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

Welcome to TD Cowen's 53rd Annual TMT Conference. I'm Josh Buchalter, semiconductor analyst. Very pleased to be joined by Jeff Palmer, Head of IR, from a really important semiconductor company, NXP.

Thanks, Josh.

Jeff, thanks for joining us.

Sure.

So to start, I wanted to maybe address the kind of elephant in the room or maybe the German not in the room. Your CEO, Kurt Sievers, announced he was stepping down from the role on the last earnings call, and you've tapped Rafael to take his place. Can you maybe provide some background on the timing of Kurt's departure and also what the Board sees in Rafael and any potential changes in style that we should be expecting?

Sure. So there's never a good time for a CEO to step down from a public company. Let's just be upfront about that. But I think one of the things that Kurt and the Board have done since Kurt's been the CEO is every year, they go through a pretty thorough succession planning process. And it's your classic what happens if Kurt gets hit by the bus model, right? And every year, going through that process, Rafael has kind of bubbled up as being the best skill set, most well rounded, most mature leader in the organization in case there was an issue that came up, right? So let's look why Kurt decided to leave. Kurt's been with the company since -- actually for 30 years, and he's been on the management team for 16 years. I've worked with him the whole time I've been here at NXP. It's a -- as you probably know, Josh, public company executive is a pretty grueling challenge. 12 hours a day, 7 days a week, it never stops. And Kurt has a family, and his youngest son finally graduated school. And I think he just made the personal decision that he was going to spend more time with his family. He's very clear that he was not going to show up in another public company as a CEO. He's taking some time for his family. So why is Rafael the right guy for the next chapter of NXP? So Rafael is -- if I was to compare and contrast Rafael and Kurt. Kurt is much more kind of a theoretical, round-the-corner strategic kind of guy. Rafael is a hardcore Silicon Valley executor, spent the early part of his own career at classic Broadcom, developing the WiFi product line there, came to NXP about 11 years ago and has built a number of different franchises that have been very successful. He built the secure mobile wallet business. That's in our Mobile business. He also owns -- pretty much anything that's not automotive, Rafael owns. So I think what he's going to bring to the organization in the next chapter is a much more laser focus on executing and delivering products on time. Not that we were not delivering products on time. It's that our end markets are changing, right? If you think about automotive, automotive had historically been kind of a 5-year design or new platform cycle. That's changing. And I think it needs a different way of looking at the problem set to really continue to execute the way we have.

Interesting. So I mean people think of -- when people think NXP, they think auto semis...

57% of the business.

And so it sounds like part of the strategic rationale is to bring what Rafael was doing in the industrial side to the auto business. Is that correct?

Correct. Well put, yes.

Okay. And then let's dive into the auto story as well. So on autos, I mean, we kind of view NXP as a tweener where there's not the huge high-content, high-voltage power semis. There's not the huge high-content ADAS processing, but maybe that's changing given your 5-nanometer SoC. Can you maybe talk through what are NXP's biggest content opportunities and why you benefit as vehicles electrify and digitize?

Yes. So I would take exception with the view of tweener, Josh, no offense.

None taken.

We are not in the power discrete market by choice. As you know, in 2016, we sold a MOSFET discrete business, which would have been the basis for going into that part of the market if we had decided to. We made the decision, power discretes were not our cup of tea for a number of reason. One, it's a very high CapEx-intensive business. It's an IDM functional business. It's not a foundry business. We -- at that time, in 2016, 2017, we saw the emergence of quite a few smaller start-ups in China in this space, and we thought the pricing was going to be very, very bloody over time. And we were kind of laughed at, like, no, no, that will never happen. And I think that has turned out to be true. So we're not in the power discrete business. So if you look at the remainder of our automotive business, 57% of NXP, almost 60% of our auto business and of the company is processor based. So it's not just MCUs, right? So it ranges from MPUs, application processors, crossovers and microcontrollers. And I'd say that we've had a very large share of the automotive processing market. Last year, TechInsight's estimates was about a $20 billion market. We have about just under 19% of that market, processing. So clearly, everything starts with processing. And so one of your questions you sent over to me was, what is this evolution of software-defined vehicles? So if you think about a modern car today, think of it as kind of a flat point-to-point architecture. Many different processors, microcontrollers from many different vendors, all connected with kind of a flat architecture. And that's worked very well for many, many decades, but it's pretty much not extensible at this point. So what the auto OEMs want to do, and this has been an evolution over the last number of years, is to build a more hierarchical processing fabric, starting with kind of use a classic networking type of topology, kind of top-of-rack switch, multiple domains in the car, which are functional areas of the car, pushing down into zonal areas of the car, which are more physical aggregations, and being able to push software down across that process in fabric in real time after you buy the car. Because nowadays, when you buy a car and you drive it off the lot, its best day is the day you drive it off the lot. And in the future, with software-defined vehicle, the idea the OEMs want to implement is the capability to add features as you drive the car. So we start off with processing and being the most important part of the automotive franchise, and we complement it with a wide variety of different kind of application-specific analog products.

Got it. And so it sounds like you see the market moving more -- in auto processing, moving more from MCUs to the APs, the crossover products and the MPUs. Like where are we in that evolution? Is that starting to benefit already? I mean you've talked about, I think, your 5-nanometer SoC shipping in the next year or so. How much can that benefit you and your -- on the content side?

So right now, today, 2024, last year we reported our -- what we call our SDV franchise, was about $1 billion, had it grown from about $500 million in '21. So a decent-sized growth. In '27, we think that will double again to about $2 billion. And as you know, from the automotive market, design-to-revenue cycles are fairly long. So we wouldn't stand up on a stage at an Analyst Day unless we had a pretty high degree of confidence in underpinning those design wins.

And can you maybe talk about the competitive dynamics as well? I mean you guys have been open about, I would say, 2 things that probably, on the surface level, look like headwinds. One, you were very conservative with how you manage the channel. That's an auto and nonauto comment. And two, there were some parts that were no longer meeting your margin threshold that you didn't want to go after anymore. How much of that is still impacting your numbers? And are we through both of those headwinds? And when should we see NXP's auto processing and broader auto franchise start to really inflect?

Well, I think it has inflected, Josh. I guess I would take disagreement. Growing the business from $500 million to $1 billion in 3 years is nontrivial, and growing it from $1 billion to $2 billion in 3 more years, again, nontrivial. So I think it is inflecting. I think what you have to realize is when we look at our total auto business, we have a number of different product categories where we are the #1 leader in that space. And when you're a leader in the space, you can't outgrow the market. So what we've done in communicating how to think about our auto business is about 30%, 40% of our business is what we call accelerated growth drivers. SDV being one of them, electrification being another, radar and connectivity. Those are kind of our accelerated growth drivers. They'll grow in the roughly 20-plus percent range over the next number of years. The core business where we have a very high market share. And these are things like in-vehicle infotainment, audio infotainment. Every car has a car radio, believe it or not, still to this day, still a very key piece of the business. Keyless entry, secure car access, in-vehicle networking, these are all areas in automotive where we're the #1 leader, and so we're not going to outgrow the fundamental market. We're going to grow similarly to it. So how we've forecasted the auto business to grow 8% to 12% over the next 3 years, that core piece of business growing kind of low single digits and the accelerated growth drivers growing in that 20-plus percent range.

Okay. But from an inventory standpoint, do you...

Two separate things. So if we're going to talk about inventory, how we manage the channel is a company-wide approach, not segment approach. And it was a lesson learned back in 2016, 1 quarter where we did not pay attention to how the channel was going from an inventory perspective. And beginning in early '17, we implemented a very systematic, highly controlled way of managing the channel. And we control what goes into the channel. We can see exactly what's there by part number, by geography, by distributor, and we get to see what actually gets shipped out of the channel by all those different metrics. I'd say we probably manage the channel stronger than any of our peers. When we started to see the most recent cycle start to top off and start to roll over, we made the decision at that time, it was best to keep inventory on our balance sheet to give us the maximum flexibility and to then put it into the channel where and when we thought it made sense. As the cycle really started to roll over, we decided we were going to hold -- use our balance sheet to hold that inventory. We're probably 2 weeks under our target. Our target in the channel is 11 weeks, and we're about 9 weeks is what we guided for the second quarter. So we still have a couple of weeks more to fill up. As a new cycle forms here, and we think we're kind of at the bottom of the cycle here, we will start to refill the channel.

So what -- I guess what metrics do you need to see to decide it's time to refill the channel? And I guess I'd ask a similar question on the auto side and the Tier 1 specifically, like how do you feel about Tier 1 inventory levels? Are they healthy? Is there still digestion to come on the auto side as well?

Yes. Let's take the channel first, and then we'll talk about the Tier 1s. So on the channel, as we've said in the past, some of the things we would look for to give us confidence that we're truly kind of grounding the trough of the cycle and what would be the early kind of crumbs on the table of a new cycle expanding would be, one, EDI feeds from the Tier 1s stop declining. They have been declining. We get EDI feeds on a regular basis. EDI feeds from the Tier 1s have been declining for 8 quarters as they were burning off inventory. We'll get into that in a second. They've stabilized. That's a good thing. Two, we are starting to see some shortages, some expedites on certain specific parts. Don't get excited. This is not huge dollar amounts, but it's a clear indication that some of the customers have not been managing their own forecast very well. So they come in with different shortages and things like that. That's a good thing. And through the channel, we're actually seeing the backlog of customers who buy in the channel, both automotive and other types of customers actually start to build. So those are all early signals to us that we are in that early stages of a new upcycle. We probably would have been much more optimistic or maybe over our skis on our most recent earnings call if we did not have the issue of the tariffs, luminous in the face. And because we did, we decided just there's no reason to make a call on something we don't have any idea about in the future with tariffs. So I think as we get more clarity on tariffs and the cycle clearly gets ground in, we will refill the channel.

All right. Well, thank you for bringing up the T-word.

Well, I think we'd get to it at some point today. It wouldn't be a semiconductor conference without it, right?

Yes. So I mean you and your peers, for the most part, were pretty clear that you're not seeing any evidence of pull-ins or changes in customer behaviors in a meaningful way. What data are you looking at to ensure that, again, you're not seeing pull-ins? And how are your customers behaving? And how do we -- I mean how does NXP have to change how it operates in this more uncertain environment?

Yes. I think that last question is hard to address. I think we're just -- we don't know what we don't know because the tariffs keep changing, at least the rules keep changing and the reaction. A lot of our customers are telling us the exact same thing. So there's a lot of unknowns right now in terms of the tariffs. So let's put that aside because we're not your macro mavens here today. In terms of like pull-ins, so how we would determine whether or not we are seeing pull-ins. So we get EDI feeds, which are electronic forecast from our major customers. And we get them weekly, biweekly from different customers, and we've been doing this for many, many, many years. So we have a history of how those curves look, if you will. When someone does a pull-in, it's kind of a classic. You see an inflection downward in an out quarter and a pop-up in the short term, right? There is no creation of new demand. There is only demand. You can moderate it if you want, but it is what it is. As of our earnings call a couple of weeks ago, we have not yet seen any pull-ins. That does not mean we may see some in future periods, but as of right now, we've not seen them.

And do you say that because they give -- they need to give you a multi-quarter forecast of demand, you would see it?

You would be able to at least see a disruption in the order patterns. So there's 2 feeds we get from customers. You get EDI feeds, which are kind of long-term forecast, and then actual purchase orders against those forecasts. Now could a customer gain the system? Sure. But we -- I don't think it helps anyone to do so. And while we're on it because I didn't answer your one other question about the Tier 1s. I think the Tier 1 on-hand inventory is better than it was last year. It has continued to come down. The challenge we have with the Tier 1s is we don't have a systematic way of looking into their inventory locations and actually measuring exactly what they have, different from the channel where we have very good clarity, right? And so we have to take them on their word on how much inventory they have. It had -- we measure it every month. We do see it coming down. It's probably closer to the trough than it was 6 months ago, 3 months ago. But are we -- can I claim all done? No, I would be foolish to do so.

And are there differences by geography? I mean most companies have said Europe still has work to do, North America and certainly China in better shape directionally. Is that match with what you're seeing?

I would agree with that, for us. China and Asia is managed through the channel, believe it or not. 40% of automotive does go through the channel. It's primarily for Asia, Japan, Korea customers. And so we have a very good clarity there. And I would agree with you, the Tier 1s are primarily North America and European type of situation.

Okay. On the trade friction standpoint, you guys, for a while, have been moving -- I think, partly because of the way your product portfolio has evolved, but moving to more of a fab-light -- fab-lighter model. Can you talk about some of the JVs that you have and why those are the right approaches for your manufacturing footprint? And again, does anything change? Or were you just early in reaction -- reacting to the new geopolitical landscape from a manufacturing standpoint?

So the term we use is hybrid manufacturing. Fab-light means something else to us. So -- but our view is hybrid manufacturing. So right now, today, 60% of all of our wafers, we buy from third-party foundries, TSMC, GlobalFoundries and a few others. 40% of what we build internally is all proprietary mixed signal processes. So that's kind of the split of the pie, so to speak. Customers don't buy from us, whether we manufacture the part or whether we have a foundry. They buy from us based on the IP that goes into the products and the system solutions they can build with our parts. So we're not hung up on saying we have to own our own fabs. We made the -- every couple of years, we look at the make versus buy. Do we build our own 300-millimeter fab? Or do we partner with someone else? And we just can never make it make sense to own our own completely. So we announced last June that we were entering into a joint venture with Vanguard to build a 300-millimeter factory in Singapore. It's a 55,000-wafer-per-month factory, the first phase. Second phase is another 45,000 wafers. We will be a 40% owner of that factory. So we will have near-cost access to wafers. It gives us a high degree of geopolitical stability. So it's not in the U.S. It's not in China. It's in a generally neutral area. And we think that will give us a very good leverage on our gross margin going forward, about 200 basis points of incremental gross margin when that factory is up and fully running probably in '28 -- '27, '28.

And how does -- I guess how do you feel about your need to have a China-for-China facility? What's your strategy there? Because I believe you have partnerships with TSM. And bigger picture, how do you feel about competing with China local vendors for the Chinese auto market, which continues to become more and more of the global auto mix?

Yes. So our China-for-China strategy is really more to provide capability for our Chinese customers. So right now today, about 36% of our revenue is shipped to China. About half of that is for multinationals, for reexport. The other half is what we call China-headquartered companies. Those companies have told us very explicitly: if you want to be a continued large supplier here, you need to have a supply chain that is basically walled off in China. Now we're not going to build a fab ourselves in China. So we're partnering with 3 different entities: TSMC in Nanjing for 16, 28-nanometer; SMIC, who we've had a very long relationship with, for higher than 28-nanometer type products; and HHGrace, which will be our mixed signal partner. So that will be our front end. It will all be foundry based. And then we own a large back-end packages assembly site in Tianjin. So right now, today, of that 17% of China-headquartered companies that buy from us, about 1/3 of that use this China-for-China manufacturing strategy. Over time, it will grow.

And how do you feel -- I think a lot of semis investors are understandably concerned about China serving -- China accounting for more of global demand. But I think you guys have been clear that, I mean, China autos -- China auto OEMs are good partners of yours.

We think so. We think they're highly innovative. They're not followers actually, and part of our China-for-China strategy is to listen to those OEMs because we think that they move faster, they've got some really good ideas, and we want to be partners with them. To your question about local competition, we do see some competition in China, but we bracket it to kind of 3 buckets. One, power discretes. Clearly, there are local Chinese competitors in that space, not our cup of tea. Two, we do see some local players in the emerging kind of what I'll call catalog-analog space, not really our portfolio space as well. One of our large peers down in Texas has been very aggressive with pricing to deny those startups any real beachhead. So that's their game. And we do see some consumer microcontroller players, guys who have maybe smaller portfolios, more geared to very consumer-type product cycles. Currently, today, in the China auto market, we do not compete with any local Chinese processor vendor. It's still the Western players: Renesas, Infineon, ST, to a lesser degree.

And how much of that is because of the pedigree in software that you're able to bring to auto customers? Or is it more at the hardware level where you feel like the local competitors can't keep up?

I think the days of just having a single product to sell a socket are behind us, especially for processors and cars. You need to bring development tools, enablement tools, firmware. More and more customers want you to kind of go up the stack with them because they -- what they want to do is they want to write software, right? They don't want to get into picking Part A versus Part B. They want more system-level solutions. And I think that folks don't realize this is not a -- so there's high barriers to entry into these businesses. We've invested in these areas for decades on end. So I think as long as we continue to innovate -- as one of the largest car companies in China has said to us, as long as you innovate, you will have a seat at the table and be a partner with us. You stop innovating, we will design you out. It's very capitalistic, to tell you the truth.

And on that note, maybe you could provide update on the BMS franchise. How much is that contributing today? Any changes in the competitive dynamics there and expectations for the next few years?

Yes. So our BMS franchise is about $500 million in '24. We expect it to grow about 15%, 20% over the next several years. We're the #2 vendor in that space. Our friends up at ADI are #1 after the acquisitions of Linear and Maxim and some of the homegrown technologies. We think we offer a very good solution. Our solution to our customers is more of a complete system, not just the analog components that sit on the battery cell, which resonates with many of our customers. So it will continue to grow quite nicely.

Okay. Per usual, I've overindexed to autos, but I don't want to neglect your other businesses, even though I did. Maybe you could talk through some of the other -- what are the key growth drivers that you expect within industrial, in particular, that are going to allow you to meet the growth target that you laid out at your Analyst Day in the fall?

Yes. So industrial -- let's compare industrial to automotive. Automotive is nice because it's kind of a homogenous market. We can kind of identify OEMs, right, different supply chains, nice and neat.

Easy to ask questions about.

Yes. And for me to answer the questions also. Industrial is completely different. It's tens of thousands of customers. It's a very, very long tail business. 80% of that business goes through the channel. Now the kind of best way we've been able to profile that for you and for investors is about 60% of our Industrial & IoT business is what we would call core industrial, things like factory automation, building automation, some energy management and a health care delivery business. The other 40% is consumer IoT, ranging from wearables, all the way to home automation. So that's where we make the demarcation line. There is no one customer or no one product type that makes up more than a few points of revenue in that reportable end market, which makes it tough for you. I wish I could just say to you, hey, Josh, follow white goods for every June and you have our business. But it's a very, very long tail type of business. It does all start with processors again, microcontrollers, application processors and crossovers, not yet so much MPUs there in that space. What we're seeing is a lot of the big industrial customers, especially in factory and building automation, wanting a lot of the same capabilities as auto customers, functional safety, right? You're talking about building robotic factories, right, where they want functional safety in the robots and they want that kind of capability. We're not proposing that we're going to go and take share from something like TI and ADI in the catalog-analog space of industrial. Our real foray into industrial is pushing in and providing a very complete solution of processors complementing with analog capabilities, complementing with connectivity and security. It's really -- it's a system approach.

So I guess in both industrial and auto, the market's moving more towards MPU, application processors or crossover family. That's a different competitive set than you traditionally have gone up against like you were just mentioning. How does NXP differentiate against the Qualcomms, the NVIDIAs that are the microprocessor companies that are looking more to service auto and industrial applications?

I think they're great companies. I think Qualcomm has done a great job of taking immediate processor franchise, right, Snapdragon franchise, and bringing it into different markets. I think they are a very good competitor. I think they have challenges in that they have certain requirements, profitability capabilities they want to achieve that maybe limit certain opportunities they can go after, but they are a very good competitor. NVIDIA, I have to be right upfront. I'm -- on NVIDIA, we don't compete with them. It's apples and oranges. We don't see them in a lot of the spaces we play.

Okay. Yes, on that note, I wanted to ask about gross margins. So you have a few -- I think you're currently below your target model. Maybe you have some tailwinds from 200-millimeter to 300-millimeter, utilization is increasing and also, I think, wafer cost pricing changes matching your ASPs. Can you walk through how meaningful those can be? Where can we expect that second half gross margins?

We're not going to give you any view about second half anything, Josh. So we're not going to go there. Our long-term model for gross margins is 57% to 63%. That's between '24 and '27. Revenue is our friend. So as revenue starts to reaccelerate on a building cycle, we'll be able to fill our factories, which were underutilized today. We're about 70% utilization today. We'd like to get those back up to the sweet spot of about 85%, right? That gives us a benefit. Two, we'd like to refill the channel. The channel is a margin-accretive go-to-market for us. We're 2 weeks below our target. We'd like to get that back up to 11 weeks consistently. Like any good semiconductor company, we have to give a certain amount of pricing concessions to our customers. We offset that by operational efficiency. Those 2 things are never timed perfectly. You give pricing at the beginning of the year, and you earn back the operational efficiency across the year. So those are kind of your levers to get from that 57% and that 63% range. In '27 -- after '27, '28 when VSMC, the fab in Singapore, is up and running, that will add another 200 basis points to our then gross margin. So if you assume we hit the midpoint, 60%, exiting '27, 200 basis points above that will be due to the VSMC fab.

All right. We're about to run up on time. But one last one I wanted to ask. I mean Kurt's historically made comments about his view that the semis industry still probably has more room to consolidate. And that -- a lot of that has been held up by regulatory headwinds. I'd be curious, do you believe Rafael has a similar view? And you guys have made a few tuck-in acquisitions over the last 6 months. Should we expect more of those? And how should we think about that from a capital allocation standpoint?

I think the comments you're making -- referring to for Kurt are he's referring to transformative M&A, large peer-sized companies getting together. And I think the environment just is not very conducive to that right at the moment. So we'll continue to look for smaller tuck-in acquisitions that bring us IP and people and access to different markets as we need to. No change. I don't think Rafael has a different view of that. I mean one of the things -- I should have mentioned this when you were asking about Rafael versus Kurt. They were both together in developing the NXP strategy. So much of the ideas that you've heard from Kurt, Rafael will execute as well.

Got it. All right. Well, Jeff, we are out of time, but we appreciate you joining our TMT Conference again this year. Thank you so much.

Great. Thanks, Josh. Appreciate it. Thank you, everyone.