Thank you, everybody, for being here. My name is Marco Lagos. I run Morgan Stanley's semiconductor investment banking business here in the United States, and I'll be moderating today's discussion with Rajesh Vashist and Beth Howe of SiTime. Before we begin, let me read the following introductory remarks here. The company has asked me to remind you that today's discussion includes forward-looking statements that involve risks, uncertainties, and assumptions, which are further described in SiTime's SEC filings, including SiTime's most recent Form 10-K, and that actual results could differ materially and adversely from those anticipated or implied. SiTime assumes no obligation and does not intend to update such forward-looking statements. For more information, please visit SiTime's investor relations web page at investor.sitime.com. So, with that out of the way, again, it's my pleasure to introduce you folks to the CEO and CFO of SiTime here.
Rajesh, thank you for being here. Beth, thank you as well. Why don't we start sort of high level? SiTime has just crossed the five-year mark as a public company in November of this year. Side note, the company went public at $13 a share. As of yesterday, it closed at $155 a share. So, it's been a nice run. Congratulations on all that success.
Thank you.
But why don't you talk us through the journey of sort of that timeline, just the last five years? And I think everybody here should know also, Rajesh has been with the company since 2007, through a lot of the journey. But what's life been as a public company? What have you accomplished? What does that journey look like over that period?
That's quite a wonderful thank you for coming in and listening to us. Just a small bio note, I've been in the Valley since 1984. I was first-time CEO in 1999, landed up taking that company public in 2005, Ikanos Communications, and then landed here, as you said, in 2007. One of the things that I have always believed in, starting with the IPO and even further back, is that timing is critical to every electronics. Timing is a heartbeat of electronics. Timing is necessary. Amazingly, no one company has dedicated themselves to solving hard timing problems across the board.
There have been companies that have done clocks, some resonators, some have done IP, some have done oscillators, but no one has ever said, committed, and said, "We're going to solve tough timing problems across the board." This is the fundamental difference between SiTime and most other companies. In 2007, it was a $5 billion market. Now, it's about a $10 billion and growing market. We have taken our SAM up to $3 billion. We were $200 million-odd last year, and we are committing to a growth of 25%-30% for this year and something similar next year. So, I think that puts us in a unique place because when we went public, there was very little ADAS, very little automated driving, so we didn't have an automotive business. At that time, we did not have an AI business.
At that time, we didn't have an Apple modem business. Now, we do. It's just been declared in the latest 16E phone that SiTime has a couple of chips in there. So, the thesis is that timing is ubiquitous. Timing gets tougher as rates, data rates, increase, latencies decrease, the environment gets more and more challenging, size, power, and SiTime is the company that wants to fulfill that. So, that's been the thesis that's taken us up all the way to where we are. So, we want to, we have only one song to sing. It's timing, differentiated timing, and that's what we do. We sing that song.
Terrific. So, Beth, this could be for you, but what does the business composition look like today? You mentioned a lot of different end markets. What is sort of the mix today? How will that evolve over time, just generally?
We have a very balanced mix of our businesses, and that's part of our strength, is that diversification of the applications in our portfolio. We have 300 to 400 different applications across three major customer segments. The first one is our communications, enterprise, and data center application area, and that was about 34% of our business in 2024. The consumer mobile IoT, 33% of our business, and then our auto, aero, industrial segment, which is another third of the business. So, again, a diversified portfolio of applications, and as Rajesh said, some of those are applications that have been with us for a long time. Others of them are newer applications. Some of them are very fast-growing applications. Others of them may be growing more at market rates.
But that diversification, everything needs timing, and that's what we want to do is differentiated timing, the innovative applications within those different segments.
Got it. So, we talked about differentiation. Specialization and timing, obviously, in and of itself is a differentiation. But what is SiTime's differentiated core competency? If you had to sort of synthesize that for the folks here, or what are they if they're multiple ones?
Right. So, for the last 70 years, that timing has been needed in semiconductors or electronics. It's made of quartz crystal. So, quartz is a non-semiconductor material. The world runs quite well on quartz. So, our thesis was that the quartz had done great, but it needed to be disrupted through semiconductor technology. So, our semiconductor technology, which is MEMS technology, which stands for Micro-Electro-Mechanical Systems, this MEMS technology is built in a fab. It's using semiconductor equipment. It uses semiconductor materials and uses our proprietary processes. So, we generated all of that, and with that, we combined it with the analog IP that we have built. So, we built our own PLLs. We built our own regulators. We built our own products. And then the third piece of IP is putting them together to make a subsystem.
Sometimes the subsystem is as little as one MEMS chip and one analog chip together, and in our most sophisticated one part, which is called Epoch product, it is three MEMS devices and two unique analog circuits, mixed signal actually circuits, for a very unique package that we produce. This is part of our technology. We've spent $hundreds of millions, literally, on R&D. We are on our fifth generation, sixth generation on MEMS, and we are probably on our fifth and sixth generation on everything else as well. We're doing something that hasn't been done before.
That is right. Okay. I mean, you talked about R&D and the massive investment you've done. Innovation here has been pretty remarkable. It's probably lost in the fact that it's timing. Timing has evolved and gotten harder, as you said. Since the second half, I think, of 2023, you've introduced 10 new platforms, 40 new products. It's remarkable. What sort of allows you to innovate and change and develop all these products over time so quickly and at such a pace?
Part of it is the definition of the product as a platform. So, when we say 10 and 40, the way to rationalize it, these are 10 individual products, which are a combination, as I said, of the analog and the MEMS, analog and the resonator MEMS. But there are many derivative products, up to 40, maybe even up to 100, that come out of these. So, we start with an architecture which is already set for some form of customization for the customer. It's programmable. Think of it like an FPGA technology. It's programmable for a customer. So, we are able to deliver that from the start, and that's how we think of coming into it.
And we are also very deeply connected to the leading providers in, for example, watches, leading providers in military fuse-grade technology, leading providers of ADAS, leading providers of low Earth orbit satellites, leading providers of AI, and we collaborate with them to develop these exceptional products. And because they need some problem solved and nobody else is solving it for them.
Got it. So, we talked about diversity of end markets. That's a lot of different go-to-markets to manage.
Indeed.
So, how at your scale? In particular, it's remarkable that you've been able to do that so effectively. How have you managed through that, and kind of how do you, your sales force, your R&D teams, how do you sort of get them focused on the right things across so many priorities?
Right. So, many people have certain favorite end markets. They like networking, telecommunications, AI, or ADAS, or whatever. Our goal is it should be a market which is a minimum of $50 million. So, the scale of the market should be for a product should be no less than 50. Our second requirement is that we should bring significant value to it as measured by the premium we charge over other products that could be used. Now, this is normally not what people do. They don't lead with price, but we lead with price to differentiate ourselves from other technologies and other solutions. Because if the customer is willing to pay our higher price, that means we're bringing enough value to it. And so, that's a particular requirement.
And then the third is we believe that a business of this scale, when I look at analog companies like TI and Analog Devices, I see that their gross margins are 65%, 70%. Their net margins are 30% plus, 40%, maybe even 40%. And I think that's a great way to build a business, and we're committed to something not at that scale, but close enough. Our goal is to make it a very profitable business with high growth and high differentiation and predictability.
Got it. So, related to that, right, there's obviously a—you mentioned sort of working with and access with the customers. Talk a little bit about sort of how you're embedded with them in developing products and thinking about their roadmaps. How early in sort of their planning cycle do you get involved? What kind of visibility, for Beth, what kind of visibility do you have as you think about sort of planning, both from the cost and the sort of revenue side of things?
Right, so this is very much a design-in business. In other words, it's the exact opposite of a commodity business. It's a design-in business. Sometimes we are involved as little as nine months away. Most of the time, we're involved a year and a half, 18 months away. And in the case of some long-term projects, it might be as much as four or five years that the development happens. We tend to have, when we look at our business, we tend to have a high degree of ability to forecast the future, particularly when we come into quarters, and I know that Beth can comment on that visibility as well, but particularly when we come into the quarters, we are in great shape.
And the fact that as early as last year, we guided for this year 25%-30% growth. That means it's those design wins that we have in our funnel that drive all that. I don't know if you wanted to add something to it.
Yeah. No, I think that's absolutely right. It is a design-win business. And so, as we look at that, the coming quarter, but even the coming quarters, and we've improved that connectivity with the customer, getting more embedded with the business, embedded with the engineers to understand how our products are progressing, and therefore getting that visibility of our design funnel, our design wins, and then the move to mass production. Again, not perfect, but we've got a lot of visibility.
Yeah. And when it comes to sort of—we've talked about the customer side and visibility with them from the supply chain on the other side, right? So, people that supply you folks, inventory building, and we've talked a lot about sort of supply chain constraints and issues over the last few years. How have you managed through that? How do you feel you've performed relative to sort of managing your inventory, supply chain, and sort of that side of the crunch, if you will?
So, if you think about our supply chain, so we get our MEMS technology from Bosch in Germany, and then TSMC is our analog technology. And as we manage that, we want to make sure that we are always able to deliver to our customers. So, we think about the lead times that go with that and ensuring that we can assure supply for these major customers of ours. So, we want to make sure that we have significant inventory to be able to respond if they have upside, if they ask for more products, etc. And because our products are programmable and because much of it is in wafer form, we are able to then—they're long-lived, so we don't have the perishability that you might with some other products.
Because they're programmable, we can, in essence, build the products and then program them for a specific customer as those orders come in, which gives us more flexibility as well.
Terrific. Okay. All right. I mean, we touched a little bit on sort of the business mix for end markets. So, let's dive into specifics and want to talk a little bit about the CED business in particular because it's exciting. But I don't know if you heard this, and I'm going to make this joke probably for the third time since morning, but AI is kind of a big deal. So, what is it about that opportunity that you, from a timing standpoint, are probably most excited about? How do you think about strategically your future within AI, what you enable within that?
Yeah. So, you mentioned CED. That is the three-letter acronym for Communications, Enterprise, and Data Center. So, while we're going to talk about AI and data centers, we should also make a plug for our communications business, which is the standard Ericsson, Nokia carrier business, which, believe it or not, is coming back and is doing fine. Our enterprise business, which is the classic Juniper, Arista, Cisco business, to those kinds of guys, that's also doing reasonably well. But as you point out, the data center business is on fire and has been for the last year and a half to two. We are in every aspect of it, right? We are in the processing part, the GPU, the CPU. We are in the switches. We are in the networks. We are in the acceleration card. And particularly, we are in all the interconnects.
We are in all the optical modules. We are all, meaning we are in the optical modules. We are in the smart cabling. We are in the AEC. So, generally speaking, SiTime has significant dollars in the racks, whether it's top of the rack, whether it's middle of the rack, whether it's the switches, and so on. So, we see this as a great opportunity. We see ourselves growing rapidly in this. I know there's been a lot of concern around the not expensive way of coming into this. We think that as far as the connectivity and the timing piece, I think that's mostly doesn't care about what kind of models we're talking about. It just is whether it's the old traditional model, old, a year and a half old traditional model, or the new two weeks, two months old, new DeepSeek-based models.
I think it's still SiTime has a role to play. So, we see for the next couple of years growth coming through this business. We also see it coming through other areas where we've not touched on outside the data center, in the edge, whether it is in IoT, whether it is in personal devices, or it is in robotics, or it is in any of those. SiTime has a significant ability to perform in automotive significant ability when and if those markets come out and come to fruition. So, we see that coming too.
So, just to double-click on edge, very important, but sort of the way a lot of folks think about AI and the opportunity around data center and everything else is training and inference, right? And it feels like there's an inflection which we're at now around the spend and where it's going to start tipping more into inference versus training. But you play both. You matter to both. Can you talk a little bit about how it matters to training, how timing and your product in particular matters to training and how it matters to inference?
I mean, it's actually a pretty simple answer. It's not a very sophisticated answer. The answer is that wherever there is need for stability of the frequency, the low jitter, it's a technical word, there's low jitter is better, better phase noise, better power, better size. Then there's a bunch of esoteric timing things, parameters that I won't go into. SiTime's job is to perform exceptionally in all of them because they get needed, whether it's in automotive, whether it's in ADAS, whether it's in missiles, whether it's in personal equipment, or indeed it's in inference versus training. I think it's not that different for us. We don't produce significantly different other than in the all of these have to perform at the requirement that there is.
Yeah. And so, it's no matter what, there's an opportunity for you guys to win.
That's right.
Beth, as I think about sort of the mix of ASPs and what they look like, or actually we haven't, but what is the mix of sort of just high-level? Rajesh, you talked about charging for the value that you deliver in that side of the equation, right, on the ASP side. On the cost side of the equation, you're also incredibly efficient. What have you sort of done and implemented? We talked about the supply chain a little bit, but what is it as a team and as a sort of cost structure, what have you implemented to actually also help you on that side of the gross margin and operating expense equation?
Okay. Well, so maybe to talk a little about the ASPs first, and then we can get into the product cost. So, our ASPs, we have a range of ASPs from $0.30, $0.40, $0.50 through $1, $2, $10-dollar products, all the way up to products that are $50 or $100 or more, depending on the application. So, consumer applications tend to be products that are typically less than $1. The data center or CED has products that range anywhere from $1, $2, $5, $10, up to probably $25 or $30. On the other end, in the aerospace and military, given the requirements of those applications, those can be $50, $100, and multiple hundreds of dollars depending on the application. So, a very wide range of ASPs depending on the customer end market and the application there. So, that's the pricing.
As Rajesh talked about, we are a premium provider. We charge a premium for the innovation, for the value that we bring to those customers. That is a core part of who we are as a company. On the product cost side, and this has been a lot of questions, because we've had so many products coming out, we're working through new products typically have to mature in the cost curve, and so when they launch, they typically, you've got to work through to improve the yields, to improve the processing, to reduce the test times. That's where we are right now, and because we've launched so many new products in the last year, year and a half or so, we've got an unusual number of products that are going through that phase.
And that combined with the fact that some of these products ramped much quicker than we would like. And so, you have a little bit of a choice. You can either get the product to market, maybe not at the optimal cost structure, but drive the revenue, get the sockets. And that's what we chose to do is really drive that revenue. And this year, we're working on the cost as those mature. Is it the same time that we're delivering for customers? Because at the end of the day, that's what we want to do. Is it the same time driving those cost curves down, maturing the yields, maturing the test times so that we can improve the gross margins over time of those new products? And it's a little bit of a where we are in our life cycle. Again, you said five years public.
We had this ability to really expand our portfolio over the last couple of years, and now we're bringing those to market. And there's some growing with that, but we'll get those costs down. And we know how to do it. It just takes a little time. And it's really that balancing act between driving the revenue and driving the cost curves.
Yeah. And speaking of a balancing act for both of those things, it's probably harder and different depending on the type of customer. So, one thing, I think, it's timely that you're here. One thing folks have, I think, generally been focused on is the phone. And are you in the phone? And so, the question is, are you in the phone? And what does that look like? And what does that opportunity in this sort of upgrade cycle look like for you folks?
Right. So, thank you for that. It's been a pretty decent-sized long journey, and as you see, there's some breakdowns of the new 16E phone, and that showed two chips from SiTime in there. That's very gratifying because I think that's good to see that people can see that, so that's what the data information says now.
That's great. And as you think about sort of, it's kind of like asking somebody what your favorite child is, right? Which end market? But the heritage of the company is a lot of comms and mobile. What do you see sort of, what have you been able to learn from that end market experience that you're seeing translating into some of the newer end markets that you're putting?
So, like any good parent, I'm going to dodge that question.
Good.
And say that it's based on the attributes that we are looking for. So, the attributes are it should be high margin in its category. We understand that consumer products are not 75% gross margin and that 50-55% margin is pretty good. But we're looking for high gross margin in its category. We're looking for high growth. We're looking for sticky, high-value product. And we're looking for multiple generations of that product in the customer's end product. So, those are the four factors because that's what gives us success and gives us exactly that diversity, Marco, that we're looking for because the world's a strange place, as we know. We know that when we went public, I was asked the question about how much automotive business we were planning to do. And I said exactly zero because automotive did not treasure the values that we brought to the fore.
Lo and behold, by 2021, we had a very nice automotive business because the market changed significantly. When we went public, of course, I said this once before, there was no AI business. Now there is. So, we are on the hunt for end markets. And there's always many of them. And there's about 30 of them that then grow at anywhere from, I'm sorry, excuse me, 20% north of that to 40, 50, 60% annually. And we're looking to isolate those and go serve them regardless of which market. Having said that, the communications enterprise data center business tends to be the leading driver of technologies which then get consumed in high-end industrial, high-end automotive, and of course, military, aerospace, defense, and even sometimes trickle down into the consumer side. So, we've seen that.
But we will look at all our markets as long as they meet those four criteria.
Yeah. And you talked about sort of the standards, if you will, right, that you have to meet. You challenge yourself to basically go after the hardest customers to please. And that's translatable into a lot of sort of other end markets in a very good way. But what do they in general, across all your customers where you do kind of have to meet these standards, do they care about the innovation and sort of the technology differentiation? Do they want reliability, quality, which is also another standard of yours? How do they kind of think about those two things back and forth? And how do you meet those?
Yeah. So, let's take a look at AI. Clearly, we all can talk about how AI is driving performance. So, all the metrics of performance are critical. But I have to tell you that when an end market explodes the way it does, supply chain, quality, reliability, the ability of a vendor to perform and be credible, I think those become very important. So, for example, in the area of optical modules, we do supply some differentiation in performance. But where we really make the difference is in supply chain. Our customers can depend upon our quality, depend upon our reliability, which is exceptionally better than our competitors. But they can also rely upon our lead times and the integrity of our supply. So, supply chain matters a lot. But in the end, we're a bunch of engineers. Even Beth is an escaped engineer from Stanford.
We're all a bunch of engineers. We want to go for the performance. We want to go for the edge of the universe.
Yeah. Where did you see the most adoption over the past year? And are there particular end markets where a lot of the acceleration and growth is going to come from over the next year? So, backward-looking, where was most of the sort of innovation? Forward-looking, where's most of the growth going to be sort of coming from?
I think as we've been talking about, the CED or AI data center has been explosive growth, right? 156% growth in Q4, 121% growth for the full year. We expect strong growth in the data center market as well as we go through 2025, and that's probably the biggest driver of our growth. But again, as we've talked about, there's a diversity of applications, so we see other opportunities for growth as well. For example, in aero and defense, those are long lead time, and they have a long gestation period, but we think that there's some really high-growth opportunities within there, whether it's the military complex or satellites or fuses and different applications there. That can be a very high-growing area for us as well and sticky over long periods of time. It takes a long time to get in them.
But once you're in, you can be in for many, many years.
Is it similar to automotive sort of from a time standpoint where you go six, seven, eight years once you're kind of designed into something? Or is it?
Probably even more. There are missile programs that we've been involved with for the past seven years already, and they probably grow another seven, eight years.
Fantastic. All right. Okay. Here's a little bit of a hypothetical, Rajesh, and we'll get into sort of more real stuff in a minute, but where's timing going? How are you going to disrupt it over the next decade, and can an atomic clock be in your pocket at some point here?
Yeah. Yeah. I mean, SiTime has always maintained that we are a timing company. SiTime, Silicon Time, we're a timing company, and we want to disrupt timing. So, our sights are not just on oscillators, which is where we started. Then we moved to clocking. We're already doing that. And we've stated that we're going to be in resonators shortly. So, that's the sort of classic timing component business. But we also believe in timing software. And we're putting out 1588, which is a timing software, out in the market that's helping our CED business and our automotive business. We also think that timing IP becomes important. Timing IP is used sometimes. So, that's one place for SiTime to go. Another place for SiTime to go is timing modules. There are Grandmaster modules that are important in this. And of course, much talked about atomic clock.
The atomic clock world is a fun place to think of, and particularly when you think of it as a small module, which is out in the future, definitely. But SiTime has its ambition set on all of that. And we think that we can do it.
Yeah. You know, I couldn't let you leave the conference without addressing something top of mind for everybody across all our ecosystem here. But we woke up yesterday to a 25% announcement on tariffs from Mexico and Canada. And then I think the day before that, we were talking about 10% on China on top of the 10% that already happened. You folks tend to be very strategic in your thinking about the world and the economy. Sort of how have you thought about sort of the change in geopolitical landscape, the sort of behavior around protectionism in the U.S., what happens with China? Just a couple of minutes just on how you've thought about that.
Yeah. So, first and foremost, we have built up a pretty decent-sized inventory in our raw materials of finished wafers that can take us through the ups and downs so that every time a tariff is put in one day and then taken out 10 days later, we don't have to respond to that. We have some stability. So, we have some stability for that. That's not going to last long. That's only about $60-$70 million worth of inventory that can only last so long. Beyond that, we have always protected our IP and our ability. So, we don't develop any significant or any R&D in places other than Silicon Valley. That's our base. And then in Netherlands and in other, call it EU or Japan or India or Taiwan, of course. And beyond that, our production is in Taiwan, TSMC. But TSMC also has a fab in Oregon.
We're in the fab in Oregon. Our fab for MEMS is strategically in Bosch in Germany near Stuttgart, and it is our technology. It's not the Bosch technology. It's our technology that we're using to develop in Germany. We have our packaging centers in Taiwan again, but we also have them in Malaysia. We have them in Singapore, and we have them in Thailand, so we're trying to spread our risks around. Having said that, if the tariffs are going all across, then I think we're all going to get impacted no matter what. There isn't much of a place to hide, particularly if there are 20% tariffs coming from China over a sustained period of time. We can count the number of products that are made, our customers' products that are made that come from China, so yeah.
Yeah, but it sounds like philosophically similar to your revenue being balanced. Your approach here is balanced, and it sets you up to get through whatever comes here.
That's right.
It's here, which is great.
That's right. And we just took a very senior person at SiTime in our management and have given him only one task. And that is to be a strategic supply chain hardener. His job is to harden our supply chains. And he brings 40 years of experience, and that's all he's going to be doing. So, it's a different world from when we went public. We didn't have a couple of wars going on. We didn't have COVID. We didn't have tariffs. And the world wasn't divided the way it is. But hey, we don't get a we have to thrive regardless of the world we are in.
Yep, and you're controlling what you can.
That's right.
That's terrific. All right. Well, we've got a couple of minutes left. I'll open it up to questions.
I have just a basic understanding question since I'm not familiar with the space. But can you just walk us through the competitive landscape, just market share, trends in terms of market share, and guessing you guys have been gaining share and guessing the competitors are more good players?
Correct.
Just help us understand from a firm standpoint. Are your customers mostly dual or multi-sourcing their timers? Or is it so you guys single-source in your application?
Right. So, the question is whether we are multi-sourced, single-sourced, and what's the competitive landscape like. So, and market share. So, if you think about it, we had a $200 million company revenue last year in a $10 billion market. So, really, from a market share point of view, we're puny. We're 2%. Even if you were to get to a billion, we'd still be 10% of the market, and I presume the market would have grown. So, at the end of the day, I don't think about it as a market share gain. I think about it as a customer satisfaction gain. And I think that many of the products that are out there, which have been around for 50, 70 years, are not able to satisfy customers' needs. So, we go where customers' needs cannot be satisfied.
So, in that sense, I think we don't have competition, even though, of course, we have competition. Quartz Crystal has a lot of competition. But in the sense of a competitive solution, when a customer wants a problem solved, SiTime is uniquely qualified in the ability to do that. As far as single-sourced, I think that it's safe to say we are anywhere between 70% to 80% single-sourced in the way that our customer depends upon it. Sometimes it's pure single-sourced, where they cannot depend on someone else. And there are some other options where they can use other people if they walk back on their performance needs or their supply chain needs. Yeah.
Terrific. Any other questions? One more? Oh, sorry. You go ahead.
One follow-up on the competition. So, you talked about the quartz Crystal. I was more curious about the MEMS. Comparable product competition, who are the names?
So, in the area of MEMS, we had competition, but the last competition at about nine other competitors from 2007 to about 2011. And then they dropped out. They couldn't make it. So, right now, in terms of volume, it's SiTime. And there's a very small business that comes out of Microchip from some acquisitions that they had done. But other than that, there are no other suppliers of any note in the market that are using MEMS, to be very specific.
Last question. Why did you choose to go with TSMC in the U.S.? Is this something your customers have asked because of supply chain hardening? Or is this something that's interesting?
Yeah. Yeah. So, the question is, why did we go to TSMC as opposed to somebody else in the US in Oregon? That Oregon fab is, I don't know, 25 years old. And we've always used it because I think it's always nice to have a spread-out supply chain. And it's pretty much an exact replica of it. It's a 180-nanometer fab. So, it's pretty much the same as what we use at TSMC in Taiwan.
Okay. All right. Well, we're at time. It feels like it flew. But thank you so much.
Thank you.
Thanks again for coming and appreciate the time.
Thank you very much.
Thank you.