] (15:51 - 16:12)

And that is, I think, a fundamental change in this software-defined vehicle era, where you're seeing a lot more collaboration and partnerships evolving, rather than the old supplier OEM relationship, isn't it? Where we get the best, like you said, the best of everyone, and one plus one equals three, rather than two.

[Benjamin] (16:13 - 17:50)

Yeah, you're totally right. The challenge with that is that the old school way of developing software was to really develop it in black boxes, and then you have to glue it all together, and it can be really complex to do. And when it finally works, you don't really want to touch it, which is why cars would stay with the same software for decades, unless there was like a recall, which is the worst reason to need to update your software, right?

Not you're bringing new value, it's because something you thought you were doing was broken. But with open platforms, and that's part of why Wind River was so important to Aptiv, was to build out an open platform that works every time and all the time. That has been key.

I'll throw one other thing out there into this conversation, which is, you know, as we were thinking about that platform, about a platform that has to work every time and all the time, one of the things we didn't want to do is reinvent the wheel. And so part of the key kind of thinking around Wind River is that they've actually already worked with aerospace and defense, and with telecom, to go through this software-defined transition. And so really a core part of the strategy is bringing that experience into automotives, rather than learn all those hard lessons all over again.

And that's especially important in a world of connected, you know, connected vehicles, where the cybersecurity matters. It's not just on-vehicle software, on-vehicle safety, it's the whole connected challenge.

[Murtuza] (17:52 - 18:29)

That is just fascinating, because you touched upon another element where the automotive industry has to open out and look elsewhere, because it tends to be quite inward, or has been in the past at least, quite inward-looking, and we know the best, and we know all the DFMEs are not assertive. Whereas now they're going beyond their industry and looking at best techniques from outside. And really, from a software perspective, that's very interesting, because it's lagging in software.

And you can learn a lot from these other industries that you mentioned that Wind River was working with, and leverage that to automotive software.

[Benjamin] (18:30 - 20:20)

Yeah, and the learning can be both ways, right? You know, automotive has its own things that it's learned the hard way, right? Driving, you know, when you get into an airplane, for example, and you fly in FAA or ICAO airspace, so I just kind of named the two agencies that govern how the rules for flying in the United States and all through Europe, right, and the rest of the world, that's pretty universal.

But if you look at, for example, the driving infrastructure, driving in the United States is, it's locale by locale, right? It's city by city. It's not even state by state.

There's tremendous differences in, you know, our autonomous driving system has to recognize elephants for some of the places where our L2++ ADAS drives. There are other places where we have to, you know, recognize other vehicles that don't look like your standard vehicle that you'd normally have on the road. So, that is a unique set of challenges in automotive.

And actually learning how to deal with new and different scenarios that you haven't previously expected or seen, that's, first of all, a great application of AI. It's a great application of deep learning and transformers. That is all stuff we can actually take back and think about, well, how do you deal with Intellicom, which is another piece of Aptiv's business through Wind River, how do you deal with unexpected changes in the network and making sure that our wireless networks are robust and work every time and all the time, just like we need our vehicles to work every time.

[Murtuza] (20:21 - 21:00)

Yes. Yes. And I think some of those challenges are very unique to automotive industries where you can't have those failures. You can't have a restart of an app. You can't have loss of connectivity, for example. Those are very, very unique to, you know, the automotive industry, some to aerospace as well.

And that was a brilliant analogy that you've drawn as well about, you know, learning from automotive where new is, you know, it's like identifying an animal like an elephant or a different vehicle like a three-wheel on the road. It's just amazing how much variety there is on the road, isn't it?

[Benjamin] (21:01 - 24:12)

Yeah. No, it really is. It really is. And, you know, I've had similar bizarre experiences. I'm a pilot, you know, as a hobby, and I once flew past the Goodyear blimp. And I had the, you know, the illusion of feeling like I was standing still in the middle, you know, thousands of feet up in the air as this huge blimp is flying by.

And the flip side is, is I've been in an autonomous car that had to navigate around another vehicle that looked like a banana. And one of the beauties about, you know, Aptiv is bringing transformers and deep learning and how our ADAS makes decisions. And one of the beauties of that is that it's really good at handling new and different scenarios that it's never seen before, as opposed to kind of the previous technology that is very well-tested and very robust and has a lot of history behind it, but at the end of the day is rules-based, which means you have to write a new rule for every scenario that you find.

And so, it's much, you know, much more challenging to handle new scenarios. So, I love the investments that we're making in AI. I'll tell you one other crazy thing, because I mentioned DevSecOps.

And one of my, you know, experiences over the last, gosh, it's a decade and a half at this point, of working on robotics is that if you don't have great tools, it is very hard to make great products. Imagine being like a watchmaker and trying to make a fine watch, but all you have is a sledgehammer, right? Everything looks like a nail.

Or, you know, a brain surgeon or a heart surgeon, you need the right tools. So, similarly, you know, Aptiv cybersecurity, Wind River's DevSecOps platform, Wind River's DevSecOps platform really provides a great end-to-end set of tools for doing rapid software development in a secure environment. So, we can bring AI in and use AI to actually improve our ability to detect vulnerabilities in our software to cybersecurity threats.

And of course, what's coming is quantum, you know, quantum cybersecurity threats and things like that. And as we build those technologies at Aptiv and Wind River, we can integrate all of that into Wind River Studio and make it available as an open platform to our customers. Part of what's really, really powerful, you know, we talked about meeting customers where they're at, they can use the parts of Wind River that make sense for them.

You know, again, some customers want the whole thing. Some customers, what they really want is the design feedback loop or the ability to simulate. It really, really varies, but that's what's so important about having these open platforms, right? You go at the right, you're limited only by the speed of your imagination.

[Murtuza] (24:15 - 24:28)

That's so fascinating that it is enabling the automotive OEMs to really address their issues as they feel ready, as they are maturing.

[Benjamin] (24:29 - 25:40)

Yeah, and it's more than just address their issues. It's also to enable them to thrive, right? You know, the OEMs, they've existed for, in some of them, for over a hundred years because they're delivering something very important, right? You need to get from place A to place Z. And as that experience becomes more and more automated, customers have more and more choice. Some customers may want to drive that whole distance.

Other customers may want to be able to get their time back and use some portion of that time to do other things. How do you do that well? How do you do that safety? And how do you monetize it? 

And so, again, by having a connected software-defined platform, it enables the OEMs to provide new, different, integrated, seamless, personalized experiences for customers, and at the same time, make money at it, right? Because they're providing their customers value.

They should be paid for the value that they're providing. Our OEM customers want to provide the customer new, different, better value versus time.

[Murtuza] (25:42 - 26:35)

That is also very, very true of the opportunities software-defined vehicles offer to OEMs, where it's not just about connectivity. It's not just about, you know, doing what they are or were doing better. It's about creating new value for themselves, enabled through this interaction with consumers, which they've never had before, to the extent that they're having now.

If I just pivot a little bit on what's happening in terms of the OEMs' transition in terms of the cost and margin pressure that they have on moving to STV, where do you think Aptiv can leverage, and where are you looking at things to help them cut down their own investments and costs on the products?

[Benjamin] (26:36 – 31:31)

Yeah, great question. Well, the first is, and you're gonna laugh because we hear this phrase often in the tech industry, which is abstraction, right? Hardware, software abstraction, physical layer, software layers, network layers, all these various different abstraction layers, right?

And it's interesting because it is a shift from thinking about domains. You know, the part of the car that plays music, the part of the car that drives the car, the part of the car that, you know, handles the lights and synchronizing the blinking lights to really thinking about zones and thinking about layers. And the tech industry has already moved to the world of layers, right?

Well, why think about layers? Why does that matter? Well, because it actually allows you to separate it. So that's why we say abstract. It allows you to separate the software application from the specific hardware that's underneath. And that is really important because the moment you have choice in what hardware you use, it means that now there's competition for that hardware.

So it provides the flexibility to our OEMs to choose what hardware makes sense for them. And so we provide abstraction layers. So when I talked about that foundation at the beginning of our conversation, what is that foundation? Well, a big piece of that foundation is the operating system and what we call middleware, which does all of the abstraction. It does all the translation. So imagine, for example, it's kind of like, for example, GPT.

GPT is amazing because it translates. You know, I can pull up a poem in Chinese and have it translated into English instantaneously or into Spanish or into German or into Vietnamese, right? Or into Hindi, like, instantly.

So similarly, Aptiv builds this abstraction layer that does all of this translation. Now, the difference, of course, is that it has to work every time and all the time, right? It has to work in a safety-critical environment. So that's part of what makes it really special. The other thing that we do is we work on a system that allows for something called mixed criticality. What does mixed criticality mean?

Well, basically, think about it as usually the way hardware-software systems and safety environments in the past were designed is you run kind of all the not-safety important stuff on one piece of hardware, and then you have an air gap, a firewall, and then on the other side, you run all your safety-critical stuff. Well, that means you have a lot of hardware that sits around doing nothing. And the reality is that just like we've gone through the experience of having lots of boxes that did very little into a few boxes that did a lot, similarly, we can use software to actually separate and keep protected safety-critical applications from things like watching movies, playing games, infotainment, controlling the lighting in your cabin, things like that, and run that on the same chip, which means that our customers can just buy one chip instead of having to buy multiple chips for the same function, and that, of course, saves them cost.

It does mean that the software is now more complicated because it does more, right? So that's why you need a great toolchain that handles that complexity so that to the developer, they're not slowed down by the added complexity. So mixed-criticality, that's another key thing.

And then, of course, the SVA that we talked about, that smart vehicle architecture, where now you're building a car with fewer boxes that do functions, it means that the amount of factory floor space that's required to assemble these vehicles is less. And so they save on direct labor as a result of our solutions. Also, we design a lot of our hardware for automated assembly.

Not only do we do that in our own factories, but for assembling those parts into vehicles, things like, for example, intercables bus bars, which are very cool. You can pick up an entire high-voltage routed bus bar and literally click it into a vehicle. And that, say, first of all, can be done by a robot, so you can save on direct labor. But also, if you think about quality control, you're no longer twisting and pulling on this harness and hoping that you don't break it, right?

[Murtuza] (31:32 - 31:37)

You no longer have the Friday built cars. 

[Benjamin] (31:38 – 31:40)

That's right. That's right. That's really cool stuff. That's really, really cool.

[Murtuza] (31:41 – 32:03)

It is amazing, because the fact that you're now having consolidated hardware as well enables you to actually have more robotics online, because otherwise, it can, I think, run at miles, isn't it, the number of cable connectors and connections that you could have otherwise in a vehicle?

[Murtuza] (32:05 - 32:14)

So, less points of failure, I guess, more consolidated hardware that can be then automated to be fitted online as well. So, a win-win. 

[Benjamin] (32:16 -32:29)

Exactly. And it really is a win-win, and that's the joy of it, right? You get to put more and more of your investment into delivering something awesome for the customer. Right? Yeah. And that's what we want to benefit at the end of all of this.

[Murtuza] (32:30 – 32:59)

Absolutely. Absolutely. And you mentioned a lot about safety as well, as one of the key features. You know, cybersecurity, you mentioned that it's been built into Active. How the scalable platform allows OEM to implement this across a product range. What is, when you talk of the architecture, how is it lending to multiple products for OEMs that can take advantage of it?

[Benjamin] (33:00 – 35:39)

Yeah, well, like we were talking about the whole, our whole ADAS system, where our architecture is a single architecture, but it scales from a very, very simple, low-cost compute and sensing, for example, you know, in a smart camera, right? Where the chip actually, the chipset that actually does the thinking is sitting inside a smart camera. For, you know, perception, or for autonomous emergency braking, or for adaptive cruise control.

And then you can scale that up, add in compute, add more compute, adding more sensors, maybe you're adding more radars. Our radars have ML in them. We use AI inside our radars, which allows them to do things like see pedestrians, that, and reliably so, that a lot of other radars can't.

So as a result, you can actually get away with delivering the same performance, but actually with a cheaper overall solution by leveraging the power of machine learning and radar put together. Radar, of course, can see through fog. It can see in the direct sun.

There are a lot of things it can do where a camera will struggle. And so our solution starts from really, really simple and then builds out to a much more full-featured, complicated solution. But the beauty is, is that the software just moves and grows.

So we can take the exact same software that is the core of the architecture and move it to a different chipset to support more and more functionality. And it's just a matter, remember, we talked about these containers, which are effectively puzzle pieces. We can just keep clicking in these puzzle pieces as you go all the way up to the very top end, premium segment L2++ solutions.

So that's that truly scalable architecture that we're using. And it's really helpful. The other thing that we do when we talk about managing cost is our architecture is an open architecture.

And that means that if you don't want, if our customer doesn't want to use our radar, but they want to use our architecture, they can use somebody else's radar. Or if they want to use a different vision solution, we actually support multiple vision solutions. And that's not just important in terms of giving our customers choice.

It turns out that regionally, there can be very, very different kind of optimal solutions, if you're operating in China versus if you're operating in Europe or if you're operating somewhere else. And so the fact that our architecture allows for clicking in different solutions, different sensors. True plug and play approach.

[Murtuza] (35:48 – 36:27)

That's amazing. We've discussed a number of questions and excellent insights there, Benjamin. I'll move on to our last question for the podcast. And I'm going to touch upon the subject of AI. I know you've mentioned a lot of time, AI and machine learning in your own architecture that you've mentioned. But I'm going to talk more generally here on how pervasive is AI across every sphere that we see? And how do you think AI will get integrated ultimately into cars for in-cabin user experience, for example?

[Benjamin] (36:28 – 38:46)

Yeah, so great question. First off, before we talk about in the vehicle, I want to make one pitch for AI before the software ever gets the vehicle. Because I think this is the hidden hero that most companies miss and don't talk about.

Which is that if your AI factory is not state of the art, then your AI products aren't going to shine. And so what, you know what I mean? It makes sense, right?

Right tool for the right job. Yeah, and so Aptiv has invested and continues to invest in that AI factory. And I'll give you a really simple, simple example because it's very clear and obvious, but there are many, many examples.

You know, you can run a large language model that is uncompressed and unquantized in the cloud using thousands and thousands of GPUs. But on the vehicle, you can't just light up more GPUs. You don't have a data center on the back of the vehicle.

And if you did, aside from how expensive it would be, you'd use a tremendous amount of power. And so we actually have at Aptiv in our AI factory, a core piece of our pipeline, is taking these algorithms, which might perform really well in the cloud, and figuring out how to compress them layer by layer by layer to run extremely optimized, super low power, and yet still super high performance on affordable, low cost, low power silicon. And that piece is the difference between a prototype and a product.

Real products do that, and that's so hard to get right. But getting that right is the reason why we're able to deliver these incredibly, frankly, cost effective, but yet high performing ADAS solutions to our customers. But you asked about the cabin. So I'll flip back to directly answering your question. I just had to make that pitch because it's so easy to miss.

[Murtuza] (38:47 – 38:52)

What you just mentioned actually seems like a true hidden gem of what you have at Aptiv.

[Benjamin] (38:53 – 38:55)

It really is.

[Murtuza] (38:56 – 39:57)

Compress and use low power silicon to operate, which, again, plays back to the cost that we've talked about for automotive, which is so critical as you know.

[Benjamin] (39:58 – 40:31) 

Yeah, actually, all right. Since I brought it up, I'll give you one other example before we talk about the cabin. At the front end of programs, Aptiv gets many tens of thousands of requirements from our customers.

Imagine how much time it takes humans to roll through all of that, and how perfect are you going to be in going through those requirements. And often, the requirements conflict with each other, right? But at the end of the day, you have to go through them, and you have to go through them in a reasonable period of time because the program has to move on. Well, whatever you miss, you end up catching later in the program. Yeah. And the further you go, the more expensive it is, right?

So another way we help our customers save cost is that we've brought large language models to recognizing requirements and being able to figure out, are they right, do they self-conflict, do they already match things we've already done? But I'll tell you the other thing that's really cool, where we run into issues, we use the LLM to figure out who is the right expert in the company, who is the best person to go and figure out how to solve that problem and resolve it. And so those are just a couple of examples of AI in the workflow is here today, and actually is amazingly powerful in the world of automotive for automotive applications before you even get to the customer just to deliver much higher quality product in a tighter timescale, which again, allows you to move at the pace of your imagination as opposed to slowly.

So in the cabin, first of all, AI is already there today in products that Aptiv provides. For example, in cabin monitoring and driver monitoring. And that's really important.

Think about the days when somebody would leave a child or possibly a pet in the vehicle. And now with cabin monitoring systems, you can do things like, no, oh, I've left something in the vehicle. Maybe I've left my purse in the vehicle.

Maybe I've left my kid in the vehicle. And get notifications and provide, you know, not just convenience features, but actually, you know, critical life-saving features. Yeah, and you know, I, it's funny because I actually, growing up, one of my parents locked their keys and a kid in the car.

And luckily the fire department came and broke the window so that they were able to get, you know, us out of the vehicle in time. And so the fact that AI is now solving this problem is actually very close to my heart and is completely awesome. Similarly, you can do driver awareness.

Is the driver having a medical issue? Is the driver falling asleep? Is, you know, those kinds of things. When it starts to really become neat is when you start tying what's going outside of the vehicle with what's going on inside the vehicle and vice versa. So for example, if the driver is not really paying attention to what's going on on the road and they're in a level two system, for example, then your ADAS can start making much more conservative decisions. In addition to notifying the driver and trying to wake up the driver, etc, your ADAS can actually, you know, help protect you.

And similarly, if you think, again, going out into the future, you can think about how can you personalize this experience? Right? If you know who's in the vehicle, then you can give them the in-cabin experience that they want.

Seamless with their devices, their applications, you know, you can start understanding what they're doing. If they're reaching into the wheel well, searching for their purse or something like that, you can turn on the light in the wheel well for them. Right? As again, as a very simple crude example, but there's just so much that can be done.

[Murtuza] (40:31 – 40:48)

Tell you an example, I was just reading about last week actually on this, where one of the Chinese implemented AI and the customer basically likes it where it greets their children by name when they enter the car. And that just shows one of the examples of AI.

[Benjamin] (40:49 – 42:57)

Murtuza, you are dead on right. And what you just showed was how the regional differences are significant because that's also my experience. I just got back from Shanghai just a few weeks ago and you literally have avatars in vehicles.

So your vehicles have a personality and you're interacting with a character. And while there are other regions that will find that creepy and go, don't want my vehicle to have a personality, like that, not that much of a personality. There are other places that are totally digital native that find that fun and exciting.

One of the things we see in the United States is a lot more karaoke showing up. We first started seeing that in China and Korea. Now it's rolling into, you start seeing that roll into the United States.

Our friends have a vehicle that their kids love to sing karaoke in on long trips. So yeah, those regional differences really, really matter. And the ability to bring AI in for all those things is great.

What that means though is the edge is really important though because your vehicle has only the hardware it has on it when it ships. And so again, back to Wind River, that's part of why we pulled in Wind River because of the 5G experience. Multiple major carriers in the United States are using Wind River as the backbone for their rollout of 5G.

But then also you need the software on the vehicle that be 5G aware. And why 5G? Because it can actually handle this low latency, high speed, reliable slicing of the network.

And so in the future, as 5G rolls out and we get network-aware or 5G aware operating systems in the cars, that will light up a whole new set of applications because now you can serve them from the edge. You don't have to serve them from the car. Yeah.

[Murtuza] (42:57 – 43:47) 

Yeah, that again, shows an amazing example of how Aptiv's bringing real value to the automotive OEMs as they progress in their STV game. So Benjamin, thank you very much for your time in this podcast. It has been absolutely fascinating listening to the different aspects that you mentioned on software-defined vehicles, as well as touching upon subjects like autonomous driving, ADAS, and also artificial intelligence that we just talked about in the cabin.

So it was truly fascinating talking to you. Great to have you on the show and hopefully we can have you again some point in the future, as the automotive journey progresses and we get more futuristic with a lot of stuff we're developing today.

[Benjamin] (43:48 – 43:58)

Well, absolutely. And thank you, Murtuza. This was super awesome to go back and forth and chat about the future of mobility, man. It's awesome stuff. Yeah. Good.

[Murtuza] (43:59 – 44:40)

You have a good day. For viewers and listeners, you can listen to our past podcast on counterpointresearch.com. We have covered a variety of topics, be it digital cockpit, artificial intelligence, electric vehicles, autonomous driving.

We have a variety of blogs, reports, features, podcasts just like these available. We're also present on multiple platforms such as Apple Podcasts, Spotify, YouTube, and YouTube Music, among others. Thank you very much indeed for listening to us.