In this episode of Industry Focus: Tech, we're switching things up a bit as Motley Fool contributor Jason Hall and analyst John Rotonti, with host Dylan Lewis, give you a breakdown of semiconductors: how semiconductors are made, the major players throughout the value chain, and the most investable ideas in the space.
To catch full episodes of all The Motley Fool's free podcasts, check out our podcast center. To get started investing, check out our quick-start guide to investing in stocks. A full transcript follows the video.
This video was recorded on June 4, 2021.
Dylan Lewis: It's Friday, April 4th, [it was actually June 4] and we're talking about the industry powering the digital revolution. I'm your host Dylan Lewis, and you're actually not going to be hearing much from me today. We put out a lot of content here at The Fool. Every now and then I'll hear a conversation and think, this is darn good. We've got to get this in front of as many people as possible. On today's show, we're going to be airing a conversation between analysts John Rotonti and Jason Hall breaking down the semiconductor industry. It originated in our Discovery Now premium live stream and then later made its way into Motley Fool Live, our U.S. member live stream. It's a fantastic overview of the industry, the shortage that's been collecting headlines, and why this space is full of so many stellar businesses. Think of it as a little sneak peak for non-members and a reminder for all of our members that there are some awesome conversations happening over at live.fool.com. Without any further ado, enjoy.
[...]
Jason Hall: So before we talk about our show today, which is going to be focused on the semiconductor industry and what's going on there, I think it'd be great if you could just spend a couple of minutes talking about what you do for The Fool and what you enjoy the most about the things that you do.
John Rotonti: Thank you, Jason. Yes. I'm a senior analyst. I've been at The Fool for seven years now and I'm the head of investor training and development. In that role I teach a lot of classes to our new analysts, that's one thing I do. In that role I also serve as a resource across our investment team. So if any member of our investing team is looking for an article or even maybe some WallStreet research on a particular company, or a model on a company, I may be able to help them with that. Third thing I do is I bring in outside guest speakers, practitioners, academics, all superstars in their fields and their area of expertise to come speak to our team internally, teach us classes, lead discussions. Sometimes I bring in a group of outside speakers and we host a round-table of panels and we get to watch these three or four amazing professional money managers have a discussion with each other for an hour and a half. That's how I spend my time as a coach. Then I'm the co-host of the morning show on Motley Fool Live, and the host of My Investing Life on Motley Fool Live, and then I'm experimenting with another show called Tech30, where I bring in some guests and interview them for 30 minutes and talk about tech and innovation.
Hall: Awesome. Well, there's no doubt that tech and innovation is the old software city in the world. The bottom line is that every company that's not using technology in their business is going to lose to their biggest competitor who is.
Rotonti: Exactly.
Hall: So that's awesome. Well, John, again, thank you so much for coming on. I can tell you that every member, whether they have listened to you talk about companies that you follow, or back when you were working on some of the premium services and providing the analysis that led to recommendations, every single member has benefited from you because of your influence on so many of the other analysts and helping out with coaching and training. For every single member I want to extend a thank you to you for that.
Rotonti: Thank you, Jason.
Hall: Onto today's show we have a fun topic lined up. One of the [...] and it's in the consequences that's happened recently, that's had all kinds of far-reaching implications across all kinds of industries is the semiconductor shortage, right? Whether it was coming out of COVID and the impact on global supply chain, natural disasters, fires, booming economy, surging demands, the proliferation of semiconductors across so many more different types of products and over. It's an amazing, interesting industry that's in a bit of a disarray in terms of meeting supply, but has so many implications that investors need to understand. John, tell us what we need to know.
Rotonti: Yeah, Jason. Regarding the shortage, you hit on most of the big ones. The biggest reason there's a semiconductor shortage right now, I think, is because of the digital acceleration that was brought on by COVID and the subsequent economic lockdown. Digital everything and the cloud was growing really fast before the pandemic, but the pandemic accelerated that growth, and those industries whether it's software, whether it's hardware, whatever it is in the digital space relies on semiconductors. When the digital transformation gets accelerated, literally by years, down into months the semiconductor manufacturers are already operating at full capacity or near full capacity and it takes time to ramp up semiconductor manufacturing. So that's a big thing. The industry was not ready for this accelerated digital transformation. You mentioned natural disasters. One major semiconductor fabrication facility was shut down by a fire, another by the decrease in taxes. Then a third thing I'll say is coming out of recessions or the economic shocks it's not uncommon to have supply constraints. It just so happened this time the supply constraint is showing up in semiconductors because of the digital acceleration we talked about and how important semiconductors are to everything digital.
Hall: It's a confluence of events. There are so many things that have occurred all at the same time that have made what was going to be a tight supply situation even worse. You hit on something that I think is really important to understand and talk about bringing these facilities up to scale and adding more capacity. I think we've all gotten spoiled when we see CrowdStrike report earnings or Zoom report earnings or Shopify, where their revenues are up 50% or 100% from period to period. When you're dealing with software or Software-as-a-Service or something on the Cloud, you can scale up very quickly, but when you think about building foundry for semiconductors and a lot of times they're very specialized, right? Making sure that you are building out the right capacity for the rights need can be really critical. It does take time. It's very time-intensive, very labor-intensive, very energy-intensive, and then there's all the logistics of this. Yeah, it's very power-intensive. But then you have the logistics too of the inputs, your feedstocks that you need, and then getting the product out and then getting a distributor to where it needs to go. It's about as old schools as it gets.
Rotonti: It takes years to build and $10 billion to $20 billion to build one modern-day advance node semiconductor manufacturing foundry.
Hall: So the one that comes to mind most immediately on that actual company that we're not really going to talk much about, but that's Intel. This is one of the companies that have earmarked these massive worth $30 billion or so to add capacity. Maybe it's more than that, but to build out more capacity in North America.
Rotonti: Yes. If you look at the three largest semiconductor manufacturing companies, Taiwan Semi (TSM +0.40%), I guess we should call them foundry. They're more accurately called foundries or fabrication facilities. But Taiwan Semi has announced they're going to spend $100 billion in CapEx over the next three years, $100 billion, Samsung that is going to spend over $100 billion over the next several years without specifying what several years means, and then you just said Intel. I thought it was $20 billion to build two fabs in Arizona, but maybe the number has gone up to $30 billion since the last.
Hall: No. I think that sounds right.
Rotonti: But it's a lot. I mean, it's a lot. Just three companies right there you're looking at, well, over $200 billion of industry CapEx in the next few years. That's a lot.
Hall: It is. Then they are going to have to be regularly spending additional capital costs to modernize the update just to continue to meet the design expectations.
Rotonti: Yep. Taiwan Semi is a leader in this. They are currently the only company globally at scale putting out five-nanometer chips, which is a leading-edge chip right now. They're going to open their first three-nanometer chip sometime in 2022. That's the timeline they've announced so far. So yeah, they're continuing to move, continuing to progress and innovate to more advanced chips.
Hall: Before we get into the specific companies it might be helpful to talk about as you think about the industry and as you think about how the buckets that the industry is within and the companies that participate in what they do, can you break that down for our viewers?
Rotonti: I can, for sure. What's remarkable is that at every stage of the semiconductor manufacturing value chain there's two or three and in some cases one global leader. We're really talking about earned oligopoly, earned duopolies, earned monopolies in some cases because as you said it's just so capital and knowledge-intensive that the barriers to entry are so high. If you look at how to make a leading-edge five-nanometer chip, for example, because these leading-edge chips they have billions of transistors on them and the chips are about the size of a stamp or a thumbnail. When I say billions, I mean billions, the Apple and one chip has 16 billion transistors. Amazon grabs a ton of chips and has close to 30 billion transistors, so tens of billions of transistors. You can't just put that together by hand. You need software. The first step is to design the blueprint for that tiny chip. How are you going to fit 20 or 30 billion transistors on it? There's two global leaders in software that do this design. It's called EDA software, Electronic Design Automation software. It's also sometimes referred to as computational design software, two global market share leaders, Cadence Design Systems and Synopsys.
This computational design software is highly complicated. It's a mix of matrix algebra, artificial intelligence, advanced geometry, and other fields of math that I can't even pronounce or understand. It's really that complicated. To work as a software engineer in one of these companies you have to have a masters or a PhD. It's not just you take a class in Python or something like that. It's super-advanced. That's the first step, you design a blueprint of what the chip is going to look like on the software to global duopoly. The next step is basically a chip. Chip-making basically has a thousand steps. But you can basically break it down into, you deposit film onto the silicon; onto the wafer and the film is different chemicals. Then you etch away or cut away at that film. Then you wash with water. You repeat that tons of times. It's more complicated than that, but you repeated tons of times.
The next step after you make this blueprint with a software, as you do the deposition, you deposit these layers of chemical onto the wafer. There's three companies that make these deposition machines, Applied Materials, Lam Research, and Tokyo Electron. But in deposition, Applied Materials is the industry leader, once again, an oligopoly, after you have done the deposition, then you take the blueprint that was created with the software and it is shined. You use a light source and you project that blueprint through a mask and through a series of mirrors onto the chip. This mask and this series of mirrors shrink down that blueprint to the size of a chip. How it's done is really magical. There's only one company that provides the light source that is shined through this mask and through these mirrors to shrink the blueprint down onto the chip. That's ASML. It's the only company on earth that has this extreme ultraviolet or EUV lithography. Basically, it's a tiny wavelength of light that is shined through this mask in these mirrors to trace or stencil the pattern onto the silicon wafer. The way it works, people thought it was impossible. Industry experts thought it was impossible but ASML figured it out. A tiny drop of molten tin is hit by carbon dioxide lasers, CO2 lasers, and the lasers vaporize the molten tin. That tin turns into a ball of plasma, shining so bright that it creates this EUV, this extreme ultraviolet light source. The science is so complicated, no other company has been able to figure it out. This process of shining these lasers onto the tin, creating this plasma ball of light happens 50,000 times a second. The reason EUV extreme ultraviolet light is so important is because extreme ultraviolet light has a tiny wavelength of light.
Hall: The wavelength. That's how you get smaller and smaller.
Rotonti: If you're putting 20 billion transistors on a chip exactly right, those transitions have to be packed really close together. The tracing, the stenciling has to be used with a really small wavelength of light. We've deposited material. There are oligopoly of companies that make those machines. We've done the EUV lithography. There's only one company that makes those machines. They cost about $200 million. Then we have to etch away or carve away at the deposition to lay down the transistors and then to connect those transistors. There's three companies that make those machines. They're called etching machines, same three, Lam Research, Applied Materials, and Tokyo Electron. But in etch, Lam Research is the leader. At each stage of this value chain, you have monopolies, duopolies or oligopolies because these companies are so unique and so important and what they're doing is really bending the laws of physics. It's almost impossible to replicate.
Hall: Well, I don't know if you're familiar with Arthur C. Clarke, the science fiction writer.
Rotonti: Sure.
Hall: -- known for his Clarke's three laws. I think the perfect application of his third law is that, "Any technology that is sufficiently advanced is indistinguishable from magic."
Rotonti: You're exactly right. Industry insiders call this process of making a chip, they call it black magic. It's really complicated. All of this happens by the way, at one other company operating at global oligopoly. All of this takes place at a foundry, at a contract manufacturing facility, and the leader there as Taiwan Semiconductor. The deposition machines, Taiwan Semi buys them. The tech machines Taiwan Semi buys them. The extreme ultraviolet lithography machines from ASML, Taiwan Semi buys them. They spend the CapEx, they buy them, they put them in these massive facilities. Massive facilities are much bigger than a football field and they go through these thousands of steps. It's called a recipe and they make these advanced chips and the process is like magic.
Hall: The business model works because if you're Apple or you're Microsoft or you're Ford, any one of these companies that's utilizing semiconductors even if you're designing them yourself, spending the capital to build out a facility to manufacture it yourself is almost nuts.
Rotonti: These companies, they're called fabulous because they don't have fabs or fabrication facilities. That's exactly right. They put the CapEx burden onto Taiwan Semi.
Hall: Right.
Rotonti: It's a mutually exclusive relationship though. Taiwan Semi is getting good economics from this as well.
Hall: That's the key. Unless you can produce massive volume for your own consumption, it can be cheaper to be fabulous. To use somebody else as manufacturer because when they operate at scale, the volumes they can produce drives down the per-unit cost, and they can be profitable, producing it for cheaper than you can do it if you did it in-house.
Rotonti: That's exactly right. The way that these chips are made when you go from one generation of chips to the next, you start with the recipe. You start with the process of knowledge from the prior generation. To go from a seven-nanometer to five-nanometer, you start with a seven-nanometer and then you tweak it from there. Taiwan Semi has done more of these advanced chips than anybody else. They have more process knowledge than anybody else.
Hall: Which is a major competitive advantage.
Rotonti: It's a major competitive advantage, Jason and it's really hard for others to catch up because you can't skip steps. Intel is behind now. They have to do a seven-nanometer because they can't figure out how to do five or three without going through that process knowledge and learning those steps as they go. That process knowledge, that IP, the library of recipes that Taiwan Semi and Samsung has built up, and Intel to an extent, it's an amazing competitive advantage.
Hall: Now that we've got a baseline of knowledge about the different parts of the industry and the levers that are tied to it, we've already been doing this for about 19 minutes, John, that's amazing. It feels like it's been 19 seconds. This is incredible. If we can spend maybe 10 minutes or so talking about some of the top companies here that you've identified that you like.
Rotonti: Two or three of them I've discussed already, so we'll be able to be efficient with the 10 minutes we have. ASML is the only company on earth that has figured out extreme ultraviolet lithography.
Hall: Right.
Rotonti: Moore's Law says that every two years or so, the number of transistors doubles. The only way to double the number of transistors on a chip if the chip sizes change is to shrink the transistors, and push them closer and closer together and then to build these 3D chips and put them on top of each other. But the only way to do this shrinking is with this really ultraviolet wavelength of light. ASML is really the company that has allowed Moore's Law to continue to progress for so long. ASML has earned a monopoly. It is truly a magical company. They are bending the laws of physics, to catch up with them is going to be really hard. Lam Research I mentioned, then Taiwan Semi, I mentioned. All unique businesses as in my opinion, are as difficult to replicate as any business on earth. I'm not going to say it's impossible, but as difficult to replicate as any business on earth.
One we haven't discussed yet is Texas Instruments (TXN 1.59%). The ticker there is TXN. Once again, back to this theme. It operates in a global oligopoly or duopoly. Texas Instruments is the largest, excuse me, analog semiconductor company in the world. Texas Instruments and the No. 2 runner-up player Analog Devices. They control about 50% of the global market share for analog chips. Texas Instruments has an extremely diverse revenue base because they sell 100,000 different products to 100,000 different customers. They sell more products, their product portfolio, their menu of products is larger than any other semiconductor company on earth, and they sell to more customers than any other company on earth. Hundred thousand products, 100,000 customers, so really diverse revenue. The question I get most often with Texas Instruments is, "Why are they selling these analog chips when everything is going digital and digital is exciting and it's fun." The reason is because TI is an amazing business, it's run by amazing leadership, and they realize that there's this analog paradox at work. You probably never heard that phrasing before. I don't think you'll read it in WallStreet Research but there's an analog paradox at work, meaning that as the world shifts to digital, the demand for analog grows in tandem.
Hall: Yeah.
Rotonti: It grows with it because analog chips, they do things like manage the power in your cellphone. Your cellphone is a digital device, but you need analog along with it. There's only a few companies really doing this at scale. Then the other great thing about analog chips, as the rest of the world is tripping over themselves getting into the digital space, analog chips have long life cycles. Texas Instruments is mainly focusing on autos and industrial.
Rotonti: The average life cycle for one of their auto chips is seven years. The average life cycle for one of their industrial chips is 10 years. So these chips don't face the risk of technological obsolescence.
Hall: When you say life cycle, you mean the period of time from when they manufacture the first one to when they manufacture the last one?
Rotonti: Yeah. That's right, or the period of time before that chip goes out of style and it needs to be updated.
Hall: Right. As compared to the chips that Apple designs and users, every new iteration of their iPhone has a new processor?
Rotonti: They're already coming out with the M2, they came out with the M1 a year ago. That's exactly right. So there's no risk to Moore's Law with Texas Instruments. If they don't sell a chip this quarter, it's got a 10-year life, they'll sell it next quarter or next year at a high-margin. In fact, Texas Instruments is selling chips that were released 30 years ago. So on average, auto chips have about a seven-year life cycle, industrial is about a 10. They're still selling some chips they designed and manufactured 30 years ago. Keywords are manufactured, they're vertically integrated. They do their manufacturing in-house, unlike digital chip companies for the most part, and unlike their competitors. They are the only analog chip company. Only analog chip company, once again, a monopoly when it comes to this. Manufacturing chips using 300-millimeter fabrication, which reduces the cost to manufacture these analog chips by 40%. So they have ultra-low cost manufacturing because they do it in-house. Manufacturing through their in-house manufacturing facilities carries incremental growth margins of 70% to 75%. You see incredible margins, incredible returns on capital from a company like this, and they're the only company manufacturing in-house in their space.
Hall: It's good to be the king. I think that's the situation with Texas Instruments.
Rotonti: It's exactly right, and because they control their supply chain more than their competitors, given the backdrop we're in right now with the shortage of semiconductors, their customers aren't experiencing as much of a shortage because they control that supply chain. They're vertically integrated. There's some shortage. There's some hotspots, they call them, but for the most part, their customers are getting the chips they need. It's a fantastic business.
Hall: Here is an interesting thing. Every single one of these companies has outperformed the S&P 500 over that period. That certainly supports two things. No. 1, the trend, the growth, the tailwind for semiconductors, and it does tend to indicate that the market supports you, John, as saying that these are the leaders in this space.
Rotonti: That would hold if you look at the Philadelphia Semiconductor Manufacturing Index and you go back 10 years, it beats the market 10 years. It beats the S&P. My point is it holds going back further, but just looking at the companies that you mentioned today, one of the reasons they beat the index, Jason, is because they weren't trading at these valuations that's some software. Amazing software companies and I love them, but the semiconductor companies weren't trading at the valuation some of these software companies were, so they haven't fallen as much. So if you look at Texas Instruments, Jason, it's down 10% from its 52-week high. Fastly is down 65%. If things aren't falling as much, it helps when it comes to outperformance.
Hall: It does. It's also a reminder that to a lesser extent, this asset rotation sell-off that we've seen from some of our more interesting names has still impacted even this industry, even though we know that there's strong demand with. It's an opportunity, I think that's the key. It's predictability and opportunity.
Rotonti: The other reason they're outperforming, this is Lam Research, this is from New Constructs. They put out really rigorous calculations of free cash flow and returns on invested capital. Look at Lam Research's returns on invested capital. Going from 2016 to 21% trailing 12-month 54%. The average return on invested capital, which is a measure of profitability and efficiency for the market, is about 12%. This is a company that's generated returns on invested capital of 54%.
Hall: So that screams demand, that screams competitive advantage, and that screams amazing economics.
Rotonti: Yeah.
Hall: Oh, by the way, we're really good at capital allocation.
Rotonti: Exactly. You look at free cash flow margins, free cash flow is over sales, 17%, 22%, 20%, 26%. Turning $0.20 of every $1 into free cash flow, pretty impressive.
Hall: Yeah, that's incredible. Those are numbers to get excited about.
Rotonti: You mentioned the demand, maybe one thing to talk about from a high level, another reason, these stocks and the industry, the Philadelphia Semiconductor Index is outperforming so well, is the thesis for the industry. There are critical infrastructure components to the digital revolution and everything digital, it's a large and growing market and so you've got the long term tailwind. There's no aspect of the modern digital economy that can function without semiconductors, zero. Cloud, artificial intelligence, machine learning, 5G video games, electric vehicles, autonomous vehicles, sensors and wearables, the Internet of Things, automation, robotics, crypto, mining, space travel, there is no aspect of the digital revolution that can exist without semiconductors. I hear a lot of investors say that it's Cloud computing that is powering the digital revolution.
Hall: But what's powering Cloud computing?
Rotonti: Exactly, Jason, and I would agree with that. The cloud is a game changer, but it's a slight oversimplification because without semis, there's no Cloud. Then you mentioned cyclicality really early. I just want to hit on that really quickly. The industry is cyclical. There's no doubt about it, but much less cyclical than in the past. Three reasons for that. One, now they're growing into everything. The cycles used to be driven by the PC cycle and the smartphone cycle. Now they're in everything, including Cloud data centers and servers and everything. Now they're in everything, that's No. 1. No. 2, there's been several waves of massive consolidation in the industry and we're in one of those waves right now. When an industry consolidates, the remaining players become more rational. Then the third reason, you mentioned earlier, now the OEMs themselves, the largest, most profitable, biggest companies in the world, are designing their own semiconductors. It's a whole new customer base; Apple, Amazon, Tesla, Google [Alphabet], Microsoft, talking about this. They're all designing their own semiconductors now. So now that market opportunity is so much bigger.
Hall: Yeah. I think that also changes the capital allocation model, because if you're no longer designing and building it, you're not really thinking about the capital investment you'll have to make to build that new design. It just completely changes the economics.
Rotonti: Totally. Then the final thing we touched on when I shared my screen, the fundamentals for semis have been misunderstood for a while. They're being better understood, but I still think they're underappreciated. If you look at a list of the best semiconductor companies and put it up against a list of the best software companies, the corporate fundamentals are every bit as good for the best semiconductor companies as they are for the best software companies. Their profitability and their returns on capital are exceptional. Their growth is not as good as software, but they are definitely growing far above GDP. The most recent downturn in the cycle was 2018 or 2019, the trough margins, so the worst the margins got in the most recent down-cycle, were better than the prior peak markets.
Hall: That's a tremendous amount.
Rotonti: Due to different industries.
Hall: Well, I think another thing that's worth mentioning too, is the way that the industry has changed. It seems like it supports more of the stakeholders in a better way, if you think about it.
Rotonti: I think so.
Hall: That is an indication that all of those strong economic factors in these businesses and the way that they're operated and the returns that they can generate for investors, seems like it predicts that that should continue to prove to be the case.
Rotonti: I would agree with that, Jason. I'm not saying they're not cyclical, but I think that there is less cyclicality than in the past. I think they will generate higher margins and higher returns on capital across a cycle than in the past, and I think over a long period of time, they will, as a group, and especially the best players that we discussed here today, we'll continue to outperform the market.
Hall: Well, you heard it right here, ladies and gentlemen. I think it'd be fair to say that a basket of these five great companies and these various parts probably work out pretty well as a good way to invest in semiconductors over the next five to 10 years.
Rotonti: I think so.
[...]
Lewis: Listeners, that does it for this episode of Industry Focus. If you have any questions or you want to reach out and say "Hey," shoot us an email at [email protected] or tweet us @MFIndustryFocus. If you're looking for more of our stuff, subscribe on iTunes or wherever you get your podcasts. As always, people on the program may own companies discussed on the show, and The Motley Fool may have formal recommendations for or against stocks mentioned, so don't buy or sell anything based solely on what you hear. Thanks to Tim Sparks for all his work behind the glass today, and thank you for listening. Until next time, Fool on!
