Doug from Fabricated Knowledge on $SITM (podcast #130)

Doug O’Laughlin, founder of fabricated knowledge, returns to the podcast to discuss his thesis on SiTime (SITM). You can find his SITM write up here and Doug’s first podcast appearance on RMBS here.

Please follow the podcast on Spotify, iTunes, or most other podcast players, as well as on YouTube if you prefer video! And please be sure to rate / review the podcast if you enjoy it, or share it with someone else who would enjoy it (more listeners is a critical part of the flywheel that keeps this Substack and podcast going!).

This podcast is brought to you by Tegus. I use and love both Tegus and Bamsec (which Tegus recently bought); you can find some of my work with Tegus in my cable deep dive here.

Disclaimer: Nothing on this podcast or on this blog is investing or financial advice; please see our full disclaimer here. The transcript below is from a third party transcription service; it’s entirely possible there are some errors in the transcript!

Andrew Walker: Hello, and welcome to Yet Another Value Podcast. I'm your host, Andrew Walker. If you like this podcast, it would mean a lot if you could rate, review, and subscribe wherever you're listening to it. With me today, I'm happy to have it on for the second time, Doug, from Fabricated Knowledge. Doug, how's it going?

Doug: It's pretty good, man. I mean it's been a volatile time in the market, especially semiconductors in the last 2 weeks. But other than that, pretty good. A fun time. Lots of interesting things going on. At least a lot of content and things to think about. Yes, that's how I've been thinking about it, at least.

Andrew: As a non-semiconductor expert, it feels like it has been a bloodbath in semiconductors. Interesting times might be putting it kindly. Anyway, let me start this podcast the way I do every podcast. First, with the disclaimer to remind everyone that nothing on this podcast is investing advice. That's always true, but I'll just remind everybody. As we said, semiconductors can be a very volatile and very kind of cyclical industry. Keep that in mind when you listen to the podcast. Please do your own work, consult a financial advisor, and all that type of stuff.

The second way I start every podcast is a pitch for you, my guest. People can go listen to the first podcast that we did together back in June on RMBS if they want the full pitch. I'm not going to do it again because that was probably the nicest, gutsiest pitch I've ever done and it might just embarrass both of us. But I'm just thrilled to have you back. I know people love the first podcast we did together. I'm super excited to have you back for the second time. Anyway, for the second time, the topic we're going to talk about today is SiTime. SiTime, I believe, is how we pronounce it. The ticker there is SITM. I'll just pause there and ask you, what is SiTime and why is it so attractive?

Doug: Okay. SiTime is a MEMS timer company. Timing goes into every single device and all of the electronic devices in the entire world. Essentially, you have to have it, your devices have to have a time, a clock so that everything can work synchronously together. Up until very recently, in 2006 when SiTime was essentially got founded. 100% of timers globally were via quartz. Quartz is the legacy incumbent technology. It's super reliable. It's been around for 100 years.

SiTime is this company that did one of like the holy grails of semiconductor technology that everyone wanted to make happen, but the technology roadmap took quite a while to get there. In 2006, they essentially put down a patent for this special type of packaging to create a MEMS resonator. Quartz resonators, essentially, are legacy technology. And to be clear, they're very good. MEMS, by the way, stands for Microelectromechanical System. Essentially, think of it like a tiny mechanical system but at a semiconductor size.

The difference between MEMS and quartz is quartz uses Piezo electric stuff where, essentially, if you put electricity into it, it resonates at a certain frequency. MEMS however does something very similar, but it resonates mechanically using energy as well. And so, think of it as like a tiny, tiny, tiny tuning fork. Like when you slap a ruler and it vibrates? Essentially, it's vibrating consistently at a frequency that you want. These timers are really important. They're super niche, but they're on all devices everywhere. The total addressable market is about $5 billion.

Andrew: Why do products need these timers inside of them?

Doug: Well, first and most important is to keep the time of day, right? A perfect example is Garmin. Garmin, smartwatches, smartphones, everything needs to have time. And so, the way time is done is, you can actually do it with a circuit, but that's actually very noisy and has a lot of electrical problems. Quartz is extremely stable and MEMS is a new technology that's becoming a lot better. Everything needs to keep track of the time for your devices to work. Imagine if one device was telling device number 2 to do something, but they did it at the wrong time and so this does it like 15 seconds late. Like what's the point, right?

All the instructions have to be going together at the same time. It's called a clock, like a clock frequency. And so, each of these clock cycles decides what actions happen. And that's like on a CPU but anyways, all in, everything needs a clock. In order for the semiconductor devices and electronic devices to work, everything has to be on time. Semiconductors are probably the most popular device that it's used in conjunction with. That's the way to think about it. Quartz is like a legacy technology, and it's actually not made of silicon versus MEMS, it's actually made of silicon. It's close to a real semiconductor if that makes sense. That's the difference technologically.

Andrew: You and I were talking for 10 minutes before the podcast about a bunch of different stuff, moving and everything. I got to tell you, I just love how passionate you are once you start talking about semiconductors and stuff. Your inflection picks up. It's contagious how passionate you are about these things. Just one last stupid question. I'm going to say iPhone, but we'll talk about Apple and iPhone later as it pertains to their side of them so don't take anything on iPhone. But somebody might say, "Oh, timing. My iPhone connects to the internet for a time. It doesn't need time in there". I would just say, you're wrong. There are obviously lots of timers in there. But please, tell me if I'm correct or incorrect, just to give everybody a very simplified example.

Doug: Yes, that's true. There's actually this whole thing called a stratum, which is this cascading network of timing and how time is distributed across the network. The time gets distributed across the network, but you also have to hold the time in each and every unit. If they're not constantly pinging, think about the data for that as well as not everything is connected to the internet all the time. It'd be like, okay, well, now we can't take a photo if it's offline. It has to be connected to the internet. Stuff like that. Every aspect of your semiconductor device or whatever has to have some kind of context of time. Yes, that's a quick example. But, I think that answers the question.

Andrew: Obviously, this is a little bit more technical podcast than a lot of the podcasts I do. But we look at stocks, we invest in companies because we're trying to generate off of it. So we're going to go into lots of technicals but let's just go high-level. The first question that I like to ask every guest, the market is a very competitive place, you're obviously looking at this company and you think it's very attractive. We'll talk about valuation and all that reasons, but just high-level, it's tough to get an edge, what do you the market is missing that you're seeing in SiTime that will lead to a kind of risk-adjusted alpha from investing in the company?

Doug: Okay. We'll just say it's like varying perceptions, right? That's kind of like what we're trying to focus on here. I would say SiTime, in particular, is probably one of the biggest poster children for things that you don't like right now in this market environment. One, it was definitely a COVID beneficiary. There's no way to say it wasn't. And two, its extremely high duration growth. Those two things are the things that markets hate right now. Third, there's kind of a near-term air gap because this is a semiconductors company's first cycle essentially and they definitely over-ship. Near-term revenue is getting absolutely hammered. The stock was doing wonderfully until the last quarter. They got it down from 50% to 35% year-over-year growth, which implies a negative Q4. Since then, almost every incremental bit of news has been very negative. We'll see if it's a match, but there's a glide down coming in this company in the next quarter, in my opinion.

To be fair, that's happening everywhere in the semiconductor space as we talk right now. AMD glided down yesterday, right? That's one of the biggest [inaudible] just do the glide down. Everyone everywhere is gliding down. SiTime is not going to be immune to this at all. What I've been noticing is that this company, essentially had this really rosy, market grows forever. They grow penetration in this market forever. I think they will continue to take a lot of market share. But all of a sudden, all of these near-term headwinds came at the exact same time. This company is extremely high-duration growth. Interest rates have massively exploded. I actually did a very full DCF, like the old-school whack. I did the whole damn thing. It's a high beta stock. Actually, the equity risk premium for this company is 16% because of the two-and-a-half beta. Market ERP is like 5%, risk-free rates, 4%. You do the math on that. It's like, "Oh, your discount rate is like 16 to 17%".

This is maybe a very academic way to think about valuation. But obviously, this matters, right? Extremely high multiple, extremely high discount rate, and extremely high equity risk premium. These are the kind of stocks that are getting hurt the worst and it has an airgap in profitability. All these things are making a very ugly short-term situation that I think is really ugly. I did some numbers of what I think they really can earn, and I'm like, thins[?] lower than the street for 2023. I think that it's ugly and it hurts my stomach to think about buying the stock, which historically, back before 2020 was like a good sign. But now, obviously, it's going to be hard. I would say the biggest fear is dead money, right? It's not going to grow a high-multiple, COVID beneficiary, and high-duration. Who wants to own that right now?

Andrew: No, that's all great. Look, just to what you're saying, the stock starts the year at $250 per share. As you and I are talking it's 85-ish per share, right? So stocks are down 65%. I always hate sell-side earnings for next year versus the market. But the more you do it, the more you [inaudible]. I don't know what you're at, if sell-side earnings are 4 and you're at 3.50 or something, right? Like, a lot of people say, "oh, you can't invest in that". It's like, "well, the stocks are down 65%". I bet you the people who are actually buying and selling stocks have probably pretty much adjusted to it. They're going to make [inaudible].

Doug: The street knows it and management is doing their best to essentially disseminate, "hey, the first half of next year is going to be an air gap". No one really knows what it looks like on the other side because they're selling it to distributors. My EPS number is $3.20 for next year. The street is at $4. That's pretty bad. And my eBITs are much worse. I have an EPS that's better than it should be because I model the income interest of their cash balance, right?

Andrew: We got 500 million of cash on the balance sheet and their free cash deposit[?]. Again, I don't want to focus too much on the near-term earnings and everything because this is a very sexy long-term story. But I did have one question, you mentioned that air gap in earnings, right? And again, that's completely understandable given everything that's going on the street. But there was one piece that was curious to me, right? At the start of the year, the Q4, they say, "hey, 2022 we're going to go 35%". They report Q1 and they say things are on fire. We're going to grow 50% this year. Then they report Q2 and they say, "oh, things are slowing down. We're going to go 35% this year." Which is based on how good Q1 was. I think they said, I can't remember. H1 was fantastic. So based on H1, that means H2 is going to be an absolute disaster. And again, the environment stuff, everyone else is gliding down but it does strike me when you find a management team that guides up and then guides down, you do start to worry like, do these guys really have a handle on the business?

Doug: Yeah, that's one of the things I'm really worried about and I guess the one place I'm going to give them a pass on is, this is their first cycle. I do think the CFO there is very thoughtful. Essentially, they are definitely sitting at the end of the bullwhip effect. I'm going to guess they haven't really worked on understanding their true end-market demand this cycle. Whereas more mature semiconductor companies, obviously are concerned about double ordering. Their biggest problem was how can we sell more stuff now. Because I think in 2021, their whole thing was, we need to have more than just a distributor relationship. So they've just been expanding, expanding, expanding. Essentially, expanding the channel and it's been one of the strongest cycles. This is their first full cycle as a Semiconductor Company and it is obviously really ugly. And I guess just like everyone else, they didn't see it coming but they also maybe got a little over their skis[?].

Andrew: We'll talk more risks later. But I do want to just go to one risk, right? Like obviously, revenue going from up 50 to down 35, next year there's going to be an air gap. The first thing you start thinking of with a semiconductor like high fixed expenses and revenue swinging like that, you start worrying about the business. We've all seen semiconductors with leverage go under, right? And just to put that to bed, the company has 580 million of cash on its balance sheet. They are free cash flow positive. Maybe they won't be next year if things get bad enough, but I don't think we're talking in the near to even medium term. I think they've got plenty of liquidity where they're not going to get cycle margin calls stopped out. Please tell me if I'm wrong, but just to put that up here to bed for anybody looking at that.

Doug: I don't think so. I think that's one of the reasons why, to be honest with you, buying SiTime is very hard. I think it's not an easy stock. It hurts. It is a high risk, for sure. It's much smaller. They were COVID beneficiaries. You definitely worry about them over-earning and maybe this is not a reasonable way for them to grow at all in the future. But then, the more you look at it, the biggest de-risker is the $580 million of cash on the balance sheet because that makes you, it's like look, this thing cannot go under, right? There's about...

Andrew: That's the famous last words, man. I can't believe you said that.

Doug: If they really screw up some capital allocation, sure. But it does have a little bit of put[?] with like $25 per share in cash and that really helps in terms of the risk-reward profile. Because it's kind of like the margin of safety thing where you have a giant cash load that really does help de-risk some of the downsides in theory. But one of the things that I'm really worried about is the gross margins. And I think that anyone who's paying attention to this company should be because it makes a 67% gross margin at its peak last quarter, which is 66.7 or whatever. That is a street high. That is as good as it gets anywhere in the entire semiconductor industry. And that's a little worrying to me. I was like, how is this even possibly sustainable?

I did a lot of work on [inaudible] economics here. And so, I would say that about 10 to 15% of their stuff is fixed. They actually are a fabulous model, which is one of the benefits of their business model versus competitors is that they don't actually own the factories. And that's one of the cool things about doing this as a semiconductor-based business instead of quartz. Because they get to use TSMC, they get to use Bosch. Bosch is their MEMS fab partner and TSMC is their semiconductor fab partner. And fabless companies are asset-light because they don't own the fabs themselves. They just put in the orders, obviously, they pay the money to the fabs and they get semiconductors in return. And then, the fabless companies have to go around, turn around, sell their products on the market and try to price competitively against the market.

So, I think the gross margin downside and margin downside are not as bad as say, a very fixed-cost company. Because they're fabless and that really helps. That's the appeal of being a fabless company. That being said, their competitors, all their gross margins expanded, their gross margins expanded. They are very sensitive to ASP increases. And I think that the gross margin is most definitely going to come down to earth a little bit, so.

Andrew: Perfect. So, let's take a step back, right? We started talking about how this is a company that does MEMS. They're taking share from quartz and stuff. So I just want to put that in perspective because I think the thing you really like about this company is you think this is a not multi-year, multi-decade growth story, right? So I guess just to start, I wrote the numbers down. If you don't remember. Just to start, how much of the market is MEMS versus quartz currently? And how are the two evolving over time? I'll start there. So that's just the market. That's not even SiTime. That's just the MEMS versus quartz and how are they growing? And then, we can talk about how that impacts SiTime.

Doug: So, MEMS is probably, the other thing too is the percentage of the market is a little bit debatable. But the number I use is about 4 to 5% today. SiTime is 90% or actually, I think 92% of all MEMS. Essentially, they don't have any meaningful competitors at all. So essentially, when you talk about MEMS, you talk about SiTime. There is about 4 to 5%, a few years ago they were like 2 to 3% and I think over the very long term, they could be something like say, 20 to 30% of the market. And when I mean, very long term, I mean 2040. I'm not saying, "hey, they are 20% of the market at the end of 2030". Which would be really aggressive. I'm saying, they continue to take share at the rate they have been over the last several years. And now, this company is part of the s-curve.

So they've been unprofitable for a really long time just because the revenue base never really covered their cost. Now, all of a sudden this company has legitimate gross profit dollars, legitimate revenue, legitimate earnings, and to reinvest in such growth and really become like a mid-cycle growth company and really chop away at the tab. There are just a lot of little things that they have against quartz because that's the big story. It's like, how are they going to disrupt quartz? And this is where this whole thing takes like if you talk to people in the industry, it takes a whole turn different because SiTime has actually been frankly, very promotional about their performance promises. They essentially will tell anyone who will listen, "we're 10x better in every way, shape, and form".

You actually do a lot of work on it, and you're like, "oh, that's some real apples and oranges". And so, the quartz guys hate it. They're like, they're just liars. They're comparing their highest end to our lowest end whatever. And so, when you try to do some real work on it, it becomes a religious war that you have to get. So it's like very hard to be very objective in this space and I've been trying to parse out what makes them when. There are a few things that really MEMS has an advantage over quartz. And so, the biggest one is miniaturization. Quartz events can only be shrunk so small as a crystal. Whereas, MEMS as a semiconductor, can go another half or maybe even half after that. It's microscopic that we're talking about the MEMS resonator itself.

It can be much smaller. And that matters for said for the total packages and stuff. The second thing is that it's programmable. Programmability is really interesting because quartz, whenever you make a quartz crystal, it takes 3 to 6 months to grow. So, it has this extremely jerky supply chain that it takes a long time for you to grow the exact crystal, cut the crystal, and then make the design. Well, imagine if you have some hiccup in that 3 to 6-month period, that's a pain in the ass. SiTime can literally press some buttons to change the frequency tomorrow. And so, they have that ability to choose their own frequency. Quartz has a product that does this, but the SP is not even comparable. And that just comes built into MEMS. So, I would say, MEMS has these little things on the edge. And also, it's improving technology.

This is the third generation of resonators and each subsequent generation has had meaningful performance increases. And when you talk to people who work in the industry in an extremely academic sense, they're like, there is a theoretical limit to how much better MEMS can get. But we are not there yet. It will take quite a bit and it'll take a lot of RNGs, which SiTime is able to spend. But it's going to improve over the coming decades and I feel very confident about that.

Andrew: Perfect. So what you've got is you've got new technology, there are debates on how much better, but probably better than the old technology. It's improving pretty rapidly. I said new technology choice, but this started in 2006, right? So we're 14 years into the cycle or 15, whatever it is. Versus, I don't know when quartz, sorry. But Quartz has been around much longer than that.

Doug: There's actually a really cool documentary in 1942, "Quartz Goes To War". It's old technology. We're talking about the 1930s or1920s, it is probably 100-year-old technology.

Andrew: New technology, more customizable, a little bit more flexible, all this other stuff. So you can see, right? we said about 4 to 5 % market share currently, you said 20% market share by 2040 or something along those lines, right? So we're talking Forex market share over 20 years and you can see why that's happening with all that. But then, I guess the second piece of that is, "hey, it's not just Forex market share, but it's the market for semiconductors". Which is growing very quickly. So you could say it's Forex's market share. But the underlying market grows, I don't know, 2x, 3x, 4x over that time. So, you're talking to the market as a whole, go ahead.

Doug: Yeah, so that's a little debatable because timing and semiconductors have a relationship. But quartz grows with volume, but it doesn't grow with price. Quartz has essentially price erosion. So let's say, semiconductors, in the long run, have done 6 or 7% or will just say 6% kegger. I think it might be more like 5. Quartz is probably doing like 3. But at the same time, I do think semiconductor devices are growing quicker than it has been. So, there is some kind of growth in this market. I would say the number I use in my whatever big tam[?], my shot of the tam[?] because everyone else takes a shot at it too. It's like 3.7% kegger, which is kind of lazy slightly above, let's say, long-term growth but nothing to sneeze at. But that market will double, eventually. So, yeah.

Andrew: Perfect. Okay, great. No, sorry, I was on mute there. That's great. And then, I guess the other thing is, so they've got this really quickly growing market, right? We just said maybe it's 5x over the next 20 years, maybe it's 10x, but it's growing a lot. The last thing is SiTime's market share, right? So, what's SiTime's market share currently? And, hence it's big, what can they expect to keep a big market share in this rapidly growing market with a pretty big addressable market?

Doug: So it's 92%, which effectively rounds up to 100%, right? That's their monopoly. So, SiTime is actually pretty interesting to me because there were like 6 startups in MEMS trying to do this exact same thing. They survived the arms race essentially. The second most successful startup ever is called the Sierra and it was bought by Microsemi. And now, that lives within microchip MCHP. One of the reasons why I think the threat of entrants in at least the medium term is pretty low is because the absolute size of the market is not that big. So, imagine investing cumulatively $200 million to take a 30% share. And 30% would just be like the most amazing entrance in a 2-year period of will say, 2025 it's like $700 million.

So, you're just really not talking really big numbers here in terms of making the quick payoff. It just really is pretty hard. Cumulatively, let's say you have to spend $200 million to catch up at microchip, to then get $200 million in revenue. And that $200 million revenue will be much less profitable than SiTime will be at that because as they scale larger. It's kind of one of these things. I think the entrant comes when the market gets bigger. And right now, the math really doesn't work because if you look at the total MEMS market, I would say it's 400 million or something like that. So 25% shares, a hundred million. You have to spend 2 times the revenue just to have a shot, and that's assuming a pretty amazing first year.

There are just a lot of hiccups. SiTime itself toiled away. They were founded in 2007 and I would say, the profitability started in 2019. It's a long road. They also were owned by a corporate who just essentially, gave the money the entire time to bridge the cash bird. Someone has to seriously put up some money and frankly, I think the down cycle is actually very nice for them. That gives them 2 years of cover, probably. In terms of pushing an entrance back.

Andrew: I guess I'll jump [inaudible], you do see a lot, especially these big tech giants with lots and lots of money, right? Apple and again we're going to talk about Apple and SiTime's history, specifically because it is interesting. But Apple, Tesla, all these guys, they do like to in-house a lot of different products and if this is something that is going to go into every one of their products, I don't think they would consider it strategic because they've already got it. But you could see how they want to in-house that at some point. Like, why wouldn't you start seeing big companies in-house this technology because it doesn't seem that great or the counter would be, "hey, yeah, maybe MEMS has a little bit of an advantage"? Do the big companies just keep access to quartz just because, "hey, we don't want SiTime to get that much-negotiating leverage on us"?

Doug: Yeah, so I think quartz is going to be around, right? My assumption here is that quartz is still 70% of the market for a very long time from now. And I think, in terms of the actual entrant for the in-housing, it's just much harder. The places that Apple and companies like that have in-house are places where there's IP that's kind of readily available off the shelf. Arm-based chips are the perfect example, right? A16 is an arm-based chip. The Apple modem that they might make was a business they bought from Intel to try that. There just aren't that many people. This is a really, really small niche and essentially, the entire company is based out of the Stanford MEMS department. SiTime just hires the Stanford MEMS department as they graduate each year. It's a really small niche and I just don't think it's strategic enough.

The list of things that Apple could outsource before let's say, $50 million a year they spend on SiTime's business, I would guess there are probably 40-50 components before then. It's just a very low priority to the niche. There isn't enough readily available IP. And it just doesn't seem like it would make sense for them to outsource this. Because that was my first question. Apple begs customers, semiconductor gets in-house. That's like always the big fear and this one is just too niche. Yeah.

Andrew: And as you said, I think they spent hundreds and hundreds of millions over multiple years to get here, and it's like cool, you did all that. You got to a couple hundred million revenue base. Yes, your gross margins are nice. But at peak margins, you kind of did what? 60, 70 million of net income. Like anybody would love that. But that were absolute peak margins. 200 million plus to get there over 15 years. That seems kind of moaty[?] almost, right?

Doug: Yeah, just so you know this is my favorite semiconductor niche moat if that makes sense? Where it's like, a lot of times, you'll have a company, this happens in semi-cap a lot where it's extremely concentrated. Actually, my favorite example is CAMTECH, a company that I used to love actually a little short right now. They essentially are this company that did PCB packaging. It was just like in this extremely niche, back market. No one cared about this packaging. All of a sudden their product really mattered and it grew like crazy. But even though it grew like crazy, we're talking about a product that grew from 100 million to 500 million or something like that.

But even then you're like, how much would you have to invest to enter this market? 200, 300 million to be a third-place competitor? Why would you do that? The math becomes really hard to justify any kind of competitive entrance. The end state for semiconductors usually ends up being like a 60-30-10 market. If you have a third player. And my assumption here is that it eventually does end up being a 60-30-10 market.

Andrew: Right now, they've got 90%. And then, I just want to go back to one of the things, SiTime was kind of the first one to win MEMS. Obviously, they have 90%. But correct me if I'm wrong. There were multiple other entrances here. I mean, I believe tens and tens of millions were sucked into each one, and all of them kind of flamed out. And SiTime is the only one who got to scale obviously 90%. Plus they got scale. They're kind of the only ones that made it.

Doug: And second place is actually within a very large company, Microchip. Which in theory should have funding. But my understanding is that they are just like, our technology roadmap is multiple years behind. And that same math still applies to Microchip even. Where they're like, "we don't really think investing $100 million for this 5-year moonshot project is really worth investing in something that's very obvious". And they have a lot of places they can invest capital very quickly all the time. And Microchip likes to acquire companies, so oftentimes they would rather spend that $500 million to acquire some niche company and roll it into their entire corporate structure. Those kinds of capital decisions make so much more sense because it's gonna be a long road. Whoever is going to have to do that has to see it all the way through. And I just don't think anyone really wants to do that at this market size today.

Andrew: Why was SiTime able to succeed, get to scale, get a product, and get 90% market share? Again, there were lots of other entrants that you listed and a lot of other entrants, why is SiTime the winner here?

Doug: The thing that I think, and this is kind of hard to know because the history is written by the victors and I'm reading the winner's version. But the thing that they really point out too is this package technology called EpiSeal. And in particular, it was an extremely novel way to design a silicon package to isolate the resonator because that's really important. MEMS resonators, we have tons of them actually. I was reading some extremely academic papers about the theoretical limits of MEMS. We have tons of MEMS resonators. We have stuff that is two hundred times better in terms of the current MEMS technology. But the problem is, you have to put it into a package and you have to get it to work, and when you're in the package has to be sealed from the outside. That's really important. And the way you seal it, it's an extreme engineering problem.

You have to have one anchor, it has to be hermetically sealed, and it has to be a certain size and you have to be able to manufacture it using current technology that is available to everyone. And so, SiTime kind of did this all. And that's called their EpiSeal technology. They have a patent on that. I don't think they have patents. I mean, I'm sure they have some patents on the actual sizing of the resonators. But the resonator shapes, that's relatively known. So, having the resonator is not the problem, having the resonator and getting it stable and isolated in a package, is where SiTime has a legitimate patent boat. They also have a deep partnership with essentially, the best company in the world. The patent was awarded to Bosch and the founders of SiTime. I think there's some kind of relationship agreement there. But SiTime and Bosch have a very intense relationship. And they have to be one of the largest customers there. So, yeah.

Andrew: Anything else you want to talk about on this? Because I was going to switch this to some other questions on customer's stuff. But anything else you want to talk about on the technology, why do they want a partnership with Bosch anything else there?

Doug: I maybe want to mention MegaChips really quickly just because it's actually interesting.

Andrew: Go ahead.

Doug: So actually, one of the parts of the story of SiTime itself is that they came out of Bosch. They were funded by private equity or venture capital rather and then MegaChips bought them in 2014, which is really weird if you ask me. Because MegaChips's Japanese conglomerate kind of sleepy bought them for $400 million. And MegaChips is actually really important because MegaChips is a cash cow. The stock went nowhere for a long time and they just manage to reinvest it into SiTime and I'm sure they had no financial constraints. They were free to lose money. And then, that little safe haven that little petri dish for like 5 years really helped kickstart them and make it through the most painful, hardest time of SiTime's business. So that was really important in terms of the path dependency. And now, we're here. SiTime's all grown up. It only took them 15 years, so.

Andrew: I'm laughing. I can't find it now, but one of the articles I looked up when I was looking it up is around the IPO. There were some snarky articles that were like, can you believe how much this tiny little company is losing? Like, can you believe this giant corporation has been funding it? But that's kind of what it took to get here. Let me switch to another question. Customers, right? Anybody who's done semiconductors, and obviously you do a heck of a lot more semiconductors than me. But customer concentration is something you want to think about. Some semis will be like, "hey, we have a kajillion customers and nobody's over 1%". Sometimes I have the semis that are, "hey, Apple is 90% of our portfolio". Right? So, I want to ask you about customer concentration here and then hint, we're going to dive a little bit deeper into their major customer and their relationship. Their historic relationship with them.

Doug: So, the big customer that really matters is Apple. And Apple is really interesting because Apple adopted them very, very early in their business. Apple was like 40% of revenue for a very long time in like 2018 and 2019. And essentially, as SiTime has grown and become more mature, they've kind of grown outside out of the Apple crutch, if that makes sense. 40% of revenue in 2018 is like the number. Now, they have tens of thousands of customers, and a lot of the customers are new to them in the last 2 years. But in particular, the thing that's really interesting is Apple all-in on MEMS really early, actually. I was actually very surprised and that's something that I think is really curious, and something that gives me some confidence in the long run of this company if that makes sense.

Because Apple is probably the most sophisticated semiconductor design company in the world. Another big customer that we really can't parse how large they are is Tesla. Tesla is a company that also all-in on MEMS. And so, these companies all-in on MEMS for I presume programmability as well as size very early and have continued to this day to be large customers. I think that will continue. And as we talked about earlier, I don't think Apple is a risk of in-housing relative to other semiconductor companies where it's a little easier to do. But now, we have a lot more uses and I think the use cases are very levered to IoT. And so, that's one of the reasons why Apple is such a big customer. Because the Apple AirPod is probably my favorite example of this. That product might not be able to be the size it is if it wasn't for stuff like MEMS.

Because quartz has the smallest package or something is like one and a half millimeters by one on quartz or something like that, whereas they do like point seven or even smaller packages. I don't have it up the top of my head, but I know that there's another halving of MEMS over quartz. And so, those kinds of devices that are really, really small and power efficient. That's where SiTime has really taken off. And so, I think that that stuff is a perfect example of the IoT long-term bulking of that.

Andrew: So you said, "hey, AirPods wouldn't be able to be as small if it wasn't for MEMS, right? And then, you said, "quartz is 1 millimeter or one and a half millimeters versus MEMS's half as small". And I hear that I'm like, oh, well, now, we're talking less than a millimeter difference. How much bigger would an AirPod be? I don't know if you know the exact number but you know, why couldn't you do AirPods as small with quartz? It doesn't seem like a millimeter would make that much difference.

Doug: Every little bit counts. Because whatever savings you get in that, you get to use it somewhere else. And that piece of quartz is on a package. It's not just one. It's not just like, "oh, that's it". Right? It's the entire thought process of we are trying to make it smaller in every way, shape, and possible. And MEMS is really the one that has it. They've already offered the smallest package in the industry. And I think that in the future, they will be able to offer even smaller packages. So, being able to be reliant on the MEMS' road map is probably better for a long-term device manufacturer for IoT. Because they're like, hey, why would we do something that we know cannot shrink anymore versus we want to use something that we know might be able to shrink soon?

Andrew: How big do you think the AirPods would be if we use quartz instead of MEMS in them?

Doug: I have no idea, man. That's really hard techno because like it's a PCB. There are tons of little devices too. I wouldn't say MEMS is the thing that made it smaller if it makes sense. MEMS is just one of the many things that made it smaller.

Andrew: But each incremental thing builds on each other. Yeah. It would have been so cool if go like, "if it wasn't for MEMS, AirPods would have to be the size of oranges or something". Yeah?

Doug: No, I don't think it's quite like that. If it was like that, then I'll be like, obviously, that would make me feel really comfortable. But I actually think it's really interesting because miniaturization is not a new thing actually. It's probably one of the oldest things in semiconductor history. It's that if you make the semiconductor smaller, it becomes more power efficient, and takes up less space. And so, MEMS has that ability, whereas quartz doesn't. Because MEMS is a semiconductor, quartz isn't. And something that I may want to talk about. I feel like I don't know when to put this in is what really...

Andrew: Just put it in right now, You got it.

Doug: Well, as I say, MEMS versus quartz. We'd also talk about something that's really interesting from the MEMS side, which is their gross margins are on an absolute basis much better than quartz. I think that's probably one of the most interesting parts of this entire story. Imagine having a price-competitive product where there are 2, and I mean, maybe MEMS tend to have a slight premium but I think if they were to be priced even. MEMS would have let's say a 10 to 20% better gross margin. Sorry, I guess 10,000 basis points or whatever. One would be a 20% gross margin the other would be a 40% gross margin. I think that is crazy.

Andrew: Yep. You mentioned 2021, I think, SiTime had a 65% gross margin. So you're just saying, hey look, if you were looking at a quartz manufacturer, they'd sell for about the same price roughly and they would have a 45% cost structure. So not only is MEMS more programmable, smaller, and all this type of stuff. It's also cheaper to manufacture. And right now, SiTime is keeping that as a gross margin for themselves. Probably because they have a 90% share of MEMS and there's only 5% of it. But in the future, you could see a scenario where maybe they accelerate their growth by cutting below quartz a little bit, but having a superior product and having the same gross margins.

Doug: I think the thing that they're trying to do, and this is my take on their strategy, which I think is very rational, is they don't want to blow it up. Why blow up your really good gross profit dollars? But when you can incrementally improve the technology because right now, there are still cost-downs going into quartz. No, going into MEMS. MEMS is still improving the cost to manufacture and that will probably continue to exist. I would say, I don't know when the end roadmap of that is. But this was invented in 2006. This is there may be their first decade of them really pursuing this challenge. I think they probably have 2 or 3 more decades of improving this manufacturing process.

And so, if you have a price that is competitive on the current market and you could continue to win back and forth when you share, and you know you can reinvest into the manufacturing to then in the future lower your crop cost even more. So you can make money now and make more money in the future. Instead of having to do the prophet nuke in the short term. And then, obviously, recuperate it in future periods, which would be really painful for investors or frankly, any company to withstand. They can slowly but surely take market share while reinvesting in our product. Meaning that they're going to eventually have a better and cheaper product over a long period of time. Quartz just doesn't have that available to them. They've been doing this for 100 years. The technology is pretty tapped out. Yeah.

Andrew: You mentioned, Apple, the largest customer. They're all in on MEMS. And that means they're basically all in SiTime, right? They're all in. But if go back to 2018, revenue drops year over year. I believe their Apple concentration goes from about 40% of sales to 15% of sales. And correct me if I'm misremembering, but that's related to the iPhone. So, that's a red flag, a question mark, anybody who's investigating the story is going to have. So, why don't you just relay the story as a whole, and then probably talk about why you don't think it applies to non-iPhone products at Apple?

Doug: Yeah, so this is actually a really good one because that was one of the first questions, right? Because revenue goes down in 2019 over 2018. Then you're like, I thought this was a, you know, a growth...

Andrew: At 20% plus growth early cycle. And you see, hey, and then a pretty good time for the industry. Revenue goes from 85 to 84 like, what the heck?

Doug: Yeah. So you're like, what the hell happened? And so, actually, I had to hunt this down. At one point in time, there's this giant MRI helium leak in a medical facility and everyone's iOS device bricks. Every single iOS device in the entire building bricks.

Andrew: You said brick not break, just to confirm.

Doug: Bricks. Yeah. Brick is essentially it completely breaks and there's no recovering if I make sense. That's kind of like slang. So, they all break. Every single iOS device breaks in this building, and then someone is like, what the hell is happening? And so, they do a test and they realize if you fill a bag filled with helium and you put your iPhone in it, it will break. And the reason for that is actually, was SiTime. And so, SiTime obviously is like, "holy shit". And iPhone is like, okay, well, they designed SiTime out for 2019. And they do that.

It's really unlikely that you're going to be in a room filled with helium. But it's just one of these things where it made some headlines and the iPhone insider was like, "how could they ship this with this clear design flaw". And so, they got designed out of the iPhone, which is a big deal for them given that Apple was 40% of the revenue. And so, fast forward a few years, if you go to the website, I believe they fixed it. It's now impervious to all small molecule gases and that would include helium to my understanding. So, going forward, this is less of a big deal. But it seems when they were designed back in, they went from 40 to 35% to 40% of revenue. So it's like a little dip but it didn't quite have the torque back. But that being said, I think, for the most part, they are still on the iPhone today. They're still [inaudible].

Andrew: So I was going to ask because my understanding was they're not in the iPhone currently. You're saying, they got designed out, but they get designed in. Because I was kind of thinking in the back of my head, "oh, maybe there's a leg up when they get fully into the iPhone or something". But do you think they're already in them?

Doug: I think they're in the iPhone. No one has a hard, like, "I don't really want to put my phone in helium to test". And maybe that isn't even the right answer, right? Because the new EpiSeal is impervious to gas. So I believe, they are in the iPhone. And the thing that I think is actually more interesting, is that Apple is a shop that is, so they're not in Samsung, especially in the Samsung modems. The Samsung RF part of the chain versus Apple right now is designing their own modems in-house a large part of something that they do that Qualcomm does. If that happens, the odds of it being SiTime are pretty high. And that would be another 200 million units, which obviously would probably come in at a low ASP. But odds are that would be material for revenue. And I think that's something because I've only been talking about near-term terribleness, but there's a real chance in Q4 of next year if they design that in. All of a sudden, they're like, "oh, by the way, we just had a large smartphone win and it's going to be material to sales". So, yeah.

Andrew: Even if it comes at a low average selling price like this is a 90% market share company and you lock up all the iPhones. There's just no space for any competitor to come in and breathe into that type of market.

Doug: Why invest in number 2? The other thing too, is where would it be fab? It's something I've been thinking a lot about is like, so the semiconductor side of it is done on TSMC. There are actually 2 devices in a MEMS oscillator. There is actually the MEMS and then there's the semiconductor circuitry to compensate for the MEMS. Super-technical, we don't have to talk about it. But one is done by Bosch, one is done by TSMC. There are a few other large-scale companies in MEMS, maybe say SD micro. But SD micro would have to go through that same problem as well. But Bosch is the number one fab for fabless company for MEMS in the world. And so, they have the largest company as their strategic partner. I think that's really, really good and for them as well. Because where else would you go?

Andrew: Just a quick question, I think we've already discussed and addressed how difficult it would be for a competitor to come in and all of that. But a quick question, sometimes semiconductors can be tough to make a switch. You know, if you want to switch from one to another, sometimes it's easy, sometimes it would be really hard. How difficult would it be? Let's say, you and I managed to go and somehow start up a business next year that was at a decent scale with competitive products beside them. How difficult would it be for Apple to switch from SiTime to our product next year?

Doug: This is actually really interesting because SiTime had this whole problem when they were trying to get everyone to swap from quartz. All the timers are on standard packages, but SiTime has one smaller package that is for IoT devices very hard to swap out of. So for example, the AirPod, watches, and probably smartphones, it's called CSP gel packaging. That's like the smallest packages only available through MEMS. Unless we offer the same exact MEMS product, that's probably hard to swap away from. And so then, it's mostly can we, the MEMS company, the MEMS startup offers the same package? If we can, then they can in theory swap back and forth. But then, the other may be friction there is qualification. This might not apply as much for phones, but when you get a semiconductor device to work, you have to test each and every single part component. And then, you're like, okay, this component works in this circumstance. We're good to go. Now, we can put the whole package together.

And so, for automotive, that's actually a little bit of a moat if that makes sense. That's actually been kind of a lot of friction for them to get adopted, especially for qualification for automotive parts. And so, that would probably, be one of the things that makes it hard for an entrant, it takes a lot of time to get qualified. But interestingly enough, one of the things that they're pitching to automotive companies, in particular, is that we know it's a pain to get any semiconductor device qualified. Well, if you qualify with one device with SiTime, then you can choose whatever frequency you want. My understanding is with quartz, you have to each device is fixed. So you have to qualify each quartz device. So let's say, you have seven different types of frequency. You could do 1 device qualify at once, do 7 different frequencies versus qualifying 7 different quartz frequencies. And if you wanted to switch from one of the frequencies for whatever reason. There's actually a really good reason why chip design is complicated.

That's my summary. So let's say, you want to swap for whatever reason. You'd have to then go back to your quartz manufacturer and say, "hey, we need a different frequency and we need to qualify it". And they're going to be like, "Okay, great. But you're going to have to give us 3 to 6 months for us to ramp up that volume". Because it takes that long to grow a crystal. Versus SiTime has inventory that they can just press a button and they're like, "oh, great, we can swap over tomorrow".

Andrew: Stupid non-engineer question, why would you switch frequency and something once you've already got it approved up and running?

Doug: Okay. So actually, I came into this literally 3 days ago with a perfect example. Frequencies interfere with each other. EMI, electromagnetic interference. And so, if something is vibrating at the same frequency in a semiconductor it can have interference and mess up. And also, sometimes there could be resonant frequency. And it will break parts of the semiconductor because everything is vibrating at a certain level. Tesla actually gave an example of their Dojo chip, which they kept breaking and they figured out it was because of the frequency. The frequency that the chip is running at is very similar to the timing frequency, so it would just create resonance in the packages and the packages would break. And so, they're like, "oh, well that's a big mistake".

So what they have to do is like, "Okay, we need to fix this". Well, in quartz, you'd have to design the whole thing. My understanding is in the package, you have different frequencies for different applications. Your radio is done at one frequency, your networking is done at another frequency, and your clock is on a different frequency, but they're all keeping time in different layers if that makes sense. Each of these data streams has to be running at this different frequency so that it could all be working together, but if they all ran to the same frequency, it would start to interfere. So you need different frequencies for different communication levels. Yeah, super complicated. That's my best. That's to me, the best I got to explain.

Andrew: I loved it. I'm starting to get a handle on it. Another silly question, all right, so we've talked about how SiTime is going to dominate, not that. But, how do they control MEMS? It doesn't seem like a competitor comes in, right? We've also talked about how MEMS is much better than quartz. I mean, you just gave the frequency example, which I think is very interesting. We've talked about how roughly similar prices, have a way better gross margin for MEMS. So if SiTime really wanted to, they could kind of cut-price and try to take a share of the price, more programmability, more customers all this type of stuff. So, if you and I were designing a product today, why would we put quartz in over MEMS? Because clearly, people are because quartz is still growing. MEMS isn't taking 80% of the market share by 2030, right? So why would we ever put quartz on a product over MEMS?

Doug: So the supply chain obviously isn't there yet. We're talking about a market that ships 20 billion units a year. You couldn't just be like, "Hey, could you give me 15 billion units"? And then be like, "oh, that's like a lot of units. We don't have that". That kind of swap can't happen mechanically just because it's just too much. Supply chains have inertia and this is something that I think I've done a decent job at. I tried really hard to do a good job because the quartz versus MEMS GHAD[?] is extremely biased on both sides. There are many places quartz is much better. And I think that there's a potential for the MEMS technology to improve to then be able to match quartz in those places, but there are places that right now, MEMS has the worst performance product. And SiTime themselves would say, "no, we're 10 times better at everything".

Andrew: Could you give an example of a place where MEMS is worse than quartz and why?

Doug: Okay. There are 2 probably, super duper cheap products, right? The absolute cheapest product you can get. That's like a Chinese device or something. It was like a cheap 6-dollar electronic gadget that's borderline junk. If you're shooting on just price alone, MEMS don't price it that way, maybe they could, but they don't. The other place is extremely high. It's called Phase noise. And so, a really complicated subject that's hard to explain. But let's just put it, a random noise happens in the signal for networking and it just enters this signal over time in this frequency.

There's a way to measure this, it's called Jitter. It's RMS, which is Root Mean Square over pico second or something like that. The best MEMS device today that I could find is like 0.1 or 0.23 change over picosecond. MEMS can do or no quartz can do much lower than that. And so, for extremely high frequency, extremely low noise products, which is mostly communications because you do not want any noise in there at all. Extremely high-end, 5G stuff. Stuff that's really, really sensitive to noise and multiplexing and stuff like that. A good example maybe is a 400G transceiver multiplexing where they're just doing crazy ridiculous noise in order for them to transmit the signal. That has to be quartz, yeah.

Andrew: Let me push back. You just gave two, not completely niche, but semi-niche examples of where quartz is better than MEMS, right? So, you gave communication and just the cheapest of cheap electronic products. The type of headphones that I buy off Amazon all the time. My wife says, "you've already got 4 of them". And I say, "no, I only have 1 because the other 3 are broken". And I'm sure this one will break at some point in the near future, right? So you gave two, but what about other products? Like wire cars? Wire cars use quartz instead of MEMS all the time, and obviously, there's capacity insurance. But why is it everyone is just banging down their door saying, "give us MEMS. Gives us MEMS. Give us MEMS, it's so much better than quartz".

Doug: The noise is another example. So, like the ethernet stuff within the car. But another example that I've been given a lot is inertia. It's the fact that quartz is very well understood. When you're designing a chip project, you're an engineer, you've been doing this for 30 years, man. Why are you going to do something new? Every bug that's ever happened to quartz has already been fixed. MEMS still is, immature in that way.

Andrew: So you've already talked about how there's this market share taking tailwind over SiTime MEMS. And one of the other things that will probably be driving market tailwinds is, "hey your 60-year-old engineer is retiring. He's getting replaced by a 25-year-old grad student". And the 25-year-old grad student doesn't have that legacy or background of doing everything on quartz. He might just say, "oh, MEMS is superior. I will learn how to do everything I design on MEMS".

Doug: Honestly, that's one of the reasons. That's probably, realistically the reason why they grow a lot of penetration. My favorite examples are Tesla and Apple are kind of the newer kids on the block. They're like the big semiconductor picture, right? And they are all-in on MEMS. Tesla's one of the newest kids on the block.

Andrew: That was the most foolish thing you said. Right at the front, when you said Apple and Tesla all-in on MEMS. You're right[?] as mentioned several times, Apple all-in on MEMS. I was like, that's the best validation, I think you can give to a product that it's probably going to take a lot of share over time. They are all in early.

Doug: Yeah, they all in super early? And you're like, "why"? And so there's, I don't know what that design choice was, but it clearly wasn't a long-term decision. And if you talk, the Tesla AI. The AI day, where they talk about their Dojo chip, the MEMS broke and you'd be like, "well, does that mean they're going to switch back to quartz"? But they're like, "no our solution is we'll just have a better MEMS resonator". And I was like, "okay, that's great". I thought that was really interesting. But at the same time, these industry shifts take a really long time. There is definitely a world where that rate of adoption really starts to accelerate randomly. I don't know, when or how that happens. That's completely out of the purview of my forecasting horizon. But I do think that there are just a lot of ways to win over the very long term because, right now, I would say the product is good enough.

High-end quartz is better in a lot of ways in terms of performance and stability. It's just the more mature product. They figured out how to do this 10 years ago. But the catch-up has been really intense. I talked a little bit in my super long piece on the low end versus the high end. The high-end, they're actually not competing in performance alone. They're actually trying to do the 80/20 strategy of like, we have 80 or 80, 90% of the performance, but we have a 40% discount. And so, they are attacking from below. And then, on the low end, they're actually attacking from above where they have. I think it's really interesting, it seems they have 3 or 4 out of 5 attributes. They'll be better than quartz, and then we'll have a 20% pricing premium.

Meaning, the only attribute there that's worse than quartz is price. So they're beating them out in performance on the low-end time of day, which is just to keep the time. And then, the stuff that's being used for communications, networking, and stuff like that, that what they're doing there is they're like, "we're trying to have a product that is very comparable". For the most part, it's probably good enough. And it's price competitive and it's programmable, and it's smaller. So they have all these other ways that shade in around. And we presume or whatever they make their fourth or fifth generation resonator. Then all of a sudden they'd be like, "and now we have absolutely better performance".

Andrew: When do you think the next generation comes up?

Doug: That's something I don't know. I'm going to guess, It's going to be like, 5. It's going to take a few years.

Andrew: Let me switch to just a couple of quick evaluation questions. Look, 70% drawdown roughly your to take, that's going to get a lot of people just looking and saying, "oh, you know, I wanted a semiconductor with some beta. It's down a lot. That seems pretty interesting. We don't have". We already talked about how bankruptcy is not a risk here, unlike some leopard semiconductors. So a lot of people say that. But then a lot of people would probably look and say, "oh, well, this is like 25 times free cash flow. Yes, Doug makes a nice story about growth". But 25 times free cash flow for a cyclical business doesn't look screaming cheap, right? It's 25 times this year and next year, which we don't even know. It sounds like numbers are coming down, right? So it might actually be your paying 40 times forward. It doesn't look crazy cheap. What would you say to someone who pushed back on that? How do you look at that, what evaluation should be here?

Doug: So, this is the hard part. I have a risk section of my write-up, and my biggest risk is trough numbers actually could go much lower than it is today. I don't know what that number is. I think something that's really important for a name like this is it's a long-term story. There's a real chance, a very real chance I'm early to this. And that's probably the biggest thing I fear if that makes sense. The things that I've done to maybe de-risk that is, I would say, my base case of numbers is much more realistic and I have them more than 30 times free cash flow for 2023. So it was expensive. That's my problem.

Andrew: Let's put aside 2023 because one thing that we can definitely get caught up on is the free cash flow number for next year and everything. It does matter. But if you start from that and work backward, what really matters is, "hey, what is the mid-cycle sustainable earnings"? And then, we talk about the growth from there and stuff. So let's ignore the next 12 months where there's an air gap and the past 12 months where everything was. What's on 2 years out, if we just said, "hey, this is the middle of the cycle"? We're not at absolute-everything-rushing-through capacity. We're not at, hey, everybody considers or whatever. What's in 2-year out mid-cycle free cash flow numbers look like for this company?

Doug: So, I would say 80% free cash flow conversion. And I would say it's five and a half dollars per share in EPS or something like that. That's kind of what I'm thinking. Actually, the other thing that I have a little problem with is, I think my numbers, my out-year numbers for the recovery are actually, is not steep enough. Let's say, five and a half, $6 per share of earnings, that's kind of what I think is the number.

Andrew: Five and a half to $6 of earnings. As we said, this is about an $85 stock. Today we're recording Friday, October 7th, the market was down quite a bit and it seemed like side him every moment another dollar down. So by the time we're done with this podcast maybe, it will be $75. Who knows? But $85, so we're talking 15x or maybe even a little cheaper than that, you're normalized to your route number. That's pretty good for a company that you think, again over the cycle should be a double-digit grower. Free cash flow lever's a little bit better than that. Am I kind of thinking about that correctly?

Doug: I think that's the right way to think about it. And I think that is something I keep trying to do because I feel like I've done a lot of work about framing the long-term really well. And I'd like absolutely just murder the crap out of the near-term numbers to be like how painful can this be? And I feel really good about the numbers that I have for the next year. I think that I'm a little mean on gross margins. Probably, meaner than I should be. I feel like I'm a little mean on the snapback. I really don't have them snap back quickly. I kind of assumed a tepid, a tepid out cycle like 20% revenue growth for 2024. I think that there are a lot of things, in the entire write-up, I've tried making a lot of conservatism through trying to, my entire job has been trying to frame how bad it could be. And I haven't put an ounce of thought into the good things that can happen, frankly. There are a lot of things that can happen that make it a lot better. The thing that I think is really interesting and is probably one of the reasons why I think it will persist over the long term. And so this is the sponsor Teague is a shout-out call.

Andrew: This podcast is sponsored by Teague's too. {inaudible].

Doug: Essentially, I did some Teague's calls on this, and the thing that I thought was really interesting is a former sales manager and a currently occurring distributor of their products was like, "I've never known anyone to swap from quartz to MEMS and back to quartz. Maybe the super low-end products, but most people keep MEMS to some degree. Maybe as a second source but they just do not swap away". This huge story actually, there's a huge beneficiary [inaudible], there's a fire for this huge part of timer ICs, and so it created this huge supply shortage and they were essentially the only one there to pick it up. So everyone rushed to design them. And so now, they have probably been introduced to more customers in 2020 and 2021 than they have cumulatively as a company.

And so now, they have all these design wins, they've been built into maybe dual sourced into. And we have a qualitative understanding that the customers tend to like and they're sticky with them. Mostly because of the supply chain programmability, it's like, hey, you can't run out of one random quartz device that's running at one frequency because they can just press some buttons and boom. Now you have all the 40 MHz devices you want. That's like the supply chain advantage that they keep talking about, but I think it's interesting. And then, the second advantage is programmability. You just have so much more flexibility whenever you're qualifying and building these designs. It's more like a semiconductor than it is this passive device that quartz is.

So this is kind of the thing that I think is really interesting because takes a big old step back. Think about s-curve penetrations, right? It is a classic 2 to 5% super early adopter. We are an early adopter curve. And then we have this, this event where there this early adaptive technology got introduced to tens of thousands of customers because of the supply constraint issue. And we know that the customers don't tend to design them out. I think there's a real chance at like s-curves don't look as linear as I projected them to be, right? When things go to 30% they don't just linearly happen at 100 pips a year of sure gain. What happens is there tends to be a freaking penetration curve where it goes [inaudible].

Andrew: It goes 2, 4, 6, and you're like 2% a year? And then it's 9, and you're like, "oh that's interesting". 15, 30, 70, boom. Done.

Doug: I have not baked that into the number. That is not even a consideration at all. And that is a possibility. Yeah

Andrew: Again, you've got Apple and Tesla all in. As you said, nobody switches back to quartz. Those are pretty powerful signals here. Okay, the last question, and then, I think we've run long. I love talking to you. The last question, so we talked about valuation talk of that. The last thing that jumps out at me and I have to ask because I know I love [inaudible] backs, right? But you've got a company here as we're talking, a 1.8 billion market cap, right? Almost $600 million in cash is on their balance sheet. They are free cash flow positive. So yes, they are cyclical but they are free cash flow positive. A third of their enterprise value is in cash, and they're not buying back shares. And then, I would add on top of that, I mean, congrats to them, they did a nice job.

But, the second thing I look at in anything that's not buying back shares, I go look at the insider transactions. And when I look at the insider transactions, every Insider all year sell, sell sell sell. Now, a lot of them are 3-digit prices that start with a 2. And again, the stock is now at 80-85. So, they were getting great sales. But as recently as last month, here's some officer who's selling a decent sluggish stock at under $100 per share. I see lots of insider selling, I see no buybacks, and no one started buying. And I say, "Hey, what is Doug seeing here that maybe the insiders aren't that they don't see enough value to buy shares either with the company's cash or their own cash"?

Doug: Yeah. We'll do the buyback versus the insiders first. The buyback question, I think it's really interesting actually because whenever they first were floated, they were 50%. So this is like a float problem. The CFO, actually kind of like the way he frames this is, they ask about buybacks, and then let's say, we don't want to buy back because we don't want to ruin our float. Our float is already...

Andrew: It's the toughest answer. I've got so many comments to tell and I know they're probably right. But it's just like, hey, your stocks are cheap, and there's nothing you can do better than buy back shares.

Doug: Yeah, I don't think it's quite the level where it's like, okay, this is just the best cash-on-cash [inaudible] they can get. I think what would be really interesting is actually if they bought another MEMS company. Because they do have a big socket and that cash is going to be used. I would frankly like to see a buyback as well or frankly, I would like to see them buy back the MegaChips stake because it's still 25% of the company.

Andrew: Yep, 5 million shares. Yeah, it's a big slug.

Doug: Yes. So it's like they could maybe take some of that off. That would be really interesting to me. I don't know. They have this big war chest. What are they going to use it on? I know they want to do some MMA[?]. They want to buy some like IP on the side. That is something I feel pretty confident that they're thinking about and looking at. And then, the buyback thing, there may be a little worried about float. I think that maybe they should just do something, I don't know, maybe a dividend? That would be crazy to me like one set of share something. But it's like, hey, it's something because I would like to see it, them to talk about it. I think that is an optionality that could be really good, but also your biggest fear is that they're going to piss it away and do something really stupid.

That is probably another huge risk that I'm very concerned about because that's right now the big put on the stock is, hey, $25 of cash a share. But you're also like, what are they going to use it for? I don't know. I would love for them to answer that. I think that would give a lot of confidence to shareholders, in the longer term. Another question on the Insiders thing, I think that one's really interesting. Because I've actually looked a lot at the granting history. They have like really hyper-aggressive grants at the beginning of this year like with PSU, some price hurdles stuff. I think they've been a little burned. They've been giving themselves a lot of shares along the way. Like a lot of shares along the way. I think them getting a little lighter, I'm not really going to chastise them, and insiders for the longest time have thought their shares were expensive to be clear.

I follow this company from a technology perspective for a very long time. One of my friends who's a really interesting semiconductor technologist is like, "hey, you should really check out this company." And I look at the share some like a hundred and fifty times earnings. Hell, no. I'll read the earnings calls, I'll keep up with it. And they did giant equity offerings all along the way. And those equity offerings obviously ended up being super awesome because they have now $580 million to share of cash on the balance sheet. But I think that insider selling continues, some of these might be 10, 40, and 41B plans. Yeah, some of these are 10P5-11 plans, so they are automated and continuous selling. That's where you're seeing most of the continuous selling from the insiders.

Andrew: Everybody's got their own thing but at the same time, if they thought it was cheap enough, you can cancel your10b5-1 at any time if you want, right? So they could cancel...

Doug: They can do prices too. Where it's under this price you don't sell. I think the rate of selling at least slowed the hell down from the beginning of this year. If you're looking at the beginning when there was a 2 in front of the triple digits stock, there was significant Insider selling the entire time. That pace is kind of slowed down, especially since we've gone to 130-100, whatever. The last insider sale I see is, there's one guy selling a nice little like a hundred fifty thousand at ninety-seven. But the insider is selling as the rate of change is really slowed down. I would love to see some insider buying [inaudible]. That's something I would like to see.

Andrew: It's sometimes you have this with companies, right? Where they're really good at capital allocation and they issue a bunch of shares when their stock is dear, right? So the stocks are at a thousand and they issue a ton of shares and all the insiders are just hammering the sell. And then, the stock comes down to five hundred and they stop selling, and they don't buy back shares. And you're kind of like, hey, these guys are good and they've got a lot of cash on their balance sheet, but they're not buying back shares. They sold a ton, they're not buying back. They sold a ton of for the personal accounts and they're not insider buying. And you're like, maybe they just kind of think it's fairly valued at this point. And if they think it's fairly valued, I should probably think it's fairly valued and I should wait till they give me the like, we are greedy. We're buying back shares. Because of the inside ownership, okay? But as the CEO, he made $11 million in 2020. 10 million of it was stock. So he shouldn't be incentivized. But it was stuck at two hundred of...

Anyway, Doug. I think we've run super, super long. I've really enjoyed having you on the podcast. I'm super excited that you're doing the Tikka T1X dive thing. I believe we're taping this on October 7th. I believe you're going to post it publicly along the side when this podcast comes out, so we'll link back and forth so everyone can see it. But Doug, thank you so much for coming on. Have a great weekend and I'm looking forward to time number 3.

Doug: Yeah, thanks for having me, man. Appreciate it. Until next time. Yeah?

[END]