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

Good morning, everyone, and welcome to Needham's Sixth Annual Virtual Automotive Technology Conference. We're very pleased to have everyone joining us again this year. My name is Raji Gill. I head up global semiconductors and automotive technology research here at Needham & Company. We're very excited to have NXPI with us for the first time at this conference, and I think it's going to be a really interesting discussion. From NXPI, we have Kurt Sievers, who's President and CEO of NXPI as well as Jeff Palmer, Vice President of Investor Relations. The format of this presentation will be basically a 35-minute fireside chat. For folks that are online, if you have questions, there's a chat box. I'll be monitoring that. So feel free to send your questions and also kind of e-mail me questions as well, and I can relay those to the audience. So with that, welcome.

So Kurt, maybe we could just first start a little bit with the macro, particularly around what you're seeing in the automotive supply chain. There's been some recent reports that capacity constraints for specific components are starting to kind of ease up. This is now basically 2 years where the automotive industry has been affected by kind of unprecedented chip shortages. So I wanted to get a sense from you in terms of kind of where we are in terms of the overall automotive supply chain.

Yes. Thanks. And first of all, good morning, Raj, and everybody listening in. Clearly, the auto supply chain has been a significant pain point over the past 1.5-plus years probably spearheaded by semiconductors. And since we are a very leading global semiconductor supplier to the auto industry, it is fair to say that NXP has certainly been right in the middle of this. Now I give you 2 answers, Raj. The one is, no, at least for the portfolio which we are serving which is actually spreading from analog all the way over to microcontrollers and microprocessors. We do not see it easing off, at least not for the remainder of this year. I mean I'm also not in a position to have all the details for next year. But while our supply capability is gradually improving from quarter to quarter to quarter, I do not see that we come into balance against demand anytime this calendar year. The other answer I would offer you is indeed a bit more differentiated, which is clearly there have been more supply chain issues compounding to the overall automotive industry, where 1.5 years ago, it was purely or almost purely around semiconductors. So the wiring harness issues, which came up after the war had started in Ukraine have been adding. And certainly, more recently, the -- all the negative implications from the Zero-COVID shutdowns around Shanghai and partially, by the way, also in the Greater Beijing area are adding to the puzzle with other things. I mean, it's also those are including semiconductor but there are also other elements. And I mean all of that is then resulting in what I personally see as a very disappointing SAAR development this year, which I think is now forecasted to only be 4% up from the super low number from last year.

So on the last earnings call, you mentioned in terms of lead times, 80% of your products now are quoted at 52 weeks or more. Is there any kind of change there? Can you talk a little bit about the specific end markets? We're hearing -- as you said, analog type parts are still extending, but microcontrollers and sensors might be getting a bit better, but that doesn't seem to be the case in your situation. So any sense in terms of where we are with lead times?

No, it actually indeed. Last earnings, we spoke about around 80% being at 52 weeks plus, which was actually worse than a quarter before. And I mean that gives you indeed the dynamic while the volumes we are shipping are going higher. So that imbalance is not a function of us being crippled from a supply perspective. Actually, I mean, we have grown like never before. It is just that the demand following the enormous content growth of electronics in the car. And I know we're going to speak about this a bit more later is it just continues to be out of balance with our supply capability. So no, I have to answer this bluntly with a no across all of our product categories. We are in that unfortunate situation. Now just to report it in perspective, Raj, it is not much different in our industrial market. I mean the second largest segment, which NXP is serving mind you 50% of NXP goes roughly into automotive, the next one in size is industrial. It's less talked about, but it's the same thing. So I just want to emphasize, this is not an automotive specific situation. It's actually much broader because it comes from fundamental capacity buckets. For example, in analog mixed signal and the microcontrollers, which are universal to automotive and other markets.

Okay. And just a question on the SAAR. So you -- in 2021, it was down to about 77 million units, maybe it goes up to about 80 million, but it's down about 20% from the high a few years ago. Do you see any kind of recovery in 2022 in terms of overall auto start production based on your kind of conversations? And can it move higher if the COVID shutdowns in Shanghai since I think they're ending now? Could that lead to a bigger rebound in 2023? And I guess the other question is, how does the transition from ICE engines to EVs to kind of help or impact this overall kind of market production?

Yes. So I mean there are indeed a couple of dynamics. So I should say, we typically indeed look at IHS as the main source of truth, when it is about forecast of the SAAR. And they would not see the SAAR coming back to these peak values of 95 million in the next couple of years. So I think only beyond '23 or '24 -- beyond '24, actually, they see it coming back. Now still, next year, IHS sees a 9% rebound over the disappointingly low levels from this year. And I hope, I would say, from today's vantage point, that, that can also be supported from a semiconductor perspective. Now I think you hit, however, the core point. And the core point is about what is the underlying mix. And I would say the mix relative to semiconductors is super relevant, both in the direction of xEVs but also in the direction of premium vehicles. And the reason is that the xEVs would typically have a 2x semiconductor content to a combustion engine car. And the premium cars, they can be 2 to 3x the semiconductor content. So that mix has shifted massively. This year, any forecast would say that 30% of the total SAAR is actually attributable to a combination of xEVs in premium cars, which means 30% of the car production has now a semiconductor opportunity, which is 2 to 3x bigger than of an average car in the past. And that 30% used to be only 15% a few years ago. So I mean that's why looking at the SAAR only to try and understand the semiconductor growth is, I believe, too little. You have to look at which cars are being built. And honestly speaking, Raj, the trends for those 2 are very different. I think the xEVs are really following a secular trend, which has to do with CO2 emissions and actually legislation in several countries. However, the premium cars is purely a function of profit maximization of the car companies. I mean they -- with being capable of building less cars, what they do is they build those cars, which print the biggest margins.

Yes. That's an interesting dynamic. You talked about in the past, the focus on premium vehicles, you're kind of benefiting kind of 2x over because not only are you seeing a higher penetration of EV units as a percentage of the total market, your -- it's actually higher end vehicles at the same time, which has more content. So just a couple of questions just on the macro, and then we'll move into the technology. You mentioned in the last call about new car dealer inventory was sitting around 27 days. It had increased a little bit, but it was still well below the average, which was 64 days. You talked -- this has the question about inventory and broadly. You talked about the Tier 1 auto suppliers and the OEMs remain very challenged where a variety of supply chain issues related to kitting to modules of vehicle assemblies. And so I wanted to get a sense in terms of distribution inventory, Tier 1 auto supply inventory and kind of OEM inventory because I know that you're working so closely with these companies, you have a really strong pulse on what's happening. Any kind of changes or anything incremental in the last 3, 4 months?

No, not really changing, but you certainly hint to the right point. I mean, for me, that fortunately or unfortunately, it's a very personal experience because there is not a day where I'm not spending time with either Tier 1 CEOs or OEM CEOs chasing parts. So I think the dealer inventory thing is just as bad as it has been, and you quoted the right numbers. To my best knowledge, it's a historic low. And at the same time, the used car prices are skyrocketing. I mean both hinting at the same implication, which is there is enormous demand for cars in North America, which just cannot be served. And by the way, and I'm sure you have the same -- just anecdotally, if you speak to friends and family in the U.S., everybody has the problem that he or she would love to get a car is just that difficult. So it's -- this is not where any inventory is sitting. There are pockets of some inventory at Tier 1s, but the numbers which have been quoted really have to be looked at in the right way. The one reason is that some of them are a result of these kidding issues where a product is missing for a week or 2 weeks, and then that product comes in and then it's flushed through, which means it's not a permanent inventory. It's just a snapshot and then it goes out again. Secondly, and I think that's the much bigger factor. The inflationary development which we have is valuing the inventory at much higher levels. So if you compare now inventory to half a year ago, from a volume perspective, it's probably not much different, but the value is just different because all of the input prices have gone up significantly. And mind you that semiconductors are a very small part of this. I mean we talked about doubling of semiconductor value in a car. Even if that is doubled, it's $1,000 per car. Now you know what the bill of material of a car is. It's so much greater than $1,000. So the semiconductor value cannot be derived from these inventory values, which you get because other stuff is so much more expensive, if you will, from a valuation perspective. Now distribution is a very interesting part of this. So distribution, indeed, distribution inventory has always been a very good measurement point to see if things are piling up or if there is very fast sell-through because of strong demand. Our distribution inventory, and that's a large -- distribution is a large channel for NXP is -- continues to be very stubbornly, I have to say, at an all-time low of 1.5 months. So 1.5 months of distribution inventory, where we have a target of 2.5 months and that target we've had for, I think, 5, 6, 7 years in a row. And when the spike in demand started 1.5 years ago, we dropped to 1.6 months and finally 1.5 months. And we just don't get off this because the sell-through is so strong. So that's why I would say that part of the equation is a clear indicator that there is no inventory building. Now Raj, I just want to make one final comment on this because, I mean, we all see a lot of data and all of the considerations about the macro. Everything I'm saying is really tuned into the automotive and industrial markets. What I'm saying cannot be taken as a proxy for the overall semi industry. I also see that mobile and PC markets are slowing. This was really an auto and industrial statement.

Got it. Yes. No, that's important to understand. Even though some investors are painting a broad brush across the board and the related to no distinction being made. All right. That's really interesting. Just kind of switching into kind of the market. You talked about -- let me ask you this, how do you define xEV and what you call premium ICE vehicles? Is that based on a certain price point? And then along the same lines, your estimate of xEV penetration in 2022, I believe, was 23% kind of moving up to 30% next year. Are there certain OEMs that are kind of moving the needle faster or slower than others?

Yes. So first of all, the definition is very straightforward. I mean, xEV is any form of electric vehicles starting from mild hybrids to mid hybrids, full hybrids and full battery electric vehicles. The reason why that matters is because all of them, starting with the mild hybrids have some form of an electric engine, which allows the car to drive electric. Maybe only 20 or 30 miles, but still the infrastructure from a semiconductor perspective for that electric engine, including the battery is existent. And I mean that's the piece which drives our semi value. So it is not only fully battery electric vehicles, it's also the hybrids which are in that xEV definition. And by the way, this 23% is forecast to grow to, I think, 29% or 30% next year. So it continues to penetrate very, very quickly. Premium vehicles is a bit more fuzzy to be honest. We use the IHS definition. I don't know what their exact definition is, but it is really about luxury cars. I mean that is the top end, the premium top end typically of the car companies. You also see the number is much smaller. I mean that was -- or is 7% of the total SAAR this year, so much less than the xEVs. I would say from an xEV penetration perspective, so who really drives it hard, let me first make a geographic comment. I would say, clearly, Europe and China are leading over the U.S. so far. That has to do, I believe, with legislation and also prices for gas. So in Europe, it is actually difficult to not have an xEV in the meantime. And in China, it's a government policy, as you know. I mean, there is really even now, I just read earlier this morning, where Shanghai has reopened on June 1. Immediately, they put out new incentives for electric vehicles in order to drive consumption, which is a good thing, if you will. The U.S. is a little bit behind, but I'm smelling that also their breakthroughs are coming because when we speak to the big OEMs and think about Ford or GM or Stellantis, clearly, the portfolio of electric vehicles, which is coming online is massive. So I think it's just a fantastic offering. So that I believe also the U.S. are going to catch up quite rapidly.

You talked about the silicon content being 2x that of ICE engines. Is there any kind of further granularity in terms of, say, hybrid versus a full onboard electric vehicle? You mentioned that you're including any type of electrical engine in that penetration. Does the content kind of vary?

Yes. So the higher the voltage, which is used, the higher the content. It is a good proxy. So the highest voltages are 800 volts on fully electric vehicles and if you have a mild hybrid, you can come all the way down to 48 volts. I mean there is a significant range. And clearly, with the high voltages, you get a much higher content. A good example is our battery management solution. It's very easy to translate. You have a high-voltage battery that has many, many cells. And on each of these cells, we have an analog front-end chip. So it's almost a linear equation. If you have high voltage, you have higher chip content because it just scales with the amount of cells. That is a statement for the BMS, but there is similar considerations for the rest. Just let me remind you of the fact, Raj, that there is more semi content to electric vehicles than only the drivetrain. Typically, these electric vehicles are made for a more tech-savvy consumer. So they have bigger displays. They have more safety and convenience electronics. It's a matter of positioning because typically, apparently, the OEMs expect that somebody who buys an electric car wants to have a very modern electronic environment in the car, which is why we benefit not only in the classic drivetrain electronics, but also with other features, which are just having a much higher penetration rate in these xEVs.

So you talked about in the past, Kurt, of that there was an architectural change happening in the OEMs. And this -- there's kind of 2 parallel architecture revolutions happening. There's the domain focus on kind of the zonal focus. And the former, the zonal focus is about scalability and more of a centralized kind of software development, while the latter is really more about wiring and connectivity. So I'm wondering when you're talking about domain versus zonal, can you maybe talk about your S32 kind of MCU family to meet some of these evolving needs? And then also talk about how you're kind of addressed on kind of the zonal focus?

Yes. So we see the 2 hand-in-hand. They are not orthogonal to each other. Both of them, by the way, are a result of the mess, which has been created by a flat network which was unmanageable going forward. The way we think about this is that domains are about a logic restructuring of the applications in the car. So you get a domain computer, which does all the body electronics, you've got another one which does all of the ADAS or one which does all of the drivetrain. And the zonal computers are a physical aggregation of compute power. So in a certain corner of the car, really physically you concentrate compute power more into 1 product. And the 2 happened in parallel. Our S32 has actually been developed initially to lead the transformation to domains and is now also pushing the zonal, which came a little later, the direction to zones. The way you got to think about it is, we have -- we are now launching and renting in significant volume, 16-nanometer-based domain computers, which are quite high performance microprocessors, where the zonal computers tend to be more classic microcontrollers which we do in 28 and 40 nanometers. However, the big advantage of this S32 platform is software compatibility between all of them. So all of the family members of this S32 family, and I think with every year, which is passing by, it gets more powerful, are working on the same instruction sets such that the biggest headache at least from my personal experience from OEMs and Tier 1s, which is software R&D, is optimized by the opportunity to reuse legacy software from 1 model to the next model. And that is an independent of a domain or a zonal computer, which is very important because functions tend to shift between them. I mean, what used to be in a domain might move to a zone, which is why it is so super important that the same code can be reused between them. And I'm proud to say, Raj, that at Computex a few weeks ago in Taiwan, I actually published that we have now the first 5-nanometer domain computer coming up. So it's test silicon still from TSMC, but also that is going to sample, I think, year after next and then probably go into production in 2025, which is the most advanced part of domain computing, with full functional safety, a pretty powerful story.

That's very interesting. And so just along the same lines around the S32 family, Kurt, that's now ramping on 60-nanometer FinFET that you talked about, it's around $300 million. That's going to double over the next 2 years. The driver of that is kind of software life cycle management. Is it correct to think that this is the same as kind of over-the-air updates? And these updates are getting larger in size, so they'll need more powerful silicon systems to enable the downloads? Or what about the upload things for like mapping or any other sensing feedback?

Yes, it's not the same, but it's very associated. Indeed, over-the-air updates, the whole idea is to refresh the performance, the electronic performance of your car during lifetime. I mean, that has always been the big issue. You buy a car, and honestly speaking, a year later, it was outdated from a performance perspective. So what now is being done is to build in domain computers, which have headroom in performance such that over the next 4, 5, 6 years of their lifetime, you download new software, which gives you additional performance or new use cases. Now the challenge is only, Raj, that this is an open door to get virus into your car. I mean the issue is the whole -- this whole system is open in a way. So the safety and security aspect of managing the over-the-air update is a pretty critical thing. I mean it's a little bit like when you update the operating system of your phone, I guess every time you do this, you hope it works well. And all your data is still there, et cetera. It's the same in the car. The difference is only that the complexity of the electronic system in the car is so much bigger than a single phone is. And what we do is we have a dedicated domain computer for what we call the gateway chip which is doing nothing but this. It sits there and manages the over-the-air updates. This is typically indeed data coming into the car. But of course, it's the same bridge which is used for data from the car being sent to the cloud. And I think some time ago, we also made an announcement with Amazon Web Services, which is running their stuff on this computer. So this, I believe, the whole idea of services will become a whole new dimension of business for the car companies, and that's why we are right in the middle with this domain computer managing that.

So if I kind of switch gears to the EV components, the full stack of EV components that you're supplying. So that includes the battery management system, the kind of e-motor control, so it is the inverters, all the propulsion domain controllers. So can you talk about how these -- the bolters converters, how are these components kind of interrelated, how they're working together? And how scalable and are these components because these are critical components for EVs.

Yes. So our focus is indeed on the power control of the EV, not the power delivery. I just say this upfront. We don't do discrete power like silicon carbide or IGBTs. Our focus is on where our strength is, which is the power control. And prominent examples are inverter control and battery management solutions and the -- and all of the microprocessors for the propulsion control. Battery management is a good example of scalability. The system is a microcontroller in the middle and then this analog front-end high-precision chips on each of the battery cells. And it is perfectly scalable because you just have more -- you have more battery cells, you have more analog front-end chips, but it always goes back to the same microcontroller. So we are just adding more analog front-end chips, which is nice from a business perspective, but it starts already on a 48-volt battery, but can go all the way up to 800-volt battery. The difference, however, to other vendors is that we offer this as a system solution. So the analog front-end chips and the microcontroller and the software comes out of one hand, which I really believe eases the scalability for our customers as if you had to take it from 2 different suppliers.

So what is kind of the average number of battery cells now in a battery pack? I know Tesla has been leading the way for many years, but I mean, it's -- where are the -- as you said it was kind of a linear relationship to the analog front on each battery cell, where do you think the number of cells are going to?

Jeff, Jeff Palmer is with me on the phone. What would you say, Jeff, is an average number? I mean I know the extremes, but I wouldn't know how to call the average. Jeff, do you know?

Actually, I don't -- not actual number of cells, Kurt. I don't -- I kind of have a dollar figure in my head, but not the number of cells. I can look into that, Raj.

Maybe let's give the dollars from the small battery to the big battery, Jeff.

I think if I remember right, kind of at the low end, BMS full system is about USD 60 kind of at the lower end. And multiples of that at the 800-volt type model. So north of $120, $140, $150, something like that, depending on volumes.

Got it. Got it. Okay.

And Raj, what that says is that the more full electric vehicles we would get, the more we benefit, obviously. I mean there is a significant gradient here between a low cell to a high cell battery.

And I think the scalability can't be overstated because if you think about what the OEMs are trying to do from a portfolio management perspective, they're trying to basically write the software, write the system development code once and have that scale across different platforms. Our friends there at Volkswagen Group are a perfect example. 800-volt [ Taycan ] sports car, that's 800 volts, that's a beast, very nice content for us, but it scales down to more of the entry-level ID platform that are maybe 400 volts or even less. So it's a nice opportunity for us.

So yes, I just wanted to tie that loop circle because the battery cell count is really interesting and just increasing. So the -- so we have that. But the BMS system specifically, is so -- is crucial for an EV. So maybe can you just talk about the importance of having a very sophisticated BMS system. Why is that important?

I think very simple for optimizing the range. I mean, the whole idea is if there is one issue with electric cars, it is range anxiety. So in order to get maximum range out of a battery, you have actually 2 levers. One is the chemistry of the battery, not our business. And the other one is the electronics, which optimize the power density, which comes out of the battery. So I think it's an absolute fundamental thing to make electric cars successful going forward.

And it's also -- both sides of the equation, Raj, there's also the charging, right? So how long, how quickly it takes you to charge or recharge your battery and then the very frugal management, as Kurt is saying, of discharging the battery to get your maximum range. And then simultaneously, while this is happening while you're driving, is the health of battery cell, right? And so that processor and that system approach we take is a holistic, almost like a load balancing kind of monitoring system of the battery health.

Yes. That's definitely. Now you talked about some use cases for wireless-based BMS systems. Can you talk about why a car company would want kind of a wireless-based BMS system? Is it to lower weight? Would ICE vehicles adopt this? Maybe you can talk about how ICE vehicles are adopting new technologies that EVs are naval like wireless BMS. Is that happening from your prospective?

First of all, all of this is only for electric vehicles. It doesn't -- I mean, there is no use case for this in a combustion engine car. We are indeed doing research and predevelopment for wireless. The ultimate idea would be to save weight and to reduce, say, complexity, which means reduced possible source for error. You have connectors and things. I mean, that can -- theoretically, that can fail. However, the big challenge is there is a lot of radiation issues by doing this, which is an EMI problem in the car. And secondly, the whole complexity with other wireless systems in the car is enormous. I mean there was a lot of wireless stuff in the car, and you just can't have that to fail. So I would put a little bit of caution on this. I know the market has a few very first trials of this running. Since we are a strong leader in that field, we also do our work there, but I would be a bit cautious when that would get fast and broad penetration. But theoretically, if it worked flawlessly and had no interference with anything else, it is obviously a nice idea.

And just kind of shifting gears to kind of automotive radar because that's the other big kind of engine of your auto business. So we're estimating kind of ADAS penetration Level 2+ to get about to 80% in 2025 when we look at a bottoms-up view. And automotive radar is going to be obviously crucial to that. So I wanted to get your sense in terms of what your expectation is in terms of number of radars when you get to kind of Level 2 plus, Level 3, Level 4? And you're on 77-gigahertz radar, I think you have about 192 channels. Is there -- what are your thoughts about higher advanced 40 radar for like imaging radar that have higher channel count. Is that something that you see happening in the future? Or do you think standard radar will be really where all the volume is going to come from?

Yes. So first of all, just to level set this, indeed, super important for NXP. I mean, we are the #1 in automotive radar. We were like $600 million revenue last year and will about double this to $1.1 billion, $1.2 billion in 2024. The growth for this is really driven by 3 factors. One is more cars which get radar per se. I mean, there is still cars which have no radar at all. Secondly, more radars per car. And thirdly, and that goes to your imaging radar, a higher performance, which means higher value per radar. All 3 are happening at the same time. For me, this is mainly around safety, Raj. So when you talked about a maturing of the ADAS penetration, I cannot see that at all for radar. I think it just goes and goes until we hit 100% because it is about fundamental safety features. So this is not ultimately -- initially, this is not at all about Level 4 or Level 5 self-driving. It is about Level 2, Level 2 plus, eventually Level 3, which is mainly around safety. The imaging radar comes from a different perspective. Imaging radar from a performance perspective touches the lower levels of LiDAR. So we actually believe that goes then indeed more in the direction of Level 3, 4, 5 that you would otherwise have to consider LiDAR. And we think from a form factor perspective and a cost perspective, LiDAR is really far from ideal and imaging radar could kill it. So 77 gigahertz clearly is the direction where to go. And for the rest, you have a whole spectrum of performance at the high-end imaging radar but also all the way down to relatively cost-competitive 1 chip integrations for [indiscernible] radar and no maturing of penetration.

Got it. I think we're out of time. Thank you so much, Kurt, Jeff. Super insightful, best of luck for the rest of the year. It's a great industry to be in. And so happy to be looking at it and analyzing it. Thank you for everyone for joining as well on the call. Appreciate it.

Thank you, Raj.

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