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

Good morning, everybody, or good afternoon, depending upon where you're sitting. I'm Ross Seymore, the Head of Semiconductor Research in the U.S. here at Deutsche Bank. Welcome to the next presentation in the 2022 DbAccess AutoTech Conference. For our presentation, we're very, very pleased to have NXP up next with us. We have 2 executives from NXP with us. We have Henri Ardevol, who is the Executive Vice President and General Manager of the company's Automotive Processing business; as well as Jeff Palmer, who is the Senior Vice President and Head of Investor Relations. So Jeff and Henri, thank you so much for joining us today. We very much appreciate it.

So Henri, why don't we start from a high level? Automotive is obviously a huge part of NXP, it's about 50% of your revenues. The automotive processing portion of that, I believe, is about 1/3 of that, so maybe 15%, 17% of the total company, something along those lines, but a big, big business, over $1 billion, maybe approaching $2 billion. So if we look at the automotive processing business, why don't we talk a little bit about the biggest growth driver in that? And we'll start with software-defined vehicles. It's a high level transition in how your customers are changing the architecture of the vehicle, and NXP plays a large role in that transition. But before we get into the NXP specific side, let's start with that transition to software-defined vehicles. Where are we today? Where are we going in the future? And what's driving this transition from your customers' point of view?

Great. Thank you, Ross, for having us here. We're really pleased, and thanks, everybody, for joining and having the conversation. I'm really looking forward to it. So yes, software-defined vehicle is kind of the ultimate goal that OEMs have. And before that, there is a series of steps. And so, you'll probably remember the vehicle architecture evolution that we discussed about a year ago migrating from a distributed setup where every function is instantiated in physical ECUs to domain controllers. And that architecture evolution is complex, it's costly. And we see OEMs evolve that step to domain controllers as platforms that leverage scale and cut across a whole ranges of vehicles. Now domain controller, that step is now a reality, it's in ramp. You can start to see the first domain controllers from NXP based on our S32 family already at the end of 2021. And the domains that we focus on are those that are necessary and core to any vehicle. And so, there are vehicle dynamics, which is your propulsion, your powertrain, your steering, everything that makes a car move and behave and the whole drive experience. Vehicle networking, which is the management of all of the flow of data, which is across the car, which is exploding a typical car today would have 15 different networks. And so the amount of traffic that is out there that needs to be managed and that needs to be handled in a secure way is exponential as well as then not just within the vehicle, but the connectivity to the cloud, right, and the linkage of services to what's happening inside the vehicle. And then the body and comfort. And so everything which is happening inside the cabin and the engagement with the user. And so those 3 domains, vehicle dynamics, vehicle networking and body and comfort, are what we call a vehicle infrastructure. And they enable all of the additional functionalities that are more scalable and updated more frequently, which would then be the other 2 domains, which are e-cockpit and ADAS. So the picture we see now is not only one where this move to domain control is happening, but there are further steps in vehicle architecture, which are taking place. One is zonal, and where domain controllers have as an objective to simplify software and fundamentally to allow the coordination and the up integration of what was done previously in disparate ECUs and to start to modernize the vehicle software. Zonalization is really about taking cost out, physical cost out of the vehicle and it's geared towards the physical network, the topology of the car, simplifying that, taking weight from those 15 different networks that are out there. And that means a category of products where the domain controllers are more processing, real-time processing intensive. Those zonal controllers are more IO kind of centric and then provide a hierarchy to the network and through that simplification take cost out. So that's the next wave, which is happening as we speak. And then the following one is already being defined, and that's the way to software-defined vehicles. And I'm sure we'll talk more about that. But in short, the objective of software-defined vehicles is to unlink the software innovation cycle from the hardware innovation cycle. And so it's really about taking the next step in simplifying the software architecture of the car and bringing further effectiveness and further efficiency where with effectiveness, it means the ability to have services throughout the lifetime of the vehicles that are available to consumer. And by efficiency, it means bringing further stability to the platform, and therefore, having something which is future proof. So a lot of movements which are taking place. I think what is very encouraging is that our vision of domain controllers has taken place. We made bold steps in terms of investing into this new category of high-performance real-time processes. We also had invested heavily into the smart actuation and the end nodes, and that is also taking a lot of -- it's taking a lot of wind in the sales, if I may say. And now we see these are the 2 evolutions of zonalization and then the move to a true software-defined vehicle, and that's very encouraging for us.

Thanks for that description. So if we just put some kind of time frames around this whether it's an NXP-specific time frame or probably more applicable to this question, more of an industry time frame, where are we in that transition to a more zonal processing and then domain processing? And then like you said, the longer-term is to go to the software-defined vehicle. How far along are we in that progression?

Yes. So domain control is something which is starting to ramp in volume now, which means that those designs have been made 2 or 3 years back. And so we started the ramp, and I think we were the first in the industry at the end of last year. It is now gearing up and really we'll go into full bloom in the '24, '25 time frame in terms of new generations of cars coming on to the market. When we look at the revenue projection that we had forecasted in our Analyst Day, the majority of our S32 revenue today is essentially smart actuation and end nodes. By 2024, maybe a quarter of that S32 revenue would then move to be domain controller. So that gives you an idea in terms of the ramp. Then zonal controllers will take place in a staggered fashion after that. So I would say zonal is maybe kind of '25, '26 ramp in terms of vehicle launches, and then full software-defined vehicles that starts '26, '27. So big evolutions. But what is important to understand also is that every OEM has their own cycle and their own rhythm. And so those guys that were first with domain controllers typically will also be first with software-defined vehicle as they have all of the learnings, those that are a little bit later -- come a little bit later into the game may focus more on to zonal and look for some kind of hybrid solution to that.

I think there's a little bit of a perception that zonal would have come before domain to architecturally, it seems like the domain processing is a bigger job, it's a higher level job, et cetera. Is there a reason why the smaller zonal processing or smaller area, more specific function, more cost-centric side is not done first? Is it you have to do the domain first for some sort of architectural reason?

Yes. I think you really need to start to rethink your whole software architecture. And so it really starts with domains with this integration and this coordination of vertical functions on a horizontal way, right? It's the move from managing your engine to managing the entire trip and the experience of the user. And that new software architecture then unlocks the possibility to go towards zonals. Now, in more detail, there are some zones today around body and comfort, which are a reality, right? But I would say there are a simple implementation of a zone, which is kind of a simple aggregator. What we see coming into the '25 and '26 time frame are much more powerful I/O aggregators and with a vision to make all of the sensors, which are available in the vehicle available to the different processes and move to a service-oriented architecture. So strictly speaking, you could say that, yes, very simple zones for body and comfort, which would be kind of very traditional microcontrollers have taken place first, but what we talk about in terms of a true vehicle architecture evolution is coming in a staggered way.

And when people hear maybe more so on the zonal side than the domain side, but when people hear on the zonal side about consolidating a bunch of the ECU sides, a lot of people would say, well, that's going to cut down the number of microcontrollers in the car, and so there's a bit of a substitution effect. Is that something that is still an aggregate positive to the revenue growth opportunity for NXP? Or is there just a misunderstanding about maybe the number of units of MCUs go down, but the dollar content would still go up? Just talk about that trade-off and how the architecture transitions and what it means NXP from a unit perspective.

Yes. I mean in terms of number of processes, right? I mean what we had -- what we have shared was, in 2020, there's about 45 processes per car and no real presence of domain and somewhat zonal controllers of these higher performance processes. Ten years afterwards, in 2030, we see actually growth in the number of processes where the average would be 60%, and that's a worldwide average. And so when we talk about advanced cars then easily go into the 100s. And then for domain controllers that would then be present on an average basis. So with the acceleration of content and with the fact also that electrification drives digitalization, and we have seen that trend very powerfully in the last 2 or 3 years. We actually see that those 2 trends are positive and additive. So yes, more processes per car. And yes, more complex processes to tackle the opportunity of domain and zonal. And so, to give you an idea, a smart actuator, simple microcontroller would probably be in the kind of $5 to $10 kind of range, a zonal controller then moves you to the $10 to $20 range. Domain controller, start to be firmly into the $20, $25 range. And then when we go to software-defined vehicles, where the idea there is to really centralize all of the compute in order to have a platform which is stable for all of these core functionalities of the vehicle infrastructure, vehicle dynamics, vehicle networking, and body and comfort, then you easily talk about double, and in the kind of $50 range. So the opportunity that we see is increasing both in terms of ASP as well as in terms of units.

Thanks for that clarification. So 2 final higher level questions for you. I would assume the answer is, this is independent of whether the drivetrain, whether it's EV or ICE, that this sort of transition is going to happen either way. But is there a difference in drivetrain choices from the OEMs in their ability or willingness to adopt this technology? Or is it somewhat independent of what the drivetrain through years?

Yes. We see a fantastic acceleration of electrification, right? And you see that, too, and it's ahead of all of the forecasts. And so that really pushes hybrid adoption. And yes, for sure, when it is a fully electric or a hybrid powertrain, then the complexity is such that I think there is no real alternative, but to implement a domain control architecture. And when we look at the next-generation of the OEMs, they very decisively are betting on to a major shift of that portfolio towards hybrids and electrification. ICE, I would say, is more of a mixed bag. So we see some OEMs who just have much more of an incremental approach to ICE and don't really want to invest much and want to pour all of the R&D resources onto hybrid and electrification. And then we see other OEMs that not very decisively are shifting the entire portfolio to these new architectures. I would say it's a bit of a mixed bag actually. Maybe just to expand that's on the powertrain basis, right? But when you think about vehicle networking, for example, that explosion of features, right, which is in the vehicle is kind of independent as to whether your powertrain is ICE or electrification. And so, those domain controllers, we see a broad adoption. So it's something that you got to segment a little further.

Got it. So last high-level question for you is, we talk about things within your purview about processing, proliferating all over the place, units and dollars, content going up with semiconductor. How do you see the OEMs offsetting the increased cost that this otherwise will represent? Is it simply that we all as consumers are going to pay more for vehicles or will things cost savings and for zonal processing, as you mentioned earlier on in our discussion be enough of an offset to somewhat minimize that increase? How is the trade-off between higher feature sets and higher price seemingly being attacked by your customers?

Yes. And maybe I'll answer that by somehow painting the picture of what the proposition of a software-defined vehicle is, right? And so there's really 2 visions that are at the heart of SDV, and what OEMs wanted to enable through this unlinking of software from hardware that we were discussing before, effectiveness and efficiency. And so, effectiveness means unlocking the opportunity for continued enhancement of the customer experience throughout the lifetime of the vehicle, why it is in the hands of the consumer. And efficiency means to direct their engineering resources to innovation rather than spend the majority of their time and expense focused on re-integrating the same feature content into just incrementally better hardware. So the first one, effectiveness is about a world where a vehicle will deliver its peak of functionality and consumer value at the end of its lifetime, a world where, unlike today, where you buy a car, it's at their peak the day it's delivered to you and then it slowly degrades. So a world where continued over the year upgrades and cloud services improve the vehicle and its experience on a continued basis. And so, obviously, that is a revenue generation opportunity for the OEMs, and it is a win-win with the consumer. The second one, efficiency is about bringing stability to the vehicle silicon architecture and maximize the reuse of the basic building blocks. And therefore, focus the R&D on new value creation versus repetitive reintegration. We've all seen multiple OEMs experiencing delays in the release of their new platforms. And typically, it's not because of a massive innovation step that they include, but because these innovation steps are burdened with a wholesale architecture tear up and redo in parallel. And so, the -- one of our customers uses the analogy of a tree, right? And so today, every new architecture is like planting a new tree. And so you nurture it and you grow it and you try to develop it, but then the next architecture comes and then you approve that tree and then you start to grow a new one instead of having the previous one, which is maturing its length and delivering fruits, which, of course, means the innovation for more advanced features and optimization of the existing ones. And so, that then means that the design of the vehicle silicon architecture becomes one of the foundational, one of the vital decisions, when an OEM wants to build their long-term platform. And so, they want to rethink their software architecture and to enable that, they want to adopt a silicon architecture that is intimately matched to the goals of that software architecture, and that will then bring a lot of additional value as well as a lot of additional efficiency, which should be a consumer benefit on both of those vectors.

Got it. Well, let's move on to the high-level stuff and start talking a little bit about more NXP-specific dynamics here. So historically, people view NXP as largely playing in the microcontroller space and maybe some application processors. But it seems like there is from a technology point of view, a wider array of processors that you're going to attack. Talk a little bit about from microcontrollers to apps processors to microprocessors, where that processing hierarchy falls within NXP's road map to attack the zonal and domain processors?

Yes. So we indeed have a very broad portfolio. And when we think of what OEMs want in terms of thinking of the long-term development of their vehicle architecture, they want depth, they want breadth, and they want scale. And so depth means to be able to deliver breakthrough our processor architectures and technology. And already 5 or 6 years ago, we took a very bold step, which was to move away the majority of our investment from traditional embedded non-volatile microcontrollers to high-performance processes in FinFET technology. That has enabled us to win that domain control wave that is now taking place. And a couple of years back, we had this vision of SDV with a number of our customers, and therefore, we decided to fundamentally invest into not just the integration of functionality into domain, but the cross integration of those domains. And that's our 5 nanometer platform. And you may have seen in August that we signed a press release, which is -- which maybe was a little cryptic, but basically, what I was saying is that a major automotive OEM, a major global automotive OEM has decided to define their entire vehicle road map based on the S32 family, which is those very high-performance processes that we just discussed, as well as the zonal aggregators. And the reason for that is the fact that this S32 family is built as a family instead of just isolated products. And so there's a tremendous amount of reuse, which is possible, which then gives them a lot of scale. So on the processing side, we've done a lot there in terms of depth and in terms of scale to have the performance as well as to have the cost effectiveness. But then the breadth also comes from the rest of our portfolio. And that is very important, not just from a revenue opportunity, but from an overall system understanding and the ability to be a true partner to the OEMs in thinking, hey, what are the potential trends and what are the potential benefits that they could think of as they go to the next architecture. And so, yes, we are well positioned in gateway, in propulsion, in body and comfort. But let's not forget, infotainment and cockpit, right, where today, we are the largest player in audio infotainment segment and with a true leadership position. And we have a very solid mid-tier e-cockpit business and traction with wireless connectivity. And everybody always talks about this kind of pillar-to-pillar, 55-inch display, which is in the car because, of course, it's very catchy. But if you step into the new 7 Series from BMW, whether it's the V8 or the hybrid, you go into the backseat and on armrest, you have a display. And that display is there to control all of the functionality that you can enjoy from the back seat, that kind of display is proliferating across the vehicle, and it's exactly that kind of mid-tier business where we have a lot of strength with our [indiscernible] franchise. Then in ADAS, we got a great franchise. And sense of fusion will be a complex situation of coopetition. But in terms of radar, we have definitely a true leadership position, which then has the vehicle architectures evolve and has so much data is spent out of these sensors opens up further opportunities to aggregate it and to bring it into a manageable way to the sense of fusion processes. And then finally, you have all of these chips that are essentially digital, well, you got to bring it to life, right? You got to bring it to life with interfaces. You've got to bring it to life with power management, you got to bring it to life with battery management systems and inverters and all kind of actuation with switches, whether they are -- can lean where we are historically and continuing to maintain in terms of wins, a very strong leadership in IBM, and a very solid and growing event at finance with business, which is far larger than anybody else in the market. And so that portfolio for us and that breadth is very important to keep next to the depth and the scale that we can bring on the process side such that we are much more meaningful partner. And even I would dare to say that silicon becomes so foundational to the road map of the OEMs that there's almost an effective vertical integration of the semiconductor manufacturer with the OEMs, and therefore, it requires a lot of trust between those 2 parties, but also it requires a tremendous amount of understanding, right, of what the system is and how it evolves. And then how to once some of the challenges, even in an analog chip or in a power management chip or in an interface chip or in the processor. And so that's why this portfolio has been architected in such a complete way.

It seems like the S32 portfolio is unique in many ways in its breadth, in its depth, et cetera, and you guys have been working on it for many, many years. Talk a little bit about what the competitive differentiation of that family is, whether it's at the relatively lower end of processing, against high-end microcontrollers or the high-end against when people think of microprocessor, central CPU companies coming down into the automotive? How do you guys compete and win business with the S32 against those 2 ends of the spectrum?

All right. So the first one we go to SDV, right, and then for the realization of that proposition. One of the hardest decisions that OEMs have to make is how much performance headroom to leave unused at the vehicle launch. Because obviously, in a world where an ECU is just a function, while you closely match the performance to the functionality. But in a world where you want to have 10 years of over-the-air upgrades in a year where you think of -- and to keep it in perspective, for this kind of vehicle infrastructure, vehicle dynamics, networking, body and comfort, think about 10s of over-the-air upgrades, 20, 30, 40, 50 over-the-air upgrades of the vehicle over its lifetime, you need to leave a significant amount of processing there. And so that requires a portfolio that has that umph, right? And that has that capability. But you need to do this in a cost-effective way. And of course, a typical semiconductor way to deal with this thing is through accelerators. And so in the S32 family, we have a lot of dedicated accelerators to make sure that the compute which is available on the chip is not consumed for networking tasks or that we have machine learning accelerators that are effective and match core processes or on the radar side that we have dedicated transform engines and all kinds of pre-processing, which leaves a lot of the CPU then available for applications. And the kind of space where we play is fundamentally the real-time space. It's what makes the car move and behave and drive. And that means that the fundamentals of automotive, which is your functional safety, your security, your robustness, your quality, your reliability become more important with time rather than less. And the use cases of these processes become more and more complex, right? When it was a processor living in a box, where you just switch off the car, and that's it. I mean that thing is off and you don't have to worry about its lifetime. Now when it's a vehicle network processor, for example, that somehow has to be always on, well, you need to manage that in a very smart way, such that you only have islands within the SoC, which are active and that then communicate with the rest of the silicon architecture in a very efficient way. So that is very hard to do with just brute force compute. And it's -- the level of interact, the time of response and the low-power consumption when the SoC is of the speed of wake-up, the determinism in terms of giving a comment and knowing that it will be performed. All of this is a domain and an expertise, which is really different from your heart called big compute, which then would be that one much more suited, of course. And that's why you see all of the big compute guys into ADAS, because it's essentially about moving data whilst what we do is essentially about moving the car, right, moving that physical object within space and time and connecting it to the cloud. And so, an analogy we often use is there is 3 levels in the car, you have the mechatronics level, you have the vehicle OS level and you have the cloud. And a lot of where we play is in the interaction between the mechatronics and the vehicle OS, and where the big compute guys would play, it's in the interaction between the vehicle OS and the cloud. And we enable the services, but they probably deliver the services.

Got it. So in the last roughly 10 minutes that we have left, I just want to transition and kind of wrap up with kind of the real-world business implications for all of this. Is the go-to-market through Tier 1s directly with the OEMs, everybody at the same time in a room in a 3 way conversation? Talk about how you get your products to the market and who makes the decisions to put NXP on the road map?

Again, a whole spectrum where you have all trays of grey from very large OEMs, not only making the decision on the silicon selection, but actually taking the ownership of the software, on top of that, which then means for us, that we have an opportunity to deliver much more software than before as well as an expansion of our portfolio. So from that end of the spectrum to OEMs that still very much rely on to Tier 1s and assembling boxes and things in the middle where OEMs do want to define the architecture and do want to define the high-level goals, but the actual selection is being done by the Tier 1. And what that means for us, it's an end go-to-market. So it's not a transition from A to B, we remain absolutely focused on the Tier 1s and definitely for the smart actuation, definitely for the zones, this is their territory, and it will remain. We remain very focused on the Tier 1s also for their platforms as they move to domain control and further where they have through their scale, the ability for the small OEMs or those that want to invest less, the ability to deliver a more complete solution. And we engage very strongly with all of the OEMs. And so, in a way, the supply crisis that we have gone through in the last 2 or 3 years has been a blessing, because it has forced the silicon manufacturers and the OEMs to get a lot closer. I mean, I still speak at least every 2 weeks and before it was every week, so it gives a little bit of a sense of the relief, it's definitely not complete. But I speak with the CPOs of all of the major OEMs around the world, right? And so when you do this for 2 years, and I do this constantly, and of course, my team does it on a daily basis, that creates a lot of intimacy, that creates a lot of opportunities for getting closer. Those conversations, which, of course, have been more supply chain kind of focus have evolved into very strong conversations on the engineering basis, and I think has accelerated actually that whole transition. So it's really an additive also on the go-to-market for us.

Maybe the last or second last question is on exclusivity and I'll hold 2 aspects into this. A lot of your peers in the semiconductor side that service the automotive market, talk a lot about their pipeline. I know that's not an NXP metric that you guys provide. But these pipelines are enormous, just 10s and 10s of billions of dollars. As I look at them, it seems like there has to be some double counting just given the sheer size of the billions in aggregate. So talk a little bit about are these design wins exclusive? Are 2 companies or 3 all winning design wins and then triple -- double, triple counting them? Just talk about how that dynamic works.

Yes. The whole topic of design wins and good luck peeling the onion with some of our peers is, of course, complex, because it all depends who is actually the decision maker and then who filtrate this? And so, you may win a large platform at a Tier 1, but if they don't have an OEM behind, then yes, you have won the opportunity to have business, but they have not won the business, and therefore, you also have not, right? And so that needs to happen, and because it's an enabler, but then you also need to go either to support the Tier 1 and winning at the OEM or just to trust that they will do a good job. And the other way around, right, with OEMs, you can be nominated in the RFQ, and they have decided that they will use your products or your portfolio. The execution of it in terms of the boards and in terms of part of the software will also be given to the Tier 1s. And so you got to be sure that you're going to support those, and that the business is not lost at the 11 Tower. So it's something which we in NXP, yes, we don't disclose those design wins numbers, but we have a lot of focus on making sure that we count them only once. Because indeed, it is very easy to get carried away with this kind of big platform numbers that actually have nothing behind and double counting or triple accounting. You win with the OEM, you have 4 or 5 Tier 1s that are bidding. You need to make sure that you're present in those 4 or 5. Now if you count in terms of design win, the OEM and the Tier 1s, then it's a monstrosity of double counting. So yes, I think some sense needs to be applied. Now you were saying about exclusivity to give me another point, which is the way our go-to-market is evolving, especially with OEMs, it's less and less of product sell and more and more a platform sell, because they are thinking about their entire software architecture and then having the underlying silicon architecture. And so, what we've built is the flexibility to instantiate products that are closely matched to dedicated OEMs software architecture. And so yes, that may result into dedicated tape-outs for specific OEMs, not necessarily an exclusivity per se, but a very close match and back to this topic of almost vertical integration where we do all of the heavy lifting of creating the SoC platform with the security, the function safety, the reliability, the robustness, the accelerators, the SoC fabric, the virtualization, the software support. But then how that is then assembled is a conversation with the OEM in terms of making sure that it will be matched to their software architecture. And in a sense, we create a unified team that has this 1 goal of delivering that value proposition. And this is not a theoretical example what I'm talking about, this is actually what has happened and what is behind that press release that we did in August, where we can't disclose too many details except that it is truly transformational in both the content and the value, the vision and will be truly, we believe, an accelerator for the market and definitely for our business also.

Great. Well, why don't we wrap up with one last question from my side. At your last analyst meeting a year ago, you talked about this segment having a 10% to 12% CAGR, going from about $1.7 billion last year to $2.4 billion -- $2.3 billion in '24. Automotive as a segment is growing significantly faster than that, about double that rate this year. Given how you laid it out, Henri, about the transition to domain and zonal than software defined. It seems like the tailwinds are only going to accelerate over that 3 year time frame. Is there anything that's gotten you less confident in your ability to grow? Because it sure seems like you have an accelerating tailwind.

I think that's a fair assessment of the market as it was and our performance. And so indeed, we had said market growing at 9%, we grow at 1.2x the market, which then gives this 10% to 12% CAGR. On the overall business, by the way, not just on S32, S32 I think we had talked about a 25% CAGR. Yes, the market is strong. Yes, we absolutely will do. And that's what we see at least this 1.2x the market. And so we feel very confident about the numbers that we -- number one, the numbers that we have delivered, and number 2, our ability to keep up with the market evolution.

And Ross, if I could just add one thing to that, if you'll remember, when we did the Analyst Day in November '21, we did not include any inflationary cost overlay on top of that. And as I think everyone in our industry realizes the last 2 years has been a significant inflationary pressure from our suppliers that, unfortunately, we've had to find ways to pass on across the whole company's portfolio. So the 10% to 12% CAGR that Henri talked about at the Analyst Day, there is an aspect that layers on top, which gives it a greater tailwind.

Perfect. We'll, Henri and Jeff, thank you so much for taking the time to join the Deutsche Bank AutoTech conference today. We're officially out of time. It's a fascinating part of your business in a general market trend that many, many investors are interested in. So we really appreciate all of your insights. So with that, we'll end this session. Thank you all for dialing in and listening in.

Thank you very much, everybody. Thank you. Bye-bye.