Doug from Fabricated Knowledge on the Semis industry and his $RMBS thesis (Podcast #112)

Doug O’Laughlin, founder of fabricated knowledge, comes on the podcast to talk about the semiconductor space in general and then dive into his thesis on Rambus (RMBS) and why the market might be missing their big growth call option.

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

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

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

Transcript begins below

Andrew Walker: Hello, and welcome to the Yet Another Value Podcast. I'm your host, Andrew Walker. If you like this podcast, it would mean a lot if you could rate, subscribe, review it wherever you're listening to it.

With me today, I'm happy to have Doug O'Laughlin. Doug is the founder of Fabricated Knowledge, a semi-conductor dedicated. I don't know if it's a Substack, a newsletter. I don't know what you call it, but Doug, how's it going?

Douglas O'Laughlin: Yeah. A research service, but yes, it is mostly through a Substack. Yes, Fabricated Knowledge is a semiconductor newsletter written with investment professionals in mind. It's a very hard place to understand. I just try to explain why it matters and you don't have to go into the technically detail I do, and then I explain in slightly better English. But it's still pretty, pretty gnarly.

Andrew: I was prepping for this podcast and I was like, "Oh, God. This is why generalists struggle in semiconductors," because it's tough. But let me start this podcast the way I do every podcast. First, disclosure to remind everyone that nothing on this podcast is investing advice. Please do your own work, consult a financial advisor, all that type of stuff. Second, the second way I started every podcast is with a pitch for you, my guest. I was telling you before this podcast started, but I am so excited to have you on because I have the best pitch I will ever have for this podcast.

A few weeks ago, I was having dinner with a friend and he's an analyst, super sharp guy. He's an analyst. He's been at four multibillion-dollar funds. He's covered semis, he's covered internet, and all this. Him and I were just talking, and he was like, "Look, I think a lot of pitch wit[?], they're flat out. They're flat-out frauds, hucksters. A lot of these guys are people who are running a PA account. They've blown up multiple times and they have no idea what they're talking about." And then he just stopped, and he said, "You know who I know is not a fraud? That mule guy who covers semiconductors. I've covered semiconductors. That guy is unbelievable. Every single buy-side shop in the world, if he was looking to get hired, would drop on a dime to hire this guy. He is fudging unbelievable at semiconductors."

When he said that, you and I, 24 hours ago, had set up this podcast and he was saying this and I was just like, "Oh my God, this is going to be the best pitch of all time." So that's my pitch. My friend's a super sharp guy. I agree with everything he said, so sharp on semiconductors. I don't know if you want to critique that or anything, but that's my pitch for me.

Doug: I'll take every compliment I can get. Thank you very much. I try pretty hard to cover this space. It's pretty hard. I would definitely say it's a little bit of a passion because it's brutal, man. I was reading something today and I was like, "Man, this isn't even English." It really is like a whole another language. But I really think one of the reasons why it's just so compelling to me is its everyday magic that is part of all of our lives every single day and people just don't pay it any due where all the software stuff has to be built on something and it starts with the hardware. I think that it's magic. The actual core semiconductor process is magic and I think everyone should learn a little bit more and be excited about how this magic runs our entire lives. We tricked sand into thinking, it's my favorite meme like we just tricked a rock into thinking, and then now we have it think for us all day. It's amazing. It's magical.

Andrew: I kind of think there's that famous line, "Software is eating the world" or whatever, but the software needs to be powered by something. As you said, hardware is power in it. We're going to talk semiconductors cycle and everything, but there's an automobile shortage right now, which was driving all these used car prices everything and everything got crazy. Correct me if I'm wrong, the reason that there was an automobile shortage was because there's so many semiconductors and autos and they couldn't get semis. So like semis were driving an automobile shortage, which was driving inflation and all these knock-on effects from semis.

Doug: Yeah. Actually, there's a multipart to the auto shortage, but I think the first and foremost is 2020 happens, right? Autos cut their orders. The first thing happens, "Hey, recession, no one's going to buy a car." They cut all their orders and they essentially become, go from the front of the queue to the last of the queue to getting a semiconductor. Meanwhile, all the cloud companies are ordering more and more and more because we're all on Zoom day by day. But then, as it goes on, they're like, "Wait, we actually can't get these chips." Almost every single one of our roadmaps have meaningful amounts of more chips, whether that's EV which means the power and all the semiconductors related to power management and higher voltages and all the complexity of that, and then also the Aida side of things.

Probably for the next decade, there is a new roadmap of new safety features and every single new safety feature has more semiconductors. There's about a quadrupling of semiconductor content and most the auto OEMs did not think about semiconductors as real. They thought of it as a pure commodity where it's like, "Hey, we put in the order, we get it." They didn't realize if we all put in orders, we all four X are content or whatever. We're just not going to have the supply availability, especially for the types of semiconductors that are ordering, which is more trailing edge, trailing edge, meaning older chips that have been around for a longer time that are not quite extremely small five, three-nanometer chips that are being built soon or tomorrow, but these older, even larger like 96-nanometer chips. They are 28 is actually probably the most popular node. They're very mature.

And so, but no one ever thought about these chips. Everyone thought that "Hey, well, we're just going to have these handy down[?] fabs." We always have capacity at the old chips and everyone ordered more old chips at the same time and they're like, "Wait, we can't just make chips don't just come out of the ground." So that was the really interesting dynamic where these old chips that were thought of as extremely commodities became a lot more important all of a sudden, all at once. So that was the two big prongs for the automobile semi shortage.

Andrew: That's perfect. Look, this podcast is going to be a little different because I want to start you're a semiconductor expert, as I said. I want to start with a general semiconductor, just overview of what's going on in the world. And then we do have a pitch that we'll do at, then we'll turn to a specific stock, but let's just keep going on the semiconductor. I'll just toss it over to you. Semiconductor is a very broad space, right?

Doug: Yeah. There's a lot of space.

Andrew: You've got hundreds of different companies, hundreds of different edges, but let's just turn it over to you. What are you seeing in the semiconductor world? What are you thinking about today? I know you had a great article on the bullwhip effect, which we certainly started talking about cars [crosstalk]

Doug: Yeah. That's even harder to talk about it. At the high level, I think it's easiest to first think about the end markets. There's really only a few big end markets. The biggest tune in the entire world is PC and phones. And then the third biggest that's starting to come up and become much more meaningful part of the pi is data centers. So data centers, the growth segment. PC is the segment that has never grown until COVID and ever since COVID, it's actually been growing. Now, it looks like it's going to shrink this year, obviously, because it's lapping of a one-time effect. Last but not least, we have phones. Phones have become way more complex over the years, way more content as we've 4g 5g. Each additional part or each additional phone has more packages, more parts of the package, and more complexity.

So that's been a huge driver on the content part, but phones are pretty topped out in the penetration curve. Everyone essentially owns a phone. So that's not exactly a growth market in terms of volume. It's a growth market in terms of value because there's higher content. PC is not a growth market in terms of units anymore because PCs have essentially been broadly penetrated ever so slightly may be in terms of value. Now, the real market that everyone is really focused on going forward in terms of volume and value is the data center. So data center is where I'm focused and most people are focused on the growth side of things.

Andrew: Can I ask you just a dumb question? You mentioned the big ones, but I do look across the world and it seems like semiconductors are everywhere, right? 30 years ago, I don't think people would've been really thinking cars for semiconductors. Now, as you mentioned, cars, smart fridges, the Amazon, Alexa, all this sort of stuff. I know there's a lot of them, but are those so small when it compares to just these giant data centers consuming hundreds and hundreds of thousands of units that they don't really move the needle or do those come into play at all?

Doug: Actually, those are the big three and then the next two big markets is automotive obviously is this huge incremental growing market with all these new units. Last but not least is IoT, IoT will probably be the fastest percentage growth going forward for the next 10 years. There's just a really subtle penetration. It's super hard to nail. You can't be like, "Hey, this is the device that's growing IoT usage." A perfect example actually was the SEC micro investor day. They talked about how this drill went from 132-bit MCU, which is just a microcontroller unit to three. That's a tripling of content, right? They're like, "Okay. It used to be just one to control the drill, essentially the motor."

Now, we have one to control the motor, one to control the power management so that the battery is more efficient since it's electric and one to control the connectivity. So it's connected to Bluetooth or WiFi or whatever. And so, that's a perfect example in my mind of the linear step of how content increases and almost exponential, especially as we connect everything to WiFi or Bluetooth. It's just entering every aspect of our lives. It's very hard to pin down the one thing that is causing the IoT thing because on the other side, you have this industrial thing, all these factories are starting to invest meaningful amounts in computer vision and even in stuff in very old world stuff like the Walmart for the world. Walmart has a GPU inside of it, running a machine-learning algorithm to see where the consumers are walking around or doing a heat map. That's very sophisticated levels of compute in something that was pretty dumb very recently.

It's just essentially wherever software is eating the world, there's a little bit of hardware that's being attached to it. So everywhere that we're starting to add smart capability, there's going to have to be a semiconductor associated with that. So just as our world becomes more digital, it just creeps in by every single way. It's everywhere. Essentially, if there's electricity in it, odds are there's probably a semiconductor in it too.

Andrew: Again, I'm a generalist. I think my history with semiconductors is generally get their face for talking investments in semiconductors, right? Because semiconductors, even though it is a secular growth market, obviously, it is still quite cyclical and we can talk bullwhip and everything later. But I think generalists tend to get really built up on semiconductors at the heights of the cycle where they say, "This is going to grow forever. These multiples look attractive," and it's famous. In a cyclical, you actually want to sell when the multiple is low and you want to buy when multiples high because when the multiple heights because earnings are depressed. Let me ask you, what do you think it is that when you're talking to generalists about semiconductors, what do you think it is where they most frequently talk to you something and they're like, "Oh, they don't understand this thing," or "Oh, they're missing something here."

Doug: One of the big ones for me is the quality of memory. The cash intensity of some of these businesses is very intense. I wrote up a Microns Investor Day and I really advise you to take a look because I constantly hear, "Hey, it's a really low price to earnings." That's the typical general strap. But if you look at a free cash flow conversion, they do not make that kind of free cash flow conversion. Earnings to free cash flow is about 50%. So that's one of the places that I really highly recommend you do a really solid look at is when does it become cash? Because that's a really important and hard part of the business. The other part is, obviously, the volatility. It's hard. Cycles are hard.

Actually, investors are not dumb. Every cycle, I would say, investors, have become smarter. The market tends to bottom one quarter or bottom to one quarter earlier than it did last time. The markets have been more forward-looking. They've been more intelligent. And so, each cycle actually doesn't get easier, it gets a little harder. You have to be a little smarter. But I do think that if you can handle some volatility, it's a pretty interesting sub-sector. I want to say both the 2000s and the 2010s is one of the best performing subindustries within the NASDAQ. It outperforms the NASDAQ over a long period of time. Obviously, there are some intense drawdowns in between, but part of the reason why is because, especially if you're a successful semiconductor, Semi cap company, you make a lot of cash, man. Those multiples get low because of the fear of the stu-quality[?].

Sometimes, a lot of your fears are usually bigger in your head than reality. A lot of times the companies get to buy back a lot of shares. The actual free cash flow per share growth over a longer period of time can be very impressive, but there could be some real troughs in between. It's very rare that you have an entire industry that grows mid-single digits, but the actual end companies have been growing gross margins, EBIT at double-digit rates. It's a pretty impressive and business that continues to still benefit from meaningful amounts of operational leverage. So as they become larger, they become more profitable, et cetera.

Andrew: I was just laughing when you said Micron as your first example because I remember back in the early 2010s when Mohnish Pabrai and David Einhorn had big positions in Micron. I think at that time, that might have been trading around book value around approaching networking capital. I can't remember. People can go pull up the stock. This is the ultimate example of a volatile stock. 2012, it's at six, 2014, it's approaching 40, 2016, it's around 10 height. Heights of the COVID boom. It's approaching a hundred and here we are today at 70. Again, I know semiconductors are a very broad industry, but where do you think we are in the cycle? Obviously, I don't think we're in early 2021 like COVID. Every slot takes everything. We're not in that boom anymore though. I think a lot of [inaudible] results, but just in the market, where do you think we are in the overall semiconductor cycle right now?

Doug: It's pretty complicated and hard to know. If you listen to the companies talk, they're like, "Hey, 2023, 2024, we're starting to be booked out by then." So to them, this shortage continues, and it kind of draws out the demand. They never really were able to match the supply needed to clear demand. And so, it's just been pulling out the cycle longer than possible. I would say most public company CEOs and management teams have been saying, "Hey, we're still going to be growing next year. The cycle is not over." The problem is some of the recent quakes in the consumer over-ordering of inventory really bodes terribly for semiconductors, right? This whip in COVID has put goods at the forefront of purchasing for the first time in a long time.
Goods include obviously stuff like phones, stuff like your smart meters, all the stuff for your house. All these miscellaneous consumer-related products have some kind of semiconductor in it. These goods are the largest percent of the pi they've been in a long time.

Now, we're starting to have the services snapback and you're starting to see companies have over-ordered inventory against this. This is the biggest fear in my opinion because now, have over-ordering at the end. Meanwhile, all the suppliers all the way up to it have been over or they're like, "Okay, well we can't meet demand. We can't meet demand. So we have to over-order our semiconductor supplies." Right now, semiconductor companies are like, "Well, we really hate you over ordering." So what we're going to make you do is you're going to have to order over the longer periods of time. It's effectively double ordering, but through lengthening the time that the orders are through instead of the intensity in a very short amount of time.

It's just been this really confusing cycle. So far, it has not cracked, but a lot of second derivative numbers are starting to crack. Now, we're starting to look at some of the numbers lower because of Shanghai.TXM was one of the first companies to move down their guidance, which is the beginning of the lowering of numbers. When the forward numbers start to lower, that's when the stocks usually start to bottom, actually, because the stocks actually reflect the cycle before it happens, then the numbers start to lower. And then the stock's bottom, actually, when the numbers stop being lowered. I don't know, man. Some of the companies are still putting out beats and raises. Some of the companies are starting to lower their earnings. It's one of the weirdest cycles ever because you could look at automotive.

I feel very confident in saying automotive will probably grow this entire time just because the content and unit problem and the fact that we have all these cars that there's just so much pent-up demand and used car prices are so high, but then you look at stuff like consumer, PC. PC is definitely not going to grow. PC is going to be a market that's going to implode, maybe not implode, that's going to be a strong word, but it's funny. The beginning of this year, everyone was like, "Well, PC might be flat." And now, everyone's saying, "PC might be down a high single-digit." PC grew in Q1. So that implies that by the end of the year, PC is going to be contracting by meaningful amounts in the exit rate in Q4. So that to me is extremely worrying, but then the bowls can say, "Hey, look at data center, data center continues to accelerate. It's growing. It's adding more revenue in absolute terms."

It's just really complicated because in the past, there really was only two-cycle. There was only really one cycle, the PC cycle. And then there was two cycles the PC and phone cycle. Now, there's the PC phone, data center, automotive, and IoT cycle. They might be able to smooth each other out, but it's starting to look like the shift from good services to goods and back to services will definitely impact semiconductors. In a long-winded way, we are probably starting to see the beginnings of a cycle, but I don't really like the downturn, but I don't have any strong evidence in saying like, "Hey, this is it. Pricing is eroding," because pricing is not eroding. They're still raising prices against all this, which is something that is different than past cycles. Also, another example of something that's different than past cycles, cycles can be expressed through capacity. There was a really interesting paper I could read, maybe I'll send it over to you for show notes, but it talks about how historically the way you grade a cycle is the capacity utilization at fabs and essentially it's never been higher, which implies it could get worse pretty quickly.

Usually, what happens is it's never been higher and then it starts to go off a high level and then the cycle turns, but we keep adding supply every single quarter and they're like utilization has never been higher. Add more supply. Utilization has never been higher. Every semi cap company, which is the company that sells the tools into these fabs continues to beat that same drum that even though they're selling more tools, utilization has never been higher. So it's, it's confusing. There's more crosswinds. It's insane.

Andrew: On the capacity thing, that's something I wanted to ask because you can have one thing where inventories get too high, right? Your customers over order, so they need to draw down their inventories for six months. It's not just semiconductors. Right now, the retailers. Last year, they couldn't give enough inventory. They ordered as much as they could. And guess what? They ordered a lot of the COVID products like the at-home stuff that people were demanding. And then today, you see target. Target's like, "Oh, shoot, we've got the fitness of equipment for people working at home," and people aren't buying that, people really want luggage. We're short on that and they get stuffed, but eventually, that will normalize, right?

So there could be a short-term inventory issue where your automaker's over-ordered, but the bigger worry I think would be the oil thing where prices are really high, everyone drill, drill, drills to hit these high prices. And then 12 months from now, everyone drilled and I say oil, not right now. I'm thinking more like in the 14th, 15 range. Everybody drilled so much. Oil prices go from a hundred to 30 because all this supply comes online. There was 90 million of demand and 90 million of supply to now there's a hundred million of supply and 90 million demand and prices go through. So my question for semi is obviously there could be some inventory issues in the short term, but the medium to longer-term issue would be, do you see any signs of them overbuilding capacity?

Doug: That's definitely the top-of-mind question because you look at semi cap and you look at the way for spending equipment. It's been the longest bull run essentially in the history of WFE spending many, many positive years. II couldn't find this much positive years on absolute like a three-year stack, unless if I looked back into 1980s when semiconductors were essentially invented as an industry. It's been a long time since you've seen the percentage numbers like that. But there are some things to push back against that thesis that I think are really interesting. I hate to say, this time is different, right? You hate those words.

Andrew: Do a shot! Take a shot! This time is perfect.

Doug: Chug, chugga, whatever you're drinking. The thing that is different this time is the economic side of Moore's law is over. One of the reasons that really got me interested in semiconductors and starting and writing about this whole space is that Moore's law was this sacred law that worked for a very long time, 30, 40 years. There's a joke that Moore's law doubles every year, but the haters of Moore's law double every year.

Andrew: Do you mind to define what Moore's law is? I'm sure most people know it, but just in case anyone else.

Doug: In a high level, essentially, you could double the capacitor or the effective speed every two years for half the price. And so, you do that for a very long time and things become a lot cheaper, quicker, and faster very quickly. That relationship, I think it was a double, so it's like 40% improvement or something each year. They did that for 30, 40 years. It's astounding.

Andrew: If you were buying computers in the 90s and 200s, you'd buy a computer in 1995 and they say, "Hey, I'm going to make the numbers of 128 megahertz processor," and then you'd buy one, two, or three earlier and they'd say, "We have a 256 megahertz processor," and it's actually a little bit cheaper to buy[?] now. That's exactly what Moore's law is right there.

Doug: Yeah. It's amazing. It's been a meaningful part of what's driven society. Our availability of information technology is very much driven on this. What's interesting is this relationship actually quietly broke down in 2012. Essentially, every year, it was cheaper to make a faster computer. That's amazing, but actually in 2012, at the end of 28 nanometers, it actually became not cheaper too. It essentially stayed the same. And then quietly, in around 2000, I want to say 2018, it actually became more expensive on a per transistor basis to make a faster computer. We actually had a u-curve where it's been going down for a super long time, we hit 2012, it stayed flat, and then now, actually, it's starting to increase again. So that's something that is very different and that is something that's not conjecture or, "Oh, well this happened, that has happened." The ability to double transistors or the effective doubling of transistors that will probably continue, but to be able to double at a cheaper price, will not continue.

Andrew: You can Google Moore's law and the second thing that pops up is Moore's law is dead, but why didn't happen? Is it we just ran into the limits of Physics, we literally couldn't fit any transistors into smaller spaces, or was there something else?

Doug: It's very philosophical, but at a high level, it became harder and harder to shrink. Once upon a time, it's called "Plane Are Shrinking." Everything shrink in 2D, in two dimensions. So literally, it was like making a smaller rectangle. Well, eventually, that just stopped working because of whatever reason, being able to get the electrical charge to really register. So they actually added what's called they added a third dimension for the first time. So that added complexity and that complexity added cost and then shrinking in the three dimensions that was nowhere the relationship shrinking in the two dimensions.

We've been shrinking in two dimensions from 1970 till now until 2012. Literally, that's when the cost relationship started to break down. You add in the third dimension and then boom, now it does cost as much. Now, we're starting to add, it's called a "FinFET," instead of this upright gate. Now, we're starting to add what's called "RibbonFET." So it's really complicated. The shapes are now instead of just two dimensions, it's like three. There's like all kinds of stuff, all kinds of materials. The precision has to be higher and everything on the increment has added cost. And so, that's where the relationship really broke down.

Andrew: Doug, billion-dollar ideas. Look, they max out on 2D, they're starting to max out on 3D, you and I, we're going to go start a semiconductor, we're going to improve on 40. We're just going to go and improve on 40.

Doug: Man, it's truly a billion-dollar idea, but unfortunately, for us, there's way smarter people that already working on it. The amazing thing is these people have been thinking, the people who work in the industry, it's magic. They've been working and thinking about this forever. What's actually interesting about the industry is that Moore's law is dead. That was very predictable and well known by law for a long time. Essentially, a lot of the problems we have going forward are well known and understandable. The solutions are kind of debated years ahead, and then eventually, they pick a path forward. I think that's actually one of the most interesting parts of the semiconductor industry that is very interesting to invest in if that makes sense.

It's very rare. You could debate the nature of cloud software, but imagine if everyone has a general roadmap and you're like, "Yeah, this is what's going to be like," there might be little kinks to iron out, but the future is pretty well known in the intermediate for the semiconductor industry. That's really interesting from an investor perspective because you can go logically, be like, "Hey, this company should benefit under this regime in the future." You know that because the entire industry has decided, "Hey, we're going to scale forward in advanced packaging," for example. Advanced packaging is the solution that we've decided that's really come to the forefront to continue to scale semiconductors at a faster and bigger, but obviously, not cheaper, but this is how we're overcoming the limits of Physics. And so, you could just be like, "Okay. Well, it's time to invest in advanced packaging." And then there's a lot of ways to express that. That's what's really interesting about it. It's a very logical well thought and long time in advance industry if that makes sense,

Andrew: It does make tons of sense. It sounds to me like, look, I think this is why pod shops love it. I think this is why as my friends said, every buy-side shop would love to have you because when you've got trends like that and stuff, it's the perfect place for long-short funds, right? You can go buy the winners of this and you can go short the structural losers of this and that can create a heck of a lot of about, but, no. Not without volatility. Not without a lot of [inaudible] [crosstalk] what we're doing.

Doug: Yeah. I was going to say, sometimes, you're like, "Man, this would totally work" But some of the names I really hated in 2021 just had these ridiculous rip verse[?]. You're like, "Am I wrong? Am I wrong? What is going on here?" If you held through, it would've very much worked out, but it's hard. It's hard to do the full cycle.

Andrew: I'll refer everyone to the disclaimer at the front of the show, nothing on this podcast is investing advice, obviously. There's stuff right now, Red box, I don't know if you're following this. So Redbox, RDBX is the ticker. They're in a merger with chicken soup for the soul.

Doug: Yes. Actually, I did follow that. It's very funny.

Andrew: Yeah. So for people who don't know, this is a company Redbox. You could go to CVS. Actually. I was with my mom the other day and she saw a red box, like the physical thing, and she's like, "Who gets DVDs at a CVS anymore?" And I was like, "I don't know, but I know they're in a merger." Anyway, the company is in financial distress, they've got this distressed merger, but because they've got a small float, the company became a meme stock somehow. The merger values them at 50 cents per share. I'll pull a number out my hat. The stock is at $10 per share now because they're just meme stocking. It's like if you could hold that forever and short it, again, shorting is risky, especially memes stocks. Please remember that. But you would theoretically make money, but guess what? You're probably going to get carried out in a body bag first.

Doug: Yeah.

Andrew: Yeah.

Doug: Well, actually, speaking of which, I remember when that press release came out, I thought it was a mistake, right?

Andrew: Everyone thought they were missing a decimal because the stocks [crosstalk] that I remember at five.

Doug: Yes, I know.

Andrew: Yeah.

Doug: Yeah. They were like, [crosstalk] "No, no, no."

Andrew: The stocks at five and the implied value was 50 cents. So everyone was like, "Oh, that's definitely missing a bit. That's a $105 company would sell themselves for 50 cents.

Doug: Yeah. But I checked the filing, I checked the press release, I was like, "No, actually you should be short this thing." Obviously, no PARO, but I remember looking at it, I was like, "This has to be a mistake." And you're like, "Um... no."

Andrew: You're not alone in saying. A lot of people texted me and were like, "Hey, is this a mistake? That seems like the most obvious short in the world." I have 15 emails that said, "This is why we don't short meme stocks. We do not short meme stocks." Anyway, neither here nor there, I've got a couple more questions on semis, and then we'll turn to the specific. First question, I've just been thinking about this a lot in the light of inflation, supply chain issues, Ukraine-Russia. Semis are, you build these huge plans, right? Hundreds of millions, billions of dollars to build them, and you build them. I have been thinking about fragility in the supply chain and stuff. I do remember that semis, I think it was Micron with flash drives in 2012 or something where there was a flood issue. One of their competitors got flooded. They were like, "We minty money." Can we talk about fragility for a second? Is there any fragility in the supply chain here?

Doug: Oh, man. Is there fragility? It's a pretty big miracle that semiconductors work. Each time you make a semiconductor, it's now thousands of steps. Each of these steps take hours long. Actually, making a chip could take months because of all these different steps that go on and each step has to be done at like 99.99999% because if it's just 99% accurate, the compounding of all those steps, essentially the chips will not work at all.

Andrew: Yup.

Doug: The precision, the time, the amount of steps, and each step is pretty important in the process. Each step is filled with extremely intense Physics, Chemistry, and Engineering for every single step. Sometimes, there's a lot of very random one-off suppliers. So one of the ones that got a lot of media press was the Neon and Zion gas. That one actually ended up being a little bit of a nothing burger, essentially because it could be recaptured by air pro like APD stuff, right? They could process and make their own Zion. There's also recapturing, there's enough inventory. So that probably the supply chain could ramp the capacity additions elsewhere, other than Russia or other than Ukraine. Ukraine had this giant manufacturing footprint because they have a lot of heavy steel production. But then on the inverse, there are real places where it breaks down and it's a legitimate shortage problem. That happens pretty frequently. I'm trying to think of the one. There was one that was much scarier than Zion and neon. It escapes me off the top of my head, but there's meaningful times.

Actually, one of my favorite ones ever was there was a specific type of epoxy plastic in the 1995 era. It was one plant in Japan that plant blew up and that was 60% of the world's supply. Everyone's like, "Well, we're screwed." You could go read articles and they did manage it much better than honestly everyone perspectively thought, which is actually the bull case on almost everything. I actually really like there should be a pessimist archive post about that. Actually, they did manage through it, but it was [crosstalk] real problem.

Andrew: The bull case is literally the smartest engineer scientists and everything in the world are a hundred percent laser-focused on this. When you've got the smartest people in the world focus on it, they somehow find a way. They find a way.

Doug: I was really amazed, but there's a lot of, at least in the semi-industry, what happens is it's pretty amazing to get the solution to work once. Every single time, they're like, "Okay, it works, don't touch it. That's it." But then they're like, "Okay, we need a second supplier." Often, and this is at least on the supplier side, it often ends up as three players with 60, 30, 10. That's a pretty common market share split. It's extremely consolidated and it's one of these places where there's not going to be a new entrant because the entire total market value of this will maybe be the cost. If you could get 50% share, that will maybe be able to pay back in a three-year period, and you're not going to get 50% share.

So these things always end up in these super stable, extremely concentrated outcomes. There's thousands of chemicals that go into these processes. It's really hard to know, but it's like whack-a-mole. Something breaks and everyone's like, "Oh my God, this is a huge deal." And then obviously, the biggest and most fragile part of the entire supply chain is the fact that it's mostly 50% and this is a number that I like guesstimate because I want to say TSMC is like 30, 40% of all found. So we'll just say 50% of all semiconductors or the ones that are really leading edge are made on a tiny island, a hundred miles offshore of China. That's the most fragile part of the entire story. What's worse is not only are all the semiconductor fabs there but all the companies that support the fab, all the companies that make the chemicals for the fab, all the knowledge and engineering that helps make the fab, all the cumulative experience is all in that island a hundred miles offshore from China. That is by far the most fragile part of the entire thing.

I am always terrified when people are like "Well, they should just invade it." And then like, "Have you heard broken nest theory?" Where it's like, "Well, they'll just rig the fabs to blow, mutually shore destruction." I'm like, "That's a dark age." To be clear, that's a dark age. The network start like I don't know, maybe I'm being a little pessimistic, but if we had that much supply go offline, it would be so bad. Everyone would massively lose forever. I don't know. It'd be like burning the library of Alexandria or something like the modern equivalent.

Andrew: The world is a scary place. Russia, Ukraine opened everyone's eyes to a lot of like, "Oh," it's really weird, crazy things can happen, especially if you're living in a dictatorship where news flow might not reach the people at the top who are making decisions accurately. I think the scenario you're alluding to would be really negative for every party involved. There's some really bad tell risk there, but the scary thing is it could happen because there is a controlled media there. So the accurate information might not reach the top.

Doug: Yeah. It's just like the world is filled with black swans, right? That is just the crazy highball events in the world that we live in. It would be really wonderful. We could always stay in the middle of the curve, but just one tail event can really screw it all up. That's one of the fragility, there's a lot of fragility in a lot of things, and there's even fragility in some of our ability to progress. It feels like it isn't a gamble. Every year, everyone is working very hard, but this is a huge treadmill with billions and billions of dollars at stake. There's so many things that just seem so impossible and such hard solutions and they come up and solve the solutions, but it's amazing the amount of progress that still now. So we'll see, but there's a lot of fragility. You only ever hear about it whenever it breaks, of course. That's always a story I feel with super complex, adaptive, crazy. It's like the supply chain, right? No one really cared about it until it just constantly breaks. Even though everyone's investing in it, it just isn't fixed.

Andrew: No one cares about it. So they can't get toilet paper, that's where people start caring about it.

Doug: Yeah.

Andrew: Last question here. One thing I always do worry about is when you've got this inflated demand from something that's in a bubble, right? I remember in 2000, Yahoo, a lot of people would say, "Oh, you can invest in them because they have actual earnings. They look kind of cheap." And then what happened was all the earnings were from tech bubble startups that burst and when those bursts, their earnings were completely gone. Obviously, this doesn't apply to every semiconductor, but I do look at what's happening with crypto right now and I say, "Hey, I do think a lot of demand, especially for probably AMD and Nvidia stuff was coming from crypto, Bitcoin mining, all this type of stuff." Maybe Bitcoin's here to stay, maybe it's not, but I can guarantee demand this quarter is going to be a lot lower than demand two quarters ago, especially for some of the really rug pulled stuff on Ethereum and stuff. How much is crypto driven demand and is there any worry that where three months from now, there's like a crypto winter and demand is just way down and people are looking to say, "Oh, we overbuilt," especially, I think that's more bleeding edge supply for the crypto mining stuff. We overbuilt supply a lot and now we're in a really overbuilt situation.

Doug: So crypto is really interesting because I even wrote up Nvidia and talking about how the E 2.0 merge moves to proof of stake and that will destroy the entire. And so, then there would be this entire backlog of cards that would hit the market, and boom, Nvidia doesn't grow. What's actually really interesting is data center is large enough this time that it will and should be able to bail them out. It's really, really, really hard to make the numbers work quite. So essentially, I assume, I did some modeling. I assume that the Q/Q decline is as bad as it was in 2018. Obviously, it's a larger base now, so maybe that's part of it. Essentially, if you assume that Q/Q revenue declines 50% in gaining or something like that sequentially, Nvidia's revenue would be flat. It wouldn't shrink. So that's a depth mill. That is really trying to kill the stock/company and you're like, "The revenue would be flat." That's the worst case.

Obviously, that would hurt the heck out of the stock a company that has extremely high inflate decks[?] like high expectations, high earning stock. But this time, the revenue probably associated with crypto mining is a lot smaller than it used to be. The ASIC side of it will definitely be hurt obviously in Bitcoin. But the Ethereum mining portion of it, I did some bubble math, some envelope math, and I think that's like 15 million GPUs or something, which is a lot, but that's a quarter. I think that's a quarter of revenue of gaming for Nvidia.

Andrew: Oh, you can correct me if I'm wrong, but I also think there has been a shortage of I do remember, especially six months ago, people were having trouble getting little Xbox's, PlayStations. There was a new Xbox, but so I would guess there's probably some demand to be made up if there's slack on the crypto side, there's probably a little bit of extra demand on the Xbox, PC. I had trouble getting a Switch at the height of COVID, maybe I can find or Switch or whatever it is. Maybe people can finally get their Switch and play some Mario party. It's all the people want, guys.

Doug: Yeah. What's funny is it's finally happened. I think one of the best ways to track that is probably the secondary MSRP price. So MSRP versus the secondary prices of GPU, and they finally have come into line, meaning supply availability is broad enough that you can buy a new GPU off the shelf. So if you've been trying to buy a GPU, an Xbox, or a PlayStation, you can buy it now, so you should go buy one.

Andrew: Too much to do in the market today and I've already got my whim[?] and I beat my wife all the time with Mario Kart.

Doug: Hopefully, Wee-man[?]

Andrew: No, it's on Switch. It's on Switch.

Doug: Okay.

Andrew: That's a new one, right?

Doug: Yeah. The Switch is the new one.

Andrew: Yeah.

Doug: But essentially, TSMC has been adding all this capacity and right now, the revenue percentage growth is one of the fastest it's ever been. So remember, TSMC is a large part of the market that essentially it is the market. When you're 50% of the market, your revenue growth approaches the entire market. It's growing 30 and change. We'll say like 35% revenue right now, which is for semiconductors, that is astounding. Now, part of that is a price increase. So it's been interesting because amid all this crypto bubble stuff, unlike last time, there was this huge data center market that is like so hungry for AI accelerators.

The products are similar enough where they are substitutes to each other where I'm sure they could do some amount of like, "Hey, this ampere-based gaming GPU, we can actually just reallocate it to an ampere-based AI accelerator." To be clear, I'm kind of bearish the pull-through demand from crypto, but I think it's not as big of a death. It's not going to kill them this time, but it could be very, very bumpy in terms of Nvidia and AMD. Actually, AMD probably matters less, because they're really levered to the server CPU. That's really their biggest driver.

Andrew: To speak to the values, talk about like, "Look, I just pulled up Bloomberg." I don't know if these numbers are right or not, but TSMC, you say, "Hey, they're growing 25, 30% year over year." I think I've got them trading at 20 times price-earnings. That's pretty attractive for a company that's growing 25, 30% per year. No, that's just high level. I haven't dove into them at all.

Doug: Yeah. Their margins are improving, so their EPS should be a little higher than that. It's pretty attractive. I would say TSMCs multiple has contracted meaningfully in taking into account the geopolitical stuff. But then, on the other side of things, every multiple is contracted. There's a lot of interesting companies, you could look at the automotive semiconductor companies. A lot of companies are starting to trade at we'll say, low teens multiples on a forward basis and they're growing. They're growing revenue. So it's pretty attractive, but obviously, you have cycle risk. How much can you hold? How much fall[?] can you have? What type of time period? It's a little bit harder than just saying, "Hey, look at this company growing quickly at a cheap earnings multiple."

Andrew: Well, speaking of attractive companies, we've been talking cycle 45 minutes. Why don't we talk a little bit Rambus? The ticker there is RMBS. You wrote it up in February, that's behind a paywall, but anyone who's interested in semiconductors should be subscribing anyway. So I'll include a link to the write-up in the show notes, but let's just talk Rambus for a few minutes. RMBS, what is it, and what makes them so interesting?

Doug: Yeah. So for context, what got it on my radar was kind of a traditional non-GAAP dark arts thing. So essentially, the management team did a meaningful upsize in how much they paid themselves. I was like, "Okay, interesting." This is a company I've actually been following for a little bit. So I knew the story, I understood.

Andrew: Can I just pause you there for one second? So just for people who don't know, dark arts is the dark arts of corporate governance. When Doug says they did a meaningful increase in what they paid themselves, this wasn't they took their salary from $500,000 to $1.5 million. They said, "Hey, we generally give ourselves 5,000 shares per year. This year, we're going to give ourselves 15,000 shares this year." What the dark arts is generally people give themselves a big increase in shares right before a lot of good news pours in and the shares go up quite a bit. Just to give people that disclosure, so they understand what you're saying here.

Doug: It was a meaningful size in how many shares. So that's what kind of ping me on it where I was like, "Okay, I really need to be tick because it's an IP company." IP companies are historically very, very frustrating to analyze because they're kind of black boxes, right? You don't really know what's going on. But this IP company, in particular, [crosstalk] you could...

Andrew: What does an IP company do? Just so people know.

Doug: Intellectual property, right? When you make a semiconductor, it takes a village. There's a lot of different parts and components. A lot of times, the technology, you're not going to go out and reinvent the wheel every single time. Rambus has certain wheels that are off the shelf when it comes to making the total package.

Andrew: Could you just give one quick example of what the wheel that they would take? They would [inaudible] [crosstalk]?

Doug: Okay. I was going to actually start with their history, which actually is really helpful. So they originally invented the Rambus, I think it's like DRM. It was essentially the prototype before DDR, which is this industry-wide standard. They created this type of synchronous, DRAM, that was one of the foundational technologies to just making RAM work. Actually, they didn't get chosen. It didn't become this ubiquitous thing. In their death rows, they sued everyone they could. And so, that what the history of Rambus is mostly known for. It's this litigious patent troll and like. Seriously, if you look from, I want to say 2002 to 2011, lawsuit with big seven companies, every single year. And so, their core IP, the stuff that they own is mostly around DRAM. DRAM is a type of memory. There's NAND, which is the memory that remembers stuff even when you turn it off and then there's DRAM, DRAM does not remember stuff when you turn it off, but it's very quick and it's very important.

In particular, they have two big segments products, which is LRDIMM, which is essentially a special type of DRAM that's mostly used for data centers.
The point of it is that essentially with a little bit of higher latency, you have an effectively quicker memory, and this is just extremely technical, but it's very attractive in hyperscalers. So that segment is growing pretty nicely right now, data center's growing very, very well this year. That segment is growing, I want to say like 40, 50% revenue. And then they have this IP business, this IP business is this long history of things they've acquired or they've owned that is all related to memory. The IP business is extremely concentrated. So their top five customers are 59% of revenue. They are SK Hynix, Micron, and Samsung Broadcom. Companies that are large. And so, these companies are not forced to but work with them because they just have the IP, it's available and it's cheaper to just pay someone 2 cents or something, a chip instead of reinventing the wheel, these foundational building blocks into making a semiconductor. They probably go into everything.

But the thing that's so interesting is that, in particular, Rambus has this IP and interconnect, in particular, for PCIe and CXL. CXL is this protocol that's coming along in the next few years, there is no revenue today, but in 2023, we should start to see the beginning of this market form. When that happens, their revenue should meaningfully accelerate because essentially, this is a market that is going to grow for a long and long time because it's essentially a re-architecture of the data center. It's pretty complicated as to why this is so compelling, but CXL is going to essentially finally unlock memory from the CPU. One of the core problems of semiconductors since the beginning, it's called the "von Neumann Bottleneck." Jon von Neumann, he's one of the OG physicists. He invented the concept of a computer. Essentially, if we had it our way, we would have infinite and infinitely fast and infinitely available memory right next to the processor because the processor like the CPU is actually much faster than the memory being able to like going out and fetching the memory and bringing it back and then cranking all the numbers and then putting it out to the memory that is actually one of the slowest parts of doing any kind of computation. In the data center, even more so, especially for stuff like machine learning, machine learning, in particular, is this huge, new extremely data-hungry application.

DRAM, in particular, is starting to become a larger, larger portion of this. But the problem is you can only stick so many DRAM sticks next to the Nvidia GPU. What's going to happen is CXL is going to essentially make this plug in play where you just plug it into the accelerator and you have infinite memory. That's going to break the von Neumann Bottleneck finally like that's the very bullish way to put it. And so, what that means is that it's going to fix this core bottleneck that's always been a problem in the semiconductor industry. This is pretty much the architecture that I think is most likely to continue going forward in machine learning. You have this huge call option, and I don't think Rambus is quite frankly the best purest way to play it. No, well, it is the purest way in terms of a public market company. I'm sure Marvell, Broadcom, or even Micron, those segments will probably do better, but you're betting on something different. Rambus, this will be material. This could be and should be a material part of their revenue going forward. How big we are going to see, but until then, the business itself. And so, that's what I think the core part of the bullishness of the grants of management to themselves, are they?

They see this market that should be in Microns, TAM estimate a 20 billion market in 2030. That's huge. Right now, it's zero revenue. Next year should be 1 billion. That S-curve of potential market size is absolutely enormous and it will be meaningful for Rambus. Meanwhile, you get that huge call option. We'll see it come, but it's going to be how a future data center is architected.

Andrew: And that caution[?], I can't remember the Micron number, you said it was a 20 billion? Million?

Doug: 20 billion by 2030. Yeah.

Andrew: 20 billion. Because they've got the IP, they will just get a right off the top little fraction of it for licensing their IP, right?

Doug: Yeah.

Andrew: That'll be super high margin.

Doug: Yes.

Andrew: Any piece of that be.

Doug: Or inversely they get bought. This frankly is an extremely strategic asset in my opinion, because of this aspect. Meanwhile, while this is all happening, their entire business is doing very well. The company itself trades it around I want to say $220 million of free cash flow on a $2.4 billion ED. We'll say like 11 times ED to free cash flow. It's growing revenue at 40% this year. I have a high conviction that yes, maybe the first half of 2023 could be bumpy, but eventually, when we get to this next level, this next leg of growth, it's going to start to accelerate with essentially a completely new business segment that is not being written in today. Today, we have this company that is pretty cheap, growing fast, because of their current business in LRDIMMs, which is a very pure-play levered to the data center market that is like one of the most attractive places you can be. But tomorrow, we will have this increasing content story in CXL that they have a very good option.

They are one of the forerunners. So probably them, Alpha Wave, Cadence, and Synopsis. Meanwhile, Cadence and Synopsis are, by the way, the logical buyers here because Cadence and Synopsis, are 800-pound gorillas, but essentially, the likes of Alpha Wave on Rambus exist because they don't want to give all their businesses to Cadence and Synopsis. And so, right now, if Cadence and Synopsis have not gone down materially on a share price basis, their earnings yield is much slower than Rambus. They could buy Rambus and it'll be almost immediately extremely free cash flow accretive to them. I think all of this is ticking to a totally different cycle than the rest of the market if it makes sense. This is an adoption story of CXL that will continue for the next 10 years. It's a very long-dated growth story, in my opinion, that is trading on a very near-term low multiple that is pretty attractive and it's just very idiosyncratic if that makes sense. That's the high-level pitch.

Andrew: It's a fantastic pitch. The first question I ask everyone is, what do you think you're seeing that the market is missing? But I think I could just infer here that what you're seeing is this big adoption story, but if you think there's something else you're seeing the market's missing, please tell me

Doug: It's CXL.

Andrew: Okay.

Doug: It's definitely this. I actually did this call today talking about CXL versus ethernet. It might not be the end all be all, but CXL is going to be this huge new growth factor going forward. I think this is actually my favorite type of story in semiconductors where a giant protocol happens, a company wins or is a large percentage is a meaningful share gainer or like a shareholder in this. Essentially, they just ride this wave of adoption. The fundamental results follow through, obviously. One of my favorites ever was the first post ever written on my Substack was on Friday. They rode this huge wave of adoption, and now they got acquired by Marvell. They're putting triple-digit numbers in Marvell, essentially.

Andrew: I was about to say, I think I remember they got acquired.

Doug: Yeah, they got acquired. Dude, I posted that, and actually, they got acquired a week later. It was ridiculous. It was pretty weird. I'm kind of actually a little salty about it. If it was standalone, I think it would be worth a lot more today, but you wouldn't send me lose some.

Andrew: I felt that way before, but it's funny for everyone I feel that way. I do have to remember that there's one where they get acquired, I'm like, "This is bullshit. This is way too cheap." And then three months later, or three years after the acquisition closed, I look and I'm like, "Oh, that stock would've been not down 90%."

Doug: Yeah. They wrote it down to zero.

Andrew: Not to throw stones at people, but I do remember at home, which was a retailer, it got acquired last year and I knew so many shareholders who were curious, who were arguing, and you can go look at the bonds, you could just look at the price of retail stocks in general. I think their shareholders are probably pretty happy they had taken off three.

Doug: Yeah. That's true. They're probably pretty happy.

Andrew: Anyway, so back to RMBS. We've already talked about what the market is missing that you're seeing, but let me push back on a few other things. The first thing, we talked free cash flow year. You mentioned 240 million or so in free cash flow. I look at this thing and I say, "Oh, well, EBIT. I'm just choosing 2021, and I know there's adjustments and stuff," but even for 2021 was 24 million and their free cash flow was they did CFO of 200 million free cash flow of a hundred million. I looked that and say, "Oh, well, it seems like a lot of this is coming from the unbuild revenue, a big working capital drawdown, and all this sort of stuff." I look at it and say like, "How sustainable is the free cash flow number?" Again, I've done work here. I know that there's IP licensing, which as you said, is black box. It can draw down at all different times, but when you say that free cash flow number, how solid do you feel that it's kind of sustainable at those levels? Because I look and say maybe free cash flows just going to go down a lot.

Doug: Yeah. That's probably the hardest part about knowing this. It is cheap optically. I do acknowledge the black box aspect of it where it is a lot of deferred revenue. Frankly, this is part of the reason why it trades, so "cheap" is it's extremely complicated on accounting basis. So the change of ASC 606 essentially ruined how they report revenue and now we have this hybrid, they report nongap revenue from the old metric, and its billings. That is probably our rightfully so the most complicated part of this thesis, in my opinion, is the fact that how sustainable is it? On the other hand, what I think about pushing back on that is it's really rare that you get a business where you're growing revenue at a top-line at this rate where it's trading at this price. So maybe this free cash flow is not as sustainable as it looks. I think that actually, EBIT will start, too.

One of my concerns is actually EBIT will start maybe their margins should not as improve as much as they have in the past because the fact that they're going to ramp all these expenses into CXL in order to make that launch work well. I think that, right now, at this point in time, I think that it's a little bit more sustainable than it's been in the past just because it's going to be mostly from products. In the interim, it's going to be more product-based, which is more predictable and more normal type of free cash flow conversion where it's like, "Hey, product gross margin, operating costs associated with it, then that flows to cash flow operations." So that's how I push back against that. I also am very cognizant of that. It's a pretty sticky story in terms of the face, the counting, and actually keeping up with it is actually very involved. I think that's maybe one of the "opportunities" why it is "so cheap" because it is a complicated stock to value if that makes sense.

Andrew: Another question. Second question I always like to ask, if they're so cheap, why aren't they buying back shares? They're pulling in a lot of free cash flow, they've got a lot of cash on the balance sheet, so if they're so cheap, management's pulled up as we can see through those equity grants, why aren't they going back and buying back shares? I do have an addendum there related to convertible notes, but I'll pause there and ask you that question.

Doug: Yeah. I'm not sure. Honestly, that makes sense that they should be buying back shares. They back[?] at their convert, correct?

Andrew: Go ahead.

Doug: So how I would argue is they're buying back shares through hybrid security instead of open market. I feel like that's pretty common where you buy back your convert first, and then you start to buy back shares. I also think that most companies right now are definitely a little bit cautious in building cash. One of the weirdest parts about this story at least is last quarter, after putting out amazing results, beating earnings, beating revenue, beating every metric they have available, they talked on the results for the rest of the year. And then in callbacks with other investors, you're like, "Okay, maybe we talked back too. We talked it down too much, it's not quite this bad," but that's part of I don't know, but I don't know why they're not buying back shares that right now, given the fact that they're issuing themselves shares. So you think they're really bullish, but they did buy back their converts. That would be my push back there.

Andrew: Yeah, that was the addendum I was going to put. They did this convert buyback and they bought back the hedges and everything. So I wasn't sure, but let me turn to my third pushback. Again, I'm just a dumb, dumb generalist, but I look at this and the first thing I look at is, Doug, says there's some dark arts equity grant, so I flipped through the proxy, I flipped through, and I just look at this and I say, "Oh, I forgot to tweet out my notes." But every transaction in the stock has been in insider sale for the past two years, right? They've been pretty hefty insider sales and insider ownership here is pretty dismal.

Doug: It's pretty de minimis. Yeah.

Andrew: The CEO, he owns, I think it's four or $5 million worth of stock. I'm trying to see it in my notes here. It's really small on the side of my screen for those on YouTube, but he has like four or $5 million worth of stock, but it's all been granted to him. He makes $5 million per year, same with the COO. These guys, basically don't own anything. I look at this and say, "Okay, well, all they do is they sell stock as soon as their stock vests. They don't own a lot." If this stock is so cheap, I have seen companies where they know they are riding a big wave, they've had something big happen to them and all the insiders just start buying shares like crazy. They're not using an MPI. They just say, "Hey, we've got this massive call option that the stock market is valuing and we want to take advantage of it." You're not seeing that there and that's kind of strange. You also mentioned there's a, I don't want to call him an activist because he doesn't own a lot of stock, but they had someone with activist history who got added to the board, or was he a board? Did he get out to support?

Doug: That one is a really weird one, so he's now an advisor. That one actually is a little bit more gray than I thought it was. I was like, "Hey, a board, an activist there." If you look at his book, Rambus is a single best-performing stock. I think they blew out of it and then he got a consulting deal to leave the board.

Andrew: So you've got this activist who gets a board advisory seat and maybe blows out of the stock. He still doesn't own a lot. All the insiders don't know anything and I just look at it and say, "Trading cheaply with this huge call option. Why aren't we seeing at least some share insider ownership, some insider buying? Why isn't this activist getting on the board in the moment the window opens just buying as much shares as possible, pushing the company to buy back shares. It just seems strange."

Doug: Yeah. I would love to see that, honestly. So far, we have not seen that. We're going to have to see over time, but I still think that maybe you're right. Maybe the correct reading on this is that these guys just like to pay themselves a lot and right now, this is the time to get paid, I guess. But I do think that regardless, I still think it's an interesting strategic option. I still think that the absolute change in grant sizing is interesting. They have been selling all along the way, but the fact that it's such a meaningful step up is an interesting signal to me against what I think is. Right now, a good market for deals for them at least to be bought because Cadence and Synopsis materially do the exact same business in their IP second[?], but Cadence and Synopsis get a 30 times earnings multiple on the market.

And so, imagine being able to the likes of Cadence and Synopsis, you buy this stock, you roll it into your stock, and boom, you get to buy something at say mid-teens earnings and then you get to turn around and write it up instantly to your share price. I mean, maybe there'll be some adjustments, the dilution cow, but it'd be unlikely that buying Rambus would be extremely materially eroding their margins, their entire business, and frankly, they're multiple. So I think that's the slightly more interesting way to think about it. I think that there's definitely a chance. I think the activist was there mostly to get that consulting agreement, but that's an aside.

Andrew: Yeah. I don't want to talk too much about that.

Doug: Yeah. Absolutely.

Andrew: It struck me as very strange.

Doug: It was very strange. You can read it for yourself.

Andrew: They just did an acquisition. It's not a major acquisition. It was 20 million Canadian, which I believe the Canadian to USD conversion rates that make it $200,000 USD.

Doug: Yeah.

Andrew: Okay, it was a bad joke. It was a very small acquisition for a company this size, but look at that acquisition, it was an acquisition worth PR and it's $20 million, I just looked at it and said, "Hey, if duck thinks this might be a sales candidate, would they be doing this kind of Bolton acquisitions while they're in the midst of maybe sailing? Or do you think it's just too early?" There's not much to read there.

Doug: First off, it was an acquisition for their CXL-based solution. It's the correct thing to buy if this is really the feature that you're going to put your hat on if that makes sense. But the second thing is they report CapEx inclusive of acquisitions. That's part of the continuing cost of this business and I think they'll be doing this kind of acquisition forever. Not forever, but that's part of how they've gotten their IP portfolio. Maybe that's the correct way to think about free cash flow. You have to account for some amount of acquisitions in there as well to make that recurring growth continue because yes, it probably is not material for, in terms of this part of the revenue, but that's part of the shaping and selling of this IP conglomerate essentially. In order for them to continue to be relevant, you're always adding in IP. I think it's the right move. Essentially, if I was Rambus, I would be buying every single CXL PCIe thing you can buy from here until much further from now.

Andrew: Just one more question on CXL, so they bought hard end. What are the odds CXL takes off, it's this huge market, as Micron says, 20 billion market and everything, and Rambus isn't getting a cut of this. For some reason, there's a work rounds to the property or people use other people's IP, does that make sense? Is there any chance of that happening?

Doug: There's a chance. It is a competitive market, so I was hoping it was a little less competitive. It did some calls, did some work. It is a competitive market. I guess one of the biggest opportunities Rambus has is actually playing against Cadence and Synopsis because Cadence and Synopsis, essentially, at least everyone has IP that can get you eventually to the same spot, but Cadence and Synopsis, unfortunately, don't support you because they're large company. They don't really care. They're an EDA software company versus Rambus is going to support you because you're their primary business. And so, one of the reasons why they would get acquired is probably because it essentially just lower competition to aid Cadence and Synopsis, whereas Rambus and Alpha Wave and the smaller players have been more nimble and been able to win business against others because of the fact that they can be more nimble and support the customers where they need it. That's the reason why I think they're going to be around because no one wants to just give the keys over to Cadence and Synopsis forever. And so, that's the reason why I think they should be able to get a meaningful part of the take.

Now, it's complicated. This is the part where it becomes really uncertain. Well, let's put it this way if the company was trading at a lot higher multiple, and it didn't have this relatively recurring IP business, I think this would be a much harder bet to make, meaningfully much harder bet to make, in my opinion, because CXL is this coop option. I was reading the CXL spec 2.0 today and yesterday and there is no revenue right now for these companies. This is a bet on the future, a reasonable bet that makes a lot of sense logically that it will happen. If CXL happens, it will make the total cost of compute in a data center much cheaper. It is just a very common sense conversion to what the next iteration of data center looks like. But in this moment in time, it is a bit of defining the future. I think it's very likely because CXL happens to be or Rambus is one of the leaders in the CXL consortium. They definitely have a lot of PCIe 5.0 IP, which is part of what CXL is built on top of, but you're making a bet on the whole ecosystem and that ecosystem has not been played out yet. We'll see. I think it's very likely. It's very highly likely and the price is what makes it able for you to sustain that bet with positive carry, but at a different price, this would be a lot harder to make.

Andrew: Let me ask a really stupid question, but when you describe how this is going to impact the data center market and how it's going to bring down costs and stuff, and then you say, Micron says it's going to be a $20 billion market by 2030. I think about just the cost and bringing down the cost, it seems like CXL should be just a much bigger market. Am I kind of missing something? Again, this is a generalist just asking, but when you say, "Bring down cost in this market," and I say, "That's a $20 trillion market or something." I don't know

Doug: That's the problem though if it really brought down costs so much by adding an incremental unit, does that really break? It's like, "Oh, we brought down the cost by adding a new cost to the total thing." It's like adding a new aspect of healthcare. It's like, "Oh, we brought down costs," but now we have another intermediary. So realistically, the thing about that 2030 TAM, that's a long time away from here. That's a big, extremely long data call. I do think that the ecosystem and momentum behind CXL is very, very real and it's very palpable right now.

Andrew: When would you start noting?

Doug: Second half of 2023.

Andrew: Second half of 2023? Okay.

Doug: Second half of 2023. Yeah. We should probably have at least a conversation or interest about design wins and the magnitude of what that means in the stock pretty soon, but you are definitely going to be paid by being forward-looking. This is a very nascent market. This is a huge, super foundational change that is going to happen. It is definitely one of the biggest incremental growth drivers that I could think of in the entire universe of semiconductors. It's going to change a lot. It's a big deal. The freeing memory from the CPU is a big deal like that is how we scale these Megatron, the GPT-3, and GPT-4. The biggest bottleneck is the DRAM. We're going to unlock that by essentially having this pooled memory option enabled by CXL. So it's going to happen, but it's not yet. We're still early days. When that becomes reflected in the story, that will meaningfully change probably the multiple and the potential durability of growth expectations of the company. I think that's probably when you get rewarded or bough out.

Andrew: The other thing here, just comedy, I was just flipping through the proxy. The CEO in a change of control, gets 17 or 18 million, which is nice and that's probably before the dark arts equity grant that you talked about, but he makes five or 6 million per year. I'd almost rather see that number be higher just like really incentivize this guy.

Doug: To be clear, I don't think the change of control is not really what I'm betting on it. I think that the downside protection scenario is that essentially if there's misexecution, it gets bought. If there is execution, it should be better off on its own.

Andrew: With low insider ownership, misexecution and activist have already stepped in here in a kind of strange way, but there has been activist interest like semiconductors and activists, there's a lot. You see it all the time. There's misexecution there. An activist will step in here at some point and enforce their hand.

You've been super generous with your time, but I always do ask, we talk semiconductors, we talked Rambus, is there anything in our conversation that you think we kind of lightly touched on that you wish we had hit harder or that we didn't touch on that you think we should have been talking and thinking about?

Doug: No, honestly, I just feel bad for some of the listeners for all the mumbo jumbo or around. I feel bad. I swear, sometimes people are like, "Dude, I have no idea what you're writing about." And I'm like, "I swear to God, I'm writing in the simplest language I can." That's always my concern.

Andrew: You and are talking about it. This is why generalists get their face ripped off when they come to semiconductors because I come over and I say, "Oh, semiconductors are in computers and you've got the history of the semiconductor, the whole supply chain, and stuff." But no, I think this has been super helpful. I've learned so much from reading your stuff. I've learned so much from this podcast. It's been super helpful. As I said before, you're a real guy doing this stuff. So Doug, thank you so much for coming on. I'm hoping before our buy-side shop snaps you up, we can get you on for another one and talk what's going on semis because it's constantly changing in some really interesting space, but Doug O'Laughlin, Fabricated Knowledge, everyone should go check it out and subscribe. But thanks so much for coming on. Looking forward to the next one.

Doug: Thanks, Andrew. I'm happy to be here. Take care.

[END]