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

All right. Welcome back. I'm Joe Moore, Morgan Stanley Semiconductor Research. Happy to have today from NXP, Jeff Palmer, SVP of IR. Jeff, welcome.

Good morning.

So maybe we could just start with recapping some of the recent Analyst Day in November, you sort of talked through a financial model through 2027. Can you give us an overview of your targets? And how you sort of thought through the guidance process in a time of low visibility?

Sure. Thanks, Joe. I appreciate to be here. So we're just talking -- I think this is probably at least the tenth year I've been here with you. So it's been a long time. So yes, we held our Analyst Day in November of last year. As you know, we hold our Analyst Days every 3 years. The 3-year window is completely artificial for Wall Street. Normally, our business tends to take a 5- to 10-year window when we make investment decisions. But for the '24 to '27 horizon model, we laid it out as follows. So baseline, we exited '24 at about $12.6 billion. We basically said that over the next 3 years, we should grow 6% to 10% CAGR. That's a 3-year CAGR. We have 4 end markets: automotive, industrial, mobile and comm infra. The Automotive and Industrial businesses are our true strategic North Star. That makes up about 75% of our business. We expect Automotive to grow in the 8% to 12% range. We expect Industrial & IoT to grow in the same range 8% to 12%. Mobile, we're a niche player in mobile. We only expect our business to grow 0% to 4%. We really kind of do one thing in mobile, that's secure mobile wallets. So we'll tend to grow more with the market. And our comment from our Communications Infrastructure group is a bit of a catch-all. There's pluses and minuses going on under the cover, but we expect that to be kind of flattish growth over the next several years. So that's at the top level. I think what's more interesting is a number of years ago, we instituted this idea of core versus accelerated growth on a per end market basis. And the reason we did that is we have many businesses where we have very high relative market share and effectively, we'll grow with the markets. And so we thought it'd be better to really kind of laser focus on those things that will grow in excess of the market. So when we look inside the total business, about 1/3 of that $12.6 billion is this accelerated growth drivers. The rest is our core business. We expect the accelerated growth drivers in aggregate to grow about 20% CAGR over the next 3 years and the core business to grow about 3%. So kind of semi rate growth. We can go into the different drivers inside of each of the segments, if you like, but continuing down at the kind of the total company model, we expect on a gross margin basis to drive gross margins into the 57% to 63% range in the '24 to '27 period. I think what's more exciting is we think that we can also put up about another 400 basis points of gross margin expansion in the '27 to '30 time frame. I'm sure we'll talk about the levers for that. Our model from an OpEx perspective doesn't change, 16% R&D, 7% SG&A. And we'd really advise investors the first thing to do is, you decide what you think the growth rate will be and then you use those percentages for OpEx. I think that's kind of off the top...

Yes. No, that's great. I guess it's always tricky to me to give 3-year growth forecast because of the starting point is cyclically so difficult. And I feel like you have a pretty good chance of outperforming when you do it at the low point in the cycle and [ underperforming it back ]. But how do you guys think about -- I know Kurt is very thoughtful and quantitative and kind of straightforward, setting targets that give the sort of aspirational things for the organization to achieve versus targets that you feel comfortable you can beat for Wall Street?

Right. So aspirational targets are what we drive the team internally, too, we're not going to share with Wall Street. And so we try to balance the numbers and the goals that we set for Wall Street to be within a realistic realm without being in a sandbag. And I think the thing to remember for most of our businesses, especially the automotive and industrial businesses, these are long design revenue cycles. So from the time we actually start to win a design till we see our first dollar revenue, it's averaged 2 to 3 years in automotive. And for industrial, it can actually be 2 to 5 years. So it's a long gestation cycle. And then the life cycle after that, for automotive is 5 to 7 years and industrial can be up to a decade. So when we go into the Analyst Days every 3 years with you folks, we have a fairly good view of what we've won, when we expect those design awards to go into production. The wild card for us, like for most of our peers, is the macro. How do -- as you said, if you decide to do an Analyst Day at the trough of the cycle, your chances of it outperforming are pretty high. We've done our Analyst Days at the peak before and it makes it kind of a challenge. So what's really -- we understand what we've been awarded. If we assume that all the design wins that we have been awarded to go to production as we expect, then the only real variable is consumer reaction to those platforms and things like that.

Great. And then on the gross margin side, I mean, you've really been the standout at protecting gross margin on the downside as we've gone into this period of turbulence. You talked about being in that 60% range. You've outperformed those targets in the past. You've actually [ rallied ] -- on gross margin, you've outperformed more than on top line. Can you just kind of talk to the general puts and takes there? And how has that held up so well? And is there a vulnerability that if this downturn ends up being extended that it goes lower?

So for those of you folks who have followed us for many years, you'll probably remember, we've really changed our cost structure over the last decade. I've been with the company now almost 15 years. When I started, our fixed cost 15 years ago, were probably about 70%, 30% variable. We have flipped that on its head. So today, our variable costs are 70%, 30% fixed. And so that gives us a really a good lever as revenue comes into the P&L. And our kind of rule of thumb that we've used with people in the past is the company is structured such that about $10 billion of revenue, we can throw off 55% gross margin. For every $1 billion of revenue on top of that, we threw off approximately 100 basis points of gross margin, all else being equal, mix and channel and what have you. So right now, in this current environment, near the trough, I don't want to say we're at the trough yet, but pretty gosh darn close. I think the levers to drive the gross margin into that 57% to 63% range are better revenue, refilling our channel which is accretive to us, increasing our utilization internally and then operational efficiencies. It's a pretty straightforward block and tackling approach over the next couple of years, but it is dependent on revenue. Revenue is our friend.

Great. And I think you're the only company that actually gives us like-for-like pricing changes from year to year, which is quite helpful to understand the role of pricing in terms of the gross margin forecast.

So over the last couple of years during the supply crunch, we did see inflationary input costs from our suppliers, which we had to pass on to our customers, which is an uncomfortable situation and abnormal for our industry. But I think as we exited '24, pricing had kind of neutralized. Our input costs were kind of flattish. Our price concessions to our customers were kind of flattish in '24. We think as we entered into '25, we're in a more normalized environment from a pricing perspective. We've already completed our price negotiations with our customers. We expect low single-digit concessions this year in '25. And what thing you have to remember is the way pricing and operational efficiency works is they're -- they are kind of phase shifted. So pricing takes effect January 1. So you do get a little bit of a hit on the front end. And then our operations teams work through the year to effectively claw back that consensus to our customers. So we think from now through probably at least '27, all else being equal, we are probably in a normalized price environment. We expect low single-digit price concessions over the next several years.

Okay. Helpful. So maybe looking at some of the segments, as you mentioned, starting with automotive, always top of mind with you guys, 8% to 12% CAGR for -- through '27, pretty healthy number. And you've talked about some of the product cycle drivers of that. Can you just talk to what gives you the confidence in that? And how does that break down?

Yes. So what I'm going to focus on is really the accelerated growth drivers inside of Auto. The accelerated growth drivers in aggregate are about 40% of the total. The things to look at first are what we call software-defined vehicles. It's our S32 MPU family and the associated products that go along with the system. That business exited '24 at about $1 billion level. We expect it to grow about 20% to 30%. This is probably one of the most exciting areas we've been focused on for many, many years. It's the whole change in the architecture of the automobile from a flat point-to-point architecture to more of a hierarchical processing fabric. This will allow our customers to partition the car into domains and be able to upgrade software over the life cycle of the car. So today, if we were to go buy a car, the very best day of that car's performance is the day you drive it off the lot. In the future with a software-defined vehicle, the OEMs will be able to upgrade and add features to the car over the life cycle of the car. Very much a Tesla-type model. So that's about $1 billion, growing 20% to 30%. The second accelerated growth driver is our radar business. It's a 77 gigahertz radar that we've figured out how to build in bulk CMOS, where in the past, many of our peers built radar devices in silicon germanium and then a processor in CMOS. And so they're always discrete parts. NXP, a number of years ago, figured out how to build the transceiver in bulk CMOS. So we're able to integrate the transceiver into the processor itself. So it allows us to get a very small die size, cost efficient, which really spurs the adoption of radar. In radar, the business is probably just under $900 million in '24. We think it will grow about 15% to 20% CAGR over the next several years. The way to think about radar, this is really kind of a 3-way multiplicative algorithm for the adoption of radar. Every year, more cars with radar, every year more nodes per car and every year more content per node. So that's kind of a multiplicative flywheel, if you will. The third growth driver is our electrification business, and we really don't -- we don't do any power discretes, like many of our peers, which is a good business for them. But we focus on the control side of electric powertrains. So our battery management business was about $500 million exiting '24. We think that will grow kind of in that 10% to 15% range, if I'm not mistaken. And it's a good business. It's -- we're #2 behind ADI and we'll continue to grow well in that marketplace. And then we introduced a fourth growth driver this year, which is what we call automotive connectivity. And this is things like Ethernet switches, Ethernet PHYs, ultra-wideband, Bluetooth WiFi in the vehicle. That probably started off in '24, at about a $400 million business.

It's really interesting. I think when companies have this kind of portfolio approach where they tell us they're transparent about the growth drivers. The first thing I do is look back 3 years and say, okay, what changed, what dropped out. And it's the same thing. It's the same thing. You're still investing in the same areas still consistency. There's nothing that you were excited about 3 years ago that's not exciting anymore. So that seems nice to see to me that it's a pretty durable set of growth drivers.

Yes. Well, I think for the automotive industry, you have to really look at the trends there over decade windows, right? If you think about our R&D engine, from the time we greenlight an R&D project until that part is available to go engage with customers, it can be 3-plus years, about 3 to 5. And so for 5 years, you're investing, you're not generating any return on that. And then you have your design to revenue cycle we talked about earlier, 2 to 3 years. So this is a long gestation business. It really wouldn't be appropriate if we showed up here every couple of years and change the goals or change what we're focused on.

Yes. And I know you've done some software acquisitions in the automotive space and things like that. Can you talk to what that might be telling us about where you might be taking the business?

Yes. So we acquired an Austrian company called TTTech Auto. TTTech Auto was a division of an Austrian company called [ TC-Tech ]. What this company offers is a product called MotionWise. It is what's known as a middleware software product. So if you think about the software-defined vehicle in the future, who is actually making the decisions on how that car actually will operate in the features and the capabilities. It's not the hardware engineers. It's actually the software architects. And so if you think of a car as a functional stack with the hardware layer at the bottom, then you have an operating system. Above the operating system, you have a middleware layer and above the middleware layer, you have the OEMs who own the application layer. So what owning TTTech Auto allows us to do is understand how the architecture of the car is going to evolve multiple years before the hardware spec is even passed to somebody like us. So think about it as kind of jumping above the stack. We will continue to operate TTTech Auto independently. We don't want to mess it up. We're not software engineers. The last thing you want us to do is mess that acquisition up. It still hasn't closed yet, so that's probably about all I can really offer at this point.

That's interesting enough. Can you talk about the role of inventory in auto currently. There's -- we went through a shortage that's as tough as I've ever seen. I tried to buy a car at the peak of that, and it was almost impossible. And I know there was a desire to put safety inventory into place in the aftermath of that. But now we're kind of on the flip side, and we're purging. What do you think is happening there? Do you think -- I think Kurt said on the call back that some inventories are criminally low. People are taking it down too far. What's the spectrum? How many people are way too low? How many people are at normal inventory levels? Is everyone depleting now? And is there still a memory of the shortage? Do you think that comes back in some form that...

Yes, a lot of questions here. So first thing to really start with, remember, about 40% of our automotive business goes through the distribution channel. Primarily, we service our Asian customers through the channel. This is in Korea, Japan, China and Southeast Asia. The other 60% is in the Western market through the Tier 1s. It's really kind of our biggest direct sales channel. I would say the channel in Asia is very clean because we operate that channel in a very -- some people might think heavy-handed way. We control it very, very tightly. And I think we've done a very good job for over the last 10 years of managing the channel. The real challenge for us in the last couple of years and going forward is the direct Tier 1s who had built up inventory during the supply crunch, maybe wrong mix, maybe too much at the time, and they're in the process of depleting that inventory. Now there's probably about -- for us, the automotive supply chain is very deep from the OEM all the way back to a company like NXP. So there's about 20 different Tier 1s we engage with. And the challenge for us is there is no automated system like there is in distribution for us to look into the inventory locations of these Tier 1s. We get that information truly by building a mosaic through conversations. And what we see is some at the one end of the spectrum are 10 to 12 weeks, which we think is relatively reasonable. We have a few large Tier 1s who they think going down as low as 2 to 4 weeks is a good idea. Now that would be fine if we didn't have -- we weren't a manufacturer of devices. Our cycle time is 3 to 6 months. So if you, as a customer, believe you can run your business at 2 weeks of inventory, you're essentially going to -- you're playing with fire, you're going to caught short on material. Because our business is inherently an application-specific sole-source business. And so if our customers don't forecast to us over a reasonable period of time, we don't know what to build. And we're not going to take the risk because we will get it wrong. If we decide what you need for 3 months from now, I guarantee we won't get it wrong.

Yes. That makes a lot of sense. The -- I guess the role of China in all of this, China has emerged as the center of innovation for EV. I think you guys have done very well there, but there's always sort of the existential threat of China silicon and if there's a tie, do you win the ties or not. But just great market for you guys, but how do you think about the long-term pushes and pulls there?

Our view of China is we have to be there. I mean it's too large a market. They're too innovative. They're too fast-moving. Their desire to add features that are enabled by silicon is just too high for us to say, oh, we're going to pull away from that market. We've done very well in China. We've been in China probably for -- probably 25 to 30 years before we were even NXP. So our brand in China is actually fairly high. I would say -- and this is maybe not to sound unpatriotic but I would say that the Chinese auto OEMs have pretty much today won the electric vehicle battle. The challenge is growth for those companies is going to be spurred by exports to the Western markets. I don't know if we will ever see a BYD car for sale here in America. But you can buy them in Europe and South America, Southeast Asia. So their growth will be led by exports.

And if they build it in Ohio, you may see it in America. I mean if they may...

Yes, I not going to comment on that, Joe.

Sorry, I'm not trying to go to politics here. But maybe shifting to Industrial & IoT. Your targets there are 9% to 14%, faster than you've grown that in the past. Can you talk about the growth drivers on the line?

Yes. So the growth rate, Joe, that's correct, it was 8% to 12% for the '24 to '27 horizon, which is actually a little lower than the last cycle that was '21 to '24. The way to think about our Industrial & IoT business is it's kind of 2 businesses along a spectrum, about, roughly speaking, 60% of that end market is what we would call core industrial. This is things like building automation, factory automation. We have a small power management health care business. And then the other 40% is a consumer IoT business. It ranges from wearables all the way up to the smart home. So that's kind of the demarcation line for us. Now, different than automotive, there are not nice, neat little accelerated growth drivers that we can give you in Industrial & IoT because it's a very, very long tail business. 80% of that end market goes through the channel. It's tens of thousands of customers. No one customer makes up more than a few points of revenue. No one end product makes up more than a few points of revenue. So there's not what I would call a nice, neat story I can tell you like we do in Auto, about SDV and radar and things like that. The fundamental though, in the Industrial & IoT business, it all -- what we're seeing is an inflection where many of the customers there are starting to adopt or wanting to adopt more software-defined products, very similar to what you see in automotive, where functional safety is important, where they want to have their products defined by the software architect. What that means for us and how it impacts us is that means there's a higher demand and desire for our processing portfolio. So when we go and engage with an industrial customer, kind of the tip of the spear when we first engage is which of our processing products do they want to adopt. And once they've decided on that and as you know, our processing family is very broad. It allows a customer to choose a product and either go up the stack or down the stack from a functionality or performance perspective. Once we've won the processor footprint, we then look for analog attach. Most of these new processes we develop are developed with a co-device for power management and things like that. So it's very much a pull along the analog attach, then we look for connectivity opportunities and then security opportunities. So more and more of these customers, these long-tail customers, they're very smart about their end markets, but they don't tend to know semiconductors per se. So what they want is for customers like -- or companies like us to bring to them complete kind of reference designs. So the processor, the analog attach, any connectivity, any security kind of bundled together with a software wrapper around it. Now I make it sound nice and simple, but the problem is you're doing multiple of these types of reference designs depending on the customer. So there's not like one reference design that we sell to everyone. And our go-to-market model is to work with that 20% of the customers that are direct, as kind of the thought leaders. These are the big Western industrial companies like Honeywell, ABB, Schneider and things like that, and understand what is the need for those kind of thought leaders in industrial. Once we truly understand that and we have expertise in that area, we can then not make it generic, at least bundle a solution together that we can give to our distribution partners to sell to that long, long tail of customers.

That's helpful. I guess -- it's really interesting when you sort of think about the science fiction elements of all of this when you sort of -- you do an S32 to build a software-defined vehicle, you're doing battery management for that vehicle. It takes 5 years for any of that to get implemented in an actual car because of the safety concerns around it. And yet you think about the applicability of those technologies to industrial robotics and things like that. It seems like there's an awful lot of transferability between the two?

Absolutely, Joe, you hit on a great point. And this is something that's kind of become clear for us over the last several years. Many of the things that we were being asked by customers 3 and 5 years ago in the automotive sector are now the same topics that were being asked in the industrial side. Definitely, they're different products. So it's not like we take the S32 and just sell it to an industrial customer because they're in different requirements. But a lot of the conceptual ideas of functional safety, of security, of connectivity are very much applicable.

Yes. Your sense of the inventory in industrial, not for NXP, but for the industry as a whole, been challenging there for quite a while. And it's been surprising how long it's going to take it to hit bottom there. Can you just talk about your visibility into that?

So given that 80% of that end market goes through the channel, we own -- we control that channel very heavily. And these tend to be smaller customers who then depend on their distribution partners for inventory management. So we, from our vantage point, don't see an inventory challenge like we see in automotive because we don't have a Tier 1 network in the industrial marketplace.

Okay. And you talked a lot about distribution in the role there. I mean you've been very careful about that. What leads to that decision? And I think the last couple of quarters, you've let that come up a little bit to try to support some of the drivers of the business. But how are you just thinking about maintaining that distribution level? And overall distribution inventory still seem high. So I feel like a lot of companies are -- not for you, but for the distributors. How are you keeping it lower?

So let's be -- we have to admit our own challenges. Back in 2015, 2016, we made some mistakes with the distribution channel, how we used to manage it. Bill Betz, who is our CFO, at that time, he was the head of FP&A. He actually started a program to build a very highly automated system at NXP that allows us to look through the distribution channel, not only what we sell to a distributor by part number, by geography, by end market, but also how much inventory the distributors have on the same type of matrix and also how sales out of the channel are looking. So we probably, I think, not to be boastful, but I think we have probably one of the best channel management strategies in the semiconductor industry. The distribution channel for us is absolutely strategic for us. One of our large peers in Texas has gone away from the distribution channel. We don't agree with that. For us, because it's such a long tail industry, the channel gives us access to those customers. Our target for distribution is to run at about 11 weeks. We've been running at about 8 weeks for some time now. As the environment starts to stabilize and get better, we will refill that channel. And that delta between 8 and 11 weeks is about $300 million. And as we've said multiple times, the channel is margin accretive. So there is a modest tailwind to margins as we see an upcycle, come together by refilling the channel.

And what's the logic of holding fewer weeks when there's lower lead times? Because you almost could argue the other way that you want more inventory.

Yes. Our view, and this is just an NXP view of the world, is, when you keep the channel tight, you tend to have a higher degree of communication with your channel partners. There's a higher level of interaction, right? When you fill the channel up, there's no reason for your distribution partner to pay you a lot of mind. They've got your products, they can do what they need to do it, off you go, and you're actually asking for information. And when you're keeping it tight, you're actually getting more real-time information.

Very interesting.

That's our thinking.

Yes. Interesting. Can you talk about the manufacturing strategy? You've -- again, in contrast maybe to your peer in Texas, who's putting a lot of capacity in the U.S. and very much focused on in-sourcing, you guys maintained kind of a hybrid structure and also a very geographically diversified structure where you're manufacturing in the regions to some degree. Can you talk about that strategy?

I think first and foremost, we don't believe our customers buy our products because where we manufacture them, right? Our expertise is in the system design and IP development. As you said, we run a hybrid manufacturing model. A decade ago, we were manufacturing about 25% of all of our wafers at the foundry. Today, we manufacture about 60% of our wafers externally and 40% internally. That 40% is also proprietary mixed-signal processes. So over the last 5 years or so, we've pushed all bulk CMOS out of our internal factories and what we run at our 3 internal 200-millimeter factories are mixed-signal proprietary processes. Now the interesting news, I think you're alluding to, Joe, is last summer, we announced a joint venture with Vanguard. We've been -- as you know with most companies, we -- our big partner is TSMC, and we use their design flows. We have been using their design flows to develop a family of mixed-signal products over the last 5 years, and we've been very successful with those products from a design win perspective. What we wanted to do is, when we engaged with TSMC was to understand, could we count on them to provide us with the capacity that we needed. And they were very clear with us. They're not really looking to build trailing edge mixed-signal capacity. They're really looking forward. And so they kind of acted as a matchmaker between NXP and Vanguard, which they own 30% of, for us to build a joint venture. So we built a joint venture that's called DSMC. It will be located in Singapore. It's a 300-millimeter mixed-signal factory running the TSMC process flows, which is really key because TSMC does not license these flows to anyone else. That's a really unique kind of invisible hand, if you will, in the joint venture. The facility has two phases to it. The first phase is a 55,000 wafer per month, 300-millimeter module. And then the second phase is an additional 45,000 wafers per month. That will probably be online, we're expecting late '27, early '28. Now what's interesting about the JV is it's not -- one, it's not consolidated into our financials. We'll invest $2.8 billion, for -- about $1.6 billion of that gives us 40% access or ownership of the facility itself. So we're getting wafers at near cost. But we are investing another $1.2 billion because we actually need more capacity in the short term than the 40% that we anticipated. What that will do for investors or how investors would see that is when that facility is up and running in 2028, it will give about a 200 basis point tailwind to our gross margin. So current model is 57% to 63%. If we just take the midpoint, let's say, we end '27 at 60%, just as an example, this will give you another 200 basis points on top of that.

Okay. Great. We have received questions from the audience. I guess we have a few. Bring the mic up. One in the fifth row and one in the third row.

You alluded to the battle for EVs kind of being over. China has won. Could you talk at all about the time frame in which you think you're going to see significant competition from Chinese semiconductor producers? Clearly, the Chinese government wants local supply into those EV manufacturers. When are they going to be a threat to your leading edge product going into high-end EVs?

Great. Okay. So we get this question almost at every single meeting we attend, which is what about local Chinese competition. So the way we've -- and we don't want to be perceived as naive or kind of sticking our heads in the sand. We expect over the next decade plus that there will be a more mature Chinese semiconductor industry. There's already a fairly decent Chinese semiconductor market. What we see when we look at it is, there's definitely more than start-ups in the power discrete space. Power discretes are a product class or fundamentally device and fab manufacturing. So they tend to go hand in hand. So we see a fairly robust power discrete market in China. We see some start-ups doing what we would term lower end analog catalog products. So this would be against maybe the lower portion of the TI or ADI portfolio. We have heard in the marketplace that TIs have been very, very aggressive with pricing to deny those start-ups and beachhead into their franchises. So again, not really our [ value weight ], but this is what we see. In the space where it might concern us is we do see some start-ups in the what we would call consumer microcontroller space. These tend to be companies that have narrower portfolios, tend to focus very much on product cycles in the consumer side of the marketplace. These are companies like Silergy or GigaDevice or what have you. And one thing about NXP, if you've followed us for a number of years, we are fine with walking away from businesses that don't meet our financial hurdle rates. And so if there is a business that is low margin, we'll walk away from it. What we don't see is we don't see local companies doing what we would call MPU or high-end processors yet today, especially in automotive. Our fundamental view is to always be paranoid, not to take anything for granted. But currently today, we currently compete with Western peers in China. And we will continue to have to innovate to be able to stay ahead.

I have also had one question about the auto MPU. So basically, you have your -- like your competitors in Europe that can offer the -- leverage their total solutions. So they have NPUs and have analog. They have sensors, they have power. And also now they are also emphasizing they have invested in software. So my first question is like how does NXP will like win the share or when over your competitors in Europe, considering in the last couple of years, like NXP because of the [ failed ] acquisition, you lose some time on R&D? That is the first question. The second question is that, you just mentioned like 60% of your OEM base, they only run the inventory 2 to 4 weeks. So how many percent within this 6% is like term business?

So I'll see if I can get both questions, 2 different ones. One of the question is how do we compete with the European peers in terms of our MPU product portfolio. So what I would say is, both of our major European peers are very good competitors, but they tend to be very focused on the microcontroller space, not on MPUs. We don't see either Infineon or ST investing in this area. As a matter of fact, we know that ST has actually walked away from any type of MPU effort. We actually don't see either of those companies investing in the type of software that we're looking at with our TTTech Auto investment. So I think maybe when you refer to software, I think the days of being able to build a microcontroller and just give it to your customer and walk away, are behind us. You have to provide enablement tools that involves tools for them to write their code, but also as much firmware as you can possibly offer to them to bundle with the silicon itself. So I'd say from that perspective in the MPU, in the processor space of automotive, I don't think that we are at a disadvantage versus our European peers. Infineon has done well with their microcontroller family, and you have to understand that their margin entitlement range is different than ours. And there was a period of time in 2016, 2017, where we walked away from a fairly large amount of business because it just didn't meet our margin structure. And we know Infineon picked that business up. They acknowledged that in their earnings call last year. So I think -- good for Infineon. They've done well in that kind of moderate level performance microcontroller space, but I think our focus is really on the MPU marketplace, 5-nanometers, 16-nanometer, 28-nanometer products. Your second question, I'm sorry, I didn't quite fully understand. I think you were asking about inventory at the Tier 1s? Did I get that correct? I want to make sure I understand correctly.

No. Not really. So you asked like 60% of your Automotive business is from the OEMs, like a Western world. Some of the OEM only place orders -- they only run 2 to 4 week inventories. So it means like every quarter, if they need something, you need to place the rush order. So I mean for this kind of term business, how many percentage of the rush order considering the...

Okay. So you're asking what percentage of our business is turns order. We haven't disclosed that. I would say, we have to remember it's our cycle time to manufacture our parts is 3 to 6 months. And given that our products are application specific and sole source, if you -- if you're a customer, if you don't forecast your product to us, we don't build it. So it's not like we can run this business long term in a turns type of environment. Right now today with lead times being relatively short and inventory on our balance sheet being a bit above target, we can operate in that environment. But it's not an environment we like to operate in, right? Because turns -- well, look, it's a signal that you're closer to the bottom than to the top. You really want to get forecasts from your customers on what to manufacture.

All right. It takes us up to the end of our time, Jeff.

Thank you very much, Joe. Appreciate it. Thank you, everyone.