Speaking the language of software and electronics in automotive – On the Move S01E07 – Transcript
Bringing a software-defined vehicle to market successfully is all about shared resources and communication. To talk about the future of vehicle development, Nand Kochhar is joined by two experts in electronics and software for part two of this three part series. Read the whole conversation or listen below.
Mike Severson
Welcome to another episode of On the Move, a Siemens podcast on the automotive industry. I’m your host Mike Severson. And I’m joined by my co-host Nand Kochhar, our VP of Automotive and Transportation. We’re keeping the ball rolling on our conversation about the software-defined vehicle and integrating traditional domains of vehicle design with software and electronics, with Michael Munsey and David Fritz.
At the end of the last episode we said the core of SDV is spanning the gap between the digital twin, how it helps OEMs develop products, how it helps validate and verify, and how it helps companies move beyond the complex system- and unit-level problems of old methodologies. Nand, is that a good recap?
Nand Kochhar
Yeah, very good points. increases the transparency and traceability across the entire ecosystem between OEMs and suppliers on the technical side. But also, it is offering new business models for our ecosystem, because most tier ones have been used to certain business models of selling ECUs as a unit, both as a hardware and software combined, but now they have to look at the business model and the payments with the OEMs and based on new roles and responsibilities. So SDV is changing those business models as well across the entire ecosystem.
David Fritz
Yeah, I think one of the big points here from the OEMs are they used to have to define all of that functionality up front and then partition it out and all pieces come together. But now, if you have these vehicles on the road for 10, 15, 20 years, then the question becomes, well, how do I make sure that some differentiating software capability five or 10 years down the road can be supported in the architecture that we’re defining, both the hardware and the software architecture? And what happens if the vehicle degrades in performance over time? Will we still be able to future-proof this? Well, with a digital twin, particularly a multi-fidelity holistic digital twin, you can answer those questions well in advance. So you don’t have a problem where you do an over-the-air update out to the fleet in the field only to find out that now 30% of the vehicles have a problem.
Mike Severson
Are we able simulate the years of use on the vehicle with this digital twin? So maybe things are working fine today, but five years down the road, over various software updates, things won’t. Do we have that capability to sort of predict what would happen in the future?
David Fritz
Absolutely. So that’s part of where the EDA tooling comes in. We’ve done things like that for quite some time for other industries, but it’s new for automotive. And often it’s one of those features that they’re happy to hear about, But the technology behind making that a reality means they have to make investments in other areas.
Michael Munsey
And one of the key things here is when we talk about the concept of the digital twin and having a comprehensive digital twin, is that the only way to be able to predict this at the entire automobile level is to understand the whole automobile. Right. Because we do a very good job of predicting how that semiconductor is going to age over time and how that might affect the performance in the propagation of signals. But unless you could then guard band your design and understand, all right, so if the semiconductor is aging, right, and those signals are going to leave the chip and go into a PCB board and then be driven into a wire harness, which is then eventually going to get to another PCB board that’s interfacing now with a hydraulic system or some other mechanical system. unless you understand the performance characteristics across the entire system and know what that is and can predict what happens as the semiconductor ages, then you’re able to guarantee that as you do these software updates, you’re still within the performance characteristics. But that requires understanding the entire system and having the digital twin of the entire system.
Nand Kochhar
Well, once again, very important point. Both, again, Michael and David touched on the electronics side of this. And Mike, as you know, in the traditional mechanical sense, whether it’s a durability, reliability, NVH calculations, the auto industry today have a suite of tools to learn about the degradation over a period of time with the number of years in service. Those are some of the things which go towards predictive performance and degradation over the years, not only just durability, but the other functions as well. So now this is bringing together the old, just purely mechanical, into electrical electronics, degradation in those areas, and the cross-domain nature along with the tying up the right levels of software becomes very important. And I think, in a way, it puts us in a very unique position from Siemens Solutions, very differentiated offerings for our customers to help them achieve this cross-domain work to bring this SDV vision together.
Mike Severson
Nand, you talked about the changing business models. And there’s this new shift to SDV. Are we seeing OEMs and tier suppliers shift to building out semiconductors in house? Or are companies still primarily relying on traditional semi-manufacturers?
Nand Kochhar
I think there’s roles and responsibilities. I think Michael started with traditionally OEMs have given requirements for the semiconductor company, they have integration guys in between before things coming back to the OEMs. That model is shifting. Of course, very few companies are going all the way to designing, developing, manufacturing their chips, but more working together with the semiconductor companies, giving them requirements and let them focus on the designing and then manufacturing is taken by a semiconductor, and then you have that roles coming, redefining of the roles and responsibility. But in the entire ecosystem, not only the tier one suppliers examples were used in the automotive space, but also the electronics space from the chip design to chip manufacturing, and then bringing it all the way into the vehicle as an ECU unit. All those are getting redefined.
David Fritz
Yeah, I think one interesting thing that we’re seeing is that a lot of the automotive companies are not ready to develop their own SOCs, but they have very small teams. And with a comprehensive digital twin with multifidelity, we’re seeing a lot of contracts whereby we help those automotive companies build the digital twin to identify what the requirements are so they could produce an RFI to a semiconductor vendor or a service company to go out and produce exactly the right part. In the absence of the digital twin, the level of requirements are not sufficient and you’ll be sold something from one of your suppliers that they’re selling to your competition. And it’s hard to differentiate in that way. So that’s something that’s different. And the next step in all this is chiplets. The whole idea behind a chiplet is you create a hardware subsystem that could manage identified software workloads, but you can combine them in any way you like, package them separately, and end up with a differentiated solution. And you don’t have to go out and have a 2,000-member SOC design team and spend hundreds of millions of dollars for each design. Again, that’s not possible without a digital twin to help you figure out what requirements are for the hardware and the software of each chiplet and for the total system that combines the chiplets together. So a lot of automotive companies see chiplets as the next big opportunity to do their own custom differentiated SOC design, running their own custom workloads to differentiate at multiple levels.
Mike Severson
okay great thank you for that let’s go to the next question shifting gears a bit here we know oems are pushing hard on software development across the entire development cycle so what are they doing to bring their supply base along for the ride and importantly how are they making sure all this added complexity doesn’t mean longer lead times, delays, etc.
David Fritz
Right. So as we said earlier, the ecosystems for these OEMs are changing. And again, the digital twin is a key part of the catalyst that allows that change to happen. So for example, if you are in the process of building a hardware and software digital twin for your next generation automotive platform, then the ecosystem partners that you choose might be different. Okay, so that’s opening up the door to a lot of startups, a lot of companies that have experience in these areas outside of automotive, and it’s also forcing a lot of the automotive suppliers to retool and think about differently. But the real problem inside the OEMs that we’re seeing is this cultural complexity. So how do they continue to develop vehicles for the next five model years using everything that’s already been put into their development pipeline? and then in parallel make this shift into SDV built around digital twin technology and then identify a point in the future where the SDV platform that is designed intersects their roadmap. And that’s a difficult challenge when you’re a very large company and you’ve been doing things pretty much the same way for the last 50, 60 years.
Michael Munsey
David, let me ask you a question along those lines, Because in other industries where we’ve seen software-defined products, there’s been the drive towards optimizing the software in the semiconductor early on, as I mentioned before, but then using that to drive selection of IP that ultimately gets used in the semiconductor design. When we talk about now the idea of chiplets and advanced packaging, do you see the industry also moving towards the case where you could optimize a system and then use that to actually drive maybe an ecosystem of some predefined chiplets that ultimately would go into the package to deliver for the solutions for the automotive industry?
David Fritz
No question. That’s the vision. We’re seeing some of these automotive suppliers putting more effort and investment into having a role in that chiplet evolution than they are in just keeping status quo business. I’m not going to name companies, but they’re pretty much out there. We’re pretty deep into the whole chiplet initiative in automotive, so we’re seeing a lot of movement there. But I think the most important thing is how do they make this shift? And some of the automotive companies are doing it via evolution. So they incrementally try to change. And those are the ones that are struggling the most. What’s interesting about SDV, as we mentioned several times here, it’s taking a holistic view of the entire system. Because one subsystem can impact the performance of another subsystem in these complex vehicles, particularly when you get into level 3, 4, and 5 of autonomy. It could be that your braking system has a major impact on the decision making that happens in a central compute of a zonal architecture. So as these get more and more complicated, what does that mean about how do I architect my software and my hardware for that next generation platform? And then what partners do I need to pull in who have experience in that area? So these automotive companies generally have very little experience in artificial intelligence. And those that do tend to not be keeping up with the latest, greatest research and papers, the sort of things that we’re seeing from Google and Microsoft and X and the list goes on and on. So it’s a process of how do you keep up with this technology and you still deliver some differentiation in the model years that are already set in process. Keeping in mind, it takes seven years from inception to get a vehicle to start production in automotive. And with this whole SDV methodology crunching that down, we expect that to go to just two or three years with an intermediate step of five years. So the speed of innovation is going to accelerate as some of these companies move into this SDV realm. And I think everybody in the automotive industry sees that and the discrepancies start to arise in different regions approaching this in a different way. In Asia, for example, they don’t have the baggage of, you know, 50, 60 years of doing things in a certain way. So they have the ability to kind of learn from those lessons and make a leap into SDV. And a lot of the research data that we’ve seen shows, you know, the rates of change and the projected spending in this SDV area over the next five to 10 years is very different between Asia, North America and Europe.
Mike Severson
Yeah, it’s really interesting that you talk about the rate of innovation, and it has accelerated exponentially. The vehicle program timelines have been compressed. Feature rollout is no longer associated with the model year. If a new feature is available, OTA enables it to be added to the vehicle. So it is a very large shift and a big challenge for all OEMs, especially the legacy OEMs, as you mentioned, with the organizational changes that are needed for this and the processes that were developed over decades.
Nand Kochhar
To be honest, that’s where exactly the digital transformation and our digital twin technologies we offer, work with our customers, help overcome some of these hurdles. and enable that speed of innovation. Because as David mentioned, the speed of innovation in different parts of the world, it is happening today. And while other parts of the world is still catching up, especially the traditional automakers versus the newcomers. So that’s the, this still twin technologies become core to get to that speed of innovation and with high quality. That’s absolutely necessary.
Mike Severson
We touched on the integration of hardware and software earlier, but does this bring suppliers into the conversation earlier than before, or does it just change the conversation that is happening at the very beginning?
David Fritz
It brings them into the conversation very early, several years earlier than before. Now, that’s not always considered a plus to certain suppliers who prefer things working in a particular way. A great example is a Tier 2 supplier that we worked with a few years ago. And we said, look, what we want to do is we want to provide a digital twin of your next generation system on chip for automotive. and we want to plug that into a much larger digital twin of the entire system such that you can use that to sell to your partners or your customers the features of your chip. The response shocked us. The response was, why would we want to do that? Because they could compare us against a competition. I think that response says everything, right? If you are the customer, you want to be able to compare against your different suppliers. You want to choose what’s best for your requirements. You don’t want your suppliers telling you what you can and cannot do. So the concept of these OEMs needing to control their own destiny is essential. And you’re seeing that with some of the top-level executive moves in these companies, looking for people who can make that direction, take control of their own destiny, control of their own ecosystem, find the right partners, bring them in early, have their partner’s knowledge integrate with this. And again, we keep harping on Digital Twin because it’s central to this, because without the ability to bring them in initially in a low-fidelity virtual environment, then it’s impossible for the suppliers to work together on one project with one OEM that they may be competing against each other for another OEM. What that does is the finger pointing that’s been going on for decades and decades, this is your problem, no, it’s your problem, goes away. And an OEM then can bring in the best partners early to help design exactly what needs to be done. Of course, those suppliers are going to kind of nudge the requirements in their direction, but that’s fine as long as you have that holistic digital twin to verify that it meets your current and future requirements. That’s fine, and you’re going to end up with a much more optimized solution in the end. In the past, you had to make decisions based on data sheets, which had absolutely no correlation whatsoever to your real workloads or your real hardware capabilities. That’s how the industry has gotten into the situation where they’ve got to dig themselves out of this hole. And digital twins support for software-defined vehicles is the right solution.
Nand Kochhar
Yeah, Mike, overall, you could say, yeah, it is bringing up their engagement up front. But it goes back to also the business models I refer to. Some OEMs will have an affordable business framework where they want suppliers as part of way up front. But it depends on their relationships and the business model they develop because they don’t have any POs or the orders to buy product from them, but they want their technologies to be coming up front. So that’s where the business model becomes very important and the relationships. The suppliers have technology, as David mentioned before, it’s not just the OEMs have technology. A lot of the technology, especially around AI, could be coming from these tier one suppliers, et cetera, as well, because they are trying to march forward and move into that, whether it’s in the electronics world or the tier one suppliers. So it all comes together and shifting left or moving forward, not only from a technology technical standpoint, but also from a business standpoint. I think for later on, BMW had their investor analyst day 2024, and they laid out this chart beautifully. And again, every company might have a different how this paradigm between OEM and T1, T1 suppliers and the semiconductors suppliers is shifting both from responsibility, accountability and the relationships.
Mike Severson
We’re going to end the episode here, but there is one more episode coming from our conversation with Michael Munsey and David Fritz. For the next episode we’ll cover one of the major drivers of the software defined vehicle – autonomy. We’ll be back soon, but make sure to subscribe so you don’t miss the next episode of On the Move when it airs.
