James Kotecki (00:00):
Hey, it's James Kotecki from CES Tech Talk, here to tell you that registration for the world's most powerful tech event is now open. What event is that? You know what event it is. Come on. The greatest minds, the most powerful brands, the most impactful technology, it all comes together at CES 2024. So, discover the tech-defining AI, transportation, startups, smart cities, digital health and solutions for a better, more sustainable planet. Register now at ces.tech. All together, all in, all on.
(00:43):
This is CES Tech Talk. I'm James Kotecki. The world's most powerful tech event, CES 2024, brings the future to Las Vegas January 9th through 12th. Today, we preview the future of connected electric autonomous vehicles, and specifically, the software that makes them possible. Jan Becker is the CEO of Apex.AI, which is, "enabling software-defined vehicles." And since CES is a nexus of technology and transportation, this is going to be a great conversation. Jan, welcome to the show.
Jan Becker (01:18):
Thank you, James, for having me on the show.
James Kotecki (01:21):
So, let's start with a basic understanding here. What does it mean when we say that Apex.AI is enabling software-defined vehicles?
Jan Becker (01:30):
So, software-defined vehicles are vehicles that work similar to smartphone, where you, after you've bought the device, can add functionality, can update functionality by downloading a new version of the operating system or by downloading apps that add new functionality to a phone. In the smartphone ecosystem, this is enabled by a software piece called software development kit, SDK, which exists for both large ecosystems, iOS and Android, and it's provided by iOS and Android, which really contains all functionality which is common to all or many apps such as downloading data, accessing the speaker, accessing the microphone, creating virtual keyboard, and so on. And that makes it really easy for the smartphone developer to develop applications. We developed such an SDK, a software development kit, for software-defined vehicles, which then enables developers of functionality for software-defined vehicles to do that much faster, much more reliable, such that those applications run safe and secure on the vehicle.
James Kotecki (02:37):
So, basically, you're creating a standard, I don't know if the right framework is toolkit or toolbox for developers to do any software-related thing with vehicles?
Jan Becker (02:49):
Yes, you can think of it as a toolbox of pre-built building blocks where the software developer grabs into the toolbox and takes the module, the predefined, pre-developed and pre-certified building block out and adds it to whatever they want to build.
James Kotecki (03:08):
The metaphor to a smartphone is an interesting one, because of course, smartphones were built from the beginning, their hardware was constructed from the beginning of their existence to be software-enabled, software-defined devices, right? Automobiles, not so much. So, if we think about the history of vehicles and Henry Ford, there's not necessarily any software in that, at least not in the conventional sense of the definition of that term. So, can you kind of take us back to understand the evolution of how software has entered vehicles over the years and how much software is in vehicles right now?
Jan Becker (03:46):
And if I may add the analogy also works in the smartphone case. 25, 30 years ago, I had a Nokia phone, where the software was preloaded on the device when I bought the device, when I bought the phone, and really never changed over the lifetime of the phone. And then, in 2007, the iPhone came out, and two years later the App Store came out. And with the App Store, the SDK. So, it's really very, very much the same. So, in the car world, we are now just living at the point of transition between the Nokia phone age and the smartphone age. So, vehicles today software is loaded onto the ECU, so the computers across the vehicle when the vehicle is built. And then, typically, for most vehicles today is actually never changed, and that goes even deeper into the vehicle architecture.
(04:39):
A vehicle that is built today contains between 100 and 300 so-called ECUs, so electronic control units. You could think of it at small computers, which then contains a software that implements a functionality. And you can think of it as one function more or less requires one ECU. So, you have one ECU for your air conditioning, one for your lighting system, one for your radio, one for your cruise control and so on. Then, you end up having between 100 and 300 depending on the vehicle. And then, it goes even further. So, those ECU are then procured by the OEMs from up to 100 different suppliers.
(05:24):
The suppliers develop those ECUs and the suppliers develop the software that runs on those ECUs. So, if now an OEM would like to change or add functionality in an ECU, they really have to go back to the supplier, typically to the tier one supplier, and request that change. That makes innovation very slow, very tedious. Update cycles of a year are actually not uncommon at all. Whereas if I go to my phone and there's an update for many apps every other week. So, we live in two completely different worlds with respect to the innovation cycles.
James Kotecki (06:05):
So, with what you're trying to do to bring some order and some modernity to this kind of software updating in vehicles, is the greater challenge technical or is it simply the ecosystem of so many different players and modes of standardization where if, theoretically, every human in the process could get on the same page it would be easy, but because there are so many different people involved, it's not?
Jan Becker (06:27):
It's actually both. It's a technology challenge and a culture challenge. On the technology challenge, in order to get to that smartphone-like ecosystem, the vehicle architecture needs to change. As I pointed out today, you have 100 to 300 ECUs, typically tiny computers spread out across the vehicle. In the future, and some of the electric vehicle manufacturers have already introduced that a few years ago, in the future, you'll have one or two really large computers, similar to the processing core of a smartphone, on which then all the different processes, all the different software that implement the functionality for vehicle will run. That's called the central architecture. And our software really enables that central architecture.
James Kotecki (07:16):
And of course, there are some automakers that are living closer to this version of the future or the present that you're describing. I'm thinking specifically of Tesla, which famously has over-the-air updates for its vehicles. That you park it in the garage overnight, you start it up again in the morning, or you don't start it, but you get in the car in the morning and a new software update has been enabled. Is that what we're talking about? Is that a model for the rest of the industry? Or is Tesla somehow unique in what it's been able to build that it's more challenging to do that elsewhere?
Jan Becker (07:47):
In terms of the technical solution, Tesla is definitely a model for the industry because they introduced that central computer in Tesla vehicle. There are typically two, one for the infotainment side, one for the autopilot side. Tesla introduced that already years ago. Tesla had the advantage to start really on a greenfield. New company, no legacy, one, maybe two vehicle lines, which makes it obviously much, much, much simpler, also from a company culture perspective, to introduce such a technology change. It's much, much harder for the large automakers, for the largest car companies of the world produce up to 10 million vehicles a year in dozens of vehicle lines. It's much, much harder for those companies to introduce such a really a radical change for the industry.
James Kotecki (08:42):
And so, what are some of the things that you're finding where you're able to have the most success in implementing these changes? I don't know if I want you to give advice to your competitors, but I guess I'm just trying to ask how are you able to deal with this? This sounds like an almost intractable challenge. How are you even approaching this?
Jan Becker (09:02):
So, it really depends on the company, our customers and how they work in general and how we can work with them. There are typically two approaches to introducing a new architecture, such a change. Either you introduce it for one new vehicle line, so you really develop a vehicle from scratch, and then you introduce it for that vehicle line. And the other change is you introduce it for one domain first. So, a domain could be the adaptive cruise control, so the whole driver assistance domain, or it could be the infotainment domain, or it could be the powertrain domain. So, you introduce it for one domain first. And then, once that works, you gradually step by step introduce it to more and more domains. Those are the two approaches.
(09:51):
Typically perceived as higher risk, but really low risk in the end is introducing it for a whole vehicle line. That requires courage, mostly on the management side, which is hard for these companies because a lot of is at stake. And managers, historically in the auto industry are typically not software experts. Car companies have often made vehicles for over 100 years. So, they're really experts in designing a vehicle, building a combustion engine and manufacturing the vehicle, but not so much software. And software simply not just from a development, but also from a company perspective requires a different culture. And traditional automotive managers are really not familiar with that culture. So, they perceive as higher risk what is unknown to them, and that really means in making a radical change on the software side is being perceived as a higher risk.
(10:53):
In practice, it's actually low risk because if you drag this out, as a result, you get what now some of the car makers are experiencing, which is due to software development issues, launches are delayed, sometimes by one, two, three years, because making those small incremental changes actually results in more issues down the road. And car companies really just can't get the software implemented in new vehicles and the new vehicles off of the production line.
James Kotecki (11:28):
And do you sense a shift in the wind here in terms of how companies will identify themselves in the future? You're mentioning auto companies have been doing this for 100 years. They're not software companies. Could we see in the future, in the next 10, 20 years, some of these companies, Ford coming out for example and saying, "You know what? We are a software company. We just happen to have platforms that have wheels on them. But primarily, the way we identify is as a software company," or will they continue to live in this kind of hardware mentality?
Jan Becker (11:58):
It's hard to foresee the future, but what I would do if I were the manager of a large company, I would actually set up two companies. I would set up one company with a software focus to develop the software for the vehicle. And we take the Ford example, I would leave Ford as Ford, because Ford still needs to make vehicles. They still need to manufacture vehicles, which requires workers in the factory and a reliable manufacturing process, which requires a different company culture. So, I would separate that in two companies.
James Kotecki (12:33):
When we talk about how these companies are transitioning, obviously we need to think about electrification as well. What extent is the software element crucial to getting the electrification piece? How are those two things linked?
Jan Becker (12:47):
On a high level, we really have three, four major trends, all three or four resulting in radical changes in the industry. Much, much more radical than what we've seen in the past 100 years. Those are the trend to software-centric systems, as we've already discussed. Then, there is a trend of electrification. Then, there is a trend of connectivity. So, vehicles being connected, also similar to a smartphone, to the internet, to the cloud. And then, there's the whole trend of more autonomy, more drivers in those vehicles. All these trends ultimately rely on software, rely on the software architecture. That's why the implementation of a clean architecture for what is now called a software-defined vehicle, is really, really so important because that's the enabler to implement those other changes.
(13:45):
You can think of it as an upfront investment that really the customers initially will not see, an upfront investment into a clean architecture, both from a computing side and from a software side, which is very costly in the beginning because you need to define and implement that first. But once it's implemented, the cost of adding new functionality is really minor. Again, coming back to the smartphone, it doesn't really cost anything to download a new app, but it costs to develop that smartphone, develop that smartphone infrastructure and the whole ecosystem around it. And then, you're at the point where you can easily upload new software to the device. The same is going on in the automotive field right now in this transition period.
James Kotecki (14:35):
Of all those software-enabled, software-dependent trends that you mentioned, there's really none that kind of captures, I think the imagination and the feeling of futuristic-ness, for lack of a better term, than autonomous driving, self-driving vehicles driving all over the place. It's been a theme at CES, and of course, far beyond that for many years. And I'm wondering from where you sit in the trenches here, when you are trying to figure out a key piece of this puzzle and a key enablement piece of this puzzle, where are we on the road to autonomy? How should we even be thinking about that?
Jan Becker (15:10):
It's funny that you asked. So, I started my career in automotive developing autonomous vehicles. I actually did that 25 years ago way before the trend of autonomous vehicles was already a thing. And I've jokingly been saying autonomous vehicles are always 20 years out. And if you talk about fully-autonomous vehicles, so a vehicle that would show up in front of my building today and wherever I could, as a human, drive a car, that vehicles would be able to take you, I think we are still 20 years out. Maybe now it's 15 or 10, but it's still a significant amount of time.
(15:50):
We are much closer though to, in regionally limited areas, to implement that. As you know, Waymo has implemented that successfully in San Francisco and in Phoenix. There's a tunnel in Las Vegas in which you can take a Tesla from one end to the other end. So, we have certain use cases in which that actually works. We have a lot of customers doing that in agriculture and mining for trucking, which are all very defined and refined use cases in which we are now actually at the point, after I've worked on it for over 20 years, where we can actually introduce products, which is great. But if you'd ask about a fully-autonomous vehicle, which we would call a level five, which can take me anywhere, that is probably still 20 years out.
James Kotecki (16:42):
I took that tunnel at CES 2023, that Tesla tunnel, and there were people driving the Teslas in that case. But your point is taken, right? It's relatively simple or simpler to automate that kind of straight-shot tunnel approach, especially with a vehicle like a Tesla that's already got some of that built in. But I want to touch on the farming piece that you mentioned, the industrial side of this. When we think about vehicles, obviously, we're not just thinking about cars that take us commuting to offices, we're thinking about farms, even thinking about vehicles on the water. And obviously, software is going to be a huge part of defining all of those as well. Are you seeing more success, more traction for what you're doing for autonomy and beyond? And those other non-pure consumer use cases, the more industrial side?
Jan Becker (17:31):
Yes, we do, and we do for two reasons. Reason number one is the use case is simpler. On a field, you typically drive in straight lines, which is a very tedious, time-consuming task. And the whole environment in technical terms we call that ODD, operational design domain, is simply simpler. On a field, you have GPS reception, you can rely on that. You can work with a simpler sensor set that makes the whole device cheaper. So, the introduction from a technology perspective is simply simpler. The other pieces compared to passenger cars, where you have largest manufacturers have volumes up to 10 million a year, you have much, much, much smaller volumes. So, the companies are more agile, the number of devices is smaller. If you need to update software, it's simpler, which makes the whole introduction simpler. So, we have a number of customers, which we've already announced in the agriculture space, which work on agriculture products which are autonomous. Same is the case in the trucking industry, in mining. So, all these off-road vehicles have much faster, shorter development times and shorter times to market.
James Kotecki (18:53):
I do want to ask another question though from the consumer perspective, because we are talking about CES, where so much consumer tech is on display. So, I am curious though, we've been talking a lot about the immediate challenges of doing this and you understand that very well. But I also am really curious what your vision is, what your most hopeful and optimistic take is on the possibilities here for consumers once we get to this promised land of a smartphone-like vehicle ecosystem with all kinds of updates happening all the time. How different can people's experiences in vehicles be? We talked about autonomy obviously, but other things that people will be able to do or that will be happening for them because of this software-enabled ecosystem.
Jan Becker (19:39):
So, once you have a software-enabled ecosystem, and Tesla is really a good example for this, it's really a smartphone-like experience. As an example, taking an electric vehicle in Germany today. At some point, you need to charge, you need to approach a charger. I roughly know when I'm going to be at a charger, I want to be at a charger and the charger is available, I connect and it actually works. Should be a seamless experience. Outside of the Tesla ecosystem, which is really a tightly vertically-integrated ecosystem. So, where one company owns the whole software across the vertical integration, that actually works really well and is very seamless. We do not unfortunately have that yet in other geographies and across other manufacturers.
(20:27):
So, that's one simple example where I, as a consumer, expect a seamless experience and I do not get that yet. So, charging is one. Just going from my office to a vehicle. Today, I either connect my smartphone to the vehicle or I need to do a new setup from scratch. That really shouldn't be necessary. I want to take my smartphone, I want to take my [inaudible 00:20:53] with me and have a seamless connection in the vehicle. Google and the Android ecosystem and Apple for the iOS ecosystem are working on this. But that is not yet a seamless experience unless I actually run the vehicle infotainment system off of my smartphone. That is another example for where I, as a consumer, expect a seamless experience, but I don't get that yet.
James Kotecki (21:16):
And it's funny, I mean the way that you're talking about consumer expectations and we think about smartphones, as you mentioned, being around since the iPhone came out in '07. We are trained to think of these experiences as being seamless. And of course, that's the expectation that we have built in, and it seems somehow wrong to us as consumers that these things don't already exist. So, I imagine that unlike the iPhone, which is able to kind of come out into this blue ocean space and kind of define a category, it's even more challenging for folks like you who have to navigate with doing these incredible technical and cultural shifts, and then having people think, "Well, yeah, I kind of assumed it could already do that."
Jan Becker (21:52):
Exactly. I couldn't agree more.
James Kotecki (21:55):
I know that Apex.AI is going to be at CES 2024. What are you planning for that?
Jan Becker (21:58):
So, CES is great for us because really across all the different pillars, people come together. For us, that's our customers, obviously, most importantly, but then also the press and investors. So, for us, CES starts on Sunday with the house of journalists, where we'll meet journalists. We'll have a press conference and talk about our products. We have a booth in the West Hall. The West Hall is actually working well for us because you have all the automotive companies really co-located in one space, which makes distances shorter and meetings more reliable. So, there, we have a booth where we'll showcase both our solutions, our partner solutions, and that has worked really well for us last year. We have a meeting space on the second floor. So, we build our booth across two levels. So, we can also meet upstairs while we showcase demos downstairs. And then, of course, we will meet with investors at CES. We're looking forward to CES.
James Kotecki (23:11):
And I have to assume you're also looking forward to just all the different meetings that you could have with folks that are adjacent to your space, but maybe outside of... The kind of collisions that happen with folks outside of your immediate worldview because they all come together at CES. And as we get close to wrapping up here, I'm just curious where you look for inspiration? What other kinds of technology or maybe sections of CES are you excited about beyond your immediate space in software-enabled mobility?
Jan Becker (23:43):
So, I really look forward to see how automotive companies this year will embrace the whole trend, the whole area of generative AI. That has been the hype, really an overhype last year. But typically, after a hype, after the peak of a hype cycle comes the phase where companies find meaningful, realistic, impactful applications of a new technology. And I really hope and expect that this is going to be the case for generative AI at CES in 2024, where I expect companies to come out with really meaningful applications, just beyond putting ChatGPT in a vehicle.
James Kotecki (24:26):
Any predictions for what that might be?
Jan Becker (24:28):
So, there are a lot of application areas we as consumers don't really think of. So, generative AI in manufacturing, generative AI in software development. So, I hope to see solutions in these two areas.
James Kotecki (24:41):
Well, Jan Becker of Apex.AI, really appreciate all of your insights and diving into these topics with us today on CES Tech Talk. Thanks so much for coming on the show and we're looking forward to seeing you at CES.
Jan Becker (24:53):
James, it's been great. Thank you so much for having me on the show.
James Kotecki (24:56):
And thank you so much for watching. That's our show for now, but there's always more tech to talk about. So, if you're joining us on YouTube, please make sure to hit that subscribe button. And if you're listening on Spotify, Apple Podcasts, iHeartMedia, or wherever you might get your podcasts, be sure to hit that follow button and/or subscribe and/or whatever it takes to keep in touch with us. You can get even more CES and prepare for Vegas at ces.tech, that's C-E-S dot T-E-C-H. Our show today is produced by Nicole Vidovich and Mason Manuel, recorded by Andrew Lin and edited by Third Spoon. I'm James Kotecki talking tech on CES Tech Talk.