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

Thanks for joining. I'm Blayne Curtis, semiconductor analyst at Barclays. Very happy to have for our next presentation, NXP Semiconductor, from the company, Kurt Sievers, President and CEO; as well as Jeff Palmer, SVP of Investor Relations. So welcome both.

I always like to start with a broader question. You guys just had your Analyst Day, you laid out an 8% to 12% top line CAGR. That's an increase from the last Analyst Day. So maybe just kind of a very broad -- looking at -- it has obviously been a challenging time, but you're clearly driving growth for NXP, regardless of the backdrop. So maybe just start there, where, Kurt, have you focused on driving the secular growth? And where do you see the biggest opportunities over the next 3 to 5 years?

Yes. So first of all, good morning, Blayne, and everybody. Thanks for having us this morning. Clearly, the 8% to 12% are indeed an acceleration against former growth levels. I mean, we come off a 5.5% growth over the past 3 years. So indeed, 8% to 12% is a significant raising of the bar relative to growth. Clearly, the weight of this is being carried by Automotive and Industrial. So Automotive, which is about 50% of the company, is forecasted to grow even between 9% and 14%, so actually above the company average. And Industrial and IoT, which is another 22% of company's revenue today, is also at 9% to 14%. So think about almost 3/4 of the company made up by Auto and Industrial growing 9% to 14% ahead of the company average. And I dare to say the largest part of the growth drivers in there are, really, secular in nature. So we don't think about this as a cyclical thing, which is also why it's important to clearly emphasize, we talk about growth from the basis of a very strong '21. I mean, we will end '21, if you saw our guidance, at around $11 billion revenue. So the 8% to 12% are starting from that strong base already.

It's a great lead-in for my next question and definitely want to drive in the -- digging to the drivers of that growth some more. But the biggest question I get is just, what is that starting point? You said off of this year, which is clearly a very strong year. I think some will look out and say, hey, look, the Semi group needs to see a correction. What gives you confidence that this is the base?

Well, I -- we talk about the 3-year growth, and we do have significant visibility definitely at least into the first year of the 3-year cycle, so being the calendar year in 2022. And I just don't see a correction at all, which means we are still on the extremely thin inventory levels both at distributors, but also at our customers to the extent we can see it all the way down to product, which is close to consumers, if you think about dealer inventories at car dealers in the U.S., for example. But it's also very, very close to us. We have NCNR orders, so this is this new way of doing longer-term business, where actually customers are giving us orders for the full calendar year, which can neither be rescheduled, nor canceled. And the size of those orders plan are more than we can actually commit to at the moment from a supply perspective. So it still very much looks like our revenue level next year, our revenue growth is kept by the supply capability. So I don't see this imbalance going completely away through next year. And that gives me for at least the first year, which is the line of sight we definitely have now, absolutely no line of sight of anything, which smells like a correction. It just feels like it's growth, it's normal growth, it's natural.

Let me dig into a little bit on your supply side. You've seen companies throughout this year get incremental supply back either internally or externally. I think your lead times for a lot of your products have slipped, maybe even as year long, right? So maybe just some comments on where your supply is and your plans to kind of catch up?

Yes. On the supply side, we have what we call a hybrid model, and hybrid means it is a mix between internal manufacturing and external manufacturing. And it is a clear differentiation between the front end and back-end manufacturing. So on the front-end side, so on the wafer side, we currently are already at almost 60% externally sourced, or you can also put it differently than ever we have built or we have need for a technology in 300 millimeters, it will always be external because we have absolutely zero intention to ever build a 300-millimeter factory. We have our own 200-millimeter factories fully utilized. We also see them having a long lifetime going forward. But with the move to the 300 millimeters for the new growth, obviously, more and more from the -- on the wafer side is going to be externally sourced. So that's currently 60, and there'll be more. On the test and assembly side, it's actually the reverse. We have about 75% internally. We also keep investing, so we are broadening that base to stay at least on 75%. And if you think about or ask about the bottlenecks, currently, I'd say the real bottleneck is wafers. We are investing significantly in the test and assembly. I think we can follow the wafer availability. So where the real bottleneck is on the wafer side, and there, it is then actually with foundry partners.

There's obviously a bunch of challenges. There was also for the MCU market, specifically, the weather you had to deal with, Renaissance had a fire. And I was just kind of curious more broadly just how you think strategically about that front-end capacity? Texas Instruments is all internal, said they were going to kind of be able to handle this better. And ultimately, they ended up probably the most short of anybody because they [ can't ] expand their own capacity. You see people kind of go different directions...

Well, I will not -- yes, I will not speak about TI specifically, Blayne. But indeed, I think through this year, which was a very tough stress test on this, it proves that the hybrid model we have is actually more resilient. And I say this again, it's not just against TI, but we will end the year with, I think, a 28%-ish year-on-year revenue growth. And from what I have most recently seen, if you compare that to our broad-based peers, which are comparable companies, we will be very much at the high end, and this is really about supply capability to start with. So I think we have all the evidence this year that this model is the right model in this environment. Actually, I have to stress that having the test and assembly in-house is a positive because that helped us actually to be flexible with wafers coming in or not coming in and moving left and right. So that hybrid model has, I think, won the stress test, so we definitely want to stick to it. I absolutely see no reason on the contrary. It's actually that I would say this year gives us even more reason to believe this is the right model going forward. However, what we do differently is, certainly, we are deepening the relationships with foundry partners. I mean, the -- say that the guarantee for success in the model going forward is, of course, that we have a very close long-term tie out with foundry partners. By the way, not only on supply, but also in technology. I mean, nowadays, of course, we speak all the time of our supply, but for the future, also availability of the latest process technology for certain applications and products will be equally important. So long story short, we absolutely want to stick to the model. It has proven right this year, but what is changing is the depth and the intimacy of the model, how closely we work with foundry partners.

Maybe -- I do want to talk about some of the drivers and move into auto, but maybe still finishing off on the supply part of the equation. I think the auto supply chain was the poster child for challenges. And kind of just thinking about -- are you getting signals that they're thinking about their supply chains differently? Maybe you can elaborate on that? And in terms of -- I think people -- a couple of companies through earnings have said, hey, customers are changing their behaviors, thinking about completing [indiscernible]. I think we've seen these rare pictures of lots of cars waiting for one part. Maybe just elaborate on that supply chain.

Yes. Probably the auto part is indeed the showcase here, while I have to emphasize that shortages in other segments are equally tough. It's just that auto has, for some reason, captured more headlines. Now, I think, there is a big learning going on. And the learning is, one, much more transparency all the way from the auto companies back to us, even back to foundries. And the much longer-term oriented planning, which had been buffered away in the past by the Tier 1 companies, to be honest. So we didn't have that planning clarity from the car companies. It was taken up by the Tier 1s and then we got it somehow translated to us from Tier 1s, and we will translate it somehow to the foundries. This whole circle is now being in constant transparent communication, which I think makes the supply chain more agile because there is more information available at each point of the chain. And the second part is the, I would say, the level of commitments [ plan ]. The car industry has a just-in-time model, and they used to work on forcing their suppliers to actually commit a forecast, but they would not commit to purchase. And that asymmetry is not very healthy, in my view. And this is currently being corrected, which is one of the other painful learnings out of the current situation. The third learning is more inventory. Clearly, and that's automotive-specific car companies, feel they should maybe have up to half a year of inventory. The half year from a semiconductor perspective simply comes from the fact that they understood that some of our products need half a year manufacturing cycle time, so between 3 and 6 months. So they feel like if they have that amount of stock, they would be more flexible to deal with fluctuations. Now, what is important to state, however, is nobody is able to build their stock currently. It's just not enough product, so the supply isn't sufficient to allow for that. Secondly, it is still unclear who's going to hold it. I mean it's not going to be the semiconductor companies because all of the mix, et cetera, I mean, they have to do this, that customers need to know what they want to have. But somewhere in this extended supply chain, I do believe, over time, will be sustainably higher inventory levels, which, of course, if you translate it back to us at some time in future, and I think it's more like 2023, it is an additional revenue opportunity for a one-time effect, of course. I mean, once that is filled, then we're going to be there. But again, it's all far away.

Obviously, it's been a challenge for this period on the supply side, but you have seen some unique demand patterns. One thing that's happened is just more focus on EVs. You've had maybe some EV start-ups kind of come to market, and I think people are looking at the way those cars are configuring and maybe some of the traditional OEMs are rethinking how they're going to lay out their cars. Maybe you can talk about that trend, and then I definitely want to dig into this more on the EV side.

Yes. So first of all, EVs clearly are one of the strongest secular growth drivers for the industry and with that for NXP. The trend has accelerated. I think we see this year between 16% and 20% portion of EVs of the total car production, which is more than what had ever been anticipated for this year, and that's going to grow significantly going forward. I think people speak about 60% plus in 2030. Secondly, that brings significant content increase for us. We talked about our battery management solutions. We revealed that we also do e-motor control such that for NXP, this is going to be a $0.5 billion business in a few years from now, with a very significant growth potential. So we are currently running on, I think, $200 million run rate and are going to go to this $500 million soon. So the point is that indeed, that drives closer collaboration between us and car companies. So you were asking for how is it working in the supply and value chain. Yes, I think this is the new engine for the business where the horsepower is coming from in future. So from a car company perspective, be it a new startup or be it a traditional company, they want to know about this. They want to own it, which is wonderful for us because it gives us the opportunity to do much more innovation directly with them. In that particular case, on battery management, by the way, also with battery companies. I mean, we do our design with -- both with the car companies and/or with battery companies, where then the battery company can offer a complete set including the electronics to the car company, which, possibly, in the one or the other case, may makes the life for traditional Tier 1 companies a bit harder, to be honest, because they are sandwiched in this situation.

I had a question, it might be EV-related or might be just broader in terms of -- another trend we heard is just people who were buying higher-end cars during this period, maybe they just had more cash. Can you count on that trend? And then in terms of an EV, you laid out initially at a teach-in with the BMS, it was 50 going to 100, and then now it's -- while it's electrification is a bigger subset, 200 going to 500. But I'm assuming a lot of these EVs are very high end and have lots other semi contents as well. Is there a way to kind of think about content per car increasing, and whether EV plays into [ this ]?

Yes. No, absolutely. So the EV content is doubling by going from a combustion engine car to an EV. So just think about going from $400 to $500 in a traditional ICE car to an EV going to $1,000 or $700, $800 to $1,000. The premium car thing blame is actually not a consumer choice item. It's a choice by the car companies to maximize profit in the current environment. So where they had all of these supply shortages on chips, what they did choose to do is to build the margin-richest cars they have, such that they would actually maximize their profits on a lower number of cars. And that indeed, in reverse, drove a much higher semi content because the mix of cars, which is being built this year, and I don't have a number, I mean it will come out somewhere next year when everything is being published. But there must be a significant -- significantly higher amount of premium costs this year, which is being built. And they can have 2.5x, 3x higher semiconductor [ bond ] to a volume average car. So when this volume average car, again, has $400 to $500, a premium car can have $1,500. So that is also one of the reasons, by the way, why we have such a stunningly high revenue growth rate this year in automotive, while the SAAR is flat. It really comes from this content boost both from the EVs as well as from the premium cars. Now, you mentioned another important point. Indeed, typically, EVs are more electronic rich than a combustion engine car, not only because of the electric drivetrain, but they are just designed for more tech-savvy people. It seems to be that consumers who buy EVs are more attracted by large screens and more ADAS features, et cetera. So the trend to EV is not only good for the electric drivetrain, but it definitely also pushes the overall envelope for semiconductor [ bond ], which is a double positive for NXP, if you will, because we benefit both from the electric drivetrain, but also from features like radar and stronger domain computing.

That's another good lead-in, because I was going to ask you about radar, it's another big driver within autos. You talked about maybe 2/3 of the cars having ADAS. But I think there's also the number of radar per car increasing. So I think the other thing that happened, there are lots of [ spec ] bringing companies public, a lot of emphasis on LiDAR. So you can also just touch on when you look at that front-facing, I'm assuming the first radar you get is front-facing to do some of this ADAS. Can you talk about, one, just radar versus LiDAR on that front-facing socket? And then what's driving the increase in radar for the more of the surrounding?

Yes. So radar for NXP is really big thing. I mean, we are at a $600 million run rate now. We talked about growing to $1.1 billion in 2024, and that is coming off a 40% plus market share, as of today, as we have from third-party reports. So that's a big deal. And indeed, the acceleration comes from more cars getting radar. There is still cars which have no radar, then to cars getting more radars, that's what you talked about, which is then typically, indeed, first, the front-facing, but then you get side radars and back radars, which do blind spot detection and that kind of applications. And then it also goes to higher value radar, which goes to your LiDAR question. So let me comment on those 2. So the higher value radar is what we call imaging radar, where we get from a performance perspective to the low end of LiDAR. And the way that works is simply you put several 5, 6, 7 radar front ends around the car. They are cascaded and they are merged in one microprocessor, which has good machine learning capabilities, to figure out a much more complete picture of the environment of the car as you would get it from 1 or 2 radars only. So it's really the combination of several radar front ends. That is what we call imaging radar, and obviously, it drives the higher silicon bond given the higher number of front ends. And we get now from a performance perspective, as I said, the low end of LiDAR. It's not as performance high as, say, a full-blown LiDAR, but it's still much smaller and much less costly, which -- and it will stay a debate for some years [ plain ]. But I -- I would still claim it is not a done deal that LiDAR will go ever into high volumes because it could be that the kind of radar I'm thinking about here is just good enough. Maybe not as brilliant as LiDAR, but just good enough, but much cheaper and much smaller. So that's one. The second one is the number of radars. I mean, I look it up here. What we said in the Analyst Day, a Level 2 car has, on average, 3 to 5 radars per car. I mean, that is really a big number. And that has been indeed the front-facing and 2 on the side in the front and 2 on the side in the back in a typical configuration. So there is 3 [ legs ] of growth, and we are -- I still believe we are just at the beginning of the radar curve. I mean, there is much more to come yet, which is fantastic.

We could keep going on autos, but we're going to run out of time. But I wanted to ask you on the IoT and Industrial because that business is almost up as much. It was very strong, and I think people look at this one. And similarly, they feel like, all right, there might have been a consumer cycle during COVID, a lot of people buying things that are consumer-oriented, off Amazon with our own. How do you assess the strength you've seen in industrial IoT? And then -- maybe start there and then we'll talk on the drivers.

So that one is indeed just repeat what I said in the beginning, 9% to 14% growth, so it's exactly the same speed as automotive. So really strong. The structure from a customer base is very different. It's a long-tail business. So we have a number of big OEM customers, but a large part of the business is literally thousands of small customers served through distribution. And that gives me a pretty confident way to reply to your question. The fact that a large part of our industrial business is going through distribution makes the distribution inventory level, a pretty good measurement on are we over shipping already or not. And we are still at 1.6 months of distribution inventory compared to a historic level of 2.4, 2.5. We have the situation now for a year, so it started actually in quarter 4 last year. And, Blayne, we are just not able to move it up. I mean, we would love to ship more into the channel, but it immediately sells. So we just haven't come off the -- the level of 1.6, which gives me a lot of confidence that, also, there is no correction in any line of sight. Now the reason why we do very valid Industrial is really our solution approach. You remember well, I mean, I remember us discussing it some time ago, we did acquire the connectivity assets from what used to be Marvell, which actually, in the end, completed that edge processing system solution capability around a very strong process of portfolio, which is the former Freescale legacy plus security plus analog attached and connectivity now from the former [ model ]. That capability of having a complete solution, including quite a bit of software, is actually unrivaled currently in the industry. And I think with all of these small customers I talked about, it is the major differentiator we have, which is driving that strong growth.

You -- clearly, connectivity, you have the broadest portfolio. Security is another area that you always have; IP, that other stone. You mentioned Edge AI as another trend at the Analyst Day. What does that mean for you? It means a lot of things to kind of people, but for NXP, what AI capabilities do you have? And what are the drivers...

Yes. The driver is actually is the Edge Compute Application, per se, because so far, most of the AI has been done in the cloud. So you would actually think about the first speech recognition systems. It actually went back to the cloud, was processed and came back to the Edge to the consumer, which is a problem from a latency perspective, and there's a problem from a privacy perspective because a lot of data is actually going to the cloud forth and back, which people don't want to share. So say, a low -- it's comparatively low performance to be fair, but the low-performance AI at the Edge with low-power consumption, which is the other big issue because most of these applications have to be wireless. This is what people need. And I give you simple application examples. Think about the robotic vacuum cleaner in your home or the machine which is cutting your lawn. These are little robots in a way, and they all have simple machine learning in order to learn their environment, in order to get better in where they are moving and behaving. And that is not something you want to do cloud-based, but it has to be at the Edge. Now typically, these machines are also wireless and have a power consumption limit. So there, our focus is provide machine learning capabilities for these low-power embedded solutions at the Edge, which is indeed a very different AI from a performance perspective relative to what you would have in a data center.

Well actually just maybe last question on kind of connectivity portfolio. UWB is an area where you have leadership. There's going to be an auto ramp. Mobile, it was featuring a few handsets. I think Apple obviously has their own flavor and is pushing some ecosystem, although I don't know, I haven't bought one of those tags yet. But kind of just talk about the evolution of the ecosystem and whether you're seeing this traction, [indiscernible], getting the phone, you can get the accessories. When is the real tipping point there?

The bottom line is we see it developing very nicely and actually much faster than what we've seen in the past with the mobile wallet, which was based on NFC. I mean there is a lot of similarities to what we used to do around security, RF and software in the mobile wallet on NFC, which, as you remember, it took 10 years plus, right? I mean that was endless. This one is very fast. So we have -- we have 160 or so ultra-wideband ecosystem partners by now across Mobile and Automotive and IoT. So I think next year is going to be a big year for ultra-wideband. It will be all across Android and iOS. It will see a significant amount of cars coming out. I mean, I can name 2 because they were publicly announced, which is the BMW IX and which is a Hyundai Genesis model. And they are also nice examples because the one relies on iOS at the moment, and the other one is relying on Android, so it shows that both systems are being covered, but there will be many more next year. We are actually currently engaged with 16 different car companies. And I dare to say -- I mean, I could never underwrite it, but I dare to say any car company which works on an ultra-wideband access system uses NXP, so I don't think there is anybody working not with NXP. But we see now also a significant number of IoT applications coming out. You mentioned the [ tags ]. There is the iOS one, but there is also a Samsung SmartTag, which is all in the market since middle of this year. We work with Tile, so Tile is coming out with a product based on ultra-wideband. We work with companies for door locks, et cetera. So I think this is -- I mean, I don't want to put it that way, but it feels almost like it's going better than we thought from an ecosystem penetration perspective. Very nice, and clearly, we can leverage the experience which we have gained in the mobile wallet earlier because there's a lot of similarities in how to do it.

With that, 30 minutes flew by. I appreciate the time of Kurt and Jeff. Thank you very much.

Blayne, I thank you. Thank you.

Take care.

Thanks, Blayne.