Good morning, everybody. Welcome to the Goldman Sachs Communacopia + Technology Conference. I'm Jim Schneider, Semiconductor Analyst here at Goldman Sachs. It's my pleasure to welcome NXP Semiconductors CEO Rafael Sotomayor and Head of Investor Relations Mike Lucarelli. Welcome, guys. Thanks for being here.
Thank you.
Thank you, Jim. Thank you for having us.
Of course. Maybe starting with a high-level question for you, Rafael. You're just a little less than one year into the job thus far. It appears that you're trying to transition the company and really pivot to this theme of physical AI to some extent. Tell us about your vision for the company, both in the short term and the longer term, especially related to that theme.
Yeah. First of all, thank you for the question. It's good to be here. Yeah, it's been almost a year now. I think what it was very apparent to me, I think, is the market is coming to us. I think my job now as a CEO is to make sure that we go to the market at the same level of speed, right? Not just to participate on this new phase of what's happening with edge devices, but just to lead. I think NXP has spent decades, right, earning credibility with edge devices. Talk about safer cars, smarter factories, more efficient infrastructure. These are systems that value performance, value security, value reliability, and now the world is changing. It's massive value creation in AI, the cloud, and this IP is now moving to the edge.
Now these devices with AI is able to perceive the world, make sense of the world, make decisions, act, become autonomous. That autonomy raises the bar in terms of engineering. These devices usually are going to operate next to us, next to the high-value assets. They don't need just security, right? They need world-class security. They don't need just functional safety. They need world-class functional safety. They don't need to just operate. They need to operate deterministically under all conditions. These are the things that NXP spent decades investing in. We're quite excited about it. Now, the pivot and the transformation that you refer to is, well, there's in three areas, right? The three areas that I'm focused on. One is focusing the portfolio towards this notion of physical AI or intelligent systems at the edge.
That means divesting things that don't point that way or doubling down on areas and talent and IP that helps us get there faster. The second thing that was a realization that intelligent systems is complex for our customers. They have compute, they have analog, they have sensors, they have all sorts of technology that needs to come together, and that complexity creates an obstacle. We needed to make sure that this obstacle gets removed. We invested heavily in systems. We are now looking at our roadmap, not only from a component perspective, but from a system perspective. That's the second change. The third one is a little bit more of an internal change, but it's just we need to keep pace with innovation and with the speed of the world. That means our intensity of execution needs to go up.
Significant effort internally and to synergize IP development, make sure that we have transversal technology that applies to different markets, that we reuse, that we don't overlap. Very, very ambitious and aggressive deployment of AI tools to actually increase productivity, and we're already start to seeing it in our output. That's the change, right? The change is moving to what intelligent is, and that's the pivot that we're making with the company.
Great. Now, I want to continue on in terms of the portfolio piece of the topic. Portfolio today is pretty heavily focused on automotive, a little over half your business today. Within that portfolio, your products span processor side of things, control plane, zonal microcontrollers, sensors, mixed signal, including radar, digital networking, Ethernet, wireless, et cetera. First of all, as you think about the portfolio within automotive, how do you feel about the portfolio as it stands today?
I think my perspective is NXP has the most comprehensive automotive roadmap and portfolio. Processors, analog, sensors and radar, connectivity, both wireless and wire. Battery management and electrification. It's a quite comprehensive portfolio, and the whole point of this comprehensive portfolio is not just to sell a bag of components. It's about really deploying those assets with a system solution, because this is a massive transformation that is happening in automotive. This adoption of, I mean, two areas. First one and most important one is the adoption of software-defined architectures, and the other one is autonomy. Both of them are having a massive transformation. The investment that we have right now, I think, is very well-positioned for these two mega trends.
Yeah. Maybe highlight some of the biggest vectors within that vertical in terms of the, you talked about them like self-driving, autonomous, move to EVs, maybe, and software-defined. Just talk about some of the products and where the fastest growth areas are within that portfolio.
Well, let's start with the most transformative one, that is the software-defined vehicle. I think the software-defined vehicle is I think it's underappreciated in terms of how the transformative aspect of adopting this technology or this architecture. The whole notion that you go from these hardware-bound static functions, to a more dynamic digital platform that can continue to be updated over time. It's creating just a massive transformation in the industry and it's bringing a lot of content growth. You're removing all these low-end hundreds of ECUs and MCUs scattered all over the place, isolated, developed by different people, different companies, to a kind of common platform where you aggregate all this with higher performance processors. The roadmap of NXP is a roadmap for that future, where all this aggregation of all these disparate ECUs are happening with high performance processors.
They are real time, and it brings new challenges and new opportunities. These processors are not just processors, they are networking products. Because now you aggregated all these ECUs, and they have to connect all these sensors. There is a massive opportunity to create a new class of products, which NXP is already ahead with 16 nm for zonal products and 5 nm for central compute platforms. I think that the roadmap of NXP, and I am quite bullish on it, is primed for the architectures that are being adopted right now by our customers. I think that is the most transformative aspect of the growth drivers that we have, which is SDV. Of course, radar, as autonomy improves and we go to Level 3, Level 4, you will need more sensors, including radar.
We have electrification as one of our growth drivers, and then connectivity, but software-defined vehicle is the largest of all of them. By the way, if you put all these accelerated growth drivers in Q2, they were very close to 50% of our revenue already. These accelerated growth drivers of automotive kind of indicate the portfolio that is tap into the secular drivers of growth in automotive. These, by the way, they were growing close to 20%. We have half of the portfolio in automotive growing at 20%+, and so quite exciting and tied to secular drivers. You have two impacts in this portfolio. You have the content growth, but I also think is underappreciated is that the leadership on these products also is going to generate share gains. These two are going to compound, and they are quite important for NXP.
Yeah. That sounds very compelling. How do you think about the risks to the portfolio? Let us say if the world moves to an automotive architecture that is far more centralized in a centralized, single large processor or a GPU-CPU combination.
Let me challenge a little bit that premise. I think, first of all, let me dispel the notion that all CPU cycles in the car are the same. If you look at the different domains, and I assume you mentioned IVI, you mentioned ADA, you mentioned core function of the vehicle. Each one of these has different technical merits, a particular purpose. The whole notion that all CPU cycles are the same, let me dispel that. Think about the scenario that you painted. You have one chip that handles all your infotainment. That handles, processes all the data for actually create autonomy, and at the same time manages all the actuators in the car. It manages all your power and all your connectivity. That just does not seem realistic. I do not think that is necessarily a threat.
I think the threat for us will be to show up, like I said, to a customer with a bag of components, instead of showing up with a customer with a system solution that leads the transformation and the architecture of the customer into SDV. I think that's where our focus is.
Yeah. Very fair. How do you think about your customer relationships, especially with the Chinese automotive OEMs? What level of competition do you see in that market from domestic chip providers, and how do you expect that level of competition to change medium term?
Yeah. Well, let's step back and say, going to give you my perspective on China. China as a market and China OEMs as customers, or Tier 1s. I think both are extremely important. Markets go up and down, and there's some weaknesses here and there, but it's a very large market that attracts a lot of innovation. People are eager to try and to iterate, and it's an attractive market just to participate in it. The OEMs in China are also very important because they're not relegated to China. They're actually winning overseas. So access to make sure that NXP wins with the winners is very important to participate at NXP. Now, the relationship with Chinese OEMs is strong, and it's getting stronger by the fact that we have a lot of innovative products that they can take advantage of.
I think it's strategically, I think, advantage, our relationship with Chinese OEMs, primarily driven by our portfolio. Now, let's admit that there is competition in China. There is an indigenous and very strong competition in different areas. Chinese players have done very well in infotainment. They've done very well with their own ADAS solutions. They've done well with power discrete. They've done well with even low-end MCUs. I will admit that, but our goal is not to compete directly with component solutions, and it's also not our goal to compete in the low end. We must take a playbook and be disciplined with our playbook, which is we must win the architecture of the customer. That's what we're doing right now.
By the way, I am quite excited about the traction that we are getting right now with our S32K5 family of products, which is a unique zonal product for these new architectures. High performance, 16 nm, embedded MRAM or embedded type of flash embedded into the product in 16 nm, quite unique. Phenomenal networking performance, world-class security. Unique product being adopted heavily, and right now in China, the POCs in K5 are quite exciting. I give you that perspective is that our playbook in China is acknowledging that we have competition, but also acknowledge the fact that innovation is the only way to win there and making sure that China competes very well with our products.
Great. Final, I had a high-level question for you. Over the last several years, you have made a number of divestitures to kind of refocus the product portfolio. You referred to that before. Acquisitions that mainly have enhanced your automotive exposure. How do you think about the portfolio more broadly beyond automotive, and would you consider a larger, more transformative acquisition that would give you some exposure to some emerging verticals, whether that is robotics, data centers, other areas that fills the AI, et cetera?
Jim, let me push back a little bit on the notion that our acquisitions were auto-driven, right? We do not consider ourselves an auto company with adjacencies. We consider ourselves a physical AI company, where auto is our biggest market. But over time, robotics, industrial is going to outgrow the auto market, and so we are positioning the company that way. If you look at the M&As that we did, if you look at Aviva Links, this is about high-speed serial interfaces, right? That, yes, it is applicable to automotive, but it is going to be used in industrial significantly. It is going to be used in robotics. If you look at the investment that we did in Kinara for AI inferencing, AI is transversal.
It applies to just about every single market and every single product. The acquisition of TTTech was all about enhancing our system solutions. Software-defined vehicles was also going to be applicable to software-defined anything. The whole notion of getting people who are very well-versed with real-time systems was something that we aspire to do. With respect to, are we going to consider something bigger and more transformative? These are things that we always evaluate, right? There is a high bar to pass if we were to actually acquire somebody. But as long as it accelerates our path, our North Star, not just revenue, but our path to a strategic direction, we would evaluate it.
Great. Now maybe turning back to your current business trends for a moment. You reported a pretty solid quarter, very constructive commentary on demand trends across most markets. How do you characterize the end demand trends as you are seeing them right now across automotive, Internet of Things, mobile, and communications? Where are you seeing the strongest orders on a relative basis?
Well, we just finished Q2, right? It was a record quarter for NXP, record quarter in automotive, record quarter in industrial. We just guided also a very strong growth. I think Q2 was close to 20%. I think once you exclude sensors, it was 20%. Similar trajectory that we grow in Q3. That kind of tells you the demand environment is strong, right? Q2 growth came from all regions in all segments. That kind of gives you confidence of the demand environment that we have. I think visibility has also improved. We have extended our lead time, so that has also helped having conversation about what is the products that our customers need and require. So it gives the visibility to make sure that we are building the right products for Q3, for Q4, for Q1 next year.
I think the demand environment today, for the most part in industrial and IoT and automotive and some of the demands that we have on comps and infra are quite healthy.
Very good. Then, in terms of the balance sheet metrics and inventory levels, I think your internal inventories were down on days, flattish in dollars this past quarter, disti inventory flattish 11 weeks. You also talked about, though, increasing book-to-bill trends, lead times, and more customer expedites on the margin. How do you sort of characterize your OEM customers' behavior at this point, their willingness to potentially increase their own inventory levels of components or not? If they have not started restock yet, what do you think might prompt that? That is normally behavior we start to see at this point in the cycle.
Yeah. I think let me address the inventory comment first. I think the inventory is under control. I think our inventory days went down to 156 days, so we are down quarter- on- quarter. The channel inventory that we have kept steady at 11 weeks, so it is quite normal, and this is consistent with our long-term growth targets. With respect to visibility on the end customer's inventory, I would say that inventories continue to be lean, and none of the forecasts and the projections incorporates this changing that we have. By the way, when you speak of OEMs, it could be Tier 1 s, it could be OEMs, but at the end of the day, a direct customer, I think that inventory is basically kind of consistent, right? The level of inventories is consistent.
What would change that, I think the fear of putting lines down, at the expense of carrying that capital, stuck with inventory. I think that would change that. The silver lining, I think, is that leads extending has helped us with visibility. The conversations, even though customers have not increased inventory, conversations are quite constructive. Right? It is having visibility into what they need, when they need it, having a conversation about the tightness of supply, because there is tightness of supply. I think that those conversations are a lot more constructive, a lot more transparency, which overcomes the fact that some of these companies are being squeezed and they are taking care of their working capital in a very disciplined way.
I think we have to acknowledge that the value of some of these products has decreased in Tier 1 s, so cash flow is important for them. Having said that, I think we work with them to make sure that we have transparency all the way from the OEM to the Tier 1 to be able to make sure that we build the right things and we deliver the right things on time.
Yeah. Excellent. Among investors, it seems like the hottest debate right now relative to analog is the topic of pricing. You have talked about starting to see some pricing increase selectively, but that has not had a net positive impact yet on your margins. Help us understand when investors should expect to see some of the biggest impact of pricing for the company and some of the product groups where those price increases or changes could be most significant.
Well, let's just talk about pricing in general, just because I think the landscape of pricing is changing even at a structural level, right? I am going to address your comment that I think was more on the reactive side of inflationary supply, but just step back and see what's happening with pricing on the landscape is very different than what happened the previous 10 years in semiconductors. I would say to start with the fact that semiconductors have become a bigger part of the value proposition of the new systems, right? They are at the core of innovation. They are the core of the value proposition, they are the core of the ROI of those devices. That carries a significant amount of weight. So innovation carries a significant amount of weight, and customers are willing to pay for that innovation, right?
Of course, I am not saying that removes price pressure or competition. All I am saying is that the heavy headwinds of price erosion on a yearly basis that you used to have back in the past 5- 10 years, is going to be different more and more going forward, just because of how critical it is to deliver the value proposition of product through semiconductors. The pace of innovation that is happening more and more, the next generation delivers more value, integrates more BOM, and by definition, that price will continue to go up through value capture. So that is a very important structural change that is happening on pricing. The second one right now is supply is tight and input costs have gone up.
Now you have more of how we react to these disruptions in the market, and both are combining into a price environment that is significantly more benign for semiconductor companies. I think the impact that we used to model in your pricing over time every year by several points of price decline, I think that is just going to get better from a semiconductor perspective. It is going to get better.
Now, the question you ask about how we see the impact, we see it already, okay? It is just gradual because you deploy price increases by having conversations with your customer. Some of them, you have contractual obligations, so there is timing associated with these price increases. You deploy gradually through some selected products and over time you kind of deploy it. So the impact to your top line in terms of these price increases is gradual, but it accumulates.
Every month it's a little bit more and more because it's not one step function, it's a gradual impact. We expect to see that more and more in Q3 and Q4 and going into next year, because it's just going to accumulate more and more.
Great. Couple questions then, market. I think we covered automotive fairly thoroughly. On data centers for a second.
Sure.
That's a market you've guided to revenue more than doubling this year versus last. It kind of appears to be shaking out to be a pretty strong business segment for the company as it mainly appears. Maybe speak to some of the specific opportunities for NXP in data center and how you're positioned there.
Yeah. The way we approach a data center is very much as an infrastructure play for us, right? We approach the data center, invest at least portfolio for us as particularly the control plane. These are top of racks, which is in processing that goes into the boards that handles and monitors the well-being of the rack. Right? These devices are very complex systems, and making sure that you manage uptime and cooling and power into these devices is fundamental for the investment of these racks to actually being up and running all the time. We invested in areas where we have strength, which is part of our industrial play. This business has grown, and will grow. We provided a guidance this year of $500 million, growing from $200 million last year, and it probably continues to kind of grow into next year around 20%.
This is based on the current design wins and the current portfolio that we have. The trend that you see right now in data centers is, if you go now, go look at what is happening on the data plane, these networking products are going from 400 Gb to 800 Gb to some of them are 1.6 Tb. When you have speeds that they are going at that level, then infrastructure required to manage those racks need to also go up in performance. There is going to be a lot of opportunity as the complexity of these racks goes up. I think there is going to be plenty of opportunities to continue to drive innovation at the control plane associated with controlling the fans, the type of connectivity that you have, the type of crypto that you actually deploy into it.
Monitoring the wellbeing of that rack is going to become more complicated, and I think there is going to be more innovation. Today, we gave you the forecast that we have with the current products, but you should expect NXP to continue to expand our product portfolio, to expand our SAM, and hopefully continue to grow faster in this market. As we make progress with the visibility of those products, we will provide updates.
Great. Let me switching gears for just a second to touch on physical AI and humanoid robots.
Yeah.
Which you kind of led with. What do you think about the available market for NXP and your portfolio in a humanoid, and sort of what are the biggest components that drive that?
Yeah. I think autonomous machines, in general, it's just a matter of not if, but when. You mentioned, obviously, humanoids, and I think humanoids are the ultimate expression of physical AI and robotics. It's probably the most complicated form factor. Today, I think there's this notion that the big challenge of robotics is just compute. Put a big brain in there and things will happen. In reality, humanoids, particularly humanoids, is a system problem. We have done analysis on what we're doing right now with robotics, and we see that the fact that intelligence for robotics, very similar to what happens. Intelligence in humanoids is very similar to what happens to intelligence in a human body. It's just very similar. It's not about one big brain, it's about having a hierarchical level of intelligence throughout the humanoid.
What this means is very similar to us, we have a big brain, we have a cerebrum that is powerful. It does a lot of reasoning. You're going to have a similar kind of reasoning layer of intelligence in robotics. Then you have the next layer that manages coordination. In our body as a cerebellum, is the ability to actually be able to walk without really thinking about it. Super low power, high performance, intelligent, but not as intelligent as our ability to reason with the body. We have a similar concept in humanoids and in robotics, which is a coordination layer. For instance, in a car, that would be the equivalent to our S32N product, which is our central compute platform, which is heavy networking, heavy compute power processing, but it's still real-time performance and low latency.
Then the third level in humanoids will be the reflex layer, which is equivalent for us for our spinal cord. Think about our ability to react very quickly to things. This in a robot will be actuators and motor controls and anything on the limbs and the nodes. Now you have three layers of intelligence, all coordinated, all aligned, and that's a system solution for robotics. Of course, things need to happen in many ways. Areas that we envision to participate in, I'll tell you, if we were to take an estimate of a western robot today from a BOM perspective, it can go from anything from $1,500 to $4,000 from a BOM perspective. I think with the current portfolio, NXP can address about half of that. These are areas where we just talk about the different processors for an AI.
By the way, there are about 300- 100 joints in a robot. Each one of them requires some kind of motor peripheral, and so motor control is a big part of it. You have connectivity, basically low latency, deterministic connectivity. So think about TSN and Ethernet. That also participate for us. Sensing, whether it's vision, whether it's radar, whether it's UWB, the ability to actually kind of figure it out what's happening around it. That's another area that we can participate on it. Battery management, obviously. I think we're quite excited. Of course, I would say this is more of a 2032 type of kind of market. Between now and then, there are other form factors that are important in robotics. Drones, AMRs, robotics in smart factories.
For us, a path to humanoids involves making sure that we are very relevant with our robotic customers today in different form factors, in form factors that are a little bit more constrained: factories, law enforcement, delivery. Things that are a little bit less challenging from a form factor, but you still apply the same principles of hierarchical AI, sensing, perception, battery management, and actuation.
Got it. Speaking as somebody who needs a bigger cerebrum, I hope you are right. Okay. I want to hit on two more common investor questions. One is on financial targets. In 2024, you laid out financial targets for 2027 for compounded revenue growth of 6%-8%, gross margins 57%- 63%, and op margins 34%- 40%. Since then, you have made some divestitures, acquisitions we just talked about. Looking forward to next year, what is your level of confidence in hitting those targets, and are there any caveats investors should be aware of when considering your performance next year?
Well, if you look at the performance that we had on the first half of the year and the guide that we gave you, I think you can probably see that I think we are on track from a revenue perspective on hitting those targets. Nothing today would indicate that this will derail. Listen, things could happen, but I think right now, the trajectory that we have on our revenue, the customer traction, the design wins that are shipping already, I think that kind of leads to very high conviction that we are going to achieve our growth targets in 2027. Growth margins, we just guided to about 58%, I think 50 .5% .
50.5% .
50.5% in Q3. We expect that leverage to continue this projection towards the 6% handle of growth margin to continue to actually increase as we grow revenue.
The operating income, I think is within inches away from 37%. I think the targets that you gave me there are well within reach. You can see that we are actually creating a lot of leverage as revenue grows, and I think in 2027, this will continue. My biggest desire is to make sure that 2027, especially towards the end, it kind of has that momentum to even bigger opportunities in 2028 because some of the things that we talked about, this new platforms for SDV and the new platform, Physical AI, they do not even start ramping until literally at the end of 2027.
For us, the results that you gave me, yes, they reflect some of the portfolio we have now, but the traction of the new portfolio, I think it is even more exciting, and it is going to reflect on some of the things that happen towards 2027.
Yeah. Okay. Final question. One question I get a lot from investors is, if we have a heck of an industry upturn ahead of us and that lasts longer and is stronger than any of us expect today, how do you think about the available capacity to satisfy that higher level of demand? Your strategy has a mix of internal fabs, external foundries-
Right.
joint venture structures. Maybe address the overall manufacturing strategy, especially the JVs, SSMC, ESMC, how they fit in your business model, and then why are those JVs a good use of shareholder capital?
All right. It is going to be a challenge for me to answer that in one minute, but let me just kind of tell you the following. When you look at a manufacturing strategy, there is three goals. One is access to supply. Very important. Two, supply resilience, and that means a lot of things, but many of our customers require diversity, and we want supply resilience. Lastly but not least, is access to cost, competitive cost. What we have established from a manufacturing strategy is that we decided to go asset light on the fab, and so right now we are 60/40, meaning 60% external, 40% internal, and we are moving that to even 80/20. That reflects the portfolio that we have. More and more of our products are moved to 16 nm, 5 nm.
That reflects a little bit of the strategy we have in the portfolio. We created two JVs. I think the JVs is a 10% ownership that we have on ESMC in Dresden in Germany and about a 40% in Vanguard. That gives us not only access to capacity, it gives us access to cost. Also structurally, we move a lot of products in 200 mm fabs into 300 mm. That also structurally is a better cost. On top of that, we decided to actually own our own backend. So our backend is about 80% internal, 20% external. We have this external fabs, internal backend, and that gives us enough flexibility to do the things that we need to have.
Excellent. I think with that, unfortunately, we are out of time. Rafael, Mike, thanks for being here. Appreciate it.
Thank you. Appreciate it.