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

All right. Good morning, everyone, and welcome to the semiconductor room at TD Cowen's 52nd Annual TMT Conference. Very pleased to be joined by Jeff Palmer, Head of IR from NXP Semiconductor. Jeff, thanks for joining us.

Sure. Thanks, Josh.

To start -- by the way, this is meant to be interactive. So if you have questions, please do wave your hands or throw something at me and we'll get a mic handed out to you to ask a question for the live transcript. But Jeff, to start, I mean, from the outside looking in it's looked like NXP has handled this cycle with more stability relative to both NXP's past and many of your peers. Can you walk through some of this actionable steps that management took to bring us what looked like visibility in a [ calm ] through a very violent storm?

Yes. Sure, Josh. I think it really breaks down into 2 things. One, we attempted to try to land or create a soft landing for our investors, for our company, right? We didn't want to have -- go over our skis and then have a big step down. So probably around mid-2022, we started to see concerning signs coming out of China. And so we made the decision at that point when we could have kept refilled our channel back 2.5 months, we kept our channel pretty lean at about 1.6 months. So as you know, half of our business transacts through the global distribution channel. So that's half the business we put a governor on so to speak. And we maintain that low channel inventory right up until our most recent earnings call, where we did indicate that we're getting a little bit more optimistic about the second half, and we will start to begin to refill the channel in a very methodical way. So our 2Q guidance contemplates raising the channel inventory from 1.6 months to 1.7. Not a lot, just kind of taking it very slow and methodically. So that's one side. The other side -- on the other half of the business, which is Direct. In Q2 '23, we started to get some feedback from our direct customers, primarily large global auto Tier 1s that they admitted they said, "Look, we know we have a contract with you for NCNR material. If you force this to take it, which is legally viable, we're just going to put that material on the shelf." And we said, "Okay." That doesn't help anybody. So we work with those customers to find a business accommodation to allow them to digest and burn off some of that on-hand inventory. And those 2 things, I think, are what allowing us to see a more soft landing here in '24 for some of our peers. Without patting ourselves on the back, if you really look at the data in 3 of our 4 segments, we saw a roughly 30% peak to trough declines. We just saw it earlier.

Got it. And then I guess, we stop there and start looking forward. What's giving you confidence? You've talked about growth in the back half of the year, half-over-half? Why is now the right time to refill the channel? What sort of signals are you seeing that's allowing you to talk to sort of a roughly flat 2024 in aggregate off of the soft first half?

It's directly from talking to our direct customers. So that kind of addresses kind of that half of the business. They are starting to go through their digestion process. We also have some very company-specific ramps that are taking place in the second half of the year. So that gives us partial clarity there. And then on the other side, on our distribution, primarily in China, we're seeing better trends, better velocity of sellout through the channel. And so we will start to refill the channel. So I think of our optimism in the second half, again being partially direct and partially channel.

Got it. And on refilling the channel. I think you've talked to getting closer to the 2.5 months normal, but not necessarily all the way there. Is that the right way to think about how you slowly build the back half of the year? And then maybe could you provide some color on the company-specific ramps in the back half? Are those primarily auto? Are they mobile? Is there a particular end vertical you're most excited about?

Sure. So in terms of the company-specific drivers, some of them are auto. We have some radar programs. And in automotive, you win a design win, you babysit it for 2 or 3 years and then you go into production. As you get close to that start-up production date, the amount of conversations with customers go up and becomes -- put some real meat to the ramp. That's what's starting to happen now. So on radar as well as some other automotive products. And then secondarily, in our Comm Infra group, we have our secure cards business, and we did have a lawsuit that was settled with Impinj. That lawsuit being settled kind of took a little bit of the cloud off of that market. And so customers are coming back in looking to start programs. So it's partially automotive, partially RFID type of tagging programs. And then on your first question about the rate of filling up the channel. So I don't think anyone should expect us to exit the year anywhere near 2.5 months. I think we're going to take this very slow and methodically. Think about 1.7 in Q2, if it sells out at a positive rate, we may raise in Q3 and raise in Q4, but we will not get anywhere near 2.5 months in 2024.

A couple more on the cycle and then I promise I'll leave it alone. Embedded in the back half growth. I think you were at 144 days of inventory on the balance sheet. Your previous target was around 105. I don't know if that's still relevant anymore.

It's about right.

Are you expecting to inventory to the back half of the year? And how should we think about the corresponding move in utilization rates that you would expect in the back half given what you've sort of guided to already?

Right. Good question. So initially, we will refill the channel from on hand inventory. We're maintaining the inventory in [ DIO form ] on our balance sheet at a lower dollar value allows us to turn it through the back end fairly quickly, right? So that will be the first source of refilling the channel. At a certain point, our ERP systems kick in and say, "Hey, you've got your inventory, your DIO down to a certain level. We need to start wafers." And so our cycle time for wafers can be anywhere from 3 to 6 months. And so sometime in the second half of the year, all things being equal being positive, we will start to see utilization start to tick up.

Got it.

But I wouldn't expect that to be significant in the second half of '24. That probably affected that would probably be more into '25.

Okay. And then sort of putting the cycle aside, last year auto -- this year and last year, auto grew roughly in line with SAAR despite you know what I think most of us agree is a pretty healthy semiconductor content growth underneath. As we get to normalization of the channel and inventory. When do you think is the point when NXP can get to the levels where we start to regrow above the SAAR? Is that sort of in the back half of the year into next year? Is that just a function of when inventory gets cleaned up, what you're expecting in the second quarter of this year?

I think it's a combination of, first, let's help our customers burn off their on-hand inventory. Let's first just ship to end demand. And so I think if we start to see a more optimistic view in global auto production in '25, you'll start to see more of the secular trends kick in then.

Got it. Okay. And moving on, I mean, we often describe NXP as sort of a tweener. Like it doesn't have high-voltage power discretes that allows you to do some spreadsheet math on content per inverter. It's not an ADAS processing pure play necessarily like a Mobileye or what NVIDIA, Qualcomm and Marvell are trying to do. So it's often harder to both quantify but also explain NXP's content story. Maybe you could spend a couple of minutes in particular, talking about where is NXP's most excited about from a content and secular standpoint? And also what are the drivers of the software-defined car that are pulling in NXP's content? And how do you benefit from that?

Great. Big question, Josh. So I think first off, when you see we're kind of between, I'd say what that question doesn't highlight we're kind of the infrastructure of the car or everywhere in the car. Yes, you're right. We don't do power discretes, and that's by choice. We find power discrete market to be CapEx intensive, lower gross margin, more than likely a price sensitive market, not something we want to participate in. So what do we do? In our audit -- end market, over half the revenue there is processors from microcontrollers all the way to MPUs. The other half of the revenue is a combination of analog and RF type of devices. So I mean that's the kind of things we do. What we're most excited by and you alluded to is this evolution of the software-defined vehicle. And what we're working with direct auto OEMs, not the Tier 1s, the auto OEMs want to have a system that allows for a more of a longer life cycle of the car. So when you drive a car off the lot today, it's pretty much stale the day you drive it off, right? I mean it is what it is, right? The idea of the OEMs is why don't we build a vehicle that has a more software life cycle type of adder, whereas you drive it off the lot, maybe in a year's time, they can download features, change different functionality of the car, add different learnings that they've created for listening, taking data from the car. And so how do we help with that? So we've introduced a family that's called the S32 CoreRide family. It's a family of NPUs ranging from 5 nanometers all the way up to 50-, 55-nanometer type microcontroller or microprocessors. The idea of the software-defined vehicle is to build a hierarchy of processing fabric, starting at the very top where you might have a vehicle computer. Then the vehicle computer will manage multiple domain processors, those domain processors are 16-nanometer products. They're in shipment today. Below that domain processor level, our zonal processors, which are more physical aggregations in the car. Those are 28-, 40-nanometer type products. And those zonal processors would manage all the edge and controllers around the car that don't need to be integrated together. Our value add is our family, the S32 family, is software compatible with that 5-nanometer part all the way down to the edge nodes.

A couple of follow-ups on that. First, I think originally, there was a perception that we -- myself included shared that software-defined car and this rearchitecture of the vehicle was going to be inherently tied to electrification and EVs.

Not true.

That was my question.

It's not true.

Thanks for getting ahead of it..

Think of the software-defined vehicle. It's maybe not the thing that you as a consumer will see when you buy the car. It's the infrastructure processing of the car, right. The plumbing, the processing plumbing. And it's not just processors, it's processors, PMIC, high-speed interfaces, ranging from Can, LIN, FlexRay up to Ethernet and to be able to distribute this data throughout the car.

And could you walk through what's the overall content expansion on a software-defined vehicle for NXP, for instance, what are the ASPs you're looking at for these -- the 5-nanometer SoC the 16-nanometer ones? And is there any level of cannibalization where you have more complex processors sucking in content from edge microcontrollers? Or that's...

There's probably some probably relatively low. So from a content perspective, that vehicle computer in most applications has 2 processors for redundancy. Each one is about a $50 adder something. The domain processors can range from roughly $20, $25, something in that range. So depending on how many of the OEM implements. You can do the math, I think, on that, Josh. And then on the zonal, you're talking kind of mid-teens.

Sorry, go ahead.

And then on the edge microcontrollers, they can range by box.

And how has the competitive environment changed as a software-defined car has become more important to your end OEM customers? And it used to be probably what, 6 or 7 auto MCU companies, but not all of them have gone below 28-nanometer for auto-grade microcontrollers. And certainly, I think you're the only 1 actually maybe 2 that have gone down to 5-nanometer. Why do you feel the need to go to that level of compute horsepower? And what's the competitive environment like in this new vehicle architecture?

In the S32 family, specifically, the only vendor that we see who technologically could do what we are doing if they decided, we think is something like a Renesas. They have a lot of deep understanding of the processing architecture of the car. We -- Infineon is primarily looking at it being a zonal player. ST coming, I think, looking in the same way. I think they both said they're not doing any type of MPU development. This is not a -- the software-defined vehicle is not somewhere we would see a Microchip or TI, right?

And do you view the 5-nanometer MPU as beachhead that sucks in content, not just for your microcontroller franchise, but also other applications like Radar, BMS? And I guess using that as a bridge, how would you rank order your excitement and the growth pipeline for those other nonprocessing centric applications in the vehicle?

Right. So the radar business is the largest of the 3 you highlighted. When we did our Analyst Day in 2021, we highlighted 3 major accelerated growth drivers in the auto segment. Radar which at that time in 2021 was about $600 million, and we anticipated it to grow to about $1.1 billion in '24. We're probably close to that should value if you would. We took a little bit of a step back in '23 with the inventory digestion. So -- but I think, secularly, that's still the largest current accelerated growth driver. The electrification was about a $200 million baseline in '21. It should have hit about $400 million exiting '24. We're far ahead of that at this point. And then on the S32 family, early production of that started at about $300 million in '21, and we expected it to be about $600 million exiting in '24. We are ahead of that as well. In terms of -- what we're most excited about, we think the software-defined vehicle in the S32 family has the largest, longest runway of growth over a longer period of time.