Entegris: Solutions for Semiconductors - [Business Breakdowns, EP.137]

0:03This is Business Breakdowns. Business Breakdowns is a series of conversations with investors and operators diving deep into a single business. For each business, we explore its history, its business model, its competitive advantages, and what makes it tick. We believe every business has lessons and secrets that investors and operators can learn from. and we are here to bring them to you. To find more episodes of Breakdowns, check out joincolossus .com. All opinions expressed by hosts and podcast guests are solely their own opinions. Hosts, podcast guests, their employers or affiliates may maintain positions in the securities discussed in this podcast.

0:45This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. This is Matt Russell, and today we are breaking down Integris. Over the past two years on Business Breakdowns, we have covered many names across the semiconductor ecosystem, and Integris is another name to add to that list. Now, what exactly do they do? They are a supplier of advanced materials and process solutions for semiconductors. In layman's terms, they are developing chemicals and filters, which are applied to all of the different materials that go into a semiconductor, allow those materials to be more pure and operate at maximum capacity.

1:28You can imagine with a small size of chips, the smallest things that might interfere with the different materials that exist within the semiconductor will slow it down and make it inoperable. And that's where Integris comes into play. Now we get into a lot of that discussion with our guest, Daniel Pilling from Sands Capital. He joins us to talk through the history of semiconductors in his own terms, what makes it such a fascinating industry to cover, not just from the Integris standpoint, but from the rest of the ecosystem as well, and what makes Integris unique as a small player in this overall huge universe dominated by major players.

2:06So please enjoy this breakdown of Integris. All right, Daniel, thank you very much for joining us here on Business Breakdowns. We have covered a few names in the semiconductor space, but it is important to rehash the importance of semiconductors each time we do it. So I thought we could start there. In your own words, what makes semiconductors such an interesting industry and such a key investment industry for you? In my opinion, you can analyze pretty much any sector simply by looking at supply and demand. Most companies function that way. So if you think about semiconductors like this, you can say, well, okay, let's think about supply.

2:48So supply has consolidated significantly about 15 to 20 years. You pick the company, NVIDIA, ASML, Integris, et cetera. All of those companies are really dominant. And without those companies, certain things just could not happen. The entire supply chain is very consolidated effectively and very oligopolistic. Now, on the demand side, I think the way you might want to think about it through is to say, what's the demand for compute? It's all about compute. So there's a great academic paper that talks about how much compute grows every year in terms of us humans requiring more, and it's about 50%.

3:25Now, I can make a very simple argument to say that over the next five, 10 years, it probably accelerates. Why? Because we have all of a sudden AI, which applies to everybody, every company, not just humans. We still have PCs, smartphones. Smartphones are still growing. We probably have self -driving cars, et cetera, et cetera. So the need for compute is probably accelerating. And then if you combine those two factors together, you say, well, okay, so supply is very consolidated in a few fairly dominant companies, and then demand is accelerating. That's a beautiful picture for an industry because that means the dominant companies are able to monetize demand effectively.

4:03And then maybe one last thing about the demand, which I think is very unique. So the cost of semiconductors is going up on the leading edge because they're trying to shrink these transistors. And usually when the cost of something goes up, economics 101, the demand goes down. But here it's very unique in a sense where the cost is actually going up, but the use case is becoming so much better that we are actually willing to pay more for it. So despite costs going up, demand is accelerating. Maybe a great example might be if all of a sudden you have a self -driving car and you save one hour a day, people are willing to spend a lot of money on that despite the chip cost.

4:41So to me, that's a very intriguing industry setup effectively for the next 5 -10 years. And it equally was over the past 10 years. Absolutely. The world certainly operates on semiconductors and everything that they power. Getting into Integris and where they fit into the overall ecosystem. Again, we've covered AMD, we've covered ASML, we've covered someone like Cadence. So very different angles. I think Integris, based on my research, is even more unique than those in terms of where they straddle the line between being a direct semiconductor business versus in the chemicals world somewhere. So maybe you could just start with an introduction in terms of what Integris actually does, what they produce, and how that fits into the ecosystem.

5:28You can think about semis in simplistic terms, like a multi -tiered pyramid of sorts. At the bottom, you have the semi -capx players, and Integres is part of that. They have the guys building the equipment, and then you have companies providing materials. Now, interestingly enough, the equipment space is much bigger in terms of market cap, because most materials companies, they're actually part of big conglomerates, and so you don't see the semi part. Integres is the only one. And this stuff then gets bought by the foundries, and then you have the chip makers on top of it, the NVIDIAs and AMDs of the world, and you mentioned AMD.

6:02Integris is right at the bottom. And the intriguing thing about being at the bottom is that effectively, no matter what Moore's Law brings out, so whether it's the PC, smartphone, and now we have AI, the companies at the bottom really benefit because that means you need more semi -capex on the one -hand side, but you also need way more chemicals and filtrations to manufacture these chips. So that's the position of Integris within the industry. Chemicals that are involved in the chips, what do the chemicals actually do? Where do they fit into the semiconductor, whether it's production process or what's their actual function for a chip?

6:40So I would highlight actually two segments here of theirs. So they do the chemicals and then they also have these filters and both are actually required. And maybe to provide sort of an example, let's start on the chemical side. So a big trend in chip manufacturing is that we went from planar architectures to 3D architectures. That's a natural step because the silicon footprint is limited. We want more compute per die, which means you have to go from planar to 3D. And I tend to joke, usually imagine you go from a city like London or DC, which is fairly planar, to New York City, which is the most 3D city you can have.

7:20And this is exactly what happens here, both in memory and logic. And then what that means for a chemical provider like Integra is actually fairly straightforward because imagine you're the cement provider. You have one cement provider in London and one in Manhattan. The guys in Manhattan are going to sell way more cement because the city is 3D. So effectively, the same thing is happening here. So chips are continuously going 3D, which means you need more chemicals. And then there's something you can measure. There's something with process steps. So there's hundreds and hundreds of process steps to manufacture these chips.

7:53And each generation has usually more process steps. And each process step requires a certain number of chemicals. I can also briefly mention the filtration side of it, because it actually is quite similar. So as you go 3D, you have more process steps, which means, again, you need more chemicals. And each chemicals need to be filtered for contaminants. And these are really, really high -quality filters. So Integra talks about the filter being 99 .999. nine. So nine nines behind the decimal in terms of being able to filter out contaminants. And then as you probably know, Moore's law is getting more complicated.

8:29The transistor is getting smaller. So you really have to get rid of the contaminants. So that means the filters are getting more important. So you have two things here. You have more process steps, driving more chemicals, and all these chemicals need to be filtered all the time. And both of the segments are effectively these growth areas because of that. That makes a lot of sense. Now, just going back in history, in terms of chemicals importance into semiconductors, has this been the case all along where there's been a need for chemicals and filtration systems involved with semiconductors? And has that changed at all?

9:05Obviously, there's more demand to the extent that you are going 3D on a per -chip basis. But I'm just curious, has there been any change over the history, which resulted in inflection and a demand for this chemicals as it relates to anything technological that changed with chip making and what's happened over the years? You always needed chemicals on filters, always, because they're sort of the building blocks to either etch or deposit stuff onto the silicon wafer. And same, you always needed to filter them. But I think what happened maybe around 2016, there was a big change in terms of the intensity of both chemicals and filters.

9:44And so during that moment in time, all of a sudden, the number of process steps to develop these ships started accelerating, which then meant that your cost per wafer went up significantly. And ballpark numbers here, I think today the wafer cost at Taiwan Semi on the leading edge is about $30 ,000 per wafer. And it used to be $3 ,000 to $5 ,000 dollars ballpark, depending on the timeframe. And a decent chunk of that is basically more process steps, more chemicals, and more filtration. And then if we boil it down to integrous, the magic that happened ultimately is to say, they used to grow at the number of wafer growth.

10:22So the more wafers you have, the more stuff you need from them. That's still the case today. But on top of that, actually, they're outgrowing the wafer growth because you need more chemicals and more filters for each new leading edge node, which then if you put it into numbers, that means they used to grow 5 % organically, which is the wafer growth, to I think now they can grow something like 8 % to 12%, which is 4 % to 6 % of organic growth. And so in my opinion, the organic growth on top of that is going to accelerate. We can debate, but that's sort of what happens. So it's become definitely much more important.

10:55And when you think about the chemicals within a chip and what that represents as a percentage of the overall cost of a chip, or we can put it in wafer terms, however you like, how big of a piece of that. And if I'm hearing you, it might be such that the cost of the chip has gone up. So the cost of what Integris is providing is maybe the same percentage of the overall cost of the chip. But I'm just curious there how important or how large of a cost as a percentage of the overall cost is the chemical as part of semiconductors? If we think about the percentages, it's hard to get precise data, but it's maybe something like 1 % to 2 % of the total wafer cost is that.

11:37And you can deduce the numbers that can take Antegra's total revenue and divide it by total wafers, gets you to something like a couple hundred dollars. And then if you look at the past few years, maybe let's say seven, eight years, I think that intensity has at least doubled on the leading edge, if not more. And I think it's going to accelerate in the future too. Ultimately, I think that cap intensity here grows similar -ish as the cap intensity grows for the semi -capex names, the applied materials, LAMs, and ASMLs of the world. I think you could make a case that maybe it outgrows it a smidge, but I'm not sure that's the case actually.

12:13But ultimately, I think it grows similarly as the other companies in terms of capx intensity. And then maybe lastly, I'd say if you talk to the fabs, the big manufacturers of these chips, let's say the cost is 1%. There have been cases in the past where there were issues with chemicals, at Taiwan Semi, for example, and that'll cost them hundreds of millions, if not billions of dollars. And then if you compare and contrast the cost versus the potential risk, it's enormous. So the ability of integrates to get designed in is very unique in that sense. It's very hard to build that trust. It's very hard to get designed in.

12:50But once you're in, you're going to grow for many, many years. So it's a nice position to be in. If you're Taiwan Semi or one of the fabs, it's not really the cost that you're looking for. You're looking at quality, yield, and the ability to supply the stuff on an ongoing basis. If it's 1 % to 2 % of overall costs, you definitely want to err towards quality. And that will make a huge difference, especially with what you're dealing with here. At the beginning of this conversation, you mentioned that their competitors are really business lines within broader conglomerates. So maybe you could talk a little bit about, one, the competitive landscape for Integris, and two, how this business came to be, why it doesn't sit under a conglomerate or a broader corporate umbrella, and how they've been able to stay independent over the years.

13:39So just some backstory in terms of how they came to be and who they're operating against. If you look at the filtration side, there used to be maybe five players in that 20 years ago. Today, it's two. And to be clear, filtration today generates about 50 % of EBIT. So it's the biggest segment. It's a duopoly market. The competitor is owned by Danaher, which is a healthcare company. And it's called PAL. It mainly actually focuses on the healthcare space because you also need a lot of filtration there. And then maybe there, you can also argue, again, it's clearly a great company, great competitor.

14:16But the focus is not semis. the focus is healthcare. If you look market share wise, we feel that Integrus is the bigger player by a decent amount. And then if you look on the leading edge, the design in, Integrus is by far the leading player. And then if you look at the chemical side, there the market shares vary. In some parts, they have 30 % market shares and some parts of 80, 90 % market share, maybe on average, something like 40, 50. What happens here over time though, is that they gain market share. And sorry, to be clear, the competitors here are sort of the BSFs of the world, DuPont, Dow, there's a bunch of Japanese chemical companies.

14:54And for all of them, semiconductors is a tiny fraction of the total, tiny. And then there's three ramifications for that, for Integris. One, they should gain market share over time, mainly because they are the only ones that focus on semi -chemicals, and they're the only ones that provide the total package of all chemicals as well as the filters and some handling equipment. Nobody else does that. And then secondly, which really amazed me, I go to a lot of industry conferences. I think it's very unique to SEMI. So this is the only industry that I've seen where you have PhD students present and it's literally 15 % of all presentations.

15:31And they're treated like gods or something. They're treated like, wow, we want to hire you. And it's incredible. So there's a lot of talent and these talents can choose where they're going, if that talent has, if you're a material scientist PhD, you have a choice and you want to do your life's work, you're much more likely to go to integrist because that's what they do versus going to a broader chemical company. And then on top of that, your PhD advisor probably has roots within integrist as well. So over time, they should grind out more market sharing chemicals as well. I think that describes quite well why I think in our opinion, the market share will go up, A.

16:08And B, they're already in a pretty good position today in terms of market shares and competition. Yeah, I think the next natural question is why they've been able to remain independent, what the origin story is here, and how they've gotten to this position that they're in today. Integris, the company itself, started in 1966 in Minneapolis, Minnesota, in a suburb. The people starting this company had a background in plastic injection molding, which then enabled them to build these wafer cassettes that were made out of Teflon, amongst other commodities. And that then enabled them over time to grow a significant market share in the handling of wafers.

16:52Then around 2000, that business IPO'd under the name Integris. And in 2005, the second big thing happened for them. they merged with a company called Micropolis, which was a spin -out out of Millipore. And Millipore, you may know this, is a filtration company that was started in the 50s. And they focused on filtrations across the board, water, healthcare, et cetera, et cetera. And then they spun out the semiconductor filtration piece, which then merged with Integris. So basically, at that moment in time, you had materials handling, you had filtration. And then in 2014, they've added the chemicals piece via an acquisition called ATMI.

17:34Then you had all three together. When you look at the company, the predecessor in Minneapolis, the founder of the company gave an interview about 25 years ago. And he was talking about a family type of culture in a cyclical industry, which is actually quite hard because when the cycle goes against you, you either have lower margins or you have to let people go. And hopefully you don't have to let people go. And if you don't, you have a really interesting culture where people really care. I think that percolated down into Integris. And then secondly, the current CEO came from Millipore, the filtration business, has been the CEO since 2012.

18:11And I think he realized that there's nothing like Integris out there that focuses just on semis and this material side. And he also realized that there should be a flywheel effect by combining chemicals, materials handling, and filters. And over time, you should drive significant synergies because they all work together. And I think that's why if you have this independent champion today now, and I hope it stays independent, I think it's some of the background, but I think it's also the CEO who's recognized he could do that. And he's been doing that for 11 years now. And can you talk a little bit about the customer base and the sales process for what they're doing?

18:51how much overlap is there in terms of the segments and the sale that's being made? What does that even look like on a year -to -year basis in terms of who's buying? And maybe just identifying who the buyers actually are from a customer perspective, if they have major exposure to particular players in the ecosystem. Let me do the second part first, and then I get back to the first part of your question. It's pretty straightforward. So biggest customers, about 50 to 60 % of revenue are the big foundries and fabs. So Taiwan Semi, Samsung, Intel, there's a bunch of memory players like Micron, Esky Hynix.

19:28There's a bunch of smaller analog fabs, text instruments, et cetera, et cetera. And then the rest of it is more like OEM type of chemical companies and distributors. And all of those need to filter the chemicals as they ship them. And all of those might also have to buy some sort of chemicals for themselves for their own product. So it's that type of mix shift, yeah? And I think to your first part, I think the design and process is really intriguing. You can't define the semiconductor industry as a three -legged stool. And there's three legs to it. I think you've done a few podcasts on some of the legs.

20:02One is the EDA guys, Cadence and Synopsis. The other is the Foundry guys, Taiwan Semi. And then the third leg is usually described as a semi -capex names, but actually includes the material name too. And all of those have to work together. So when LAM comes up with a new chamber, they will have a deep, deep, deep relationship with Integris for years to design both the filters and the chemicals within that chamber. Then LAM has to talk to Cadence, which then designs the stuff into their EDA tools, and obviously to Taiwan Semi, which then actually sort of runs the process. And the reason why I mentioned all of this, Integris gets designed in years in advance.

20:42It's a very, very R &D heavy process over the entirety of their business. And in my sense, this design and stuff has accelerated for them because again, they provide everything and they're just as integral part as anybody else in a three -legged stool. So it's hard to get into that stool, but once you're in it, it's really nice. Yeah, I guess there's a certain stickiness that's related to that. But when you're involved there, does that mean for that chip series, whatever that vintage is, that you are then the partner of choice until basically it's extinct and you move on to the next series? Is that the right way to frame it just in terms of the stickiness of that business?

21:26Absolutely. So once you're designed in at a node, you're designed in. Because most likely you will have created a special type of filter. You may have created a special type of chemical just for that process. which then works really well with the equipment, which then works really well with both Taiwan Semi and Cadence. So once you're designed in, you're done. Now, the second step is then for the new node business, you have a very unique advantage. You have a lot of data in terms of how did my filters react on the previous node? How did my chemicals react on the previous node? And what does that mean for the 10 -year roadmap that Taiwan Semi talks about?

22:00What do I need to tweak? How do I need to talk to LAM, ASML, AMET, et cetera, et cetera? It's very likely you're going to gain market share and gain more design wins on top of that versus previous history. Now, obviously, the key risk here is that if you make a big mistake, if you can't supply, if your big customers don't start trusting you, you have a real problem. But that would be an execution issue. But on a standalone basis, in terms of your competitive mode and in terms of your ability to gain share and to stay where you are, it's actually really a nice position. And over their history, have they had any execution issues or events which have resulted in big drop -offs beyond just traditional cyclicality?

22:42Luckily, no. I would argue they've done well over time in terms of not having these issues. The issue I mentioned before was actually a Japanese supplier of some commodity materials that were contaminated and delivered to Taiwan Semi. And if you look at their cyclicality over time, it's been mainly driven by wafer growth rather than having issues to execute. On that cyclicality, semiconductor business historically was a very cyclical business, which fell into the traditional capital returns, capital framework in terms of seeing that capex come in, that supply, seeing the resulting demand adjustments.

23:23how have they fared relative to the rest of the industry? Are they any more or less cyclical than the rest of the market? There is a perception in the market and in general that Integris is less cyclical than average. If you think about the semi -capics materials, I'll elaborate on that. There's pros and cons here. Truly, I mean, it's a consumable business. And so basically, whenever you manufacture a chip, you're going to consume materials from Integris. And that on average, at least over long periods of time means that it's a bit less cyclical than maybe semi -CAPEX names. Maybe to take a number, if you look at 2009, I think the average semi -CAPEX company revenue was maybe down 50%.

24:06Integris was down 30%. So slightly better. Still pretty cyclical, but slightly better. What I would say though, is that actually every 10 years or so, they do exhibit above average cyclicality, at least if you look at broader companies. And that happened this year, it happened in 2009, and it happened in 2001. And the reason for that is basically you have a de -stock event. The utilization at the fabs goes down, memory prices may go down, and then you have a year of less wafer output. And that's the key driver of cyclicality. And that seems to happen every 10 years in a big fashion. And it happened And this year, it happened 15 years ago, and then it happened 2001.

24:47But it doesn't happen that often. Whereas maybe on the semi -cap side, I think the cycles are more like maybe every three years or so, three, four years, something like that. And then as it relates to their own pricing power, we'll start to get into the financial model a bit here. They are operating with major players, extremely large businesses. They are involved in that design phase. But how do they defend themselves just in terms of getting pricing power, maintaining some margin profile? Talk a little bit about that historically, how they've been able to trend when obviously some of their customers have a lot of power in these relationships.

25:27So how has that fared for them as a business? Industry -wide, what I really like about semiconductors is that what you enable is so valuable that actually many companies have rising gross margins. Taiwan Semi, NVIDIA, ASML, LAMP, they all seem to benefit, although it's ultimately oligopoly on top of oligopoly on top of oligopoly. But somehow they all seem to benefit. Yeah, it's because we just keep buying iPhones and loading up AI chatbots. The other end of the spectrum is what's going up in price, perhaps. Yeah, that's right, actually. And I agree. And so maybe if we then hone in on Integris, they've done decently well.