Monolithic Power Systems, Inc. (MPWR) Earnings Call Transcript & Summary

All right. Good morning. If I could ask for people to start finding their seats, we'll begin here shortly. All right. First of all, welcome to Monolithic Power's 2025 Investor Day. For both people in the room and online, it's going to be a great day. For those of you who don't know me, my name is Tony Balow, and I'll be helping MC the event today. First, I have a few housekeeping things to go through. First, please take note of our safe harbor statement. A full list of risks and uncertainties can be found on our website. Second, if I could ask you all to please silence your cell phones for the event. And maybe most importantly, the bathrooms are through the door path registration and down the hallway. All right. Finally, let me just cover the agenda for today and what we'll be going through. First, we'll start with Michael, who will give some words of welcome. And then what we'll do is move into a series of presentations from our product line managers, who will talk about what they're doing in the world of power and other things in data center, in automotive, robotics and more. At that point, we'll take a short break, at which point, Michael will come back on stage and talk about some of the innovation that MPS is planning for the future. Bernie will come up and update the financial model, and then we'll go through some Q&A. After Q&A, I'd invite everybody to grab some food and join us in our demo area, where we have a number of stations that showcase MPS innovation and technology. With that, we'll start with a short video, and then we'll have Michael. [Presentation]

Well, first of all, thanks for coming and you will not regret it when you're coming. For those people that didn't come over here, they will be regretted, okay? And it's all about -- you know all these numbers. And -- but I still have to remind you some of the numbers -- give you a summary. And most importantly, we're going to have a lot of fun. And in the last few days, maybe you think of artificial intelligence. We're talking about real intelligence here. And that's what you're going to see. And it's all about what's coming and how we can improve our life and how we can do better, a lot of things. What today is tend to be -- okay, I want to intend to be short, very short talking. You see the real stuff, okay? And let's go through this quickly. First, here is -- well, let me -- 10, 12 different things goes into my minds. I don't know which one will come first. Maybe, instead of I do that, before I forget, this last meeting was 2018, just 7 years ago. There's a lot of factors causing us, do the 7 years later, pandemic was one of them. And hopefully, next time, 7 years, without not going to do 7 years. We have more exciting things to show you sooner. And so today, we want to -- I want to emphasize that, okay, what we did in the 2018 in the back of the room here. We are also here. And what we said, what we presented. And there's a lot of impressive, I have to say myself, okay, we have a lot of impressive pages. And for probably most of you don't have it, okay? And -- but we will remind you, a couple of pages, okay? And some of the stuff we didn't really good, really successful. And we also talk about the things that are not very accessible, okay? And I have to be fair. Well, this one, you know the number. 25% CAGR. I didn't hear anything. I guess is fair. I guess this is fair, the 24%, you didn't it, okay? And if you look at the -- this is the 21st year. It's 7 years from 2018. If you look back from IPO days, that number, that's the reason you didn't product, was even higher. So 3x 7 years, yes, we slowed down a little bit, okay. All right. And we took 1 million shares, about -- roughly about 1 million shares off the table and increased our dividend, we started with like $0.25 or $0.30. I don't remember exactly. now it's over $1.50. And total since 2018 is $1.5 billion returned to our shareholders. All right. So since then, we grow and this is 2018. Now we have something over 32 or 34 offices all over the world. And the R&D centers, the purple ones, we grew the R&D center. And last, the manufacturing. Before we are a lot more concentrated and look now, we grew out of one place. And so MPS keep evolving. We have a global footprint. And we're manufacturing in -- we have manufacturing partners outside of China. And also more importantly, MPS doesn't make anything. We can portable to wherever we can find a manufacturer. MPS is the key, all the things we do here is our knowledge. And we don't manufacture anything. We're making -- using our knowledge and making money. We do a lot of products and have a lot of funds and it makes a lot of money. So do you guys. And if you own the share for 7 years. And okay, next one, since then, we have 30,000 customers. And I'll talk about it later, okay, 30,000 is [indiscernible], I will cover that later. And this is geographically, we're very even. All right. So what is MPS growth drivers? Well, you need a lot of people have -- want to talking about artificial intelligence power. This is all the area, it's all our baby. We are equally -- we're not biased to anything. Even AI, we talk about it. We didn't know it, where it's coming. As long as we do everything the best. Some of the stuff listed here may be the hardest things for next year. We don't know. As long as we do the best, we have a chance. So -- and that's the segment where we're involved. And now look at this busy page. We have over 4,000 products. And this is not intent for you to remember. But this is intention. We talk about only 50% to date. These are the product -- but the other product, I said equally important. And so now the segue is -- we -- I let the rest of the people talking about this 50% of our business, how we -- how the sausage is cooked. All right. Okay. Jinghai?

Thank you, Michael. Welcome, everyone. It's so great to see you all. And my name is Jinghai, and I'm in charge of computing power for Monolithic Power Systems. I might know some of you guys in the past and the past and the past that past Analyst Day. I still remember in 2013, when I have the first Analyst Day presentation, I think the crowd is maybe 20% of this and look at this crowd. So welcome. Today, I'm going to take all of you through the journey of power -- computing power evolutions. The computing power has been a challenge for the industry through the years -- throughout the years. As you guys have seen in the GTC the last couple of days, one is the GPU and the other topic is power, how to solve the power problem, and that's who we are. So let's travel back to 2018 when I presented the story and we predicted what's going to happen in the next several years. And let's go. Everyone remember this page. This is the slide I presented in 2018 right on this stage. And back then, we are seeing something changing. There's transforming of power industry that we kind of feel it. We can feel that the power eventually is going to go to a point that there's no solution actually back then going to solve the issues, the problems that the computing, the GPUs and the CPUs are facing. So Michael used to call this the Jinghai's Law. So what is Jinghai's Law? Let me explain it. So we predicted, we predicted. Every 2 to 3 years, the processor, especially the AI process, GPUs, CPUs, the power level is going to increase by 100%, basically double the power level. Imagine when the GPU has more power, they're going to take more space on the board. And when that happens, we're left for power. We're left for us, we're shrinking, right? The space available for power is shrinking. And that follows the same law that every 2 to 3 years, the space available for us is half. And that's a big problem. To solve that problem, every power suppliers, vendors, IC manufacturer like us, we all scratch our head, try to see what is the best solution we can offer to the market. So of course, pushing the frequency to reduce the size of our converters itself and more importantly, how can we change the architecture of the power delivery. So that was the next slide. We actually predict this before -- even before 2018. I remember I presented this in APAC 2016 or '17, and we're proposing let's do multiple stage to solve this problem. And the 12-volt power delivery is not going to support enough juice to the CPU. Let's do 48-volt. And the 48-volt opened up a new area for us to invest, and this is our DCX converters back then. And the structure change really changed the whole power delivery industry and look at the GPUs in the market right now. It's all this stage, 2 stage. Another interesting slide we presented 2018. Back then, a lot of people are focusing on CPUs. And everyone knows Grantley, Purley, Whitley, actually, in this room, there's investor friends just asking me, "Hey, MPS, are you familiar with Whitley? What's after Whitley? So it's all about those buzzwords. My take on this is, that's great, but let's look at the different dimension. This is a dimension I was -- we were focusing on, not just focusing on CPUs. The data center is going to transform to AI data center. At some point, all the other processes, the GPUs, I always make analogy. CPU is more like managers or directors in the company. A GPU is all the workers. We need the workers to do all the heavy lifting job. And then we focus on GPU, then at the GPU, we know there's Ampere, there's Hopper, there's Blackwell. So that's the road map we're focusing on. And then we also look at the ARM processors and the tensors, the TPUs. Now you imagine back in 2018, there's a huge map in my brain, not just this dimension, but it's 2 dimensional. Every category has road maps and MPS is heavily investing on that. That's how we be so successful after 5, 6 years. Another nice picture. Everyone remember this picture. That's the first SXM card NVIDIA announced. And you can see that the power is almost half of the size. And that's a GPU, that's MPS. So we look at this picture, and we immediately realized we're going to do something, because if we continue to follow the same power solution approach, it's not going to fit. Again, back to Jinghai's law, you're not going to have enough room once that middle piece grow by double the size. So look, we have -- that's the scorecard. And I was surprised. I was surprised. Back then it was like $24 million, $25 million, and we were just touching the surface. And all of a sudden, in 2021, it doubled and 2024 is a big jump. And I look back, try to figure out what did I do right? What did we do right to make this happen. And I think there's -- I always use this 3-D methodology. 3 D means define the ecosystem, design the right part and deliver the part. So those 3 D is so important. The 3 pillars are always critical to us. We're not just focusing on design the part, but we want to focus on define, develop the ecosystem with our customers, try to solve their problems. I think by solving our customers' problems, and then we design, execute we deliver. That's how we make it successful. So we always had 6 months to a year ahead of everyone else, try to solve the problem. So now let's zoom out. Let's zoom out. We did this architecture change 48-volt solution and also the monolithic design, as always. And right now, at this stage, MPS is focusing on further optimize, improve our Intelli-Phase design. That's the key to the success. Then we also introduced our ZPD technology for the next generations of power -- computing power. But then what's next? Everyone has the answer because the data center is going to grow, direct power is going to -- I wouldn't even say grow, going to explore, right, from 100 kilowatts to 600 kilowatts. So with all those power, imagine how you handle all the power flow into the rack and then how we deliver directly to the processor. So that's where we're going to focus. This chart looks like Blue Angels taking off. So everyone look at this chart, 2 years ago, I was still thinking, yes, with NVIDIA, AMD GPU beyond the 1,000 watts. Is that even possible? and how can we handle this and that. But all of a sudden, without even noticing we're at 2,000. And we're even beyond that. So again, it's very hard for us to imagine what's going to happen. But then the power just keep growing up. And what that means, it will drive this power growth will drive the fundamental VR solution and the density. And also, we will drive this to a different structure, which I will talk about later. So this is what we do now. MPS do all the power solutions to power the core directly. And this solution, whether you want to do a discrete design with Intelli-Phase with our digital controller or you want to integrate everything together, make a module. And furthermore, since you do the module, nobody wants to put capacity -- or there's no room for the caps underneath the processor and we integrate all the caps into the module. And this is what we do now, and this is what we are providing also at this stage to provide the 48- to 12-volt conversion, and this is really our focus. Our focus is to take 800 volts coming to the rack and how we convert it down to the next level. And that will solve a big problem, I would tell you in the later slides. Now look at the modules. Back then in 2018, we just announced this beautiful module at the bottom. And back then, it was, wow, how can we do that? It was so small, everything integrated. But within a year or 2, it follows Jinghai's law again, we doubled the density. So within 2 years. And we didn't stop there, because we understand the need for the vertical power delivery, and we know the height of the module need to shrink, and we also understand the space of the capacitor is gone occupied by those VRs, so we need to integrate everything in there. So we jumped from 0.75 amps per millimeter square to 1.5 amps millimeter square in a few years. And then -- we didn't stop there. So I'm very excited to see this baby to come to earth and it's 3 amps per millimeter square, very small design and everything is integrated, everything integrated. So we're going to announce that in a few months. And by the next time you guys come in for next year's Investor Day, and this is going to be right here. So what drives this? How can we convert that into this such a small device. Here's the answer. MPS founding technology is monolithic. So by doing monolithic, we integrate all the pieces, all the converting the power, the control, the brain, the muscles everything into one die. Look at that. Isn't that beautiful? It's so simple. Simple is the best. When you put everything in one die and you manage to communicate in between those silicons, and you minimize so much hassle in the design. And you don't have to worry about interconnections. You don't have to worry about the noise coupling and just seems very elegant. And this is a discrete solution going to be look like. So monolithic is the best way to do power integration to improve the density. And we didn't stop there. With all those great modules, what really help us is now we can move the power stage laterally resides on side of the processor. With huge PDN losses, 150 microns, all the current has to go through that to the processor. Boom, we move to the back. And it's not a brand-new concept. But the way how we implement it is very novel because our power stage is so small that we can even have space for the caps and also the PDN can be dramatically reduced by 5x. Again, we didn't stop there. So the next move is can we put more process more power stage and controller all into one block, very thing compact and including all those capacitors. Now it will be very simple. You go there and you take over the whole space at the bottom, and there's no caps, the cap layer is right here and use the BGA and we call the direct power ball mapping. So all the power, the plus/minus go straight to the GPU, CPU power pings and there's no lateral power flow whatsoever. So that's provide us the superior PDN performance. What's next? Let's just pause here. Try to imagine what does it even mean a 600-kilowatt rack. A typical rack is the size of my height, a little bit further than me, then it's right there, and that rack going to take in 600-kilowatt and burn it. The only output of that is heat nothing else. 600-kilowatt, if I do the math correctly, that means -- all right, a typical household oven it's about 1,500 watts. And that means you need over 150 ovens in that rack. It's all burning at the same time. Imagine that. And MPS, we need to provide a solution to take the power in and deliver it to each individual GPUs, 144 of them, and with the precision, accuracy and with speed. That's the mission we're taking. Are we ready? Is the industry ready to do that? All right. So this is the similar slide we show. I can tell you today, yes, we are ready to take the challenge. We're -- we are set to deliver a product in the next few quarters. And that part will convert a plus/minus 400 volts to the 48-volt or 50-volt given or take with 6-kilowatt and above power, imagine that power density. And this is what we're doing today. With that, let's recap. It's a $4 billion SAM industry and still growing by this latest power shift from 100-kilowatt per rack to 600-kilowatt even higher. So there's more opportunity there. And today, we're providing all the powers for our CPU or server board, our GPU cards pretty much everywhere but we didn't stop there. Our next mission is how can we deliver the rack power from the input -- high-voltage input to this. And beyond that, what can -- what else can MPS help for the data center. I don't have a clear picture for the data center -- entire data center power right now. But like the short video said earlier, MPS, we are powering possibilities. So we don't want this possibility. There's a lot of possibilities here. And the data center may not look like this in the next 5 years. And that's what we are here to support them. Again, it's all about power, power and the power. That's my journey of computing power evolution. I hope you guys enjoyed it. Thank you, everyone.

Okay. Good morning, everybody. I'm Allen Chen. I have the good fortune to lead automotive here at MPS. It is fitting actually that automotive is going after AI and data center because, in fact, Jinghai's law, the famous law that you guys all now know, actually applies to automotive as well. The thing is, it happens in automotive about 3 to 4 years after AI data center. So the one thing about MPS is Jinghai and his team, they develop these incredible technologies, these solutions. They deploy them, they deliver them at scale. He does his 3 Ds. And then 3 years later, we had the fourth D. We divert them to automotive. So it's happening right now for the compute power, and it's also happening in this transition from 12-volt to 48 volts. We're going to talk about that today. But I think we're going to cover 3 things in the next 15 minutes. I'm going to go back and reflect back on what I told you was going to happen in our business back in 2018 and how we did. We're going to take a look at how our automotive business looks like today. And then finally, we're going to look to the future. Some of the major trends shaping the future of mobility and where MPS plays a critical role in this space. So let's start with what I told you was going to happen in 2018. I said 3 major things. The first thing was that we were going to grow our business beyond the $53 million that we had done that previous year with a nice beautiful pipeline of design wins. The second thing was that we're going to do that by growing our customer base as well, pushing into the upper echelons of the global Tier 1 community. The third thing I told you was that we were going to diversify our business across several major application categories. So let's take a look at how we did. On the first one, I'm very, very proud to show this. Over a decade of consistent growth, 44% CAGR since 2013, about 34% since 2018. Compare this to the overall automotive semiconductor market, which was about 1/4 of this overall growth rate. So this really is actually echoing the overall MPS story. People find out about us for the first time. They learn about a technology, they try it out, they love it and they come back for more. And this is that MPS story in the automotive world. But really, this is still just the beginning. The second point about growing our customer base. We've tripled our customer base now from just about 50 customers to almost 150 over the last 7 years. And this is a mix of the major Tier 1 suppliers. It's the automakers directly, and it's even some of the new EV makers. In fact, we have some of our cherished customers here. They're going to be showing you some cars afterwards, really, really cool technology out there. On the final thing, so growing our business across 5 major applications. So you hear Michael, you hear Bernie talking about ADAS wins, digital cockpit wins in our earnings calls. What is this? In ADAS, we power millions upon millions of cameras, now radar as well. And of course, we are the power solution for many of the ADAS SoCs. Today, it's domain controllers, tomorrow it's going to be more integrated. Digital cockpits, you can think about USB charging, now it's all Type C, very high power, but most importantly, infotainment systems. So about 1 in 3 or 1 in 4 infotainment systems among some of the top automakers in the world is all powered by MPS. Finally, lighting is anything that is dynamic. It has these cool animations. You're going to see one in the back after this. Motor control is body control. If it has a motor, we can drive it. And then the final one, electrification. This is actually a new growth category for us. This is not a material portion of our current business, but we have a lot of wonderful future activities happening there. Okay. So that's a -- that's a look back at how we did versus what we told you in 2018, we basically deliver on every single one of our commitments. So here is where MPS stands today. Annually, we're delivering over 1 billion devices, automotive grade into the market. This took us first 10 years of our history, we did 1 billion in total. Now we're shipping that on an annual basis, and this number is going to continue to grow. We're also now recognized as what they call a preferred supplier amongst many of the top global OEMs they recognize us for the technology, for our ability to move very fast. And in some cases, we save their buts back in 2022. So recognized as a preferred supplier. That means that when they go to Tier 1s, and they say, hey, choose power for your solutions, we're on that list of the top preferred. And then finally, we've built out a portfolio We now have over 1,200 solutions across all of this to offer end-to-end solutions for customers. So if you're doing an infotainment system, if you do an ADAS domain controller, it can be all MPS. We built a very, very nice situation for today. So let's take a look at what's coming next. So basically, all the automakers are talking about software-defined vehicles, SDV, right? This really is just the latest kind of way to call what we previously in 2018 were referring to as autonomous, connected, electrified mobility. Today, they call SDV. I'll give you an example. I actually just bought a car a couple of months ago that is sort of this first generation of SDV. Just last week, I got in my car to come to the office, and I get this message on the center screen. It said, hey, congratulations and surprise, last night, we installed an update. Today, your audio system just improved dramatically and you get Level 3 hands-free driving on the freeway. Think about that. Your car doesn't just get worse as soon as you drive it off a lot, it actually evolves. It gets better. It enables really cool features over time. That's the promise of SDV. We're just now at the beginning stages of this evolution. But there's a couple of things that have to happen in order for this to really thrive the way automakers have in their goals. 5 major trends. So let's talk about it. On the very first car compute. So think about the 100-plus ECUs that today exists throughout your vehicle, that's all consolidating into one massive computation box. The ADAS and IVI functionality, that's combining into one system. In some cases, even one chip, we call it the Super chip. And at the same time, of course, the power demands, Jinghai's law again, is doubling to 10x in some cases from one generation to the next. So the need for power is ever present and even more urgent for this application. And at the same time, the landscape is crazy crowded. It's super crowded. You've got, of course, Silicon Valley titans. You've got regional upstarts. And even some automakers are saying, "Hey, I want to control my silicon destiny. So I'm going to do this myself." And the common thread across all 3 is that MPS is engaged with all of them. I'll give you an example of why. So DrMOS, remember that monolithic image that Jinghai showed you guys. Same exact thing showed up for automotive a couple of years ago. MPS was the first to release this for automotive and to deploy this at scale. And today, millions of cars on the road run, thanks to MPS DrMOS. With our monolithic technology, we can shrink it, have higher power performance and so on. And like some of our competition is now finally catching up to our first generation. We're like 2 to 3 generations ahead. We're actually moving on to vertical power. So talk about having to cram more stuff into the same size. This is what's going to enable a lot of these guys to achieve those goals. But it doesn't just stop there. DrMOS is the gateway. That's the first thing. We've now crafted an entire solution. It's 2 dozen kind of devices that work in concert. It's DrMOS, it's safety monitoring supervision kind of functionality into an overall subsystem. And this is what folks can use to quickly enable their power strategy for a lot of tomorrow's car compute. One of our partners actually did an assessment with us once. They said, what would be the equivalent cost? If I tried to do this myself and build this from scratch, we saved them something north of $200 million when they leverage this sort of solution that we've carefully crafted over 8 years. Okay. The next thing, ADAS sensor, the proliferation of sensors are throughout the vehicle, both outside and inside. Level 3 cars and robotaxis are here. And I mean that literally, we've got 3 of them sitting outside. You're going to see them after this. But in order to achieve that level of self-driving capability, hands-free driving, most automakers are saying, you know what, I need to rely on a multimodal strategy. What does that mean? It means camera, radar, LiDAR, ultrasonic, maybe thermal imaging in some cases. LiDAR, in particular for me is very interesting these days because even just a couple of years ago, you guys probably know this. It was perceived as way too expensive, a ridiculous option, not even remotely considered for cars but the cost for this have dramatically fallen. In fact, some LiDAR sensors today are on par or even cheaper than airbag sensors. So if you have this option, you absolutely can consider adopting this. We've got a great LiDAR technology demo next door from one of the top makers in the world. So what's our proposition for ADAS sensor? So it's this, using our latest generation process technology, we're able to offer solutions that are 1/4 the size of the nearest competitor. One of the biggest barriers for software-defined vehicles ironically is software. That's actually the hardest piece for a lot of the automakers. So anything that we can do to help cut down on that development expense, development time is actually a huge benefit for these guys. With these kind of solutions, we can enable something called artifact free sensing. We have very, very low noise, so they don't have to use software to additionally filter that out. Okay, the digital cockpit. So your in-cabin experience. It's very simple. It's about more, more speakers, more motors, more screens, more bespoke of an experience for the passengers as they're on their journey. But it's -- the car is not actually getting bigger, right? The cabin doesn't get any bigger. The space that you have to cram all these additional electronics that doesn't get any bigger. So how do they achieve these kind of goals and enable more technology inside the vehicle while maintaining the same size. Well, here is our solution for this. So this actually leverages two things. One is, again, that process technology, enabling us to build more with less, but the second is our latest generation of power conversion algorithms, which fundamentally enables you to deliver very fast performance but dramatically reduce your external components. So this is the way that our automaker customers can dramatically reduce their size, cram more into the same form factor, simply do more with less. Oh, and that's not it. Song as well after this is going to be showing you guys a whole new category of products that today we don't participate which is audio in the vehicle. We have some really fantastic, you're going to hear it, literally hear it next door afterwards. And that's an entire something like $600 million market that MPS does not today participate in. Okay, two more things. First, high voltage. So contrary to what you're reading in the news, EVs are not dead, okay. They're actually not slowing down. They're still growing. They're simply growing different rates in different regions. But EVs for sure are still continuing to gain adoption. The trend here is also the move to higher voltages. We're going from 400-volt architectures now pushing into 800-volt architectures. This enables faster charging types. This enables higher performance. And what's also happening is that automakers now are actually doing vertical integration. So they're actually developing a lot of these subsystems themselves because this is how you differentiate the car. This is how you say, "Hey, what used to be a Supercharge V6. Now it's my traction inverter. It's even my onboard charger. So the name of the game here is system solutions because they're not power experts necessarily, certainly not power conversion experts. So we come in, we say, here's our entire module software is preprogrammed in here, go ahead and use it, they can quickly deploy, get to market faster. Another example of this, we'll be sampling this later this year, is our latest automotive BMS solution. We have embedded algorithms. This works in concert with a series of these devices. It dramatically reduces the software overhead otherwise needed to do your battery management solution -- and with this kind of solution, you can actually increase the EV range. And then Chris, actually, after this, is going to be speaking of some more of the battery management technologies. Finally, 48 volts. There's one thing I want you to take away, it's that 48 volts is not a question of if it is merely now a challenge of when it is going to happen the 12-volt power architecture that exists today. That's a dinosaur, okay? That came out in the 1950s, and it survived through 70-something years of automotive development. Think about what a car in the 1950s had electronics wise. And think about what a car of today is, right? The physical wire harnessing of vehicles is insufficient. It's actually impossible to power most of the demand that these guys have. They have to go to the higher voltage, because by going a higher voltage, you go and shrink your cabling, a dramatic reduction in the weight and the -- when you couple this with Zonal, a much smaller, simpler architecture, this absolutely has to happen. And depending on the automaker, you either hear them publicly saying they're going to do this or quietly behind the scenes, I think they're scheming to enable this. But the thing I ask you also with 48 volts is once people finally realize they're ready to do it, and they absolutely must do this. They need to get to market in a very short amount of time. So they have to have modules system kind of solutions, and that's what we're offering. You'll see this next door. This is a drop in. It's about the size of this little clicker that is the entire power stage to take you from 48 volts down to a low voltage domain. Three of these guys is sufficient to power the entire automotive electronics demand for the car. It's a very, very nice solution; quick, quick drop in; very little software development overhead. So what's the through line across all 5 of these applications. And every single one of these instances, the need for power is increasing significantly, sometimes 2x, sometimes 10x depending on the system. But it's not just power that converts or distributes from point A to point B anymore. This power needs to be safe. It needs to be smart. It needs to be scalable, and most importantly, it needs to be systems-oriented, total systems oriented. So look, I'm really proud and humbled by what our team has accomplished. We've grown eightfold in 7 years to over $400 million. But we look at the opportunity in front of us. It's from the power content expansion. It's from these additional opportunities these kind of products that we never even participated in. We're looking at a SAM that's over $10 billion. So truly, for automotive, this is still just the beginning of our journey. Thank you.

Hello. Good morning, ladies and gentlemen. I'm Song. I manage Standard Analog frontline. Today, I want to tell you some very amazing MPS products and technology and how they enable the next generation of intelligent robot. Automation is a big market, a multibillion-dollar market and with different segments. By the way, MPS plays in all these segments. We offer microcontroller, power management IC, motor driver, BMS and sensors. But today, I want to focus on one segment that everyone is talking about, which is robotics. So we have a total solution for both motion control and power management for different types of robots. But you may notice in the robotic industry, the growth is really driven by the humanoid robots. So today, I want to explain how MPS placed in this fast-growing market segment. Okay. So MPS AI power solutions powers the brain of this robot. And the ADAS solution, Allen just mentioned, also used here. And the heart of this robot is a battery pack, which is managed and monitored by the BMS IC. So the brains and hearts are very important for robots, but robot also needs muscle to move around. So that's where you see all these joints, shoulder, elbow, hip and knees and also hands. All those are driven by actuators. Each actuator has an MPS microcontroller, model driver, DC/DC and position sensor. Overall, we have about 45 MPS products worth over $150 inside this humanoid. But unlike the industry robots or service robots, the humanoid robot presents some interesting and also technical challenges. So I want to explain what those challenges are and how MPS can shape the future of robotic motion by solving those challenges. The first major challenge is actually the size. Humanoid robots resemble human being, so they need to have the size of a real human being. Today, actually, I want to introduce a friend we invited. So I don't know if he's already chatting. Okay. So maybe last thing if -- okay. So he's going to say hello to everyone, and -- so -- yes, so this is my $150 friend. Hello. All right. Okay. You can see this is -- I'm talking about a real human being. This is a real kid. Okay. Thank you. So here's the challenge. If I look at the hand, right? So there are about 15 actuators inside this hand. Each actuator there's a coreless motor of this size. So we have to fit all the electronics, the microcontroller, the motor driver, DC/DC and sensor into this small motor. So how can we do it? So only MPS can do it because we have the highly integrated solution. We also have the innovative package. By combining this, MPS can make the smallest coreless motor in the world. So the next big challenge is really the precision control. For robot, humanoid robots, they're supposed to perform some very dedicated fine tasks, for example, picking up an egg without breaking it. So that requires precision control for both the speed and position, right? I want to use this video to demonstrate. You can see here, right, how by combining the MPS best-in-class sensor technology and our control algorithm, you can achieve this kind of accuracy and consistency, so which allows the humanoid robots to perform all these fine delegate movements with high precision, with consistency, right? So that's how we did it. Okay. So what's in the future? So MPS is not just a semiconductor IC company. So our expertise actually goes beyond semiconductor. We have our own in-house motor design experts and mechanical design experts. We don't make motors or actuators, but those guys, they help us invent and design all these innovative motors and actuators. We even have a physicist conducting all the fundamental research on magnetics. In addition, we partner with all the major global robotic companies, both in U.S. and in China. We are always at the leading edge of the technology revolution. But MPS expertise is not just limited to motion control. I think Allen just mentioned the ADAS solution for physical AI that can also be applied to robotics for autonomous robots. And also MPS is a wireless charging expert. Later, you will see in the demo area, we have a 3-kilowatt wireless charging station that can charge robots wirelessly. So with all this competence and expertise, I'm confident that MPS can really shape the future of intelligent robot. But now I want to switch gears. So robotics is emerging market as MPS is a major player here. I want to talk about another market, which is different. This is the audio market. The audio is a very mature market. There are many competitors in this market. How does MPS play in this market? So about 13 months ago, AXIGN. So this is a Dutch start-up company. They are the leader in digital audio processing. They joined MPS about 13 months ago. After joining MPS, we can really leverage all the MPS power technology and their digital processing capability. This opens up another $1 billion new market for MPS. So it's true that there are many players in this market. But many competitors, they compete in this mass market consumer segment, which they trace off the audio performance for cost. But MPS is different. Our focus is really on the high-end consumer, automotive and also professional audio segments. So in these market segments, the audio performance is the most important consideration. So let me explain why MPS is better than our competitors. First and foremost, so MPS offer the best high-fidelity audio performance. So I remember Michael once said about this audio performance from AXIGN. I remember he used 3 phases. So the first is rich texture. Second is wide, dynamic. Third one, I forgot. So I think it's a clear spatial resolution. So Michael is a real audio file. So I think he has an expert opinion, right? So that's why we say high fidelity audio performance is very important. The other thing, if you look at all the amplifiers, the high-power amplifiers, they dissipate a lot of power, generates a lot of heat. When you look at the high-end amplifiers, they always have this huge metal heat things attached to them to dissipate the heat. But very few people realize actually a large percentage of that heat is due to the idle power loss when the amplifier is not playing music. So MPS has a very unique spatial modulation scheme. So that reduced the power loss by 90%. So this allows us to remove all these heat things, make the amplifier much smaller, much lighter. We also have this, what do we call intelligent noise cancellation, and this is very important when you listen to music while driving. because MPS, we have a multi-core micro DSP like a mesh network we use in the audio processor for streaming and for this noise cancellation, right? So this technology can selectively cancel the road noise, engine noise without compromising your driving safety. So that's very important, which means you can actually listen to the studio quality music on the road. So that's a big deal. Last but not least, so MPS, our solution provides a cost saving in the system cost and space saving. and also very easy to use. So the end result is this. So I brought this up on stage before the presentation. Actually, I don't smoke, this is not my cigarette. So this is a highly integrated, high-performance audio amplifier module as well as this size. So this is revolutionary because it's replacing the expensive and big amplifiers. So yesterday, my boss, Maurice, he asked me a question, how big is big, how expensive is expensive? I couldn't answer. So I asked Michael for help. And he -- today, he bought his McIntosh amplifier here from home. So that amplifier costs $6,000 and weighs 200 pounds. So now I have the answer. How big 200 pounds, how expensive, $6,000. Today, in the demo area, you're going to see how this small guy is replacing that expensive McIntosh amplifier. You will have the opportunity to enjoy this unmatched sound. But you will ask me what this unrestricted freedom means. This morning, Michael has to ask someone to help him lift up that amplifier and put into a car and drove here. But for this guy, you can essentially put into your pocket, you can go anywhere you want. That's how we define unrestricted freedom. So that's the end of my presentation. Thank you so much.

All right. Hello, everyone. Thanks again for coming. Just a time check. I think we're probably about, what, 2/3 of the way, almost done, right, with the speeches. So anyways, I appreciate you paying attention. I want to give you an update on battery management. So some of you were here in 2018. I had literally just started working here, and I put together a plan for what I was going to do with this product line. Now 2 things that I talked about. One was that we were going to create a brand-new product category called BMS. Allen talked about this, others talked about it. This is not a power chip, so totally different than what MPS typically does. I can -- I'm going to speak about that in a second, but we did it. We have it. We're making money from it. We're on generation 3 now. So that was a success. And then the second thing I told you was on the battery charger side. So this is more of a power chip, right? So the -- what I said was we were going to go after notebook, smartphone computing applications with a smaller integrated FET solution. So the result of that, there were some good and some mixed results. On the positive side, we've just released our first laptop notebook charger hitting the market now with first customer designs going to production now. That is doing very well and I'm very proud of that. On the smartphone side, it was kind of a funny story, because I remember going to Michael presenting this product to upper management and saying, okay, this is what we're going to do, huge market. Everyone knows that. And he said, okay, well, be careful because high risk, high reward, but at the same point in time, we only have this -- a couple of chips in this entire phone, right? So we do this. Revenue goes way up. I was thinking, okay, this is great. I feel good, and then it just evaporates. So I think the purpose of this meeting is we want to be frank with you, okay? We're a company that likes to take educated guesses and calculated risks. I mean you saw from the previous presentations, some of these risks pay off massively. If you miss 100% of the shots you don't take. So I want to continue showing what we've done, but that was kind of the history. Okay. So the message here is the green energy transition. Why I picked this is because last time we talked, a lot of semiconductor companies focusing on battery management just focus on smartphones. battery chargers, smartphones, that's the biggest market. That's what you have to attack. But now everything has changed. It's not just EVs that are prevalent, energy storage. I mean, all the way from the point of electricity generation from wind or solar, these massive grid-level energy storage systems are storing that energy. And I mean, you can listen to the previous Tesla earnings call where Elon said, I always told you this is going to grow faster than automotive. It is growing faster. This market is huge, especially since a lot of solar farms are already installed and they don't have storage. So that means the electricity is intermittent. So this is a huge thing that we're going after. And as Jinghai mentioned, it goes into the data center as well. Data center has its own battery backup. These huge boxes are going to end up being at the data center on-premises. Your home, right? You've got the Powerwall or other type of products, your home battery. And then all of that trickles down to your portable devices, your vehicle. So it's now a complete ecosystem. Batteries are everywhere. So that is -- that's the message today in terms of what -- we have products for all of this stuff. Going back through what happened from last time until now. So we said we are going to come out with our first BMS product. And again, this goes inside of the battery pack. It monitors, protects temperature voltage current. It's a monitoring device, a precision accuracy device. So we did that. By the next year, we had shipped over 1 million units. We also came out with a fuel gauge, which happens to have the best accuracy in the industry. It is a separate chip that calculates how much run time you have, state of charge. Then we secured major design wins in this grid storage market. which was huge because, as I've said, that's the fastest-growing market. And now 2024 last year, we released our second-generation part. We are now on our third-generation part sampling. So basically, you want to have product velocity. You want to constantly show your customers, you're innovating, you're incrementing, you're adding new features, this type of stuff. We have several design wins at the largest e-bike manufacturers in the world. And then flashing forward today, what are we doing now? Well, obviously, more newer BMS with new features and higher voltage, this type of thing. But also, we are sampling an active balance solution. So you'll see in the demo section about this. But basically, the idea is you have a ton of batteries in series, and they're not all the same capacity. They're all the same state of charge. So traditionally, you just dissipate that energy and waste it. Active balancer can take the high one, move it to the low one, vice versa, shuttle that energy around, longer run time, longer battery life, very important for energy storage and other large battery applications. And then finally, the turnkey BMS. So one of the common threads of these conversations we've had today is that MPS is not just developing ICs because there's a lot of people developing ICs. We are developing full complete modules, full boards with software, all of this stuff. Why? Because that enables our customers to design faster and increases the market we can attack, the tail customers, if you will. Okay. So why are we winning? So as I said, more than just one chip, right? This is what goes in the battery pack, whether you're talking about the server backup, energy storage, what have you, similar. So we actually produce this ourselves. We did all the design. We did all the software, the whole thing. So instead of just selling 1 chip, 2 chip, we offer this complete solution. This, I think, is a great idea because we leveraged this back with my previous job at Qualcomm. We would build the entire phone. What did that -- what advantage did that give us? We knew all the issues that customers would have before they had them. We knew what we should do better like with new products and innovations because we did the entire system. We weren't just selling one part into a system the customer is designing. So that's a big advantage, but clearly, other people could do this. They may be behind, whatever, but they could do this. But what we have to do and what we are doing is constantly using these platforms to offer something differentiated. That means higher voltage BMS, right, where the competitor might take 2 chips, we can do it in 1. better accuracy fuel gauging, active balancing, all of these companion devices that increase the SAM of this type of a system. Okay. So where are we going now? So as I mentioned, this is where we are currently playing. These markets are growing, as you know, and we have new products sampling right now and later on this year to expand what we can attack in these markets. But there's many other things that are huge that we are also developing stuff for, a little bit farther down the road. But for example, for notebook, this has BMS and charger sockets. We already have the charger. We're developing on the BMS side. Automotive BMS, Allen mentioned that earlier, power tools and appliances. The common trend is everything is going higher voltage, higher power, exactly what Jinghai said in the server, Allen said in automotive, it's all the same situation. So circling back on the battery charger and all that other stuff was about the BMS. So the battery charger, a lot of people think this is a charger, what you plug in. That is an AC/DC, different terminology. The charger goes in the laptop. It's the thing that's charging the battery or in the smartphone, taking the power from that, regulating it to the battery. So what we've done here is we've created a solution that is essentially half the size of the existing solutions. All of the existing solutions on the market are discrete FET controllers. You've seen this in the other presentations where everyone is showing, yes, here's what the competitor is doing, here's what we're doing, all the FETs are inside this one thing, right? So I don't have to reexplain that. But what is the benefit, right? So this delivers much faster charging time and half the size because the thermals are better. And in addition to that, it's much higher efficiency. We have customers coming to us saying, "Hey, we actually care about the power consumption from the wall when we're plugged in." This is higher efficiency because it's all integrated. So that helps them meet their ESG goals, for example. The other thing is it can enable smaller and thinner designs. Now I want to touch on the fact that you might say, okay, well, what happened to you guys in the smartphone charger could happen to you in this space, too. What is the difference? The difference is in smartphone, we were selling a charger in one system that we do not control. We don't make the PMIC. We don't make the CPU. We don't make any of that stuff. It would be the equivalent of looking at what Qualcomm does. They make the entire thing, right? So it's very easy for them to get other sockets in that market. For the notebook, MPS has everything, right? I mean, Bob over there, he's been growing this market for years. I think we have almost every provider designed in for core power. We have graphics power, point of loads, the USB stuff, the backlighting. So for us, it's actually a huge advantage. Go to the customer and say, you want a complete solution, you want a solution that works really well together, that's sticky. The relationship is good, this type of thing. So it's a completely different story for us. Okay. So I'm looking at this clock and yesterday, it took me 20 minutes to do this presentation, and now it's only 10. So I'm wondering what I didn't say. So anyway, thank you.

The schedule is that we have a break, and we take a poll. So do we want a break or want to keep going. Keep going. All right. We keep going there. Great. We go through some highlights, dim light, and I'll go through the low light. Okay. And -- there's a lot of good things, okay? And -- but something didn't happen, okay? And I show you that well, it's not quite envisioning, okay? Show you the low light first. It doesn't go. So remember this one, I was very, very enthusiastic about it. The reason I talked to -- okay, I just remind everybody, we have -- I was impressed myself, I even made AI algorithms, that's back in 2018. And I thought this ideas and we use some kind of intelligence and we came at making product, and this is based on algorithm software. And put on the website, our customer can program the product. And we give -- we can program -- they can order it, program multiple parts. And we can ship it -- the programming here. We can ship the product in less than a week time. I checked with our customers, I checked with everybody. It's a great idea. It didn't happen, okay? So my analogy is we know there's a lot of fish in the Pacific Ocean. I dropped a hook in there. I didn't hook anything. So it has to be -- there's a lot of reasons why it didn't happen, okay? And the serendipity comes out of this is that we know like how we're going to increase customers. And from that time, we have less than a few -- just a few thousand customers. I don't have exactly numbers. Now we talk about 30,000. So that's the story around and dim light to turn into a highlight. And it is true. We want to keep diversifying. The more customer, the better it is, more stable the company is. And we're committing in the directions. And for e-commerce, we still believe that. And we -- from this, we learned, we created MPS now. And that is the reason we increased so many customers. This is since 2018. Okay? That's one low dim light, okay? And the other ones not so like, okay, we mentioned about this. And we didn't note. We really don't know what is the market segment is all about. I know in my opinion, okay, when I built the house in 2012 or so, there's a lot of things missing in the market. I can't pinpoint what should be done in there. And a lot of software, a lot of building automations and a lot of big companies want to do a building automation, all failed. And this building, we have all the heat -- thermal controls and the air conditioning and the air quality, you can't control it. It's so bad. And all the other building is in a similar -- facing the same kind of problems and even go beyond that, I'm looking at how we're going to install these electricities. You have electrician, you have AC wires and 110 in U.S. and in Europe is 240, 220. And all of these, these are 100 years old technologies. Why do we need that? And finally, I think we can do it. We figured it out. It's ease of use and its ease of use for many, many different sensors and temperature sensor, why do you need a thermostat? You don't need it. These are all our customers. And why do you need an access point? Why do you need, okay, a motion detector? All these sensors are very low cost. It's about $1 or $2, finished product. Why we're not adopting these? And by the way, these are all our customers. And also the lighting, I talk about AC, the most of the power in the modern building is the AC and also lighting. Why there has to be run in high voltage AC to DC, which we provide a product for that. Can we go just -- there's no reason for it. We can provide DC into lighting, into that entire building. We talk about 48 volts, right? That's going to happen. And we can provide MPS now making MCU. We can easily program the MCU communicate to these sensors. So this is what we want. This is what we want to do. We want to control that. Actually go beyond that, it's life quality. Your lighting, your air, your sound. Sound doesn't mean only music you hear. It's your environment, noise. Why do you have to cancel the noise cancellation only on your ear? How about we cancel in air? We can do the same things. We have those technology. Sound just presented in the car. How about we present in the room. We can filter out the noise. We don't want to hear that. And it's actually go beyond the building automation is to improve your life quality. The lighting, why do you have to -- under this sunlight is changing the color from day to dawn, from morning to in the middle of the day. And your circadian rhythm is dependent on the sunlight. You spend 90% of the time in the building and you have one kind of color. And can we change that? We can change that very easily, okay, why didn't do it. So we start to do this. We start to figure out ideas. And the key is cheap rate. And same time, in this room, since we have about an hour or so, you take a guess what's the CO2 level arises 1,000% from before you come in until now, 1,000% and increase a huge amount. And so what does the CO2 level do to you? You don't think it cleared. Next meeting, I want you to think clear, you can ask better questions. Okay. So this is what we do, and we made a box, all based on MPS product. And you will see it. We're going to demo it today. And I can tell you another story of Eric Sanders. And he joined the company. The reason he joined the company is we -- I start to do for my home, changing it and do for the company, and he does a subcontracting, do all the work. And he has a lot of experience for intelligence building. And it's nightmare. It doesn't work. It's never been popular. And so I told him this concept, I'm going to make this box. He said he cannot join the company. So you will talk to him today. He knows a lot more than I do, and you'll see the demo today. So what's the advantage? Almost forgot. Okay. What's the advantage of that? Any box is scalable and you have -- can connect multiple boxes. And for medium-sized home, one box, you can cover everything. You just plug in the wire and runs on the default mode, everything works. You have a control pad on the wall and energy saving. This is 10% to 25% energy saving because you get rid of all these -- just AC to DC. We do a high-efficiency AC to DC central rather than discrete. And 30%, this is a conservative number. Think about it. All these commercial buildings, especially, they need a metal conduit. All the AC wires had high in the metal pipe, if you don't know what the conduit is. And these are DC, we don't need that. So we converted this area into this and using just DC wires. And so we talk about comfortable wellbeing. That's only, not only this today, we don't have a, we don't have air controlled. And in the next few months for particularly this building, we will convert that. And also with MPS, we're building this product. We're building this product for MPS use first. We're building in China, have multiple offices, 2 in Netherlands, 1 in Spain, 1 in Taiwan and here also. We're using our product. We eat our own dog food. And do this -- make this happen, improve the lab quality here within the building. That's what we want to do. And at the same time, if you have 1 million square feet building, that will enable its own ecosystem. All these are our customers, they were going to work with us to make this thing happen. So how do we really monetize this one? I know there's a lot of value. We need a partnership. We don't have those sales channel, but these are the value. Okay. And next thing. That's another low light, okay, we can turn into a highlight, ultrasound, we didn't -- 2018, didn't talk about it, but we -- 2018, 2017, we talked about our ultrasound chips. And guess what, we generate less than $1 million revenue, okay? And we -- the reason is, okay, there's a lot of reasons, okay? Maybe we have a lazy sales force, we're not marketing -- address the marketing correctly, our customers' design times for too long, there's a lot of reasons, okay? But here is we -- only 3 guys in the product line. We make the whole chip sets for ultrasound, except GPU and these FPGA product. And we have a new architect. Those guys writing -- making their own chip, write their own software. And here we go. We made a full system. I tested myself. I don't have a plug in. And you will see it today. And this is a full system. And -- this is about 3 years ago, 2 years ago. So what's the specification of the system that they asked me, I said that build the best one. Build the best one and the best. And the best is over $200,000, $200,000 system for cardiac. And MPS, we can do a software, nobody sell everything in one system. And we can make a vascular and cardiac. And it's all software configurable. And as a matter of fact, you see it today, it's all there. And so our imaging quality, I can't say better, okay? Well, I can say, definitely, it's not worse, okay? Some are better, some are similar, just similar enough to see it, okay? So see clear. It's equivalent to these $200,000 systems. And there's a lot of them. This kind of low-cost system enable another market segment. I call it AI enablement because you can do a remote diagnostics, remote medicines. These are how we commercialize it. And we don't know. But there's a system there, we can be through a partnership, through -- we do -- we sell ourselves, okay, build the sales force and build and sell it. And you go through FDA approval, which is very -- this type of a product is very easy to -- it's not invasive. And I keep threatening it. I sell $2,000 on Amazon as a consumer device. Just [indiscernible] everyone. And -- this system, it's all these things, worst case, okay, we sell chips and [indiscernible]. And it is a system we can licensing it, we can do a partnership, so we can do all kind of opportunities. The key is that we monetize all these 3 people's efforts. They spent 2 years and spent a little over net of a little over $1 million. So that's my excitement coming to office to see all of this happen. And so it's a lot of veterinary opportunity, which doesn't require FDA approvals, okay? And the future telemedicine, remote diagnostics, and I think it's the consumers. I want one in my homes, not this big. We can be even making it smaller. This is everything inside. I want to make the portable ones. Put it inside, I can check myself. It's like a blood pressure meters. That's the next thing we're going to do, okay? For the next analyst meeting, we will do that. Okay. And today, actually, you can check yours, whether you have a clock or not. All right. Next week. We will go through quickly.

Hi, everyone. My name is Bernie Blegen. I think by now, many of you have been here. I see a lot of familiar faces from 7 years ago. And it's just really -- it means a lot because that means that you stayed with MPS. And for a lot of the newer faces out here, I hope we'll have an Investor Day before the next 7 years, but we look forward to seeing you again and that you will stay with MPS and that you heard something interesting today. One thing that you all know very well, if you've listened to our calls, is we have what I call the Michael and Bernie show. And it's sort of point and counterpoint. And I just want to not correct, but just add to what Michael just said. We do not have a lazy sales force. So Back in February, we provided guidance and it looked like this. And today, we're providing guidance and it looks like this. We've increased the revenue outlook for Q1, and we've narrowed the bandwidth there. And it doesn't necessarily say that the back half of the year has changed materially. We don't know anything new about that. But at least the momentum going into the year is remaining ahead of schedule, and that's a very positive story. When we look at how we compare back to 2018, and this has been a very interesting week for me because we've heard so many stories, particularly from the product managers and what's come across is that there's some very consistent themes, and I want to pick on a few of them. One is we heard from Jinghai that he had a vision 7 years ago for what he saw the opportunity for MPS and for the broader market in the data center. And it's interesting because you roll forward 7 years later and the vision really has not changed that much. It's been adapted to the new market conditions, but the vision is the same. And there's nothing better than having a good story that ages so well as he told. The next is we had Allen who got up here in automotive. And oh, by the way, Allen will send -- sell you a [ ZEEKR ] out here, just talked and we got the paperwork lined up. Any case, he actually gave a very good story saying that a lot of the technology we develop, derivative products come from that. And so a lot of the same themes that you heard as far as like higher power, 48-volt, on and on, these are derivative technologies. So we create a platform technology and then you're able to apply it to all of these other end markets. We had Song who got up here. And one of the things that might have been underappreciated is the fact that on robotics now, AI enablement, along with ADAS and actuators is what's really going to make this market explode. Well, hopefully, the product didn't explode, but the market. And it's that level of integration that we do that differentiates ourselves from the competition, the system-level solutions, and that is really important. In fact, we even saw Song's new friend who came and joined us for a little bit here. And we also proved that Song is indeed very human. What was that third thing again? Finally, we had Chris who came up here and showed us that not only can you try something against Michael's best wishes and fail, but you still have a job 7 years later. But that also goes to the culture of we're risk takers. And as Michael said on the e-commerce, maybe that didn't work out the way we believed. But you know what, we learned from it. We course corrected. We're going to improve it. We found new attributes that we can monetize. And that's also part of the MPS story. So when I look back at our numbers, maybe like yourself or like me, if you look at it, we were at $0.5 billion. I couldn't even like with that, that seems so recent. I couldn't believe we're so small. And when you kind of do the report card against the financial model from 2018, we wanted to -- we aspire -- I used to say aspirational a lot when I first became CFO, to have 20% year-over-year growth rate, and it proved out that we're over 25%. When you look at the gross margin, what we're really trying to solve for here, and I'm going to make this more complex than it really should be, but we want to have a range that we have the freedom in order to experiment with new markets and new products and new technologies that may not be within our corporate average, but over time, are going to accelerate our revenue growth. And ultimately, the combination of those 2 will create and generate the highest free cash flow for our investors. When you look at operating expense as a percent of revenue, well, unfortunately, the 50% to 60% of annual revenue growth, we'll put that in fiction category. But there's a very good story behind that. And what that is, is that we continue to invest. I think you saw in Michael's presentation, the number of different locations, the partnerships that we've developed. I'll show you a slide in a second on the headcount and all the partnerships we have. And we continue to invest because as you're going to see in the slide here in a little bit, the products that we have under development today are the future revenue streams that enable that top line growth. And again, that's all part of what is the highest level of free cash flow we can generate for the investors. And then finally, the operating margin. We show expansion, consistent. And capital allocation, particularly over here, augmented by the stock buyback is an important component because we have not historically done large acquisitions. We do IP or we do talent tuck-ins. But outside of that, this has been a very organic story. So we believe that it's important that we make those returns of capital to you. And if I keep hitting this button long enough, I will flip. So this is a chart that you already saw when Michael presented. Again, and a point I want to make is that even though we only grew 1.5% there, we outperformed the market by over 10 percentage points. And that's what we continue to do. And during that period of time, when we only grew 1.5%, we still expanded our R&D headcount by 8%. So that's how we invest even if the economic conditions don't favor that level of investment. And obviously, I want to take credit for the 25% or 24.9% and the return of capital to you. But I also want to take a few other elements here, and this is important to look at. The employee base that we have has more than doubled in that time. You can see as far as major sites or operations that they've doubled. When you look at foundry partners and capacity, and this is incredibly important because what we want to do is be able to provide to meet our customers' requirements. And so we demonstrated particularly during the supply-demand imbalance back in '21 and '22 that we had capacity, and we're able to provide parts where our competition couldn't. And that became a huge market opportunity as far as share gains, which is continuing. It's ongoing. It wasn't one and done. As far as being able to deliver the highest level of innovation, a reliable product, to be able to exceed customers' expectations and be able to deliver parts so they don't have capacity anxiety. That's a winning formula for our customers. It's a winning formula for our investors. You can see our products have more than doubled. Our patents have nearly doubled. And as far as the quarterly dividend, even I can't do the math, but it looks good to me. The floating bubbles. I love this chart. So what we're trying to do here is demonstrate exactly what I was talking about earlier is that MPS over 5 years has had the highest growth rate. So it's one thing to say 24.9% versus the market that grew at X percent. It's another thing to see how much that really looks like and how did the competition do. And it's interesting because I hear a lot of people and they say, well, TI, they've got their own fabs or other companies, they have a lower cost model than you. And we operate within a gross margin for a particular reason, and it's to generate to accelerate our rate of revenue growth. And I think that, that formula, when you look at how we're differentiated from the competition, it proves itself in this one chart. getting better at clicking. The key to it, and we've talked about this in the past, is the diversification by end market. And here, you can see that when Jinghai was presenting back in 2018, enterprise data only represented 7% of our overall portfolio. And I've had people who have come to me and say, Bernie, except for enterprise data or AI, you'd just be middle of the pack like everybody else. But we're not. In fact, we were able to grow when all the other companies were not because we had made investments 7, 10 years ago that timed with the market. We didn't know exactly that AI was going to -- ChatGPT was going to grow the way it did. All we're doing is it's portfolio management it's most base element. If we put enough bets on the table and we make them low cost and we support them and we learn from that experience, we will win. And the outcome -- it's interesting because we don't actually go in and design what next quarter's numbers are going to look like. It's a byproduct of everything that you've heard from all of our product managers in Michael. So it's great. I have the easiest job in the company. I just report the good news. This to me is a very interesting and telling slide. And I hope that after this and you reflect on what you've heard today that maybe this could be a benchmark to say what differentiates MPS. And it goes like this is we develop new technologies, and this is not all inclusive. This is just meant to be a representative list. And that -- the technology is what drives our SAM expansion. And if you look at all of these, you can go back even to all of the different earnings calls that Michael and I have had, and we've told you this is coming down the pike. The length of these lines show how long we were committed to the development of these different products here along the way. And so when you come up to here, you go, yes, I just heard about the silicon inverters for green energy. In fact, I'm looking here, the BMS and the active balancer, those are all Chris' business. They're right there. They're really poised just to start to take off and generate revenue. You can go back here. This is Michael, the analog to digital converter. That's the -- we have the first offering is for security systems, and now we're going into the medical devices, the ultrasound. So all of these, you go down here, we were developing our own Class D audio. And now we've extended that development with the addition of the AXIGN R&D engineers. So this really drives it. And when you look ahead, again, representative of everything that you've already heard as far as power isolation, multiphase architecture and automotive BMS, it continues to drive. We're not done yet. We're just starting. Again, here, you look at the diversified SAM expansion and growing from $10 million to $16 million and now up to over $27 billion. So again, that's just a reflection of the fact that we are so diversified. And as Michael said at the beginning, we've only showcased half of the market opportunities that we're going after right now. So it's still -- it's -- if we had more time and maybe next time when we run SAP, we have there. Okay. So how does this all land? I'm here for one thing only, really, to give you the new model. But the new model looks a lot like the old model. We're targeting growth of 10% to 15% above the market. If the market is growing at 5%, we get to 15%, 20%. So that's largely unchanged from what we've been talking about. So what's interesting about that, and Michael referred to it, is a lot of companies face the law of large numbers. It's hard to grow because you're off a bigger base. But because we're so diversified, we're going to so many new markets and even in the markets that we're in, we occupy such a small percent of market share that it's all just green space that we can grow into. On the gross margin, again, we want to keep the flexibility in order to drive the highest free cash flows and at the same time, be able to develop new products, which today may not offer a gross margin at our rate, but that will over time. And it's very important because we've talked about today, what are some of those growth drivers, integrated solutions. We were talking about everything that has higher power and has to be smaller. And for our customers, they are demanding more integration. Allen used the number that we saved one of our customers that wanted to go with their own $200 million of foregone development costs that they didn't have to incur. So we're approaching this. We're not a pass-through for these other third-party components. We're selling designs and our total cost of ownership to our customer is being recognized and valued. And the gross margin allows us to have that freedom. I did up the operating expense that we'll still have margin expansion, maybe not as aspirational as in the previous model. And then it's very important to us that we stay steady and being able to return up to 50% of our excess free cash flow back to our investors. So with that, if I've done this correctly and I hit the button properly, this to the Q&A.

All right. So for Q&A, we'll have Michael join Bernie back on stage. There'll be a couple of runners with mics in the audience. Please use the mic so that we can get the question onto the live stream as well. I'd ask everybody to just announce their name and their firm and one question only so we can get as many as we can. We start right over here with Ross.

Ross Seymore from Deutsche Bank. Just a question on the ED segment and the power management. On one hand, you had Jinghai's law and all of that. The complexity goes up, the power goes up, the size goes down, et cetera. On the other hand, the market is concerned about the fact that more competition is coming in. And it seems like a lot of the incremental competitors are happy to operate at significantly lower performance and/or price and margins than you guys. So how do we reconcile the fact that the performance needs are going up, but the profitability of going into that market is going down? Is it something that's commoditizing? Or are we just waiting for the next phase of growth that will reward your innovation?

Yes. I don't see it a commoditized because this is a very fast-growing market segment. And the performance in customers pay for performance. What does the performance mean? And not the unit price is a part of it, of course. And it's a total -- in the end, it's a total ownership, total ownership cost. And you may pay us, okay, you may buy a more power-efficient problem. We're not there yet. We still try to figure out, increase the efficiency, reduce the heat. We're still at that stage. And in the end, really the total -- the winner will be the total cost of ownership because you save electricity. And that's the market.

If I can add just one more quick, and I apologize. One of the things that we do is we innovate faster than the competition. And where we bring our technology to intersect with the opportunity is when you have a change in the market. It's not just -- we're not going for pin-for-pin replacement. We're going for ADAS. We're the first to the market. We're going with 48-volt, we're first to the market. These are examples. So we always get the first-mover advantage. And then how our customers decide that they want to operate after they're in the market, that's up to them. We don't control that. But we always look for being first.

Joshua Buchalter from TD Cowen. Michael, Bernie, Tony, thank you for hosting a very informative day. A lot stayed the same in the story since the 2018 Analyst Day, but maybe following up on Ross's question, it does feel like the competitive environment has picked up as you've been early into these exciting markets, but now there's a bigger seemingly target on your back. I think you used to say you call yourself like the gorillas and it can't punch down to you or something. How are you seeing your competitors sort of react as these markets have taken off and you've seemingly captured a lot of value in very important markets, in particular, in the AI one?

Yes. Those kind of things, I'll let the numbers speak itself, okay? And MPS, okay, if you remember the history of MPS, we from nobody, 20 years ago. And all our competitor, maybe doubled, aged, a lot older. And early days, our investors keep asking me, okay, why all these large companies going to do it, you cannot do it. They cannot do it, you can do it. And I'm tired of answering that question. Okay. And so here, all the people here, we have a lot of them making our products and lots of them making a lot of money. That's all it is, okay? We have a lot of questions.

Rick Schafer, Oppenheimer. Kind of a 2-parter, Michael. The first is you guys were really clear how accelerator power is increasing dramatically each generation. And so I'm curious, your view, Michael, on what the theoretical limit or is there a theoretical limit to how long we can stay on lateral power versus making a leap to vertical because obviously, there's a big hold out there. And then the second part of my question because it seems like it'd be a huge TAM expander. We're getting pretty close. I mean some of the new CPUs for server from AMD and Intel are pushing 500 watts. So how close are we to seeing that 12-volt to 48-volt conversion for x86 CPU and server?

Okay. Let me answer the second part first. The 48 volts is absolutely happening now. And we're shipping those products. And from now on, I think Jinghai can answer better later. And we have 3 sections and you can talk to all the people here in the demo side. So Jinghai can answer you better. And from now on, everything is going to AI, it's all 48 volts. And go in the next couple of years, it's a 400 volts DC, not AC anymore. And that Jinghai has talked about. The other thing, what was the first one -- first part of my question?

Can you hear me? Sorry. So yes, my first part of my question was just is there a theoretical limit? Like how much longer can you extend the life of...

Yes, I will let Jinghai to answer from the module side. The fundamental is the same as everything else, the transistor level. How can we improve our transistor. It's like a MOS law. We call it Jinghai law, it's the module density, power density. It's the transistor from MPS starting, I can tell you the numbers, okay, I amazing myself. I started this. I made the first transistor, which is about 20% smaller than -- or more than 80% smaller than our competitors on the market. That was late 1990s, okay, and early 2000s. And that's how we win the product. Since then, the current product, we have it, it reduced another 90%. Just imagine. The journey I started, now MPS, all the R&D people doing these reduced another 90%. So we have 1% of the size. So keep going, MPS keep inventing this new type of transistor. And same as the rest of the industries for digital circuitries, for GPUs, for CPUs, same journey. We're in a different path. Okay.

Tore Svanberg from Stifel. Michael, maybe I can follow up on that because the way I looked at the slides earlier was that you're going from XPU, GPU power management to rack power. And a lot of your competitors are probably still focused on the GPU power, right? So is there anything else that you can share with us without obviously talking about trade secrets and things like that? Because it does feel like you are now moving to a more total solution for rack power, including power supply, higher voltages and so on and so forth. And what do you need to invest in process technologies in order to deliver a total power solution for rack power?

That's a good one. Okay. Yes, we already said, we increased the current density to 3 amps millimeter cube. For these GPU power some memory power and the CPU power delivery, okay? We already moved that direction. And the other direction, as Jinghai has said, okay, we're going to announce a product, we're going to sample our products in the next few months is 400 volts DC dropped down to 48 volts. And I can tell you this, I'm leading a project going AC to 400 volts DC. And so what kind of technology? We talk about it in 2017, '18, we are working on a silicon carbide project, and we are shipping those products now. And not for mass market. MPS is never going for a mass market, okay? And unless we can dominate. And we go for the mass market, there's a clear reasons, okay, we can control it, okay? Otherwise, diversified. And silicon carbide we produce, it's not in a selling as a dumb power device, a passive power device. We integrated it with our controllers. That is the value we do with them go even beyond this 400 volt DC. There's a lot of other applications to it, other than data center. Yes.

Gary Mobley at Loop Capital. Bernie, you are guiding revenue at the midpoint, 2% higher than your prior outlook for the first quarter. But I also hear from you, you don't have seemingly as much visibility to the second half of the year. So my questions really are what's driving the upside in the first quarter? Was it broad market strength? Was it better-than-expected share in the enterprise data market? And then how does the rest of the year set up? Yes, this is my question.

Sure. Let me give a backdrop that there's the old Chinese proverb and curse, may you live in interesting times. And I would probably offer that we're all getting sort of whiplash. It's not just me, it's everybody in this room with the changes or potential changes or threatened changes in tariffs and trade policies and things like that. Those are all outside of our control. So when I look at Q1, I'm actually able to see what's in my backlog. What POs do I have for what products. And as I said, initially, in the year, we have stronger early momentum than had been anticipated. And there are a couple of exceptions. I won't talk about them now. But overall, it's pretty broad strength that we're seeing. And when you look past what I have as far as backlog, it does get a little bit more blurry. I mean I can point to indicators where we fundamentally believe in our total numbers as far as what we can achieve for the year. But until I can see something that's fundamentally changed and it hasn't good or bad, I'd rather stay neutral in the second half.

Let me spin it the other way. Let me -- wow, 10 days from earnings call now.

Yes.

We don't tell you some numbers out there look stupid. We have to tell you something. So our job is that we want to have a very consistent result. I don't want to disappoint you, I don't want to look ourselves foolish, okay? That's what Bernie think, oh, we should see only 10 days. I didn't remember. I don't know that. That's -- so don't get excited. Okay. Everything else stayed the same.

We wanted to make it worth your while to come out here. So these are the numbers.

Thanks for hosting Great Analyst Day. It's Will Stein from Truist Securities. Bernie, you mentioned tariffs a moment ago. And I wonder that's also sort of 2 parts. Are you seeing any sort of secondary effect that's driving demand volatility among the customers because of that? And then the primary effect as it relates to your foundry strategy, you talked about, I think, going from 5 to 13 partners. I assume that's factories, not companies, but maybe it's companies, maybe you're going to have more factories. You used to talk about China Plus One. Can you just linger on that topic for a moment and talk about your ability to supply in the light of sort of volatile, unpredictable tariffs?

Yes. These, as Bernie said, is out of our control, but we control our destiny. Do we want to manufacturing in one place? The answer is no. And it doesn't matter, you have a tariff where you don't have a tariff, okay. It really doesn't matter. We need to diversify our risk. And we did it even before this trade war happened. A long, long time ago, that was actually, by the way, we run out of supply, not like we ran out of capacity. We diversified. And remember, MPS don't make anything. We doesn't make anything. MPS doesn't make anything. Wherever the capacity is, wherever the manufacturer is, the best manufacturer is, we go for it. So we don't make anything. Our supply as long as we have availability, we'll go for those just a normal business growth, business expansions, capacity expansions.

If I can add to that real quick is that nobody knows what the future really looks like. And -- but what I can offer is because we built up this diversified capacity and diversified R&D capability worldwide that MPS has proven to be able to be nimble and we can move more quickly either in a defensive move because there's been a change in the market or as we capitalize in the supply-demand imbalance back in '21 and '22, an opportunity. And so I think that when you look at positioning, while we can't tell you what the future is going to look like, we can tell you that we can be the fast first company to take advantage of it.

Maybe a little longer term and then a little shorter term. But I want to go to the slide you said that was important, kind of listed all the different kind of product areas and that. And as you look through that slide, are there any particular areas of focus, meaning in the past, the history of the growth of the company, sometimes there's been big areas of growth. ED AI power was one of them. is the view that it's going to be more diversified as we look over the next kind of 2 or 3 years. And I don't want to come to a comment, understanding in the shorter term, understanding that there's a lot of uncertainty in there, but you're confident in your own abilities. How you -- how confident you are in our abilities, meaning the numbers that we have out there because they're our numbers, not your numbers? And how does that look for the year?

Again, I keep saying, we don't -- we cannot -- we know some market trend. Of course, we know it, okay? And early, all the people will said here, that's how we're predicting. And at the same time, a lot of times, I put the brake on this. I want to have a very diversified growth and growth customers and also expanding our product portfolio. And you remember in 2016 or '17, we started, we're going to do data converters and 2018 or '19. And we're going to do MCU. And all these products is very fundamental for our future growth. We're not picking on one area. And so that will defeat our purpose. We will be very, very -- you will not like us, our stock going up and down. We did it go up and down anyways. That's right. Other people say that we are the victim of our own success, okay, so be it. And so we -- earlier, I said that are presenting all these products, okay, and we talk about 50% of these other ones, it may come in the next couple of years, some things become very hard. As long as we enjoy doing that, we provide the best things, best performance, cost performance. And one day, these things are going to happen, and we will be there. So that's our strategy is our culture. So the next one is, Quinn?

Joe, would you assist?

Quinn Bolton with Needham. I guess a 2-parter. First, on the data center, the high-voltage data center opportunities you're looking at AC to 400 volt and then 400 volt to 48 volt. Can you just give us a sense how much of the TAM would be AC to 48-volt? You've been in 48-volt to 1 volt for a while. You've been very successful. It seems this higher voltage opportunity could be significant. And maybe I missed it, but I don't see sort of a relative sizing of the market you're moving into for data center versus where you've been? And then I've got a follow-up question.

Yes. Frankly, I don't know, okay? All the data centers will move into that in that type of architecture. And Jinghai probably can tell you more, okay? It's always big. And this is at the very beginning, okay? All right. So second question?

Two quick one on that, though. But I mean going back to building out your tool set and the technologies that we showed, we've been doing -- we talked about power isolation, I think, about 2019, 2018, 2019. And that is fundamental to being able to do the high power that's required in the rack and in the data center. So again, it's building out the portfolios and it's understanding our customers' most difficult challenges, and that's where we succeed.

And also, it's to do this type of development work, it's not contradicted to what we do somewhere else. The building itself is going to be DC power. We are pushing that in the directions because there's a building automations, the box that we made it. And it's all going to be okay, changing the way we look at supplying power.

The second question as you move into the higher power, higher voltage markets, I want to come back to this idea of discrete solutions versus integrated. I think discrete, you've proven at lower voltage, lower power is more efficient. But as we move to higher power Infineon and some of the discrete FET guys will say, "Hey, we'll put multiple die in a package, but each die is process optimized and so we can have the best FETs in that package." And as things are typically on running at high power, they'll say that, that gives them advantage. And I was hoping you could sort of respond to those kind of...

Yes, quite -- you always raise the good technical questions, okay? It is. Okay. Once the power -- once you have the power -- have the need for power density, the integrations, the better power device is all kinds. And also efficiency. Efficiency, you can trade-off with size, okay? And some of the ones like the energy storage for power grid, is less care about the size, okay? And -- but they do care accuracy. And other ones is like in the data center, absolutely, they care about size, and they care about the heat generations. And so again, MPS is getting to a silicon carbide business. And we are not -- these are 20 years ago, we did this integrated silicon -- integrated power device on silicon. And now okay, our power device is a lot better than the discrete one. And in the high power in the silicon carbide, so far, I see all the numbers were a lot better. So it's the same journey to me.

Adam Rich from Vaughan Nelson. Obviously, you're an innovation company, and you look back the past 7 years, you exceeded your revenue. And so either your pace of innovation was faster or your return on your R&D was higher. Any insight into what that was? And then looking out over the next 3 to 5 years, do you see anything that can accelerate pace or increase your return on R&D?

Maybe Bernie can answer better. I don't know -- we don't do a return on investment. But we do -- we know all the projects. I don't believe big R&D dollar creates a better result. Most of the time, I feel inversely proportional. I'm not joking, okay? And in our business, I'm not talking about every other company. For MPS, 3 people developed this ultrasound. I was told to have taken a real ultrasound company, take 50 people in 5 years to do it. Of course, they are not stupid people, the system screw it up. That's efficiency we're talking about. And for last 7 years, especially the last couple of years, we didn't know AI is going to take off when the ChatGPT took off, so we're going to grow 24% or 25%, I think we're lucky. But if you're looking back, 21 years ago when the first time IPOs, we actually follow the same trend. And those CAGR probably higher. I didn't calculate. And we keep doing the best. We're keeping betting on the right area. And those things will happen. It may not be Jinghai said, okay, it was 2019, we're going to intercept our customers must use us and actually a couple of years later, but it did happen more than we expected.

If I can add a couple of comments here. So earlier, I made the observation that I have the easiest job in the company, and it's true. And the reason for that is everything that we did 7 years ago is what's coming to market today. And we've been able to.

Including e-commerce.

It's on the horizon. And when you look at how we're different from a lot of our competitors, particularly in analog, people come here and they want to pitch an idea. Well, guess what? We do not have an ROI model in the company. It's true. So when we have a go/no-go decision, it's based on 2 criteria. One, is it fun? And is it -- two, is it engineering differentiated? Is it solving a problem in a different way than our competition? So it's not like does it hit the ROI? And then another aspect is we're fabless. We don't make anything. we're getting compensated for our ideas for innovation and our execution. And so as a result, we don't have a high capital investment. It's really human capital investment and then making sure we don't put up some cumbersome administrative process that slows people down from having fun.

Maybe one follow-up. How about from your customers, are you seeing their pace accelerate? Obviously, you've had auto market, a huge change and how fast new products are introduced. Are you seeing that in other markets? And maybe could that accelerate you all.

I wish they can go even faster. So that's -- I wish they go. They don't move very fast. They are moving, but not moving, better than they're not moving, okay? If it's a not moving market, okay, MPS will have more trouble, because we advanced product, they were adapting to those products. And if it's a market fast-growing company, you look at it, all these fast-growing company, we are with them. And for those that are more stagnant and more supply chains emphasized. And we made a pretty good inroad, and a growth in there, but it's much slower.

By the way, I understand the tempo of the room, and it's because the carbon dioxide has gone up yet another 100%.

Yes. Not the next time, okay. I think we have a lot more time to answer questions. I'm not going to finish, we have some food and more importantly the demos. And some of them will blow out of your mind, okay, this one. You will see those products. You will see those products. And so a couple of more questions. Okay.

Joe Quatrochi from Wells Fargo. Right. You talked about -- I think on the slide I saw you've got foundry capacity for $4 billion. And as we think about just modest growth for the industry between now and '27 and your outperformance, like can you talk about the comfortability of just increasing that capacity and like the planning that goes into that, thinking about just the revenue growth opportunity?

I'll take it. So there's a lot of things that I'd probably say that I'd rather be lucky than good. And we came up with an idea years ago that we wanted -- we were at the time, probably at that $500 million level. And we said we want to have capacity of $2 billion. And then I said something like Q3 of '21 or something like that. Well, everybody made their models to go exactly to that. What we do is we pick large targets, and then we get everybody aligned around delivering that, because it has to be a company-wide organization effort. So we picked the target. It's big, it's audacious, and we have a track record of executing against it. So that's all that went into picking that particular number.

This is [indiscernible] from [ E20 Capital ]. So because there's too many business questions, so I'd like to ask 2 technical questions. So firstly, like for the BCD technologies, it seems like the [indiscernible] of MPS. So in the IR slides a couple of years ago, you will keep showing it you have progress to shrink a lot on BCD products. But it seems like 2 years ago, this slide is gone. So I'm wondering like is the progress has hit the wall and you cannot shrink the large size anymore and especially considering you have a competitor in Phoenix just announced the BCD technology improvement last year. I mean does MPS still leading this technology or like you just like on par with others?

We're far ahead. I can tell you we are far ahead. And -- we removed -- we didn't deliberately remove it because we have so many things to show, right? I thought you guys are getting sick and tired of seeing the same things. So that's the reason we show you something new, okay? And I said earlier, okay, and the competitors and we -- how we shrink and what is the -- is there a war, okay? For improving the power density, it's fundamentally, this goes to a transistor level. So we keep improving that one.

Let me take one more add there is I had a funny conversation with one of the engineers who is responsible for continuous improvement. And back then, we had a chart and it said BCD2, BCD3, BCD4 and then we had all these things of what the new attributes were and what it gave us. They're better, faster, smaller, higher integration with software, things like that. And I went to them and said, when's BCD7 coming out? He goes 7, I want BCD 64. It's like...

Let me answer the other way, how much you can't think of what's the room, okay? Why using 60-nanometer, maybe 45 equipment, okay? There's a lot more room we can improve on. We're always using the trading edge product, a trading edge equipment. And that's our model. And our competitors have all these captive fabs, they're over the 90 nanometers, for power, well over 90 meters, 90 nanometers. And we will keep using the smaller node just take advantage of it.

And last comment for me is that 10 years ago, BCD was a primary differentiator for a number of reasons. So just like it was the BCD technology 10 years ago, modules and system-level integrated solutions is what differentiates us today.

Go for last question.

I have one follow-up. So also technical. So because when Jinghai mentioned about the ED solutions, so you keep thinking about shrink -- improve the power density. So to me, like you will make the die smaller and because you have the monolithic solution. But like there's a market or like a customer is saying like when you keep shrinking the die size, when you put the driver and the DrMOS very close, then you test it and thermal will impact each other. So I mean like that means for the future ED solutions, actually, the improvement to get the higher power density actually can only go for packaging side. For the silicon side, just you can have minor improvement.

Well, what is the -- our power density is still lower than what the GPU does consume. That's total consumption. And when you say we generate on each phase, Intelli-Phase or those each modules, and we're far less than the GPU consumption is, very, very little. So crosstalk on the thermals, it's not a factor at all. So we don't have that problem. All right. We don't have it. Good. All right.

All right. For those of you online, this concludes our presentation for today. For those in the room, I invite you to get some food and enjoy the demo area we have right over here. Thank you very much.