Chit Chat Stocks - A Primer On Investing In Semiconductors (History, Supply Chain, Big Winners) - $TSM $ASML $CDNS

Episode Date: July 15, 2026

On this episode of Chit Chat Stocks, Brett and Ryan dive into the semiconductor industry to give an introductory overview of the sector, including multiple stock case studies. We discuss: (00:00) Introduction (04:04) What Is a Semiconductor? Basic Explanation and Analogy (15:11) History (19:54) Industry Cycles, Leading Companies, and Market Dynamics (30:09) Case Study: ASML and EUV Lithography Machines (38:03) Case Study: Taiwan Semiconductor (TSMC) and Advanced Nodes (53:00) Cadence Design Systems and Semiconductor Software (01:08:13) Listener Questions ***************************************************** Subscribe to our newsletter, Emerging Moats: emergingmoats.com  ********************************************************************* Chit Chat Stocks is presented by Interactive Brokers. Get professional pricing, global access, and premier technology with the best brokerage for investors today:  https://www.interactivebrokers.com/  Interactive Brokers is a member of SIPC.  ********************************************************************* Fiscal.ai is building the future of financial data. With custom charts, AI-generated research reports, and endless analytical tools, you can get up to speed on any stock around the globe. All for a reasonable price.  Use our LINK and get 15% off any premium plan: ⁠https://fiscal.ai/chitchat  ********************************************************************* Disclosure: Chit Chat Stocks hosts and guests are not financial advisors, and nothing they say on this show is formal advice or a recommendation. Learn more about your ad choices. Visit megaphone.fm/adchoices

For the past three years, IBKR individual clients averaged 24.3% annually, beating the S&P 500's 23.1%. Lower costs and 170-plus global markets matter. Interactive brokers, member SIPC, visit ibkr.com slash performance. Welcome to Chit Chat Stocks. On this show, hosts Ryan Henderson and Brett Schaefer analyze businesses and riff on the world of investing. As a quick reminder, Chitchat Stocks is a CCM Media Group podcast. Anything discussed on Chitchat Stocks by Ryan, Brett, or any other podcast guest is not formal advice or recommendation. Now, please enjoy this episode.

Welcome into the Chitchat Stocks podcast, a podcast to help you find your next great investment. My name is Brett Schaefer, and I'm joined by my co-host Ryan Henderson to bring you another special Wednesday episode. We are doing another stock market sector theme. This was a new style of episode. We came up with 2026. We think listeners like it, so we're going to keep doing it, I think, over at least the next few years. We've covered so far stock exchanges. That was a very fun episode. We've covered the space and defense space economy. We did companies like, what was it even? Planet Labs, I think, maybe Black Sky Technologies, Redwire, a lot of interesting startups within that space or sector, no pun intended. But today, we

are diving into what is probably the hottest sector at the moment, semiconductors. They've rapidly grown, which one might argue, I was trying to do some analysis myself, and by By market cap, I think unless you include all financial services around the world into one sector, which I think that is probably disingenuous, you'd want to separate insurance from banks and stock, stock brokerages, all that stuff. Semiconductors are the largest by market cap in the world. And given the importance of the sector, there have been massive winners going back decades famously nvidia has delivered a 500 000 percent gain since 1999 or 37 percent cagger through to today ryan you do a lot of these stats for fiscal ai i think that is the best

performing stock since their ipo i believe it would be yeah from 1999 yeah all right tsmc or taiwan semiconductor company we'll be talking about in a case study today an 18 cagger since 1997. ASML, another case study, we were doing a 26% CAGR since 1995, which is 150,000% GAID. I mean, this is a sector with total market caps in the tens of trillions. McKinsey pegs total revenue at $775 billion with projections for growth to $1.6 trillion in the near future. That's definitely depending on the AI build-out, things like that. Obviously, things can change. If we go back 1960, the semiconductor market was very nascent and well under $1 billion in revenue. Estimates vary on the true size, but it is this decades-long level of sustainable growth through market cycles,

through the changing of the economy, and really providing the computing backbone for the internet, for smartphones, for computers, for PCs, for AI, for self-driving cars, for electric vehicles. If you have reasonable starting prices, like a lot of these stocks here, massive growth runways and margin expansion, that's how you get these massive winners. We're going to discuss the sector as a whole, the history, and then go through three separate stock case studies to try and identify characteristics that deliver these massive returns so we can learn and the listeners can learn to make better future investments. And we're going to conclude to look at ranking, I guess, our favorite three semiconductor businesses.

But before I start, an important reminder. If you like this show even a bit, follow us on your podcast player of choice, subscribe to our newsletter, and join our free chat community, which is in the show notes, and give us a five-star review wherever you are listening. Ryan, I'm going to let you kick things off. Did you do any confirmation on NVIDIA being the best performing stock of the 21st century?

They are. NVIDIA is the best performer. second is monster beverage and third i believe is actually decker's outdoor which is the uh they sell like hokas which that always catches me off guard yeah anyway um yes nvidia trounces the returns of every other company over the last 27 years so um yes they are by far the best performer before brett dives into some of the history of the industry i'm going to talk briefly about what a semiconductor actually is i know some people are going to roll their eyes uh people that know this industry are probably going to roll their eyes um hopefully not too many electrical engineers listening if you are skip this part yeah but i'll i'll try to give my analogy and

And try to give the basic explanation for what a semiconductor does and then what are the actual elements of the supply chain that allow a semiconductor to go from a drawing to a physical microchip that powers the modern internet today. I'm going to start with a quote. A semiconductor is a material with electrical conductivity that falls between a conductor, like copper, which allows electricity to flow freely, and an insulator, like glass, which blocks electrical flow. This intermediate property allows semiconductors to precisely regulate and switch electrical currents, making them the fundamental building blocks of all modern microchips and electronic devices. The analogy that I have heard that I think is somewhat helpful is that of the smart doggy door.

So if you have a doggy door that's always wide open, just a hole in the wall, obviously anything can walk right through it. So you can think of that as sort of the copper wire in this case. On the flip side, if you have a brick wall, nothing will ever get through it. So think of that as sort of the rubber or the glass, something that stops electricity from getting through. But the best solution is a smart doggy door, one that as the dog approaches, the door opens, and when the dog is gone, it stays shut. That is the equivalent of a semiconductor. The material defaults to acting like an insulator and blocking the current, but when you apply a small amount of energy, it instantly switches to a conductor and the electricity passes.

When you connect, today, billions of these smart doggy doors together in an organized layered grid on a piece of silicone, you can form what we now know as a modern microchip. These microchips can then execute code because they convert written software instructions into a physical sequence of flowing electrical currents. And that is the basics of sort of what the modern economy rests on. And while simple at its core, obviously the designs and the building of these chips has become incredibly advanced and one of like the modern marvels of the world that we're able to produce these things at such microscopic levels. But let's go through what the semiconductor supply chain actually looks like. So sort of the first stage, you have the blueprints. This is basically the IP. it's chip design firms like nvidia for example that employ teams of engineers that are writing

custom code to describe how a chip should behave this design process includes buying licensing rights from pre-made chip parts for companies like arm for example i think arm has thousands of patents for specific designs especially with the cpu core i believe is one of one of their primary popular patents and then you combine those with your own custom logic as well and you build your own chip designs it also often requires using eda electronic design automation software like synopsis or cadence design to simulate the chip so that the design firms employ a bunch of engineers buy patents or rights to patents from other chip firms. They use software to simulate, they use custom code. And once the digital blueprint layout is finalized, most companies

send them to a factory as a massive file. Actually, all design firms send them to a factory. Sometimes that factory just happens to be in-house or under the same umbrella. The companies that specialize in chip design specifically include businesses like nvidia amd broadcom qualcomm those are probably the biggest by market cap sort of the second stage here is what i would call manufacturing pre prerequisites so this is what the actual manufacturers need in order to uh turn these ideas into physical atoms so before the designs can even actually be built there's a massive supply chain on the raw material side of things most notably acquiring silicon so quartz sand is mined and melted down into raw silicon the sill the silicon is grown

into a massive cylindrical crystal called an ingot or ingot and then specialized suppliers then slice these ingots with diamond saws into ultra thin circular discs called silicon wafers if you've ever watched a video of like a taiwan semiconductor factory you'll typically see these circular discs almost looks like a dvd just way thinner and that is the basic or the the base for chips side note about 70 to 80 percent of the world's silicone comes from china so they are the largest miner uh by far and there are some obviously there's other countries that do it as well uh but yeah that's that's by by far and away the largest is china along with silicon factories also have to buy a ton of specialty chemicals the reasoning for each chemical is

kind of technical i tried to look into each one but it's it's sort of an endless uh list of chemicals that you need for different elements of the uh manufacturing process ultimately it comes down to semiconductors are incredibly small so any microscopic impurity can cause the entire chip to fail so there's a whole bunch of chemicals used for layering and controlling contamination that's that's sort of the basics and then the other critical element of a factory being able to bring their clients designs into reality is equipment if you watch a tour video like the one i was describing earlier of type 1 semiconductor or any fabrication facility you will see factories are full of some of the most advanced industrial machinery ever made

These include EUV machines and DUV, extreme or – what's the D stand for? I'm not sure. I can't remember. I'll look it up. Ultraviolet light. EUV and DUV are kind of the machines that ASML is known for. Deep. Deep.

Extreme and deep. There we go. And those machines can cost upwards of $350 million a piece. There's also machines from companies like Applied Materials and Lam Research, which help with etching or layering. And so these are all kind of the suppliers that go into a factory. So just to go through it one more time, big silicon providers are basically China. And the companies that are most known are Shinetsu Chemical and Sumco. Also, big specialty chemical providers are companies like Merck Germany. Note that Merck Germany and Merck are not the same. Apparently, they split during World War II, I think. And DuPont, they're another one of the big specialty chemical providers. And then the major equipment suppliers are companies like ASML Applied Materials and Lamb Research.

The next stage is the fabrication itself. Fabrication facilities or manufacturers put all of the steps we just mentioned above together in one building or in many buildings, depending on the size of the company. the most notable of these is taiwan semiconductor but there are some other pure play foundry businesses as well so umc is one they're also based in taiwan global foundries which i believe was the foundry side of the business to advanced micro devices amd okay long time ago they spun it off i think uh and intel is trying to get into this they're trying to pivot from design plus fabrication to splitting that off having two separate businesses which is proving much harder regardless of what their stock is doing right now it's proving much harder than maybe many thought and then yeah so brett just alluded to it intel that's an example of a more integrated provider

so this used to be the model where you would do the designs and the fabrication both under the same roof in the early days of the semiconductor industry now it's kind of been separated out and everyone has their own specialty for the most part but samsung intel are probably the two biggest where they do both in-house and then the last component is distribution so once fabrication is complete and the wafers have been diced into individual cpus or gpus or in the case of google tpus they are sent to the customers or to manufacturing companies like foxconn If you're wondering where you've heard the name Foxconn, they are the largest manufacturer for Apple or assembler, I should say, for Apple. I believe they account for around 70% of iPhone manufacturing.

There's also testing, packaging, and assembly in there. Some fabs do that themselves. Some pass them along. But, yeah, that's kind of the gist of it. When you look through that whole supply chain, I was thinking about this. What's the best part as an investor to own in that supply chain? What is the highest margin, highest growth potential? And I don't think it's any one element of the supply chain. It's, it's the companies that have created something that no one else can, or at least have such an advantage that no one else can can catch them. So, and that exists all along the supply chain. And so design even, you know, NVIDIA's GPUs have been hard to catch. And obviously there's workarounds depending on the workload that you're trying to create for these chips for.

You've seen this with big tech rolling out their own chips. But, you know, companies have big advantages with design. ASML has big advantages with equipment. They're light years ahead. No pun intended there with the light part. And then Taiwan Semiconductor is – they've created a – not necessarily monopoly, but 90 percent of the leading nodes are manufactured by Taiwan Semiconductor. So – and I'm sure the software design too. All of that is if you've created something that's specialized, way advanced, that's where you're going to get the highest margins.

It's not from any specific segment of the supply chain, if that makes sense. You research your investments, you analyze markets, you manage risk, but did you research your broker? For the past three years, IBKR individual clients averaged an annual return of 24.3% compared to 23.1% on the S&P 500. IBKR's lower trading costs, competitive rates, efficient execution, and access to more than 170 global markets helped investors keep more of what they earn and put more capital to work. Over time, the broker you choose matters. Interactive Brokers member SIPC. If you care about performance, find out why the best informed investors choose Interactive Brokers at ibkr.com slash performance. Okay, let's move into the history. I'll try to

go through exactly how this industry was developed so listeners can help better understand where we are at today, the challenges of the industry, why the leaders have developed. For anyone looking at comprehensive history and the geopolitical ramifications ryan mentioned this whole industry is global and it is turning into a national security uh priority risk for many of the countries involved many mainly china and the united states as well as their allies i would recommend people read chip war uh provides a nice global industry perspective so if we go back to who invented the not necessarily semiconductors but the transistor and the modern computer chips we have to look at bell labs which was a another first half of the 21st century as i wrote here

the first half is the 20th century a joint research effort between western electric and at&t it's called bell labs and in 1947 the lab invented the first transistor which is what ryan basically mentioned, the on-off switch using the semiconducting materials. Old computers, which were used to do very simple computations, were vacuum tubes. They used vacuum tubes to do these calculations, and they were large, bulky, and pretty easy to be disruptive. You came up with another product like the semiconductor transistor. And if you look at this, this basic transistor, which, gosh, I forgot to put down the person who invented it. It's on the tip of my tongue. It's the basis for semiconductor computers today. Instead of one transistor, I looked this

up, the NVIDIA Rubin chip is going to have 336 billion transistors, which as many listeners will know, will take all the information that is fed into them and get them down to the binary one or zero decisions of the on-off switch decisions. And that helps power the AI and computers and smartphones and everything out there today. Now, in 1954, the first silicon-based transistor was made. 1958, importantly, the integrated circuit was invented by Texas Instruments and Fairchild Semiconductor, which is one of the bases of the founders of Silicon Valley. And I think people from Fahrenheit Semiconductor went to form Intel, I think.

And this enabled a huge cost efficiencies, performance improvements, and other factors. If you think integrated circuit, it's really in the name. You integrate everything together and build kind of the computer on a single circuit. In 1965, the founder of Intel, Gordon

Moore, came up with what's known as Moore's Law, stating that the number of transistors his engineers could put on an integrated circuit doubled every two years while at the same time, the cost of producing them fell. So you have a combination of decreasing costs and an increase in performance over the last, what is this, 65 years, 60 years?

That meant that computers would quickly become accessible compared to the gigantic mainframes of the day. This progress, which has largely held up through 2026, is why we have relatively cheap iPhones with better processing power than the entirety of NASA during the Moon program. Brings me great pleasure that people are using this, hopefully to listen to our show but also doom scroll on tiktok uh eventually though if moore's

law holds up decades from now we're going to be able to jam even more overly more intelligent processing power into tiny devices um this is the vision i think for smart glasses even though i kind of find them as we've discussed on this show more related to testable it's why they have that vision these technology companies because eventually you'll be able to if moore's law holds up put tiny and tiny you know the same processing power of an iphone within a tiny side part of your uh glasses and it feels simple like but i think people sometimes try to get too complicated with the semiconductor industry from an investor perspective but there's a reason the industry has delivered long-term growth going from under one billion dollars in revenue to

775 billion dollars today and probably 10x that in the decades ahead because the total roi of having these insanely good computational devices is significantly higher than their cost. Even though the sector is collectively putting tens of billions of dollars annually, maybe hundreds of billions into factories, design, software, and given capital, return on invested capital remains high. And the overall profit pool is large across pretty much the entire supply chain. Of course, we're testing this ROI threshold by throwing about a trillion dollars at the AI black box. But if any sector can deliver that returns, I think it's the combination of this. They can deliver the healthcare miracles, complex informational chatbots, and all the other

stuff that are promised. Well, the ROI may still be positive if they're throwing trillions and trillions of dollars at the industry. Let's see. We've talked about the geopolitical stuff. I think maybe we skip that as we try to get through these case studies. But if we look at maybe the leaders of the industry. Historically, the semiconductor industry has been cyclical with some subsectors like memory, which is Micron, Samsung, SK Hynix, the popular Korean stocks of right now. They've been more cyclical than others. However, over the long term, it's been a cyclical industry with long term secular tailwinds as demand from customers has gone up and to the right over the decades. The companies have been able to put up sustainable leadership positions or monopolies

in the niches have delivered phenomenal returns to shareholders, as we discussed in the opening. If we look at the early days, 70s, 80s, and 90s, this was Texas Instruments, Intel, 2000s. We had Samsung, Qualcomm, Broadcom emerge in the last 20, 25 years. Taiwan Semiconductor, NVIDIA, and ASML have come to dominate the industry. And in recent years, we've seen a surge from the big three memory players like Samsung, SK Hynix, and Micron. I mean, we can even go down more in the supply chain. Applied Materials has been a leader for a long time. There's software from Cadence and Synopsys. There's architectural blueprints from Arm Holdings. If you look across and down the supply chain, you see virtually every company out there. You can use our friends at Fiscal AI and do

this. They all deliver fantastic gross margins through the cycle, even if they are selling to each other. One person has 70% gross margins. They sell to another provider who then sells to someone else who has they have 60 gross margins that another person sells that's another 60 gross margins and the question you might ask is what is it about the semiconductor industry that prevents composition from driving down margins and i think it's the fact that the roy for the global economy is so massive is there anything else ryan you think besides the fact that hey we could sell everything for pretty penny everyone makes out well and end customer who buys the $1,000 iPhone is still happy.

No, I think that's the important part here. As you look through the financials and you look at the growth of what seems like everyone in the industry, and to be clear, there are a lot of semiconductor companies that have fallen by the wayside and failed. But you look at the just market cap explosion of semiconductor companies combined, I think you could argue it still pales in comparison to the benefits that human beings have received from the the growth of this industry it's you know i can tap my device in my pocket a few times and have anything in the world delivered to my door in a few hours that's that's pretty nice and a lot

of that is because of the evolution of semiconductors do we want to talk more law and some of the disruption potential sure i think if you're looking at industry like this you might want to ask well there's been these leaders over the years that deliver these long-term returns is there any risk from new technologies completely upending the industry what's an example of this i can't even think maybe in like consumer packaged goods coffee getting disrupted by something else things along those lines, soda getting disrupted by natural things, things of that nature, or maybe tobacco, tobacco getting disrupted by new nicotine devices. Is there something on the horizon in semiconductors that can completely make them obsolete? And I think

there are questions out there about Moore's law continuing. TSMC is down to, I think, a two nanometer node. For anyone that's wondering, one nanometer is equivalent to 0.00, I'm going to get this right, 0, 0, 0, 0, 1 centimeters. And with these tiny transistors, the gate for the transistor is only a few atoms thick. And given the probabilistic nature of single electrons, it sometimes ignores the gate, which can present an operational issue. They've worked around this. But again, they're working at extremely microscopic levels at this point. And there are concerns that Moore's Law will continue. So all around, people, this is why when you get the hype around companies like the quantum

computing, or I wrote quantum computing semiconductor, basically quantum computing technology, is why these stocks, even though they really generally don't have any business models at the moment, there's a dozen out there, maybe even dozens if you count ones that are privately held. It's because if they can do what they promise, that could, and this is a huge emphasis on the if, if the technology can be commercialized, you could see the semiconductor industry replaced with a quantum computing industry within a decade. It could provide massive value there. Now, despite the consistent presence of, say, a genius disrupting everything, or a team of geniuses disrupting the current state of semiconductors, finding something better,

I still have high confidence that the profits from the semiconductor market, maybe if we exclude the absurd stuff coming out of memory at this exact moment, will be larger every decade moving forward. Of course, there's a small chance every year that there's some revolution out there that completely upends the computing market, makes semiconductors obsolete. It would make probably, what, plenty of the companies on this list obsolete that we're talking about today. But when we look at them and we want to look at some of these stocks in the case studies,

we want to understand, despite this tiny risk, the attractiveness of these businesses, why they've delivered such strong returns, and what we can learn from them, and whether there's any opportunities out there today. This is close to a potentially top in the cycle, or one may argue there's some elevated multiples, as we're going to look at here, but I think it's fascinating nonetheless. Maybe there'll be stocks to keep on the watch list. And if anything, maybe you can also apply it to other sectors to try to find winning stocks, monopoly positions, things like that.

Yeah, it's sometimes, it just boggles the mind when you think about how small these things are And the manufacturing supply chain that's been built, it's really astounding what we've gotten to. Let's go through – I think that gives a pretty good foundation for where we're at today, where semiconductors are at. To paint a picture of the current memory chip boom, maybe we could call it temporary or the beginning of something even larger. sk hynix samsung and micron in 2026 will generate so those are the memory chip the three primary memory chip players they will generate more in profits than all of big tech combined which are their primary customers it's it's pretty astounding to think about

how much they are going to earn. I believe Micron is going to earn more in 2026 than they earned from 2000 to 2025 combined. So we're in quite a time to be doing this episode. Let's go through a few case studies. Brett, you've got two. I've got one. Who are we starting with?

We're starting with ASML, and I will mention, we'll maybe discuss the greediness of the memory chip companies at the conclusion of this podcast. But yeah, let's keep things moving, get ourselves on schedule here. It's one of the companies, ASML, that I mentioned in the introduction. They were originally a backwater of the semiconductor equipment field, but they've slowly innovated their way out and gotten to the top of the food chain market cap-wise of the subsector with a current market cap of $675 billion.

And even if you looked at the stock after its development and full commercialization of the EUV lithography machines that were the sole reason that 5 nanometer and below transistors are possible, or 5 nanometer gaps, if I'm using the wrong nomenclature, don't sue me, electrical engineers. Even so, over the last decade, you could still have seen a 34% total return CAGR, which leads me to believe maybe the lesson is that it's always early in semiconductors. This could be the exact wrong time to say that, especially with ASML stock trading where it is. There's actually a book called Focus, the ASML way that's on my reading list. I think people, anyone interested in this company or the semi-cap space should probably look at that. But if we go

to their journey, they began their journey in 1985. They were spinoff from Phillips to research and design the emerging lithography systems used to print semiconductors designs. Think of a lithography like a light projector that puts the designs onto the actual semiconducting material. Now, the first 10 years were extremely shaky for the business, which isn't uncommon with new technologies and required a lot of help from Philips funding, that type of things. But they developed a system called the PAS-5500 and were able to go fully independent as a publicly traded company in 1995. By the 2000s, they were making inroads with new customers using its dual immersion machines that help with manufacturing efficiency but the key part of

the story is in 2010 when the first extreme ultraviolet lithography machine prototype was shipped to a customer proving that the technology which used again it's in the name extreme ultraviolet lithography light to print tiny uh basically designs onto these semiconductors and And around the same time, the largest players in the space like TSMC, Intel, and Samsung actually invested in ASML to secure the future of EUV technology. It was unclear at this time whether EUV would even work, but ASML made the gamble and it paid off handsomely. The investments from the supply chain were perhaps a good hint at the time.

Maybe if we're looking back and trying to do a case study on this, like if we're sitting in the 2010 to 2015 period and trying to analyze whether ASML is a buy or not before they get this nice growth runway from EUV machines. If you said, hey, Intel, Samsung, TSMC, the leaders in the industry, the leading customers of ASML and the semiconductor equipment companies, if they're investing in this company, maybe it's a good sign that we have more of a margin of safety than the stock price, the financials, the balance sheet would suggest. And it is these EUV machines that allowed the advanced semiconductor designs to be printed, as I mentioned. And it's why ASML is able to charge hundreds of millions of dollars for each machine sold to customers because no one

else has been able to copy them. And not even not even China, I guess, which had kind of a national priority to do so. And you also get the services revenue on top every year after I have a screenshot here from our friends at fiscal AI, but I made it too small. I believe if I'm going to remember from 2012 through to today, maybe Ryan's able to read on his computer screen, I think the services revenue which is essentially helping the manufacturers operate and service the machines has grown at 18 percent tagger since 2012 that's more of a recurring revenue where you sell the machine once 400 million dollars but you get this recurring stream over the next decade plus of hey we have someone coming to your facilities helping you with these things making sure they

all work perfectly um and asml you know they were successful and they've secured its position as the only supplier of EUV machines because, well, the United States, the West, sorry, they secured themselves as the only supplier of EUV machines in the world and basically to the West because they had smart people and they've been on a revolutionary technology.

However, there are other reasons ASML maintains its dominance in EUV, which recurring guest Leandro from Best Acres Stocks discussed in his newsletter, and I agree with all these takes. One, there are exclusive supply chain partners for asml now this is downstream from them i think no upstream if i'm using the analogy correctly only for example asml only has access to some of the advanced subcomponents that make up an

euv machine so if you wanted to copy them you not only have to copy the design that asml has but you have to copy and get the same sort of mirrors or light sources or lasers that asml was either acquired or had exclusive agreements on. Second, there's switching costs from the services revenue from operating the machines. You have a whole factory set up to operate with EUV machines as one of the most important subcomponents of the assembly line.

And if you wanted to pull that and put something new in, that might disrupt the whole thing. And these companies like TSMC, Intel, and what have you are investing tens of billions of dollars into each factory, pulling out EUV and replacing with something else would be quite difficult and you're not going to get disrupted overnight. And as well, they work hand in hand with their supply chain partners like TSMC. I think the success of both these companies over the last decade are really because of each other. Leandra also has a great quote about investing in the semiconductor space

from back in 2024. I think it's still relevant to today. Quote, investing in semiconductors is not about forecasting the next application that will require semiconductors. Instead, it is about understanding and feeling comfortable with the fact that whatever comes, more and more advanced semiconductor content will be needed in the future. How many people foresaw smartphones two decades ago? How many people foresaw cars with $2,000 in semiconductor content? I don't know the answer, but neither of these forecasts were needed to understand that technology and thus semis would play an increasingly important role in the future. I think that sums it up nicely what prevents asml ryan is there questions that probably we aren't qualified to ask but what

prevents asml from getting quote asmled in the future yeah it's a good question i think the current gap between them and i mean if it were to happen i think it would maybe have to come from new technology entirely not necessarily someone catching up uh on the euv front whether it's the exclusive agreements with suppliers or the uh just current technological advantage and process advantage i think that's an understated part of all these companies we're going to talk about it with taiwan semiconductor the actual processes that they've developed and iterated on for building and improving their equipment and manufacturing capacity and the i'm when i'm thinking of taiwan semiconductor i mean more buildings and the processes involved

there that's something that it's hard to do from scratch like you can't just catch up on a whim i think these i think asml taiwan semiconductor some of these companies that have built massive leads are going to be very very difficult to catch with their existing uh business lines maybe if there's some new technology that disrupts the uv that that's how they would get asmled but i don't i i don't know enough to tell you what it would be yeah i think maybe no one does except for some of the smartest engineers out there that are working on these type of things right now ASML trades at a PE of

58 it's probably closer to trim territory for long term shareholders than buy territory it's what I'd like to own in the future it's probably a solid bet to look out for if the AI boom turns to bust but I'd say I have the PE chart

here Ryan you can kind of look and I like doing this with the case studies where was the best risk reward in ASML's history. And I kind of think maybe the 2010 to 2015 range when it was trading at a fairly low price-to-earnings ratio, I think sub-20, especially on a forward basis. But there's also just over a year ago in the summer of 2025, when it was back down to a PE of 25, when it's about to have this

massive growth runway from the AI infrastructure build-out and returned since then. Well, the PE's gone from 20 uh 25 to 61 so okay what i'm saying is you had phenomenal returns to say 2021 2023 2024 you might think oh i missed the boat on asml there still could be opportunities out there the market can get shaky the market can get quite volatile don't think you you won't have ever have an opportunity to invest in these type of companies yeah i mean the cat was out of the bag with ai last summer and the stock traded down to 25 times earnings down from what was i think 50 times earnings a year prior two years prior so yes opportunities will present themselves but i would say um last summer we should have picked up on it i think you actually asked me like why

why shouldn't we own asml i remember you i think we talked about that on a podcast and i said it's it's too big for me it's not you know i felt like the cat was out of the bag the story's been told and it's i was completely wrong um and yeah i would guess that would have been the best time honestly to buy it because there was no risk to the business model at the time like there was nothing saying this business is going to be disrupted or anything like that and you knew there was another catalyst for growth for the semiconductor ecosystem broadly so i'd argue yeah that was probably one of the easiest times to buy let's talk about my first case study. Taiwan Semiconductor, known as TSMC, they were officially founded in 1987. There's a

lot of great history on the business that's been written about and spoken about in interviews. Most of it, I believe, is included in the book, The Chip War. So I recommend giving that a read, but I'll go through some brief history. The company was founded as a joint initiative between the Taiwanese government, Dutch electronics giant Philips, and a group of private investors from what i understand it was basically the taiwan government saying like to a bunch of the large taiwanese companies at the time like hey you need to sort of strong arming them you need to invest in this um but morris chang who is now he's a legendary figure in the history of the chip microchip evolution and i would argue probably

one of the most important figures in the history of semiconductors overall uh he was recruited for a while i think basically the taiwan government was trying to recruit him for 10 years uh to build out their entire semiconductor business taiwan made it like a leading initiative like we want to be a leader in semiconductors um and chang was kind of the one they honed in on chang I think he worked at Texas Instruments for about 25 years. I believe he won a Nobel Prize. He was widely considered one of the brightest people in the semiconductor space. He was passed over for the CEO job at Texas Instruments. And that is when I think that was late 70s, maybe 1980. that's when he decided to step away so he joined a company initially after stepping away called general instrument but in 1985 he finally accepted taiwan's recruiting efforts and the proposal for him was chang would virtually get a blank check to lead a government-backed research and development

organization known as itry industrial technology research institute and he could go sort of any direction he wanted with it uh chang apparently evaluated taiwan's strengths and weaknesses at the time and determined that they would not really be able to compete as an integrated device manufacturer he thought they didn't have they didn't have the knowledge and designs they didn't have the sales relationships with customers so instead he proposed the what was revolutionary concept at the time of being a pure play foundry business and it took tons and tons of capital to to become this but when you are partnered with the government that certainly helps and it gave them the capital they needed side note in the capital raising process chang was turned down

by both intel and texas instruments before receiving an investment from phillips which i find kind of interesting given the competition that we've seen from intel and taiwan semiconductor over the last couple of four decades but yeah big miss for both those companies here's a quote i found from a good write-up what chang foresaw was how each step of the chip development process would become almost exponentially more expensive and complex as the node shrank keeping up with each step under one roof soon enough became very challenging creating an explosion of fabulous companies meaning no in-house fabs well there's not no but there are far less in-house fabs while part of this part of the success is certainly just down to incentives like customers

if you're apple you don't want there's more incentive to work with a pure play foundry than someone who could also compete with you on the chip design front so intel if they've got the design firm and they've got the foundry not only are they going to prioritize their own chips but there's you know potential competition there too um anyways the other part that really let taiwan semiconductor take a step ahead of the rest of the competition was came sort of in 2022 with the bet on EUV. So in 2012, Intel was actually the largest investor in ASML at the time. And EUV technology, as Brett just mentioned, was sort of in its infancy. Intel's management team apparently grew very impatient waiting for the EUV technology to be commercial ready. So they attempted instead

to push their older duv machines to their absolute physical limits this gets into sort of the nitty-gritty and the technicalities with their manufacturing process so i'll do my best to summarize it but apparently this effort to push their duv machines to their limits backfired big time the process became incredibly complex prone to errors and their manufacturing yields plummeted This delayed their 10 nanometer node by roughly four years and completely stalled their roadmap. This 10 nanometer node for Intel would have been a 2.7 times increase to their density. So from the previous notes, so we talked about Moore's Law, how the goal is to shrink the size of a chip by half every two years. This was them kind of going for the home run and saying we can increase it 2.7 times. Taiwan Semiconductor, on the other hand, went a little less aggressive.

So they, I guess, had a more pragmatic goal of 2x-ing the density. here's a quote uh as to sort of what what the difference is so because duv light waves are too wide to print sub 10 nanometer features in a single pass manufacturers must use multi-patterning splitting a pattern across multiple masks and exposing the silicon multiple times tsmc decided to stick with double patterning uh instead of quad patterning quad patterning is apparently what intel went with to try to sort of create that monumental leap to the 10 nanometer node because it's the equivalent of intel was swinging for the fences while tsmc was trying to hit singles and doubles tsmc because they went with the double patterning it helped keep their

process simpler and it boosted yields significantly during that time then once they stabilized the seven nanometer, they introduced EUV incrementally to print just a few critical layers. This allowed TSMC's engineers to master the temperamental EUV technology slowly without risking their entire production line. I'm stealing some of that from a summary of it, but I think it makes sense. Basically, they were a little more pragmatic, a little more patient, layered in EUV technology slowly and they sort of mastered it so and that's i think an important thing to distinguish it's one thing to just have access to euv technology but how you integrate it is sort of the the difference in your manufacturing productivity relative to peers so intel and

taiwan semiconductor both have access to asml's machines but because taiwan semiconductor had a more pragmatic more patient approach they were able to integrate it in a way uh that was much more effective and gave them a major advantage which they've pressed since taiwan semiconductor today is an absolute choke point in the semiconductor supply chain industry estimates indicate that taiwan semiconductor manufactures over 90 percent of the world's leading edge sub seven nanometer advanced chips which is what modern ai compute basic primarily requires apparently i'm sure there's other chips as well but efficiency ryan i think that's what it comes down to yeah basically like if they had if an nvidia chip had to be 10 nanometer it would be

not nearly as good so the fact that they can get the most yeah it's just energy output computational output kind of that whole matrix yeah and it would when i think about the moat today it would be very difficult to replicate the tech advantage for the tech and process advantage that tsmc has built but it would also be difficult to replicate the production capacity advantage they are massive they've spent 220 billion dollars in capex over the last since 2021 most companies can't do that so they've got right i guess most companies maybe they could but they shouldn't do it um but a lot of them also just can't so they've it's been a huge advantage for them they've got facilities in japan taiwan they're building one in arizona as well i think

which is already operational, if I'm not mistaken. Maybe it's still in production or being developed, but it's a massive advantage that they continue to expand with investment. So you might be asking, okay, they've spent $221 billion on CapEx. Is that really an advantage? What if they're overbuilding? What if demand for the most advanced chips slows down and we're at sort of a temporary peak in customer demand for AI chips? That is a fair concern.

And you look at Taiwan Semiconductor over the last couple of years, revenue has completely accelerated. They were growing sort of 10%, 20%. I mean, this is the largest chip manufacturer in the world. It's hard for them to grow a whole lot faster without just drastically raising prices because they have to build out capacity and it takes time. And they grew 41% year over year last quarter. It's astounding growth rates for their size. Analysts think they will double revenue over the next three years. Now, my question would be, like, I think it's a logical concern to say maybe demand is overstated in the short run.

but it's the whole i can't remember who said it initially but like only the paranoid survive it seems like that's kind of cc ways uh mentality today that's intel intel invented that well that's so they stole it that yeah it's funny that they stole it from intel here's a quote from the fourth quarter conference call from 2025 uh from cc way i think it's fair this really kind of stood out to me. Someone asked him about AI demand. He said, you essentially are trying to ask me whether AI demand is real or not. I am also very nervous about it, you bet, because we have to invest about $52 to $56 billion for the CapEx, right? If we didn't do it carefully, that would be a big disaster for TSMC for sure. So of course, I spent a lot of time in the last three to four months

talking to my customers and then, and their customers, customers, because I want to make sure that my customer's demand is real. I talked to those cloud service providers, all of them. Their answer is I am quite satisfied with their answer. Actually, they show me the evidence that the AI really helps their business. So they grow their business successfully and he or she in their financial return. So I also double check their financial status. They are very rich. Now, part of this is kind of funny translation but uh it's just to say like he's asking the same questions i think he's probably as concerned as anybody about whether or not spending 55 billion dollars is going to be worth it just because you're concerned about something doesn't mean

it won't happen though true but don't you think they're booking these orders prior to the build out oh yeah it'll be fine through the long haul there's gonna be a down cycle there's got there has to be at some point uh the people said the same thing in 2022 and we can look at well the revenue looked okay wait now you have your yeah revenue dipped a little bit i mean they have a very strong position through the cycle the bigger concern for me if i're an investor i actually am I'm an investor. I think I own a very small stake, but it trades at 30 times EBIT, which is historically their most expensive – around their most expensive multiple. Margins are up. Yeah. Record profits, record multiple. It does feel like you're susceptible to a big drawdown here.

yeah i agree i agree but good business and it's really fascinating that the last 15 years now have been not entirely but a lot has been determined by choices from players throughout the space on what they thought of or did with the euv technology all right yeah let's move to my case study number two cadence design systems uh it's one of the two premier software providers for semiconductor manufacturing design really want to look at the history of the business since i've never looked at the software side of the semiconductor supply chain uh here is what the company says on its website because i think many listeners are thinking well what exactly do they provide it says quote cadence is a market leader in ai and digital twins pioneering the

application of computational software to accelerate innovation in the engineering design of silicon to systems. Our design solutions, based on Cadence's intelligent system design strategy, are essential for the world's leading semiconductor and systems companies to build their next generation products, from chips to full electrical mechanical systems. And the range of markets includes hyperscale computing, mobile communications, automotive, aerospace, industrial life sciences, and robotics. Interestingly, Cadence has underperformed some of the other players here since 1990. It has, quote, only delivered a 12.5% total return, which is still fantastic, but underperformed some of the other winners out there. And it has done so with a PE

that is currently sky high. We're at 50, or sorry, sorry, 85, not even 55, 85. Revenue has grown at a 7.5% annual rate since 2005, but 12% from the bottom of the GFC in 2009. Now, why are these software programs so valuable? One, long lead times. Getting a prototype of an advanced chip can take

a while, meaning if you can test, iterate, and validate virtually through Cadence's software, peak performance can be achieved much faster when the first real-world chip is made by TSMC. You have these virtual testings essentially, and

instead of going six months to the factory and say, make us this prototype, come back and see if the physical one works, they probably do hundreds and hundreds of iterations within the software program, make sure they get it as right as they think they can. Then they finally get a physical product to be made and test that before it goes to mass productions. The second thing is real physics simulation. Through decades of building up its expertise in the software program, Cadence has extremely realistic software programs for the chip world, similar to the Autodesk or the Dassault systems in engineering. And three, the reasons people use them is saving money. Even though they pay, I'm sure, who would be the largest customer here, NVIDIA? They probably pay

hundreds of millions to Cadence every year. The ROI from the save money, the speed, um just the way you can design much faster within these software programs is you just have a fantastic roi there and you save a ton of money you know you the same thing as building a bridge automobile engines skyscrapers you want to make sure it's as safe as operates as well as it properly can and follows proper regulations with semiconductors you also want to make sure that ip is being used properly like with arm and things like that now it's much more complicated than but one could simply say that Cadence allows NVIDIA, Apple, other players to build the chips in the virtual world before sending them to the factory. It is the first step in the semiconductor

supply chain, which gets slingshotted down to the other foundries before a smartphone is assembled and put in your pocket. Competition mainly comes from Synopsys. It's pretty much a duopoly here, although there's some other players, and Synopsys does $8.67 billion in revenue versus $5.5 billion for Cadence, but Synopsys just bought Ansys, which is a real physics simulation company that does have a large amount of non-semiconductor revenue. I think an investor could probably simply say that the world is moving towards more virtual validation here. It's going to always get more software-like in the design and testing, and it's going to make Cadence, if they maintain their R&D and just maintain their lead as the leadership position here, that's going to be more valuable to

designers every year as well as the diversification of design efforts from new entrants such as hyperscalers so you have more people designing chips okay that's more cadence subscriptions most likely and this should give the company an expanding addressable market secular growth pricing power i feel like it makes it a nice candidate for durable growth but the last interesting part is that the company and the cadence story despite its fantastic business model, they almost went bankrupt. This is a lesson in partnering with a good management team and a good company at the same time. In the 2000s, cadence shifted to a subscription and rateable license. At the same time, the telecom and internet bust led to a down cycle in the

industry. And in 2008, management made a hostile takeover of a competitor called Mentor Graphics right at the time of the GFC. And then some accounting issues got piled on and the executive team abruptly left. The stock fell to $2 a share in 2009 when Lip Bhutan, who is currently the Intel CEO, came from the board of directors to lead the business. He focused on what customers wanted, got cost discipline, went back to organic R&D spending instead of M&A to simplify things, and the stock price began to recover. Really a tale as old as time. We've seen that plenty of times before. The stock price is now not $2 or $5 a share. It is $378. What are the disruptions risk for cadence i'd say vertical integration maybe i think a company like nvidia you might

argue could they design their own software would it be worth it there are apparently some internal design stuff at apple and google and some other places but so far there's been no signs of major disruption here there's the synopsis acquisition of ansys is combining to create a better product uh potentially because you have the complications with things like robotics evs even ai dandy center solutions where you need a comprehensive simulation for fluid dynamics heat as well as the electrical engineering in the actual computer chip cadence is building this internally but synopsis combined with ansys could be a much better value proposition now last quarter and look at the numbers here cadence your revenue 24 which is higher than the long-term

average due to you guessed it opera or artificial intelligence and operating margin is at 28 percent over the last 12 months could they get higher i think probably that they i would guess given the fact that they only generate 5.5 billion dollars in revenue compared to the rest of the industry into what as the equipment companies in the tens of billions nvidia's of course in the hundreds of billions, TSMC in the hundreds of billions, Intel might get back there someday. For the value they're providing, I feel like they can add some really strong pricing power here. They maybe won't take it over three years. Let's hope they don't pull FICO or else everyone will get mad or a micron. But that probably makes the business model very attractive and you still have the

secular tailwinds. I would guess the $1.56 billion in the last 12 months operating earnings could And over the next decade or so, $5 billion doesn't feel out of the question. But the problem is the market cap today is $100 billion. So hard to make the math work given the fact that they're not an explosive hyper growth company. Yeah. To me, I just think this is like an Autodesk or a Dassault Systems for semiconductors. It feels like it would be mission critical. And you think about $5 billion in a land of trillions in semiconductor revenue, it does feel like a small take, assuming that they are powering a lot of the designs. so yeah i would say there's probably room for them to expand it the let's talk our three favorite companies here no this is not our three favorite stocks at the moment because all the companies we just talked

about are trading at near their highest multiples of the last decade so again not stock recommendations But what are your three favorite businesses in the semiconductor sector that you would add to your watch list and be interested in buying in the next downturn? I'm going to go number one, ASML. I think they have the best position. I've read them things about lithography. What is it called? Like penetration within the company. So using it less as a percentage of the manufacturing process versus other equipment makers.

I don't know what to make of that. but it seems like i would think it's still mission critical second one will be tsmc and third one which we didn't cover today applied materials i don't necessarily like the design firms because i think there's much more susceptibility to disruption intel which also manufactured got disrupted by nvidia nvidia get disrupted by tpus there's just so much competition in that area where if you look at lithography there's no one if you look at advanced nodes there's no one that's why and then applied materials niche i think they're exclusive in a lot of things as well with those three you have no competition what's not to like i would say

asml and tsmc are on my list as well taiwan semiconductor would probably squeak above ASML for me purely just on understanding of the business like I maybe not the business but understanding of the direction

I feel like there's less technology disruption risk to Taiwan Semiconductor than there is to ASML again I don't know EUV space and any electrical engineers might be laughing at that but But on the half chance that EUV gets used last or there's some other process that is used in any further advanced nodes, Taiwan Semiconductor is still a beneficiary there.

The third one for me, just because I don't know applied materials as well or some of the equipment providers, I like the software companies, the EDA software businesses, Cadence, Synopsys. I think they're going to show much less cyclicality in terms of revenue. over a full cycle but again both are very expensive at the moment so yeah uh three best businesses for me taiwan semiconductor asml and then i'll go cadence design okay we have a couple of listener questions i want to have there was a lot so we're not gonna get all of them guys uh but i appreciate everyone on the sub stack chat doing that uh let's see first one what is the real mountain semiconductors i think we talked about that already uh let's how about this one how should investors underwrite nvidia if ai demand remains strong but margins eventually normalize

would that's would this still justify the valuation what are we trading at what's the nvidia pe let's look it up quickly 32 i i just think there's a risk that their revenue is cut in half right the they are yeah they're highly susceptible in some sort of a downturn the other part here is the the circular agreements where they fund customers or they invest in customers and then the customers purchase chips feels like bad vendor financing it just gives me kind of an icky feeling that they're having to prop up revenue a little bit.

So look, if this AI bull market, AI demand growth continues for the next five years, NVIDIA will be earning

probably four times more than they are today. But it's not it's not necessarily locked in like like it's almost the more they earn the more incentive for their largest customers to invest in custom solutions same with memory maybe not custom solutions but go into other providers all right another one what part of the semiconductor value chain has the best five to ten year risk adjusted return fab fabulous design memory equipment eda software advanced packaging i don't know but i think i

have a i have a bet on what the worst area will be and that is memory i think they're being extremely greedy tsmc could do the same thing but they're not because they want to preserve a long-term customer relation asml could do the same thing they could charge a billion per machine but they're not and it's because they don't want to incentivize people to end their relationship yeah the it's just harder when you're in a more competitive industry because if your

biggest competitor is doing it then why won't you why won't you capitalize on it it just tells you the stocks are going down 90% at some point yeah I would be curious what the barriers to entry are for building memory chips

like compare that to a GPU how much harder is a gpu design to build like our micron samsung and sk hynix are they in some irreplaceable territory like can no one build what they've built if it takes three years of investment yeah this is going to be a maybe not them even if they earn outrageous profits over the next couple years they're going to have challenging times ahead okay a couple more quick uh for anyone on the watch list company to look at nvmi the company is called nova it is a semi-cap equipment company that helps fabs get the most yield from each silicon wafer sort of like kalak only 13 billion

dollar market cap could be another small cap of the week we look at could be another company we look at we'll put that on the watch list uh someone with people talking about geopolitical risk i'd say quickly if tsmc goes down everyone goes down with them so i don't know why tsmc gets a it doesn't anymore i used to have that uh multiple like discount uh someone was said they're astounded at the rate tsmc has been able to ramp up production over the last two years well there was a down cycle coming out of 2022 2023 but at the same time they were diversifying that supply chain by investing in the West, such as the giant Phoenix area,

$100 billion plus manufacturing facility. So the question he has there is the way to track what is TSMC capacity now and under construction? Just keep it simple and look at CapEx. And then there's some other one here from people. I'm sorry, there was a very long question about Qualcomm. We are not qualified to answer. So I think we can close things out here.

Ryan, any closing thoughts on the semiconductor space? Thank you, everyone, for the list of questions. this is it's it's fun to do this because it's like a marvel of human ingenuity and creativity and what we're capable of and studying that is it's just exciting to see how far we've come but i'm yet to find anything in here that's going to be in my portfolio at the moment everything feels a little rich and i'm not there's definitely elements of the semiconductor industry where i don't truly understand all the competitive dynamics especially on the design side yeah design i don't like as well uh i'm gonna say it right now i'm gonna put to the audience uh i'm making a pact with the audience in the next down cycle i'm going to invest

in some of the high quality companies um i guess i'll probably write about it on the emerging notes newsletter but that is my pact with ryan and the listeners i'll hold you to it but maybe what if the industry isn't cyclical anymore like people are saying well well then i guess my loss starting to starting to sound like an indicator there brett exactly i hope i hope i jinx it i hope i jinx it all right i think that's gonna do it you want to take us out or you want me uh i can do it as a disclosure we are not financial advisors anything we say on the show is not formal advice or recommendation. Ryan, I, or any podcast guests may hold securities discussed in this podcast, may have held them in the past, and may buy, sell, or hold them

in the future. Thank you, everyone, for tuning in. I hope you learned a lot from this episode, and we'll see you next time.