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

And gentlemen, introducing the Executive Vice President and Chief Technology Officer of NXP. Lars Reger.

Hello, everyone. Good afternoon and a big thanks for having me here. Truly a special moment for me being here in Taiwan with all of you, and I would like to take the next hour to take you on a journey towards a brighter future and, of course, a journey throughout the semiconductor industry. And with all of that, let me start with my journey with NXP. I joined NXP 16 years ago in 2008. And to tell you very frankly, I thought by that I have made a mistake of my life career wise. I joined the company in 2008 that was fully hit by the economic crisis just spun out of Philip semiconductors. So searching a little bit for the identity who are we, NXPiers. And I called my wife, I still recall, middle of the year and said, Darling stop packing the bags, we don't relocate. I'll be back soon this thing is going sidewise here. And you hear it in my voice, how proud I am to be now standing here as the CTO. So the chief [ nerd ] of a company that has 35,000 employees, 12,000 engineers is sparkling from innovation present in 30 countries, amongst others, also very strongly here in Taiwan. And of course, what is making me even more proud is that 2-weeks ago before the first time hit a company valuation of USD 70 billion. And this USD 70 billion, I'm telling you not to break about a number. No. What is an analyst evaluation? An analyst evaluation is basically a forecast of a potential, what can this company capture in value in the future? What can this company generate in future? How can this company help drive this world to a brighter future. And this is, of course, what fills me with pride if others are seeing this. And hey, this is a great story for NXP over the last 16-years, but this is not a story. It is not a story of NXP standalone. This is a story of an ecosystem. This is a story of strong ecosystem partners, and I see a couple of familiar faces here in the first row. Ecosystem partners that are very, very strong here in Taiwan, especially as well -- and I know exactly we are standing on many of your shoulders as you are supplying us as you partner with us or as you are our customers and as you help us working the ecosystem going forward. So big thanks for all of that partnership and looking forward to the next years in that industry, where we strongly need that partnership. And I'm also proud that I'm not only here at Computex, but I'm also here for InnoVEX, so the start-up convention that is running in parallel to Computex with 400 startups from 100 nations because also what we try to do is, of course, we try to breed unicorns. So in other words, the next big companies over the next 1, 2 decades going forward. We need them as our suppliers, of course, as our partners, again, maybe they have interesting technology that is relevant for guys like us or, of course, if they are big, if they are growing in the way like we want it, then ideally, they are our customers and use our silicon. So there is a reason. There is a reason why we have coined the slogan "NXP brighter together." It is all of us, it is us and you, it is an ecosystem. And what can this ecosystem do going forward? Well, we can try to make this planet a better world. And the world ahead, for sure, is different. You have seen a lot of keynotes in the last days. Everyone is forecasting there is disruptive change coming. The only funny thing is the innovation drivers are not changed -- that much changing. I had one slide that I used already in 2008, and that is behind you. Yes, our [indiscernible] teams have found better pictures, but the innovation trends, the drivers here on that slide are unchanged. They [ unchanged ] since 2008. It was clear that we will get a climate change. It was clear that we are facing natural resource shortages. It was clear that we have aging demographics that more people want to live in mega cities, that everyone wants to be connected in the world. Mind you, 2008 was the year of the smartphone, but everyone want already there to be connected and everyone wanted to live in a safe and secure world. So that is all unchanged. And it is good that we have these innovation drivers because if we can bring solutions to all of these problems, if I can solve your problems, you will give me money for that. And this is unchanged in the last 10,000 years. This is business. So for all of us, this is ample of opportunity to go into areas of good manufacturing or people healing. It is work that we are doing in the infrastructure around us, how we live, how we be moved. So it is all of that. A nice foundation for continued business and business success. Now how is that going to look like in future? Let me try to jump with you a little bit ahead. I'm in the industry since 27 years, 16 years with NXP. How is the world going to look like in the next 27 years. So how if you join me on a journey into the year 2050. So Lars is getting a little bit older, 2050, Lars has an age of 80 years already. And the key question, of course, that is bugging me is -- what is the world looking like in 2050? Of course, I don't have a glass ball. If I would know the details, I would be rich already today and I would retire tomorrow. So it's just an assumption, of course, that I try to bring to you, and I try to take you on that journey so that we can dream together because if we can dream it, we can build it. What we will see is very likely a couple of core assumptions. We will see an economic power shift. So the big populations today, the emerging countries, China, India, a lot of Asian countries, -- these big populations emerging countries are emerged countries by 2050, a lot of wealth in those economies. So that is all a geographical economic power shift, of course. Then one other thing that is different to our thinking today. The world population is already past its peak point. The population is declining. Energy supply, the energy transition is likely behind us. We are using mainly renewable energy sources, decentralized but well networked all around us. And this is a stable energy supply grid, so pollution, exhaust gas and so on. That is no topics that we will see by then. Mobility, it's very likely very, very differently. So mobility will be safe, efficient, different type of vehicles, different form factors. Today and a very, very different means of transportation. And the biggest shift that will touch all of us is our jobs. I will be retired. So I will not care that much about it anymore. But the jobs, the manual tasks, the simple jobs have disappeared. They are completely robotized. The ones of you who are still working in 2050 will mainly work in caring jobs where the human component is the value proposition or in operated jobs, basically observing devices or being in control centers, mission-controlled centers. Or in other words, there is a huge potential in the next 25 years to complete change these ecosystems, to completely transition into that world that anticipates and automates. What you will take for granted in 2050, is that you don't care for your house anymore. Your house is completely around you. Your house knows how you are doing, your house knows when you need to do -- and it needs to do maintenance, not only sending out a lawn mowing robot, I mean that is a technology of today. but your house is completely climatizing, filling the fridge. It is like our old descriptions of Paradise? There's always food, always warm, everything seamless and barrier-free. So if you are moving in your house, you don't know the concept of door locks anymore because you are moving and the doors are always open, but only for you, burglars cannot get in. They are completely shielded. And this is a seamless experience when I talk about the world that anticipates and automates. Mobility. We have lived now for 150 years on horse carriages with a combustion engine. So in other words, the concept of the horse carriage has not changed. There's always a coach man in front operating the horses driving the steering wheel. Forget it, this is not what our kids will think of when they think of driving in 2050. The driving task is completely automated. But these cars, these transport capsules, maybe driving flying or being on the rails. These transport capsules have become convenience spaces. Like already today, in some Asian markets already, we see this. People are leaving the flat where the kids are playing in a noisy way, go to the cars have a convenient space to watch a video or take a phone call. So this will be a totally different concept. And like in your living room, these convenience spaces are upgrading themselves. You have a new app, you have a new feature, maybe you're changing your real estate a little bit bigger display or whatsoever. This is upgradable systems and also a concept that I hope we get rid of by 2050, traffic trends. Just think of some of the mega cities, 500 meters outside of the building here, how many traffic jams do we have because there is good transport in some of the cities, there is pizza delivery, parcel delivery, all of these type of things that are by 60% of the traffic congesting all the roads. If you can automate the entire world, then these type of things are, of course, happening in the off-peak hours at night, definitely not in the rush hour and you have a totally seamless traffic flow. And again, -- how are our factories looking like? Our factories will be clean and automated, okay? And there is only maintenance workers in the factories, workers that come to repair or to upgrade a manufacturing robot, there will not be normal operation workers in the factories anymore. These operation workers are outside in a mission-controlled center are sitting in front of 2 or 3 screens. They don't wear the blue color shirts anymore, but they are sitting there as dispatches, as observers and only interact if something goes wrong, that the robots cannot handle. All the rest is automated. And I see the first faces here in the audience saying, should I believe him. I don't know. This year, what I have here on the slide is not a dream. This is happening, and I have seen it -- and I tell you where I have seen it. I've seen it here in Taiwan. This is what TSMC is already doing. If you are looking into TSMC's most modern fabs, what you see is a lot of maintenance workers in the clean room, highly automated systems, high-speed transportation systems of silicon wafers and the mission-control center outside of the clean room. And TSMC is maybe one of the first companies in the world because the semiconductor factories are so capital intense and so heavy to invest in that this far advanced concepts are taking place for the first time, but Taiwan is one of the frontrunners in all of that. Now I told you, I've already jumped to the year 2050 and let me quickly introduce you to another member of the Reger family. This is Keira. -- my then -- 16-year-old granddaughter. And Keira is going to ask all sorts of questions to Grandpa. Hey, Grandpa Lars, Grandma has said that you have been pretty heavily working in the industry of all these robots. And that you have talked since 50 years already about a world that anticipates and automate, Grandma even said that when you started working, you didn't have a mobile phone. So yes, Keira, that's true, right? But can you tell me how your use was? how your job was? how was your career in this [ industry ]? How do you call it semiconductor industry grandpa? Yes, Keira, I can tell you. Over the last 40 years, they had been interesting developments. And let me give you the story. Grandpa was actively involved in the transformation of the industry, of the world to digital twins. Then we had the great robot awakening. And to awake the robots, we have to do the system brand shift. Keira's interesting headlines Grandpa but what does this exactly mean? Well, Keira, let me tell you. When I was a young engineer, we in the semiconductor industry, computer chips, we were working on pretty boring electronics. We worked on laptops, Keira that is portable computers. Keira there have been even desktops. So can you imagine desktops was a different word for a computer, so bulky and heavy that you could not carry it. Unthinkable today in 2050, of course. We had the first portable telephones, we had separate computers for gaming. That's also a different concept than in 2050. And we had audio and video electronics. That was the entire industry between 2000 and 2010. And then Keira, they came this green phase. The green phase was the on-demand world, so some smart engineer at the U.S. West Coast took a portable mobile phone and the portable computer stuffed it into each other and called it a smartphone. So we were proud of our smartphones by then. Smartphones were data display devices. And we had, for the first time, Keira. We had big storage and big compute in the cloud. And it's big storage and big compute and these data display devices. They helped us to create an on-demand world. Whenever someone needed a taxi, you could press a button on your mobile phone and the taxi would come. Whenever I want to -- someone wanted to have food, you could order it on your phone. And whenever Grandpa wanted to switch on the auxiliary heating in his car, he could press a button and the heating would switch on. So in other words, Keira this on-demand world is a world that is remote-control world, but human beings had to work. And then Keira this entire market exploded only in the blue phase. So 30 years ago, only in the early 2020s because that was the transition where we moved into the robotics age. We took all the machines that we had to operate manually and move them into robots that autonomously could work. And for that, we started creating digital twins. And these digital twins were complete mappings of me in the cloud, so my state of health, my state of wealth, my entire physical correct mapping in the cloud. And what we also did Keira is, we did this for everything around us. We did this for industrial manufacturing. We did this for health care, for hospitals. We did this for smart homes, we develop smart homes only by 2020, and we did this [ forecast ]. And then we tried to combine all of these digital twins. And that was the first step. But Keira, having all of our staff somewhere in the digital world didn't give us anything. There had been visionary guys. Maybe you still recall that old tool Facebook, Mark Zuckerberg invented it and Mark Zuckerberg started talking about a concept that he called Metaverse. Jensen Huang, an AI evangelist of these times. He talked about digital twins and he called them Omniverse. But seriously, without the next steps, these 2 gentlemen would only have been funny digital game developers. So what do you need? Keira, what we were needing, they are all these smart connected devices, the tentacles of the Omniverse and of the Metaverse. The reach out into the analog world. The data transition is happening in the analog world, get it into the digital twins, digital twins, talk to each other, optimize each other and bring that entire know-how back into the analog world and react there. So in other words, we had to wake up billions of robots around us. And then Keira looks and says, hey, Grandpa. Really, all of these robots came from that idea, Yes, Keira, a similar concept. Yes, but Grandpa. I mean they all have so many different form factors. Billions of different form factors. And you talk about one family of robots. Yes, Keira, because our concept that we had in mind was that all of these robots follow the same way how to be built. All of these robots were sensing their environment in the analog world. All of these robots are connected to the cloud and they're getting data from there. All of these robots were computing an interesting thought when they were thinking. They were trying to anticipate what has to happen next. And then they were sending this concept to the arms and legs of these robots. So if we only would build sensing, connect, act into our robots, we are done, not entirely because Keira, you would only hand over tasks to a robot, then you can trust that robot. And therefore, safety and security or in normal people's language, trust has to land in the machines. So Keira we could enable these different robots only because we could master the sensing, thinking, connecting and acting and we could bring safety and security into these robots. And NXP tried since 30 years to be a world champion in these 6 areas. And then what we try to do is we try to find scalable platforms to address all the different form factors of these robots in sensing, connect, act, safety and security, and this is how we could wake up all these robots. And Keira is looking at me, shining eye say Grandpa, that's a great story. But Grandpa, you're a little bit slow. What do you mean Keira. Grandpa I mean, you needed 30-years to wake up all these robots doesn't sound it went all that seamlessly, did it? I would say, yes, Keira, you hit a sore point because not everything worked that well, especially in the beginning. We knew how to build acting machines. We had 150 of engineering of machines under our belt. This is what we knew. The connectivity part we had solved in the 1990s and 2000s, largely, also not a big miracle. Well, we made our mistakes that was on the thinking part and we needed to catch up that was on the sensing part. And Keira says, "Grandpa, what are the mistakes that you made in the thinking?" Well, Keira, let me briefly show you what happened in 2016. When people, for the first time, understood the power of AI and machine learning, everyone was freaking out. That was really the gold digger spirit, AI everywhere. If you have a problem, AI will solve it to you -- for you. Everything is seamless, is easy because AI is taking over the world. And the biggest newspapers by this time, 2016, we're hammering out articles like car industry is changing. In 4 years from now, most of the vehicles are autonomously driven. Journalists were asking me, "Hey, Lars, when are your kids driving autonomously to the kindergarten in a car without a steering wheel?" And 4 years later, the same newspapers were sending out articles like, "The techies have overpromised and underdelivered." Let's get sober. Autonomous cars only come in a couple of decades. It's not all that easy. It's a mistake. It doesn't work that way. So big depression in the market. Four years. So we got it totally wrong in the first step. And Keira is asking, "Hey, Grandpa. What was the reason?" Well, Keira, we made a mistake. We thought it is enough that an AI system is just looking over our shoulder, is training how we drive. You hand over the key, and it's all good and it's all easy how to continue. We should have known better, because also for your father, Keira, my son, Jonas, I could have worked and saved a lot of money if this concept would have been true. Because what I would have done at Jonas' 18th birthday, I would have said, Jonas, you are an intelligent system. You have seen now 18 years how daddy is driving. Here's the key to my sports car, enjoy the day. This is precisely how it did not work. I spent tons of money for Jonas to make his driving license. He had to pass a deterministic driving test, theoretical and practical, test on the road, and only then he was allowed to drive. How come we came to the conclusion that an AI system doesn't need to pass a driving license and we just can't throw over a key and say, yes, you have seen how some somehow drives, get going. There was a fundamental mistake in the overall setup because we did not put trust, safety and security into the system, and we had to rework that entire ecosystem. So Grandpa, how did you solve it? Well, Keira, this is what I earlier meant with a brain shift. The brain shift came in or the need that we found that we have to reengineer the brains before we put them into robots, and that was quite a challenge. But what do techies normally do? We are looking around and say, okay, where exactly can we copy a solution and move it into the market? So how do we find blueprints for these robots and how do we continue from there? Well, Keira, and the closest example, look at me, I'm an 80-year-old biological robot. And I have a very simple architecture. I have 3 parts of my robot brain. The orange part is my wiring harness, my data transformation network. And this data transformation network has reflexes. The responsibility for acting on reflexes is in the orange part. This is not an intelligent system, but it is ultrafast and ultrareliable. And if a robot stumbles here on stage, this reflex system just says straighten your legs. It does not know that I fell over, but straighten your legs and stabilize yourself. The green part is decoupled from that. And it's in tech language, a highly functional safe system, real-time system, and that is caring for heartbeat, temperature control, stability control, yes, complicated tasks in motions. All of these type of things, and I can operate all of these tasks. While I'm standing on a stage talking to 1,000 people and thinking of the next sentence that the blue part, the non-real-time part but the creative part should crank out. And you see it already here on that concept, Keira. The blue part is big and very energy-hungry. But for transportation and motion, you largely need orange and green and rule-based systems. Otherwise, insects could not move. So an ant is doing transportation tasks and mosquito is not flying straight into the wall. Only on orange and green. They don't have blue. They have simple trapping algorithms and you can trap insects pretty easily because they can get stuck in simple algorithms and don't find enough creative solutions to climb out of a trap. Again, there you need the bluish part. But for us, human beings, that bluish part is often needed for playing piano, for writing love letters. And for most of the robots, you need the capability to write love letters. You want functional safe transportation. So in other words, the high magic of creating robots is largely in orange and in green. And of course, what we want to do is we want to have then the scalability again for all the different form factors. Building platforms that can address this area. And Keira, this is what we tried to do in the early 2020s. Like in your human body, you have a blood circuit, your energy provision system. For a robot, this is your electron provisioning system, your power management, and you need to make sure that you never get a congestion, that always the electrons are supplied. And why do you need to make sure that this happens? Because otherwise, you get a stroke or a heart attack or whatever. So for your robots, this green part, functional safe power management ICs. Functional safe power supply and efficient power supply is the absolute foundation for that robot building and we brought out a complete platform for functional safe power management ICs in the early '20s. And you have your neural system because otherwise, you cannot talk to your fingers and your hands of your robot. And there is very simple neural information transportation systems. [indiscernible], for example. There is for industrial robots and for transportation robots more and more Ethernet-based systems, time-sensitive networking, so real-time networking systems coming over all different data rates, from 10 megabit per second to 10 gigabit per second. We needed data switches. We needed physical layers, but a bunch of LEGO blocks also there, a bunch of components to build these robots. Then, of course, the heavy lifting came. We introduced in our company in the early 2020s, so precisely 2024, a platform that we call the CoreRide platform. So a family of similar products from small microcontrollers with embedded memory, in those days, 14-nanometer technology, to 5-nanometer microprocessor units, customizable IP, really heavy-duty, but all highly functional safe electronics that can do the task of this surprise-free, real-time execution of big amounts of data. And as they came all from the same family offspring, we could start to leverage one software platform. So one way of programming this hardware. And just to make it very clear, what is a robot? A robot is a machine, just a piece of hardware, but it is a piece of customizable hardware. And how do you customize your hardware? You program it. And this is the reason for existence for software in this entire machinery discussion. So a software-defined robot is software-defined because it has to be flexible. So in other words, if you have these building blocks, Keira, you can start mapping it back on to the brain. And now you say, okay, there is the network information dispatching layer. There is the orange, the green part, and, Grandpa, for the blue part, we need AI accelerators. Yes, Keira, and they had been already in the early 2020s, somewhere between 50 and 200 companies out there that did AI accelerators that we could use. And Keira, in 2023, I saw the first time that customers were taking these architectures and by bringing them into the market. For example, a Chinese company, Leapmotor, you see it exactly the orange and green part on one PCB in the bottom part, the blue part on an AI-enabled chip in the upper part. We started experimenting already in 2019 with guys like Continental on what we called a blue box. You see this reference kit from NXP there. The S32G, so a big gateway function. Let's say, gateway is the green part of the brain. [ Four ] PCI Express slots for AI accelerators that can land on the unit and bring your own AI accelerator, plug it in and you can expand your brain as much as you want into the bluish part. And then, also in 2024, what we saw is companies from Australia for the first time getting secured -- sorry, safety certifications on these type of architectures. So Applied EV is a design company. They are building these rolling skateboards, functional safety certified on exactly that type of architecture, functional safe green part and orange part of the brain and then enhance in a less functional safe setup the AI accelerators on top. And this is not the only 3 that we saw there, but it was taken over very, very quickly by a lot of the customers in the market. It was just the zeitgeist of that time, it was the most logical thing to do, Keira. And Keira said, "Wow. Great. Grandpa, thanks. So that at least explains a certain part of that delay. And with that brain reengineering, you solved it all." No. Keira, to be very honest, we had even a bigger challenge in front of us. And that was the challenge that these robots were inferior to humans. For example, in '23, we had already self-driving cars, level 3 cars, so with a highway assistance system out there on the roads in Germany: the Mercedes S-Class and the BMW 7 Series. The only issue was these cars could only drive up to 60 kilometers an hour on a highway. And no one, hardly anyone, would buy a system that can only drive 60 kilometers on a highway. I mean -- and you go manual driving and you speed again, especially we didn't have speed limits by then on the German highways. So Grandpa could go 250 kilometers an hour on that highway. So that was too fun, 4x the speed of these autonomous systems. "So what was the reason? I mean the brains were there, Grandpa? And what were the reasons why these cars couldn't drive faster?" Well, Keira, the issue was these cars were more shortsighted than Grandpa himself. So these cars needed better eyes. These cars needed better sensors. Keira says, "Okay, I understand you, Grandpa. But what was your vision? How did you try to solve it?" Well, Keira, I went back to my childhood dreams. And I was looking at my idols. And these idols were superheroes. And like the guys in Hogwarts, I was dreaming of having a magic wand, make one funny move and the spell, and objects could be influenced in my environment. Like the [ Jedis ], I wanted to have telepathic capabilities. I wanted to know what goods are doing outside of my line of sight. Like Superman, I wanted to have X-ray vision. Looking across a wall, looking into rain, fog and snow where my eyes could not help me anymore. I wanted to understand what is in front of me. And like Douglas Adams, in Hitchhiker's Guide to the Galaxy road, [indiscernible] was using a thing called Babel fish. And you could stick it into his ear and then suddenly, he could understand all languages of the universe that were spoken around him. With these 4 super sensors, superpowers, my robots would be unbeatable. So we started working on that, Keira. For influencing the objects in your proximity, we use the thing that we call ultrawide band. My car key moved into my watch. When I was approaching my car, the car opened up without me taking any action, but just by being there, by measuring the [indiscernible]. And you have this today in your house, you move in your light, your sound, your temperature, it's following you. All the door unlock and lock is done by your superpower, your magic wand at your wrist. Knowing what is happening 10 cars ahead of you in real time, we do buy a V2X car-to-car communication, cellular or DSRC-based that was a battle in the 2010 years, yes? But there was a lot of discussion. If these cars, 10 cars ahead of me is doing an emergency braking, I would know it in my car 2 milliseconds later, and I can react before I see the brake lights of the car in front of me. So that was really, really important for reaction, for safety, for arranging the world around me in a proper way. We are working on high-resolution radar. We try to bend the rules of physics as much as possible there. If radar would be as good as laser scanners -- we have laser scanners in the early 2020s. If radar would be as good as my eyes or as laser scanners, how cool would that be because radar is working in rain and fog. I have my X-ray vision. And we grouped together the big industry players of the 2020s and we defined a standard, a common language matter so that all smart connected devices have one way to talk to each other. Or in other words, what we did is in physics for this high-resolution radar, let me just give you 2 examples, Keira. We took 3 radar, sensors off-the-shelf radar sensors, and we synchronized them and we synchronize them via Ethernet. And if you can make sure that this standard, cheap off-the-shelf radar sensors are sending and receiving at the same phase, so coherently, then what I can build is out of 3, 10 centimeters antennas, I can build a virtual 2-meter antenna. And in radar, antenna size is a resolution. So for the first time with a partner company and its P could show at CES 2024 that we can measure 2 of you walking in front of me over a distance of up to 250 meters. If I could manage obstacles, obstacle detection over 250 meters, then I can afford driving speed of more than 130 kilometers an hour on a highway. So Keira, this is how we reached the inflection point of highway assistance systems with the right brains are in the cars and now you all -- we all could buy them because it made sense for us. We had highway Level 3 driving systems for 130 kilometers an hour on the highways and that was the speed of these days. Second example, this meta standard that I was talking about. Keira, what we did there is we had connected homes in 2020. But my father and my mother, Keira, they stopped buying smart home equipment, connected home equipment because they were afraid. Whatever they buy, they can throw away the year thereafter. They had sound boxes that did not talk to the energy management system. The energy management system did not understand the window blinders, the window blinders did not know where the electric vehicle is so we had connected homes, but we did not have smart homes. And by this language unification, suddenly use cases came into play, but the solar cells on Grandpa's house were producing energy. And then your arms, [ Leah ], was switching on the hair dryer. And the energy management system in the house knew this is a hairdryer that needs 2,000 watts for 5 minutes. So switch off the fridge. The fridge is keeping the temperature for 8 hours, switch off the washing machine for 5 minutes. Let the hairdryer run on the solar energy and afterwards, you're going back to normal operations for all the other energy users. And of course, it was not only a language that we introduced, but it was also a passport that we introduced because you didn't want to have crazy equipment in your house. The washing machine had to have a passport and had to say, hey, believe me, I'm the washing machine. The hairdryer, I had to say, and here's my passport, I'm the hairdryer. And the solar cells or the energy management system had to say, and you know what, guys, I have authority above you all because I'm the energy management system. So these entire security certifications, passports and common language, they helped us generate this ecosystem. And what we at NXP did for that in 2023, we brought the first meta-certified controllers out there next to the [indiscernible] microcontroller and microprocessor family that we are using a lot of industrial devices and in a lot of display-centric devices. We had the first radios because this meta standard was living on ZigBee, Bluetooth and WiFi transmission standards. So we did not invent new transmission standards, but the common language. We did not invent new phone lines, but we decided that we talk in one language, in English, over those phone lines and we could understand each other. And this is how this meta came into life. So Keira, what Grandpa did over the last 50 years, largely not over the last 10 years because I was retired already, but in the 40 years before. Grandpa just worked on one simple task. Grandpa worked on converting machines into robots. And it took me and the industry 40 years. And this architectural transition was the first and most important part. So the way how machines were built had to be completely redone. We had to invent robot brains, trustworthy, safe and real-time parts of brains. Then we had to give these robots superior sensors. We had somehow to manage to build in trust. And we had to find the ability to scale that in an energy-efficient way because also one of the problems in the early days were that all of these systems could easily consume thousands of watts, highly inefficient. Think of it like there would be human beings out there with a 1-meter skull and this brain would be so big and so energy-hungry that our bodies cannot carry it. Evolution has done this optimization part for biology. And I think we have done a pretty decent job doing that parts for robots only in 50 years. But Keira, very honestly, there was only one thing, and that one thing that you could learn is we had to switch from one company does it all and standalone like in the machine building days to we have to do it together. There is no such thing like one big cable is doing everything under one roof. It was an ecosystem play. It was a play where we needed the industry players, where we needed academia from research and where we needed a lot of help from politicians also globally because also one nation could not do it standalone. This was like 100 years earlier flying to the moon. This is hardly payable by one nation. And that, Keira, was maybe the most important learning for me in the last 50 years, and it will be the most important learning for you going forward. What you can dream, you can build. If you can't dream it, don't even try it. And for all of you here in the audience, maybe I'm just a dreamer, but I'm not the only one. And if you join me on that journey, our industry will be a lot of fun. Thank you.