Hi. Thanks, everyone, for joining. I'm Tom O'Malley, Semi and SemiCap analyst here at Barclays. Pleasure to have Bill Betz and Jeff Palmer with NXP. Thanks for being here, guys.
Great. Thanks, Tom.
Hey, why don't I start off, Tom?
Absolutely.
We've got Bill with us today, our CFO. He's got a bit of laryngitis, so I'm going to probably take most of the questions. So I apologize upfront for that, but he'll take the hard ones, I'm sure, if he has to.
I'll chime in. As you can hear, my voice is a little not so well.
That's no problem at all. So why don't we start with just a broader topic? So a thematic for the conference here is AI. How are you intersecting that spending trend? Where are you on that journey today? And maybe talk about your portfolio and where you're focused on bringing those big amounts of dollars into kind of the NXP umbrella.
Great, so I know AI, Bill, that has just been a hot topic for a couple of years now, and it's been fun to watch. Unfortunately, we don't really participate in the data center, so that's really not our bailiwick from a portfolio perspective. Where we're really focused on in terms of AI is bringing what we term intelligence to the edge, so this is really in the broad industrial edge market as well as in the automotive market.
As you know, we just recently acquired a company called Kinara, and Kinara is a Silicon Valley startup that has developed an NPU for accelerating large language models. We've been able to take that NPU and made it with our industry-leading i.MX application processor to create some fairly interesting edge applications.
And if you think about what customers on the edge want, is they basically want an ability to run these large language models, but without having to go to the cloud to run it. So kind of in a non-wired or wireless type of environment. And so to give you an example, maybe kind of paint a picture. So we have a large industrial customer who has a pretty robust aftermarket maintenan ce business.
And so these are very heavy-duty industrial pieces of equipment, many, many years old. The manuals to operate these pieces of equipment are like you can fill a shelf. And so what happens is, how does a person in the field figure out what's going on with that piece of equipment? So we worked with this customer to develop a handheld, voice-activated, very inexpensive kind of computer.
And this computer, call it roughly a $500 computer for the customer type of a thing. Inside of it is a wireless, excuse me, a voice-activated system using our i.MX application processor and the Kinara NPU. And what a customer can do, the field technician can do, he can say, "Hey, I have model XYZ here, and I see the yellow light blinking every 10 seconds, and I don't really remember how
to fix it." And so he voice-activates that, and he gets the response back. And so you might say, "Well, that doesn't seem really interesting." But if you think about the industrial market, which is very fragmented, thousands and thousands of customers, very different opportunities. And so there's not a one-size-fits-all like there's in the data center.
I say the data center is more of a homogeneous market, a smaller number of customers building out the data centers. The industrial market is tens of thousands of customers where we already have known capabilities. We have go-to-market reach. We have a good channel for dealing with them, and these applications are not one-size-fits-all, so there's a lot of customer enablement and working with them to exactly fit a solution to their requirement, so very different than I'd say what you see in the data center market, and we're pretty excited about it, so if you think about the industrial market overall, it was about $32 billion, kind of a SAM, if you would, last year, 2024, going to about $45 billion by 2027, and so that's just our current market.
So, if you think we're able to add intelligence to the edge in that SAM, it's probably some adder. We've not got a figure for it at this point. It's still very early days. We probably won't see initial revenue from the Kinara NPU till late 2027, early 2028, so that's our play on intelligence at the edge, Tom.
So it's already intersected. You feel like you can always bring technology to the market today. But in terms of the crossover, in terms of revenue contribution over the next several years, it's kind of early days, wait and see.
Very well put, Tom. And I think the thing that's key to understand is we already have a very large footprint with our i.MX application processor in industrial. It's the most well-used apps processor in industrial applications. So we already have the customer relationships, the situational experience, and understanding.
And it's really working with them because they're watching the build-out of the data centers. They're watching what you can do with ChatGPT and other things. And they're thinking, "How can we leverage that in our businesses?" But their businesses are different than a data center.
Gotcha. All right. So that's the macro side. Let's dive in a little bit more to the company-specific side. So we've heard your comments on NXP's performance in the broader macro environment over the last couple of weeks, both in industrial and in auto. Can you talk about what you're seeing in the market today? It sounded from your recent comments like things were stable to maybe slightly improving. Maybe give us an update on how things are trending as we get to today.
Yeah. I think actually we're more optimistic on our Q3 call. We're more optimistic than we were on our Q2 call. We clearly see that this prior eight-plus quarter cyclical peak to trough has been pretty tough. We think that things have troughed probably in Q1 of this year across the business.
We start to see accelerating trends in our automotive business, and in automotive, what we were looking for is the digestion of on-hand inventory at the Western Tier 1, so this is in North America and Europe, the big Tier 1s. That makes up about 60% of our automotive business, right? And so that excess inventory had been built up during the prior cycle and during post-COVID, and it was taking longer than we had anticipated to digest it, so that really has started to become clearer as being addressed here in Q3 and into Q4.
In our industrial market, that business actually troughed Q1 of last year or the year before, I think Q1 of 2023. It's just been a slow grind improving. Now, into Q4 this year, we did guide our Industrial & IoT business up 10% sequentially, but that was on company-specific design wins that we had won a number of periods ago, and we're finally coming into production.
Gotcha. And then.
Maybe if I just add, what I would say is the metrics that we measure, they continue to improve. So customer escalations continue to increase. Our backlogs continue to improve, both NXP and our distribution backlogs. And so we're feeling more and more confident that we're slowly getting back to that normal environment that we were waiting for quite a while.
So marrying that with what you're doing with the channel, you've talked about moving from nine weeks to 11 weeks. When you got to the call last quarter, you got a lot of questions on why hasn't the channel increased. And it was kind of a shoot you either way, right? If you don't increase the channel, then the demand profile isn't good. If you do increase the channel, like no one else is seeing that, right? But you talked about a dual dynamic, right? There's the sell-in, and then there's the sell-through, right?
Correct.
Could you talk about, do you see customer inventory build through the channel? Essentially, are people pulling more from the channel than you can put in, or are you just deciding not to put more into the channel?
Well, one of the KPIs that Bill alluded to that we have a unique ability to see is we can see end customer backlog through distribution. So let's say, Tom, you were a customer buying from one of the big disties. I can see your backlog and see how you're building backlog. I can't modify it, but I can observe it.
We can get that data almost daily. If you think about these customers, kind of mid-tier industrial customers, IoT customers buying through distribution, they're not going to be building their backlog if they have excess inventory on hand. They're not going to be building backlog if they don't see their end markets improving. And so we're seeing basically backlog of end customers in distribution actually improving quarter on quarter. And actually, so that's a positive sign for us.
So our goal to your specific question about sell-in versus sell-out, if we wanted to just fill the channel and get back to 11 weeks kind of ignorantly, we could do that automatically. Just put whatever product we want in there. But we want to be much more strategic in what product we put into the channel. So our goal is to put in high-velocity sell-through products, products that are in high demand that when we put them into the channel, we have a very high confidence they're going to go out of the channel either during the quarter or very early into the next quarter. So that's our focus. I think to be really clear about this weeks of inventory in the channel, our stated goal, and I want to be real clear about this, is we want to get back to 11 weeks.
That's what our systems are built around. Anything below 11 weeks, actually, we have to override our ERP systems internally. And so when Rafael gave the guidance for Q4, he made the statement, "We'll fluctuate between 9 and 10 weeks going into the fourth quarter." And the reason for that statement of fluctuation is if you put a product in and it sells out, you could still be at nine weeks, even though you're putting more product in.
You could put products in and it tips over and no, it's 10 weeks. But I think the key thing here is don't get too hung up about how many weeks. We want to get back to 11 weeks. We've managed the channel, I think, better than most companies. We have a very systematic methodology for managing the channel. We see what goes into the channel on a daily basis.
We see what's in the channel every day, and we see what sells out of the channel every day. And I think we're unique in that aspect.
So as sell-siders do often, you'll take some information you give us, and we'll try to extrapolate it into the next quarter, right? So on the call as well, you offered up normal seasonality, kind of down high single digits into Q1. And that wasn't guidance. You were just offering up, "This is what normal seasonality is." We marry that with the idea that you're moving from nine weeks to 11 weeks, and we think, "Oh, well, isn't the natural reaction here that they're doing a bit better than that because the channel helps them out?" Could you talk about seasonal trends into Q1, why this year may be a little different than other years, or you can just leave it at what you're at?
Well, so I'll be the bad IR guy here for a second. So my view is we only should guide one quarter at a time. But at any points of discontinuity, either when a cycle peaks or when a cycle troughs, you want to give a little bit more comfort to the sell side and to the owners of the stock. And so what we decided to do when asked about, "Hey, tell us about Q1," our best advice was to think of it as kind of historically seasonal trends into Q1 from Q4, right? Normally, you see our mobile business does tend to trough in the first half of every year and accelerate in the second half. You do see automotive trends in China be a little bit weaker in the first quarter after pushing to strong year-ends.
And so those are just some of the things we wanted to give. We're not going to go into specifics about each segment or margins or anything like that, but we just want to give you something that you can model.
Perfect. So why don't we talk about the things that you talk about more generally in the long-term guidance that you guys kind of give out? So you've talked about more and more of your business moving to higher value solutions with your customers. As we move into 2026, could you talk about certain areas that you're really excited about and where you think you may see some growth drivers, particularly in the auto space? We've talked as the year has gone along about areas where there may be a little bit faster growth than the general car market. So why don't we spend a little time?
Yeah. So last year in November 2024, we had our analyst day. And then during that event, we outlined a number of accelerated growth drivers in both automotive and industrial & IoT. With automotive being over 55% of the revenue, we know that's a real focus. So we highlighted four very specific accelerated growth drivers in automotive. The first being our software-defined vehicle efforts.
This includes our S32 products, our automotive Ethernet products, some of our software enablement, things like that. That business was $1 billion in 2024, up from $500 million in 2021. We expect that business to grow to be about $2 billion by 2027. It's a global business. Meaning it's not just tied to one OEM in one geography. It's across the world. Every major OEM globally has a software-defined vehicle architecture on their roadmap, but they're at different places, if you will.
So we're very excited about that. The second area would be our 77 gigahertz radar. So that's our play in the ADAS domain, if you would. That business was just under $900 million in 2024, up from about $600 million in 2021. And we expect it to grow to about $1.3 billion or so out in 2027. We're a market leader in 77 gigahertz radar. It continues to be an evolving market.
And kind of the algorithm we explain to investors is think about every year there's more cars with radar. Every year there are more radar nodes per car. And every year customers are driving us to add more functionality to each node, which we get paid for. So that's kind of the multiplicative model that we advise you to think about in the growth of that business. The third area of accelerated growth drivers in automotive is electrification.
This is really our play on battery management systems, high-voltage gate drivers for xEVs. That business was about $500 million in 2021, I mean 2024, excuse me. And it was up from about $250 million in 2021. So a nice growth in the last three years. We think that'll be just under $900 million in 2027. That's the only piece of business that we have that is exclusively just tied to the powertrain of EVs.
We do many, many other functions in EVs and ICE vehicles, but we do get asked about what is your exposure just to EVs. And then lastly, we introduced a new accelerated growth driver called connectivity. And what this really is, is smart car access using Ultra-Wideband technology, as well as in-cabin Wi-Fi and Bluetooth Low Energy type applications. That was about $400 million in 2024, up from $100 million in 2021.
And we think it will grow to about $700 million in 2027. So a nice curve of growth. Now, when we gave these outlooks for growth for our automotive business, which we said would be 8%-12% 2024-2027, we knew that 2025 was going to be a challenging year. And it has turned out to be, maybe a little more so in terms of inventory digestion. So it does mean in that that we will have to grow above trend line in 2026 and 2027 to hit the target.
Okay. And then maybe talking on the industrial side, you just laid out the auto drivers. I think industrial is a little more difficult just because it's just much more broad facing. Maybe talk about your differentiation there and also products you're excited about for the next couple of years.
Yeah. So in industrial IoT, the kind of the tip of our spear is really our processing portfolio. So in almost all of our businesses, about half of the revenue is tied to processors. In industrial, it's driven by our i.MX application processors, our RT crossovers, and then our general purpose microco ntrollers.
It's not as clear-cut as the auto business where I can say there's this function and that function. As you said, Tom, it's a very diversified business. But we've seen that there's probably a little over $500 million of accelerated growth drivers, new application processors, new connectivity, new security products that are kind of doing very well. And we think that should grow in a very reasonable way through the next couple of years.
So a question that I get a lot, and I think one that is yet to be fully answered in the stock, is how have you guys managed to do a bit better in terms of your auto business over the cycle, at least from a channel perspective? And if you plot out long-term growth trajectories, it does look like you guys have hung in a little bit better than peers. People ask, "Is that channel management?
Are you going to see some sort of cliff over the next several quarters?" I feel like you guys field that question well as well. So I guess looking into the out year, I think that the general outlook for auto is modest to maybe slightly better than modest growth, right? So nothing heroic.
When you guys look at your business and how that compares to the broader SAR environment, what are your kind of initial takes? Why would you be better than that? And why has your strategy that you've kind of deployed so far, which has looked pretty good, going to work again here in this out year?
Okay, so let me take them in some reverse order there. So first off, while we have to focus on global production SAR, that's not really what drives the business. What drives our business is content per vehicle. So you're right. Probably SAR in the next couple of years is going to be low single digits.
We think against that, our content per vehicle on average will be mid to high single digits, at or on top of SAR. So it's not. We get asked a lot of times, "Well, if SAR is up X, why is your business up better than that?" Because it's not just tied to units. It's content per vehicle.
In terms of your first comment about the channel, I think we have a long history with how we manage the channel. This comes from lessons learned back in 2016.
We put in place a very highly automated system in the channel where we can measure what goes into the channel, what is actually in the channel, and what sells out of the channel on a geo by geo basis, disty by disty, product by product. So it's a massive amount of data. We get that data daily. And as a management team, we analyze it on a weekly basis. I think the way we managed the channel through COVID into the peak of this prior cycle showed our ability to have a very steady hand in how the channel's managed. And so as we started to see signals that the top was kind of happening in the prior cycle, we decided to pull the reins back on the channel, not put as much inventory into the channel as some of our peers were doing, right?
Because you recognize revenue on selling into the channel. So we held back. So our peak was probably less than some of our peers, but it allowed us to very, very steadily manage the channel that started to roll over into the trough. Now we're going to manage from the trough back up to the next peak in a very similar strategic, very steady manner.
So two, I would say, non-NXP specific items that could come up in 2026 are, one, China, which people ask about a lot. How much does China impact your business? Do you have domestic supply coming online? Do you get concerned about that? And then the second, which is talked about a lot, is memory, which is a newer one where you're seeing global memory shortages. Auto and industrial tend to be much more specific, so it doesn't impact as much, but I would love to get your opinions kind of on both of those.
Yeah. So, Tom, nothing to update on the memory comment, right? I mean, it's something we observe. We're not seeing the shortage of memory or the pricing of memory affect our order book or our expectations. So, really can't comment on that. In terms of China, China is a very important market for us.
It's about 39% of our revenue last quarter. The way we like to describe that business, and that's on a sell-to basis. So, about half of our China business is what we would term multinationals for re-export out of China. So, they're manufacturing in China, export around the world. The other half of the business is for China-headquartered companies. I would say those China-headquartered companies are starting to be much more sensitive about having a segregated supply chain inside of China.
And a couple of years ago, we started to put in place the capability to offer that to our customers. So on the front end, that includes working with three front-end fab partners, TSMC in Nanjing for 16 and 28 nanometer, SMIC for any other bulk CMOS above 28, and then HH Grace for mixed- signal. On the back end, package test assembly, we have one of our largest package test assembly sites in Tianjin.
So we have a completely segregated supply flow in China if our customers there want it. Of that 17%-18% of our revenue that is China-headquartered companies, about a third use that segregated supply chain. It's not something we're forcing them to do. It's an option for them.
Okay, so 6% of your business gives or swinging from there.
Fairly close.
Gotcha. Okay. So we went into auto. We talked about China. How about industrial & IoT? I asked after China because you do have a bit of a unique exposure in your industrial IoT business. It's not as much broad facing. You have a higher concentration of IoT than I'd say most. Could you talk into next year? I look at the PMI data. It still is contractionary. Is there a way that you can offset maybe dampened global trends into next year because of some higher growth areas in that IoT business in particular?
So our industrial IoT business may be a little different than what you remember. So about 40% of our industrial IoT business is what we call consumer IoT. 60% is actually industrial. So it's actually more industrial facing. Even with that being said, it is a super long-tail business, tens of thousands of customers. Our product portfolio is very processor-oriented.
And what's going to give us confidence that we can achieve our long-term growth are actually company-specific design wins. Again, similar to SAR, PMI is important. But if all I say to you, Tom, is, "Oh, my business is just going to flop with the wind with PMI," how interesting is that, right? We have to make our own luck, if you will.
Gotcha. Maybe pivoting a little bit to the operating model. So you laid out, I think, a very clear gross margin trajectory. Could you remind us what you said about the revenue dollars to get to the target model and then the incremental dollars that come in? What does that mean for gross margins and that progression?
Yeah. So our model is 57%-63% non-GAAP gross margin. The rule of thumb that we've given is for every $1 billion of incremental revenue on an annual basis, we should see about 100 basis points of gross margin improvement.
And so if you think just swag by 2027, we hit $15 billion, that means we should be at 60% gross margin. Now, beyond 2027, we have levers to take that up into the 60s plus range. The most obvious one of those, one we've talked the most about, is our joint venture that we've invested in with Vanguard in Singapore. That's called VSMC. When that fab, it's a 300 millimeter mixed-signal factory in Singapore. We basically will be a 40% equity owner of that factory. When that factory is up and fully loaded in 2028, that will add about 200 basis points to our then gross margin.
So if you just assume our NXP exits 2027 at 60%, exiting 2028, we should be at about 62. And then there are other levers that also can get layered on top of it. As you know, you've probably heard Bill talk about rationalizing our internal factory footprint. We have three 8-inch mixed-signal factories, two in the U.S. and one in the Netherlands. These are 35-plus-year-old factories. They're close to the end of their useful life. And so we have kicked off a process to start to rationalize and decommission those factories over the next several years.
Helpful. Why don't we pivot to capital allocation? So you talked about the acquisitions that you just made. Clearly, times right now are a bit turbulent in just the broader macro. You're saying things are improving off the bottom. How do you feel about stock buybacks? How do you feel about capital returns? Maybe remind us on your strategy and where you would direct capital from.
Yep. So our capital return strategy is fairly simple. We're going to aim to return 100% of all excess free cash flow to our owners through a combination of dividends and buybacks. We kind of view a dividend as like debt. So there's not a lot to talk about there. It's going to be consistent.
As the environment improves, we may consider increasing the dividend. So on the buybacks, so long as our net debt to trailing 12-month adjusted EBITDA leverage is at two times or below, we'll be in the market buying the stock.
Very simple.
It is very simple.
Yes. In terms of future acquisitions as well, I think you've done a good job of kind of both addressing the software needs as well as the hardware needs. If you look at that portfolio today, I think you're probably always looking for ways to get better. But where do you see an area in which you could probably tack on future acquisitions? Or is there a certain area in which you would wish you could do an acquisition, but you can't? How about M&A in general?
Every year, we review the portfolio in excruciating detail. We tend to look at our portfolio along two axes: growth, growth potential, and relative market share along the Y axis. We try to fit in our different products into that kind of four-box quadrant of high growth, high RMS, high growth, lower RMS, so on and so forth.
I would say the three M&A deals that we did this year, they kind of fall into that high growth potential, low relative market share, early days. You're seeding investments for the future. Our accelerated growth drivers that we talked about earlier, I would put those in the high growth, high relative market share quadrant. Then the other kind of two quadrants of lower growth and a variation on RMS is kind of our core business. Every year, we look at our portfolio and say, "What fits?
Where do we want to put more R&D dollars into? What doesn't fit? What do we want to de-emphasize?" You saw us announce earlier this summer that we were going to sell an automotive sensor business to ST Micro. And the reason we made that decision, it's a very nice business. It's a little bit below the growth rate we want to achieve and target. It's a little bit below our gross margin targets, but it doesn't make it a bad business.
So that kind of falls into that lower growth, lower relative market share. And so we said, "Look, this is going to fit better in someone else's portfolio." We got a very, it was a $300 million-a-year business. We got $950 million. Not too bad. And that allowed us to actually fund those three other acquisitions we did.
And you've heard Bill talk about in the past about OpEx and making room for fitting in acquisitions. The idea is we have a pretty strong model for how to think about OpEx. 23% of revenues are OpEx model. That's our guidelines, if you will. We didn't think it was right for us to buy three companies, pour it into the OpEx expense without taking some actions on the rest of the company. That's how we're always thinking about M&A. If we're going to buy something, it's got to fit inside the model.
Perfect. Well, we've covered a lot. Really appreciate you both being here and enjoy the next couple of days.
Great. Thanks, Tom. Appreciate it. Thank you everybody.