NXP Semiconductors N.V. (NXPI) Earnings Call Transcript & Summary

Well, good morning, good afternoon. Thank you for joining Evercore ISI's Autotech and AI Forum. My name is C.J. Muse, semiconductor analyst here. Very pleased to be hosting NXP Semiconductors. With us today, we have Henri Ardevol, SVP, GM of Auto Processing; and Jeff Palmer, Vice President, Investor Relations. We've got, I think, 35 minutes fireside chat. And so very warm welcome to Henri. Thank you for joining us.

Thank you, C.J. Looking forward.

Excellent. So just to start off, big picture, you've embarked on a strategy to push the automotive processing market to 1 with domain zonal controllers. Can you speak to what's driving the platform-based approach and why NXP is the company capable of providing the right solution for this evolution?

Yes, absolutely. So yes, I mean, a few years back already, I think about 4 years ago, we recognized that vehicle software and networking requirements would outpace the traditional MCUs. And so we began to refresh our portfolio to try to meet and anticipate these trends. And they really have gone into 2 directions. On one side to go to a class of processes, which is much more performant. And that allows to migrate the strategy of the management of all of the different sources of power and torque in new vehicles with hybrid cars. And now it is a much more complicated management than when you had just a single engine and missing resource of power, integrate the behavior of the car with the chassis management and do this in ways to optimize the range, to optimize the experience of the car. And on one side, also recognizing that there's more and more gadgets that are coming more and more features that are coming into the car. Just briefly before we're talking about our steering wheel and how a steering wheel today have so many different switches and displays and controls. And all of those require a class of microcontrollers that are much more low power and are able to be plugged into these new architectures without taxing it too much. So there is big shifts, big changes, which are taking place. And in the mind of the OEMs, they really have 2 kind of problems that they're trying to solve. On one side, how do they simplify the software because you have so many more requirements, you have so many more features that you're trying to cram in that it becomes very, very hard to integrate all of those and to verify and validate that the system continues to function. And on another side, how do you simplify the wiring and just the physical connectivity of all of these additional functions that you want to have in the car? How do you not have multiple networks stacked on top of each other and have at the end of physical topology, which is physically manageable and also, frankly, which is cost effective? And so when you move -- when you think about these migrations to even a higher class of processes, they actually do help a lot with software in terms of integrating the management of it. And when you think of this much more electronic power and modes, they also plug in a much easier way into your network. And then I'm sure we'll talk about it. There's also evolutions into the physical topology of the car. So some big shifts are taking place there and some big problems that the OEMs are trying to solve. And we've done our best to try to anticipate. In semiconductors, we need to try to spot the trends many years in advance, and we've taken some bold decisions, and we're now seeing through a lot of the conversations with OEMs that it's the direction that they're actually looking for.

So when we discussed some of the challenges here, I think some of the largest ones were integration, access to data and software. And so perhaps you could walk through each of those and how you've come up with a solution that provides the right solution at the right time for automakers.

Yes, sure. So yes, I mean, if we start with software, right? I mean we're moving from hundreds of millions of lines of codes, which is already huge, towards 1 billion lines of code by 2030. And so that presents massive challenges in terms of making sure that, as we're saying, the system is very viable. And the core attributes that the consumer expects from a car of it is going to be robust. It is going to be safe. It is going to be reliable. It's not going to break down. It's going to have a high quality. And it's not going to get cracked, it has a high degree of security. Those key expectations from consumers, I would say, not only continue to exist but actually are even stronger now because everybody is more aware of the risks. And so this poses an additional level of challenge on to the software architecture, which is to meet all of this in a very robust way. And so the way that the industry is moving and how we are responding to it is by finding -- by defining domains, which are -- which go across the entire vehicle. And so those domains would be your propulsion, how the car moves and stops, your chassis management, how the car behaves, networking, how you manage all of the flow of data that goes across the car. How do you ensure the security of it, how do you distribute and manage the privileges of that and also how you connect to the external world. And finally, body which is the connectivity of all of these different controls that you have inside the car. So each one of those areas becomes then a domain that you want to start to manage as a whole. And this is where you start to want to have in order to realize that a category of processes which is much more performant than just a single microcontroller that would do a single function. And in order to have those processes at that right performance level, you need a few things. On the one side, they need to move to much more of the leading edge in terms of technology. And so this is why we moved our entire portfolio to FinFET to 16 FinFET right now. And we are even collaborating with TSMC to make 5-nanometer automotive qualified. So we're taking the right steps in terms of technology to have, one, which is going to be future proof, and it's going to have the attributes of performance and power, which are required. But then on the design of the chip, because they are going to be managing an entire domain, they need to have the capability to run multiple applications in parallel. And so you need to design your chip fabric for what we call freedom of interference between applications. So of course, you have multiple [ cores ], but it's not just that 1 application is running in each of those cars. But the management of all of the different IP blocks, the management of all of the interfaces, so the entire fabric design of the SoC is the key to enable that freedom of interference and therefore, to enable that ECU consolidation and that ability to move the management of the domain from the edge to the center.

That's great. I guess as a side note here, thinking about, like you said, 16 FinFET and now working with TSMC on 5-nanometer. Curious, I think NXP has lots of different businesses. So obviously, you're benefiting from your digital networking business. Is that a competitive advantage for you versus your automotive peers?

Yes. So indeed, networking is a huge trend. And you have all of these packets of data that are going across the car. And you want to manage the secure and safe distribution of all of those. And for this, you need a few attributes. On one side, you need the ability to manage a lot of bandwidth of data. And that indeed is an asset that we have in the packet processing from our digital networking from communications side. You also need to have -- to ensure the security of it. And that security goes both in terms of the rights management of your traffic, which again, come from digital networking, but also everything to do with encryption and side channel production, which is a lot of know-how that we have from our [ smart ] business. Now that's not enough. It's required. It's necessary, but it's far from sufficient. You also need to ensure function and safety. So you need to make -- to ensure that this is going to be a platform, which is going to function with the level of robustness and reliability, which is needed. And typically, your high safety processes in automotive are rather simple devices, are rather simple MCUs. And so what we have done is we have embedded that level of function and safety into these high-performance MPUs. And so how do we do this? Well, it's things like taking beyond cause and then building them into a safe platform. And so it's this SoC fabric that we are talking about, which ensures the freedom of interference, to also look at all of the key steps, research, your clock, your power, your memory protection, your [indiscernible] and enable that for safe competition with the right level of redundancy and the right level of assurance. So there's quite a few layers that you need to stack in order to have processes that will meet these requirements of domain controllers for automotive specifically.

Thank you for that. I guess I wanted to combine 2 questions into one. So in prior discussions, you've emphasized the competitive understanding of how the entire system works as necessary for success. And then the second point would be, as you think about the changes that are underway for auto processing. How do the work responsibilities evolve for Tier 1s and Tier 2s? What does the future look like in your mind?

Yes. So Indeed, the conversation, and I think the answer is probably going to be a little mix on your 2 questions because they touch each other. The conversation that we're having with our customers and with our customers' customers, the OEMs, becomes a much more system-level conversation than a function-level conversation. So whereas before, you simply have an RFQ for, hey, I just need now a new body controller. Now the conversation is, how am I going to work collaboratively and define an architecture to manage a complete domain. And so what we find is that in order to add value to this conversation, you need to have been exposed to all of the different bits and pieces because otherwise, you're just kind of a provider of raw technology you're not really a partner. And this is where we can truly differentiate. We have been the supplier of the very first microcontroller in the industry with a very, very fast engine control and the very, very fast automatic braking systems. And so since then, we've developed a tremendous understanding and know-how and application base on both the traditional domains of your vehicle dynamics and your chassis, but also evolving our portfolio for the new demands like electrification and battery management systems, et cetera. So taking this and being able to then having that complete view of the system, enables us to define together with the OEMs, what the requirements for the next generations are going to be. And that is quite unique, right? If you just come from a background of raw processing of brute force, the discussion is going to be very short because you need to go into a lot of depth to understand the criticality of our customers' requirement. And a good example would be real time, right? Your car is a vehicle which is moving through space and time. and you have tremendous requirements in terms of ensuring real time of your compute. But these real-time requirements also depend on what kind of applications are you talking about. And so you don't necessarily need to have the hardest category of real time all the time. But when you do need to have real time for your braking or for some engine function, you need to ensure that this is there. So there's quite a level of depth and sophistication that you need to have the background for that in order to be able to incorporate it into our products and then accelerate the OEM instead of asking too many questions. Now the logical question after that, which is the one you were asking was what about the Tier 1s, right? And what is their role and how does that whole thing evolve. And my experience in the last couple of years is it's evolving in a very collaborative way, actually. So maybe a couple of years ago, I think maybe 3 years ago, everybody thought, hey, robo taxis are going to be in the future. And in 2021, we just have an app, and nobody will own a car. A couple of years ago, I think everybody was thinking, well, the OEMs are going to take everything over. They're going to do all of the software and the Tier 1s are just going to be. ODM and module manufacturers. Now everybody recognizes when all down a second, this system is very complex. We need to organize it. We need to build a hierarchy. There are subsystem within that, and there's a tremendous amount of engineering that needs to go into that. And so the Tier 1s are also organizing themselves more around these subsystems. They have a tremendous amount, generally if the big ones of software engineers that can truly contribute to execute that architecture definition of the OEM and then work with us in terms of providing a software platform, which enables that work efficiently. So the work distribution becomes much more the OEM. OEMs are taking a much more ownership in the vehicle architecture definition. They are becoming much more deliberate into that, which means that they need to understand the whole system, but they don't need to execute the whole system themselves. They are then working collaboratively with the Tier 1s, of course, OEMs have to do more software themselves. And then we are there working with the both of them.

That's great. I get a lot of questions for you on that one. So I guess a 2-part question. The first 1 for what you offer? Do you consider that kind of an open system that can work with other competitor solutions? So Tier 1s or OEMs can pick and choose. And then I guess the second part is, do you think the technological changes that you're making in terms of moving to domains, larger domains. Do you think that will propel some sort of standard based kind of methodology for the automotive industry over time?

Okay. So we are absolutely a believer in open platforms. And our approach as a company since forever, I would say, always has been open platforms are good for the market. We are a leader in this market, and so our role is more to grow the market rather than to try to grab a little bit more market share. So we'd rather have a large market share of a very fast-growing market then to try to control something which is going to stagnate because people feel trapped. So open platform, absolutely. Now this said, we bring value beyond single components to our customers, and that goes in 2 directions. On one side, when we think just of pure processes. So we talk, okay, you have those domain controllers that manage the strategy. You have the zonal controllers that are acting more as a sensor server, if you want, and that organize the physical topology of the car, and you have these end nodes with all of the controls with your battery management system or your inverter actuation. We are providing for all of those devices from the most sophisticated MPU to the lowest power MCU, a consistent software interface. Such that the way that the chip behave, the reset the clock and the debug and the security framework and the safety framework is consistent. And so that really helps the OEMs in the integration. So that's one layer of value. The other layer of value is on the system approach. So when you think of power control systems for an xEV powertrain where you're going to have a DC-DC converter, you're going to have an inverter for your electric motor control unit, you're going to have a battery management system, you're going to have an overall part and demand controller. You probably still, in many cases, are going to have a combustion engine control unit. Each of those requires a combination of a processor with a companion chip to do a lot of the analog management, which started as a PMIC as a power management IC, but now a lot of analog features are integrated into that. You need a lot of IVN transceivers for your CAN/LIN, your Ethernet bus, you need gate drivers. If you control a motor, you need analog front ends. And this is what my sister business line in automotive analog is providing. And what we do is we integrate those into a system solution. So what does that mean for customers? It means, on 1 side, it's integrated on the software level. So we pre-integrate and we deliver a complete software package where everything is already validated. But also -- next to that, we also integrate the function safety framework. Because when you think of the function safety of a subsystem like a DC-DC converter, it is not just about what you process. It is also about how you sense and how you actuate. And so that requires these analog components. We integrate all of those into 1 consistent and coherent function safety framework, which again accelerates greatly the time to market of our customer and the OEM. So yes, very much open systems, but at the same time, they are very strong ways to add value and to be a partner that doesn't simply deliver upon solutions but collaborate on the entire system definition.

So I have a question from the audience that kind of fits what we were just discussing. OEMs are talking about having more direct relationship with semiconductor companies. How might this change things for you?

Very positively. This whole supply crisis in ways on everybody's mind actually has brought everyone a lot closer together. And as much as everyone intellectually was thinking the entire innovation of OEMs of car manufacturers has moved from mechanics to electronics. In reality, their understanding of the value chain of that world needed quite a lot of depth. And I personally have conversations with the Board of Management of almost literally every major OEM on the planet on a weekly or biweekly basis. And yes, a lot of those conversations are on supply, but a lot of -- through these conversations, they are a lot about understanding the dynamics of each other, they are a lot about understanding the added value of each other. And they are very soon move into -- well, actually, we can do a lot more stuff together than what we thought. And so that really accelerates the conversations on, well, let's take the next step in vehicle architecture because clearly, there is more potential there than what they might have realized. And also, if we just take the brutal fact of supply, moving along into more advanced nodes, also has more supply capacity. And so that is another layer about the value that we bring with our portfolio, which is you're going to be into that part of the value chain of semiconductors where most of the investments are going and where hopefully, we'll be able to satisfy the demand without the little dramas that we are living on a daily basis.

That's great. I guess if you think about sizing the market, I know in a previous presentation, you talked about 10% growth CAGR for the auto processing market. $4.5 billion to $6 billion into 2023. You've got -- had previously 24% market share. And I think you as a company have talked about growing 1.2x the market. So in a very kind of ballpark type of perspective, it looks like you think this business can grow from like $1 billion to $1.7-plus billion over kind of a 3-year time frame. Is that the right kind of math we should be using?

Yes. And this is a part of our overall play, which is really focused on to that core vehicle platform. We have a lot of additional growth factors in NXP that touch on that, which are, for example, our radar play, which are, for example, infotainment, with audio or on the display side. So when we think of that core vehicle infrastructure, then yes, it's a market that we anticipate to grow from $4.5 billion to $6 billion. Whereas the growth going to come from, on one side, clearly electrification. So moving from 13% to 26% of costs. Every time you have a car moving to some form of electrification, typically, it carries some enhanced degree of digitalization because everybody wants their controls to be much more digital. And so that brings an acceleration for our end loads. Networking is going to grow at a very strong pace. And then domain control, particularly for xEV and integrated chassis management. So that acceleration that we are going to see in terms of our market penetration is going to be driven by these several waves. In terms of how to think of those and that's this new portfolio that we're discussing here in this chart has the brand name of S32 covering across domain controllers, zonal controllers and end nodes. It starts with the end nodes because you need to have all of these controls, you need to have all of these features. So that is a growth which is really starting now, and it's going to be a big part of this 25% of our overall revenue in 2023 that we are highlighting for the S32 product family. It will continue. But at the same time, we have a tremendous amount of design wins that are being won and that are being now in development with the OEMs for networking and xEV domain control that will start to ramp but they really will start in volume from 2023, 2024. And so they will provide more of the next wave of revenue growth beyond even what we are discussing here right now.

So just to follow up a bit on that. If you think about accelerating EV, what does that mean for the auto processing market? Is that an incremental driver on top of kind of for combustible engine?

Yes. I mean very soon, we will have about 50% of all cars that will be sort of -- some sort of xEV. And as we are saying, it is driving the digitalization of the vehicle together with it. So we see multiple benefits to it. One, the sophistication of the processing and the amount of processes that you need is much higher than when you just had a single engine with a gas tank, one. Two, you need a lot of peripherals what we're talking about before, right, these gate drivers and IVN and PMICs and analog front ends. And we have a pretty unique portfolio for that. And then three, it's driving all of these end nodes to become digital. Everybody that has an xEV wants to have a cool car, right? Nobody wants to have a car with a mechanical seat and roll down windows. He really wants to have the latest and greatest. And that's this MCU part of our portfolio, which we really dedicated for this internal power for these kind of applications. So 3 waves in 1 is essentially what we're seeing in terms of benefit for NXP with electrification.

That's great. So in our last 5 minutes, I guess I wanted to combine 2 kind of questions together. So when -- just talking before, the evolving nature of customer engagement and how the future is not iPhone on wheels, right? And you talked about safety, long-term quality, assured supply. And obviously, that puts you, NXP, in a great position with your customers. The second point would be that obviously, you've been working in the prior few years on current solutions today and clearly, today, you're working on future solutions that I think could push in the single-chip platform type direction. So in the last kind of few minutes that we have, I guess, perhaps you could kind of speak to competitive positioning, given your breadth of offering and kind of the future engagement and future potential for what you can do within your purview of auto processing?

Yes. So we made a bold decision on our road map. -- to move away from this kind of mid-range microcontrollers, which was the norm a few years back. And we made that bold decision to move on 1 side to this high-performance multi-application MPUs with the highest degree of functional safety. And these [ ultra ] power end nodes for the integration of all of these additional controls. That bet clearly is paying off. We see that the adoption of domains, if anything, is accelerating, being fueled by the OEM now really considering their usage of semiconductors and their architectural semiconductors much more holistically and much more deliberately probably than what may have happened in the in the past period. And so the road map validated consistently very good traction with multiple customers. How will that further evolve in 2 directions. One direction is you have this intermediate level in the hierarchy, what we call zonal controllers. And these zonal controllers aim to simplify the physical network topology, where we have a very strong footprint in analog and to unlock all of the data from these different end nodes. So that's yet again a new category of sophisticated microcontrollers that is going to appear and where we are starting to direct a lot of our road map. And on the domain side, we see that these different domains as they -- the strategy has moved from the edge to the center. These different domains, the trend is for them to work together. And so the fact that we have a consistent software architecture means that we are able to have them collaborate. We have the ability to have them work as 1 logical unit and over time, even integrate with the core attributes that we had discussed. That's one of the reasons as to why we've developed our road map in 5-nanometer. The other reason is that just for the domains, we see that the processing requirements within still a limited power budget continue to increase. So we intend to continue to have a thought leadership kind of solutions, bring silicon early to customers in order to facilitate their thinking and the prototyping. And we believe that this will continue to both accelerate the market as well as fuel waves of growth for NXP. So last word, maybe, the importance of safety, we should never underestimate right, in developing a modern car. So we've been investing in that for decades. It doesn't come easily. It is a multidisciplinary problem that is trying to solve. So you need to have this know-how at the application level, this deep application understanding. And we believe that this combination of aggressive road map with deep application understanding with the world's best, frankly, we can say, function safety and security will be pretty unique as we move forward. So we are very excited about the current outlook.

Excellent. Well, best of luck with all of that. And Henri and Jeff, thank you both very much for your time. Always a pleasure to host NXP and learn as much as we can. And I think today was great. So thank you.

Thank you.

Thanks C.J. appreciate it.