Arm: The Silicon Blueprint - [Business Breakdowns, EP.200]

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

0:45This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. I'm Zach Fuss, and today we're breaking down Arm Holdings. Arm designs the architecture powering billions of devices, from smartphones and data centers to IoT devices and automotive systems. In this episode, we'll explore Arm's unique value proposition, how it thrives as a licensing giant in a market dominated by leading -edge manufacturers. We'll discuss its business model, the partnerships that drive its growth, and its role in enabling companies like Apple, NVIDIA, and Qualcomm.

1:19We will also unpack Arm's business history, including its acquisition by SoftBank, its failed takeover by NVIDIA, and its IPO earlier this year. Orm currently sports a $150 billion market cap, with sales approaching $5 billion, a rather robust 30 times revenue multiple. This valuation is, of course, predicated on its 90 % plus gross margins, complemented by Orm's dominant market share in mobile, and its increasing share gains in markets like automotive, cloud computing, and IoT, with its technology being licensed in nearly 30 billion chips annually. To break down ARM, I am joined by Jay Goldberg, who is the CEO and lead analyst at D2D Advisory, a technology and strategy consultancy, and who has spent various roles in the semiconductor ecosystem.

2:06We hope you enjoy this conversation. All right, Jay, it is great to have you back. The world continues to change and evolve, particularly as it comes to your area of expertise, semiconductors. Today, we're talking about ARM. So to start things off, I know this is a loaded question and it has a fair degree of complexity, may require a bit of a semiconductor one -on -one education for our audience, but let's start with a brief overview of Arm's business model, what exactly they do, and how they fit into the broader semiconductor ecosystem. So I'm going to give you an analogy on Arm to give an intuitive sense of what they do.

2:44This is not a 100 % perfect analogy, but it's what we work with. Arm licenses its intellectual property, its IP, to companies that design chips. So they don't make chips themselves and they license their IP to companies like Qualcomm or NVIDIA or Broadcom, these big chip companies who then design that IP into their own chips, which then get manufactured at TSMC or one of the foundries. So ARM is fairly early in the process of this whole semiconductor flow. So the way to think about ARM is that the IP that they provide is almost like a blueprint, but a special type of blueprint. So imagine you're an architect and you're designing a house.

3:23As an architect, you differentiate yourself by how the house looks. What's the curb appeal? How does the light flow? These big architectural features. Typically as an architect, you don't get rewarded much because you have the best designed plumbing. You don't have the best designed bathrooms. Those are important in a house, but that's not really where you as the architect differentiate yourself. And especially in lots of places, bathrooms are generally very standard. And what you could do is just take the generic blueprints for a standard bathroom, and you're going to copy and paste those around your design.

3:56That frees you up to design the parts of the house that really matter, that are really going to drive your business as an architect forward. I think it's a good analogy for how ARM interacts with its customers. There's a certain amount of low -level plumbing work that needs to be in lots of types of chips. Certain types of math, certain types of functions need to be in all chips, but there's no way that a Qualcomm or a Broadcom or NVIDIA really differentiates around those. It's a basic math things that need to get done. They license that IP from ARM, and then they work that into their chip, and then they can differentiate on things.

4:29For Qualcomm, it's how their modems work, how their communication systems work, or for NVIDIA, how their graphics and AI processing works. ARM plays a vital function, providing a really important piece of the overall functioning of a chip, we're still allowing the flexibility for their customers, their licensees, to design their chips as they see fit. So what they do is they license that IP, their business model works essentially where they charge an upfront license payment, and then they take a royalty, some amount of money per chip that their licensee actually ships. That's a really helpful analogy to frame it.

4:58In my research, clearly GPUs have ruled the day, at least in the last, call it, 12 to 18 months. My understanding is that this is a CPU -oriented business. So help us to better understand the importance of their CPU versus the GPU architecture that people are working with today and the interplay between those two in a future world state. The CPU is very much a general purpose chip. It can run any type of workload. It can run the low -level functions of the keyboard and the mouse and the hard drive for your computer. It can handle the operating system and the applications. it's a general purpose jack -of -all -trades chip.

5:35But because of the geometry and physics of semiconductors, you can always design a chip that does a single one of those tasks better than the CPU. So in the case of GPUs, they were first developed to run graphics really, really well. The CPU could run basic graphics, but a GPU can do graphics much better than a CPU. The GPU can't run the operating system well. It can't do all the low -level functions as well. It can do graphics really well. So in engineering, you're trading off, what do you need the chip to do? You make those trade -offs and you design different tasks to different chips. And with GPUs now becoming important tools for AI, these AI accelerator chips are even more finely tuned towards doing AI math.

6:13And that's similar to GPU math. We don't need to get into it. That's part of the NVIDIA story. In regards to ARM, what's common to all this is you have certain general purpose functions that ARM sits at the heart of all those. So you don't need necessarily an ARM IP, an ARM core inside of a GPU, because that's not what the GPU is there to do. But you're going to want it inside the CPU. You're going to want that functionality. So in most compute systems, you're always going to need some control node managing all the multiple tasks. So even in these big NVIDIA systems that have eight or 72 GPU cards all linked together, there's going to be some number of CPUs in those as well.

6:48So the way to think about ARM in relation to the growth of AI, it's obviously the hot topic now, is GPUs are exploding. We're seeing so many more GPUs being sold, NVIDIA is doing really well, selling all those GPUs around the world, there is some degree of CPU attached to that. And depending on where they source the CPU from, that can often be an ARM -based CPU. And in fact, NVIDIA probably prefers to sell its own CPU, Grace chip. It often attaches those to its GPU systems, and that CPU is ARM -based. So after we establish their unique business model and the CPU versus GPU dynamics, I guess what I want to really try to wrap it all around is that core technology.

7:26What is it about ARM's IP and their business model, how it designs chips that differ from their competition. So ARM really faces two elements of competition. Historically, they have been seen as in competition with x86. And I should have to take a step back and say, ARM's IP is sometimes called instruction set architecture, an ISA. And that's just a framework for how chips are supposed to handle these types of math, these sorts of problems. 40 years ago, there were dozens of different ISAs. And over time, we've winnowed down to two. There's the ARM architecture, and then there's the x86 architecture.

8:01And today, there's two companies that provide x86 architecture, AMD and Intel. Intel started it, and then AMD joined along. And ARM is the other instructional set architecture, which historically wasn't used for computers, wasn't used for PCs, really got its big boost from smartphones and mobile. That's changing. We can get into that. historically, you'd use x86 in a PC and you'd use ARM in a smartphone. Most recently, we've seen another ISA enter the fold called RISC -V, R -I -S -C -V, RISC -V. And that scene is an open source, not exactly open source, but it's open. It's not owned by one company.

8:37It's designed by a consortium that anybody can contribute to and use freely. So RISC -V is out there as an alternative to ARM. It's still very early days for RISC -V. So it's on the, I'd call it an emergent potential competitor as opposed to certainly not a dire threat to ARM anytime soon. So if we think about how we got here and work from the start, my understanding is that the general familiarity with ARM came through the early day success they had with Apple. But how did they go from this, what I would call, niche player to someone that's so dominant in the architecture of the future today?

9:09What is the business story? ARM came out of England, out of the UK. There was a company called Acorn Computers. And this is back in the 70s when everybody was making computers. And Acorn was selling computers through other people's brands, typically like the BBC. And they had a couple of hit products and they realized that to be competitive, they wanted to design their own chip. So they tasked the team with designing a chip as a proto -CPU of the day. And that was the kernel of what would become ARM. Eventually, they split that off. The chip design team and the hardware team went their separate ways.

9:39And the genesis of ARM, the company today, was originally called Acorn Risk Machines and later Advanced Risk Machines. We can talk about what risk means later. But the point is, it was part of Acorn. It got spun off into a three -way joint venture between Acorn, VLSI, chip company of its day, and Apple. Because Apple originally wanted to use this Arm IP for the Newton products. The Newton didn't do particularly well, but Arm actually did pretty well. It made a lot of money from Newton, which it then used to grow and grow its part of portfolio. And from the get -go, Arm was an IP licensor. They stopped making their own chips when they spun off, And they licensed their IP first to Apple and then to many others.

10:17And slowly but surely, that grew. And then they started adding customers. I think after Newton, their first few big customers were fax machines, designing IP for fax machines, of all things. And then I think what really tricked them into the beginning of this big growth trajectory was Nokia. Nokia and their chip partner, Texas Instruments, started using ARM in the late 80s or early 90s. And the first ARM -powered mobile phone was actually the 8110, which is the Matrix phone. That was the first ARM -powered mobile phone. And very quickly, other mobile phone companies started to realize the appeal of using ARM IP, ARM architecture in mobile phones.

10:51And the reputation I got from very early on was that it was much more power efficient than alternatives, especially something like x86, which at that point really evolved into something that was meant for a device that was powered, a laptop or a PC that had pretty access to power. Mobile phones were so power conscious that ARM had a lot of appeal. So first, it was in a lot of feature phones. Over time, their capabilities increased. And then what really, really kicked it into high gear was 2007, the launch of the iPhone and this explosion in smartphones, because suddenly you needed very, very complicated, very advanced processors for your phone.

11:24And the best way to do that was to start with an ARM IP at the core of the phone. So to answer your question, really the explosion came with the growth of mobile. So as history dictates here, ARM went on to dominate mobile devices. Presumably they have a growing presence in other markets, automotive, Internet of Things. My guess is that if I walk through my living room, there's probably eight or nine devices that have some ARM architecture license to them. But I guess most topical today is this increasing demand for AI and machine learning. So I guess in contemplating why a business today that is running at close to $5 billion of revenue has $150 billion market cap, it kind of begs the question, what the heck is going on here?

12:05So it's a few things. First is the arm of today is very, very different than the arm of just even a few years ago. For a long time, they had talked about expanding into new markets. By the mid -2000 teens, they were very much a smartphone -driven company. 90 -something percent of the revenue was derived from smartphone chips. In the last decade, but especially in the last three, four years, they have made huge inroads into other markets, the most important of which is probably the data center, where they're helping all the hyperscalers, Amazon, Microsoft, Google, Facebook, Ali Baba, Byte, Tencent, the big Super 7, have all now designed their own CPUs to run web workloads.

12:41And those are all based on ARM chips. And this is an alternative to using expensive Intel and AMD data center chips. These companies went out and designed their own chips. The natural place for any of those companies to start with would be ARM, because there's really no other alternative. So I'd say data center is first and foremost, their biggest growth opportunity. IoT is out there. IoT is a messy market. They're doing okay there, but nobody does stellar in IoT. It's too complicated and too messy. Automotive is certainly really, really interesting. I think they've made good inroads there with a lot of their partners.

13:10A lot of their licensees are doing well there. Automotive takes a long time to mature. These are long product cycles, five to 10 years to develop a chip into production for automotive, as opposed to one or two years for everybody else. So we'll see how that goes, but it's looking pretty good right now. So ARM is doing well, expanding into pretty much everywhere. And I think you're right. AI is such the story of the day. And it's not quite as clean a story there, because like I said, they're not necessarily going to have an ARM core in every GPU or every AI accelerator that's out there, but there is some attach rate.

13:40You'll need CPUs, you'll need other control functions. There's ARM cores in networking chips that sit next to all this that connect all these AI servers together. So there's a lot of content going into that's adjacent to the core NVIDIA AI accelerators. And ARM is absolutely benefiting from that hugely. But I personally think that ARM's valuation is justified by more than just AI. It's this broad expansion into so many other things. So maybe to back up a little bit, I know there's kind of this classic debate amongst semi -enthusiasts around RISC versus CISC and these two fundamental architectural approaches in processor design.

14:16It feels important here. So maybe we should just spend a little bit of time explaining those two and why ARM ultimately adopted a RISC approach and how that contributed to success. It's funny, this risk versus CISC debate is something that's been in computer science topics for a long time. Back in the early days of compute, this debate was incredibly important. I remember I took some computer science classes in the 90s, and that was the topic of the day was risk versus CISC, and which is better? Lots of famous moments around it. But at heart, it's two ways to think about how you architect a chip.

14:50And RISC stands for reduced instruction set, and CISC, or C -I -S -C stands for complex instruction set. And the basic idea is if you have a chip and the chip runs on zeros and ones, but a level above that, there are assembly language, there's human readable code that the chip then translates into zeros and ones. And the idea behind CISC is that each of the critical mathematical functions you want to implement in silicon, you have a separate instruction for those. And as opposed to RISC, where there's a much smaller number of instructions you can give the chip. And if you want to do more complex things, you have to string a few of those simpler instructions together.

15:25So the simplest way to think about this trivial example, it's not quite perfect, but works is in a CISC architecture, you might have something like a square root function, where it calculates the square root, which knows how to do that math, and it moves the zeros and ones around to do the square root of a number. In RISC, you'd have to do a series of division, subtraction, addition steps in order to get that same command implemented. And people at the time used to get kind of religious about which one was better. And the truth is, I don't think you could say that one is better than the other.

15:54There are just different situations when one works better than the other. There's always trade -offs. This is engineering. There's always trade -offs between what is good, what's right for the application in front of you. What ended up happening was risk initially was very, very memory dependent. You needed a lot of memory because you had to remember all those steps. And at the time, Sys looked more appealing because it didn't need as much memory and memory was very expensive. Over time, on -chip memory got very cheap very, very quickly, and that made risk much more attractive. And because it was this reduced instruction set, you can say it ended up requiring less power.

16:25That's oversimplifying it a bit, but at the heart of it, you needed less power to do these calculations because the steps involved in the complex instruction sets were complicated by design, and those ended up chewing up more power. So that's how RISC got its initial reputation for being much more power efficient than CISC. What ended up happening in history is x86 took the CISC path, the complex path, and ARM took the RISC reduced instruction set path. And that's how they diverged way back in the late 70s, early 80s. And it's humorous to me that it's still a topic that keeps coming up. We have the RISC -V project, which is very explicitly RISC -based, coming on stream today saying, hey, we have all the benefits of RISC.

17:05And I think as much as this is about low -level chip interactions, what it really, really meant, what this difference between CISC and RISC meant was RISC allowed users a degree of flexibility in designing their chips. It wasn't just the power savings that made RISC appealing. It was much more about the flexibility. I need RISC to do something slightly different than what x86 has done. I'm going to take this other approach and I'm going to optimize for power savings. I'm going to optimize for this function and that function. And that flexibility was really important because one of the critical differences between x86 and ARM is that x86 is owned by two companies effectively, AMD and Intel.

17:40while ARM is available to be licensed by anybody. It's very, very hard to license x86. Anybody with enough money can get an ARM license. So then to get a step further, one of the interesting things about RISC -V, this open source project, is that it is free for anyone to use or at least to take the code. So it's even cheaper and more available to ARM and it is by design even more flexible. You can really shape this around and use RISC -V in ways that you can't even do with ARM, which itself is already pretty flexible. Now, there are drawbacks to that. Probably you don't have to get into them. There are software complexities down the stream from that.

18:15But at heart, really, the debate is around flexibility of instruction sets to customize, tailor for each individual chip designer, each licensee to do as they see fit. It's not every day that you approach a business that trades at 30 times sales that has this classic innovator's dilemma where there's this open source, hypothetically free to use, nothing obviously is free. competitor. But I think before we go into the competitive landscape and the dynamics around ARM and the x86 and RISC -V, I just want to better understand what that partnership and licensing model is really like. How does that model work going forward in a world where there's so much competition?

18:54And what are the key advantages to licensing its tech to Apple and Qualcomm and Samsung and others versus vertically integrating the business in a way that could help to maybe make their competitive advantage more durable. This is a little complex to answer. Let me talk about historically, and then I'll speculate on where things are going. Historically, it was essential in ARM's early days that they didn't manufacture chips. Part of the problem when you manufacture a chip is, first, someone has to do the design. That's expensive. You have to hire designers to do that. And that's what Qualcomm and NVIDIA do.

19:24They design their chips. And then once you've designed your chip, you have to pay someone else to manufacture it, typically. So Qualcomm or Broadcom will pay TSMC to do the manufacturing for it. So that's a lot of upfront expense. There's a lot of working capital involved. You have to pay for mask sets. You have to build inventory. You have to take ownership of all those chips. That's a lot of expense. And for ARM, back in the early days, it just wasn't financially viable. It was much simpler to just license their IP. It was also something the market really needed. So I think that was driven originally by necessity.

19:54And the earliest CEO was dead set on this model. He saw the opportunity and grabbed it and really ran with it. And then as time emerged, you got this big ecosystem of lots of other customers, lots of licensees designing their chips that really helped grow the ecosystem. There was a period in the early 2000s when there was a big debate between x86 and ARM. And this was around mobile mostly, but other areas as well, where ARM was able to succeed because it certainly did its own set of R &D and innovation. But then they had hundreds of licensees who are also innovating and exploring the market and finding every segment and every niche.

20:30And that drove a huge amount of volume of chip designs versus x86, which was Intel and AMD. Big capable companies, but limits to how much they could explore. So ARM was able to, beyond the technical merits and the power savings available at ARM and Spones, they had this giant ecosystem, which was very, very powerful in expanding the university dressable market. And then over time, because of the rise of phones, it also drove volume to the foundries. It really powered the growth of TSMC into the beast it is today. So you have this big ecosystem effect that worked really well in its favor. So now we're at a point today, though, where the market is consolidated a lot.

21:07There's always lots of competition in semis. The question is, does ARM want to build its own chips? And it's certainly, I think, in the realm of the possible where they're going to move up a step and design their own chips. I don't think they're going to go so far as to necessarily put their own label on a chip and design it and sell it into the market like a merchant solution. But I do think it's very possible that they will take a lot of steps to help their customers, their licensees, bring ARM chips to reality. Because there's a step in here that's important. In between having ARM IP and actually then sending it to the foundry to manufacture, there's a lot of steps that have to take place.

21:43We call it hardening the IP, taking the IP from just digital files and good ideas and translating that into the design for chip. That's a fairly cumbersome process. And I think ARM is now taking a lot of steps to make that process easier, accelerate it. And I think what that will probably do is open the door to new customers who may not design chips today, but with ARM's help, they can bring those chips to market faster, better, in ways that probably weren't possible before, just relying on merchant partners. And that's certainly what we're seeing with the hyperscalers. ARM has done a lot to help those companies design their chips.

22:16And I think they're going to take a few more steps and get very, very close to doing a full design of their own chips. And when you consider those steps to designing chips and partnering with other technology companies that participate in other parts of the ecosystem, who else are you bringing together for this project than ultimately spec yourself into something which I assume has extremely high switching costs on a go -forward basis? So I think that the chief complexity here is in the software realm because chips don't exist in isolation. you build a chip so that it can run some form of software.

22:48There are certainly incremental steps in the design process, but I think what's more critical than any of that is the software that's going to run on these chips. And a big part of the ARM story of the last decade has been the amount of work that software companies are doing to make their software run better on ARM architecture. So in theory, you have all these different ARM chips out there in the data center. Like I said, Amazon has one, Microsoft has one, Facebook has one. There's a fair degree of compatibility that if you have software that runs on Microsoft's CPU, you can fairly easily port that over to run on an Amazon CPU.