Good morning,
Today’s Stratechery Interview is with Qualcomm CEO Cristiano Amon. Amon first joined Qualcomm in 1995 as an engineer, and in 2021 became only the 4th chief executive in Qualcomm’s history.
Qualcomm is most known for and widely associated with mobile phones; the problem for the company from a valuation perspective is that mobile is a replacement market, with limited prospects for growth. Amon has sought to tackle this reality head on, re-fashioning Qualcomm as a semiconductor company with growth opportunities in a multitude of adjacent industries, including auto, PCs, and the Internet of Things (and, of course, generative AI). We cover all of these topics, as well as Amon’s background, what it’s like to become CEO of a company you have worked for your whole life, and what mistakes Qualcomm has made — or misconceptions there may be about events in the past.
One topic we did not cover was the ongoing litigation between Arm and Qualcomm; this was my decision, as I assumed that Amon would be limited in what he could say about an ongoing legal dispute, and had plenty of other topics to cover.
To listen to this interview as a podcast, click the link at the top of this email to add Stratechery to your podcast player.
On to the interview:
An Interview with Qualcomm CEO Cristiano Amon
This interview is lightly edited for clarity.
Amon Background
Cristiano Amon, welcome to Stratechery.
CA: Happy to be here, Ben.
So you may not know this, but we are kindred spirits of sort, in that we have both endured quarantine in Taiwan. However, I had to stick in a hotel room with two kids for fourteen days, and then a dorm room for eighteen, while you only had to do three days, I bet you didn’t realize how good life is as a CEO.
CA: That’s good. Look, I think you’re probably the first one that I talked to that had been to Taiwan right at the beginning of the pandemic and been in quarantine, so I can relate to that. But I tell you that three days is better than 14.
Yes, and with two kids, you missed that important detail! Obviously, we’re here to talk about Qualcomm, but before we get to that, I’d love to know more about you. You’re from Brazil, how did you get started in technology? How did you come to the United States? How’d you get started at Qualcomm? I want the life story, albeit in somewhat abbreviated form.
CA: Sounds good. I’ll do my best to keep it short. But yes, born and raised in Brazil, went to engineering school in Brazil. I think when I was going to college, I remember my father said, “You have three choices. You can do anything you want, as long as it’s engineering school, medical school, or law school.” From the three, I was very attracted to engineering, my dad is also an electrical engineer, so I went to engineering school there and I came out of school exactly at an interesting period of time. It was in the early ’90s when cellular was starting — it was the beginning, the late ’80s, beginning of the ’90s was the beginning of cellular communications — and I started my career working in cellular, and I joined a company, it was a Japanese company called NEC. I started working in the subsidiary in Brazil and then after working there for a little over a year, they transferred me to the headquarters. I actually left Brazil for the first time to go to Japan in 1994.
At the time, I was working in cellular, I was a system engineer working on infrastructure, and as we started thinking of going from analog to digital cellular, we saw that there was some markets that wanted CDMA technology that was being developed by this new and small company called Qualcomm. That’s how I got to meet Qualcomm and Qualcomm ended up asking me to join the team, and I joined Qualcomm in 1995 as an engineer, and became a CEO on the 30th of June of 2021.
You just skipped over a big chunk of time, but I am actually interested — obviously, given the fact I now have ten years of daily opinions about what tech companies should do under my belt, you would be unsurprised to know that as a low level Apple or Microsoft employee, I had takes about what I would do if I were CEO. Today, as you noted, you’re the CEO of a company that you first started working for twenty-eight years ago and I’m curious, at what moment did you start thinking, “Man, if I was in charge, this is what I would do”? Did that thought occur to you back in 1995?
CA: I had a lot of those thoughts as I went on my journey, but never really thought that will actually become a reality. But the good thing about the Qualcomm culture is, I think the company’s always been very open to new ideas, and has a very unique collaboration culture, not very hierarchical. So I did have many opportunities to express to the CEOs what my opinion was about things and it didn’t hurt me in my career. But I kind of relate to what you said and there’s many times I said, “Well, if I was CEO, here’s what I would do” and then now I have a chance to do it.
But what is interesting about my entire career in the company, is what’s happening with the company right now, and what I’m doing as CEO is, this is one company that has had an incredible ability to reinvent themselves over those many years in technology. I think Qualcomm started as a satellite communication company, then went to CDMA, that’s what put Qualcomm on the map. Then from CDMA was to transition into becoming a chip company and also then the transition from CDMA to FDMA, going from becoming a modem-only company to a computing company, and we have all sorts of opportunities and challenges along the way.
If you look where we are right now, we’re becoming an automotive company, an Internet of Things company, and working in a number of different areas, and I’ve been fortunate enough to be in my career in the company, actually my entire career since I started at NEC, to see every single transition of wireless, from analog, to 2G, to 3G, to 4G, to 5G, and hopefully 6G, I think in the not-so-distant future, and also be able to see how mobile technology has transformed so many parts of society. I kind of feel privileged in some way to have had that career at Qualcomm. Still, I didn’t expect to become CEO, but I’m happy as CEO.
It’s kind of interesting, because sometimes when a new CEO comes in, it’s an opportunity to throw old management under the bus to a degree. I don’t mean this in a bad way, but rather it’s like you have a chance to curry early favor with investors and partners, because it’s a one-shot chance to admit to any mistakes that were made and blame it on the last guy, right? It’s like this one time Get Out Of Jail Free card. But it’s interesting, you’re the fourth CEO in Qualcomm’s history and obviously it was sort of a family business, particularly with the first two CEOs. When you took over, did you feel like, “Well no, I’m a company man, we’re just pushing forward”? This idea of becoming the boss of a company, you’ve been there for so many years, how did you balance that, “Look, this is a new start, this is a fresh beginning, even though I’ve been here for so long”?
CA: That’s a very good question, actually. So two answers. The first answer is, if I look in my career progression within Qualcomm, and I was managing the semiconductor business, became president of the company in 2018, and I’ve been growing as the company is growing. So I have been involved with many other aspects of the many transformations at Qualcomm and I don’t think it’s a good idea to throw myself under the bus.
Right. That’s the tension, right?
CA: Yes. Now the second part of the conversation though, that’s a very interesting question, which is, everybody that’s been in a company for a long time, you have been doing many different things and as you become CEO, you have to have the ability also to say, to stop and think, “Is this the right strategy? Should we be doing something different?”. I had to do that as well, which is part of an ongoing thing in this company, to always be able to try to predict the future, and adapt, and adjust to what that’s going to be, so I’m not immune or having to do that as well, even though I’ve been here for a very long time.
Licensing and Semiconductors
One of the longstanding criticisms of Qualcomm is, or critiques from an investor perspective, is that it’s sort of a legal company with an engineering arm attached, at least in the latter years. And yes, Qualcomm Technologies helped create the modern cellular industry, and still drives most of the revenue, but the profits and all the margin is from the licensing arm, which was very aggressive in making sure it collected its chunk of flesh. There have even been calls from investors over time that say, “Look, just split up the company. We actually just want the profits from licensing,” or whatever. Obviously, you’ve come in and basically made a real double down on “No, we are a technology company, that’s what we’re going to invest in”. Do you think there was a bit where that characterization was unfair and not an accurate representation? Or do you think Qualcomm maybe was a little bit stuck in a rut in that regard?
CA: No, I think that characterization is not fair, and it’s old. I think that’s the best way for me to describe it — that’s old news. At the end of the day, Qualcomm, and we’re very proud of this, and it will continue to be a big part of the company, but Qualcomm is probably the most successful IP licensing company in the world, we have the most successful intellectual property licensing business. If you look at the scale and the standard essential patents, that has been Qualcomm contributions to all of the wireless standards, we monetize that R&D through a licensing business and I think the licensing business had been, I think the best way for me to describe this is being battle tested in old markets.
(laughing) I think that is a very fair characterization of it. By the way, I think it’s fair to point out, as much as Qualcomm is sometimes in the news for, going back a couple years now to be clear, but your fights with Apple, and the FTC, and the Chinese government, and things along those lines, to your point, you did invent a bunch of this stuff and deserve to be compensated for it and the world has benefited from having access to those technologies. I do think it’s important to say that I agree with you, it is unfair to a degree, not just in technology versus licensing, but the validity of a licensing business, so I’m with you there.
CA: It’s kind of bigger than that, actually. The way we should think about it, what if Qualcomm was a company that got all of the inventions and kept it to ourselves, and we’re the only one building products, and we’re the only company that will offer those kinds of products and technology? Instead, the licensing model enables any company to get access to a technology because we contributed to the standards, and participate and if you look at all the companies there today, which are extremely valuable because of what they do in the cellular industry, it’s in part enabled by this business model that we actually contribute our technology to standards. But anyway, I’ll go back to your question.
I will spare you any counter arguments about pricing by device as opposed to chip, or things along those lines. But anyhow, continue.
CA: No, but look, I think this model, I’ll go back to what I said, it’s been battle-tested, it stood all of its challenges. It’s now probably the most stable time of our licensing business, but I want to go back to your question — why do I say this is old news? The licensing business was the very first scale business of Qualcomm, but Qualcomm started to also develop a number of different technologies on the semiconductor business and today the semiconductor business actually is the largest contributor to the profitability of the company. The operating margin of the semiconductor business is larger than the licensing business but more importantly, the semiconductor business is the growth engine of Qualcomm.
It’s important to understand how did we get there? So the interesting thing is, as we think of technologies ahead of time, and usually we think about technologies about ten years before it’s going to get commercialized and we take risks. I think an example right now is we’ve been working for more than a decade on merging of physical digital spaces for virtual reality, augmented reality, we’re still in its early days, and as we think about those technologies, we realize, for those technologies to exist, we need to create some products. For example, when you think about 4G, and 4G was about bringing broadband to a cell phone, we realize, what are you going to do with broadband if you don’t have a computer in the palm of your hand? Where’s this broadband going to be useful for? So we had to figure out how to develop a bunch of technologies in semis that allow you to actually have a computer that fits in the palm of your hand. We were the first company in the world to come up with a CPU that could run at one gigahertz speed in a battery power device and have the battery last all day. Now we’re doing the same thing with AI.
Qualcomm has changed a lot and I’ll summarize the answer to this question by saying the growth engine of the company is semis, and we realize that the technology that we create, which is basically three pillars, it’s everything wireless communication, everything high performance computing for a low power device, and then artificial intelligence, we will have new end markets beyond mobile. That’s exactly what we’re doing. We saw a huge opportunity in automotive, huge opportunity entered in the PC space, future mixed reality devices, broadband devices, Internet of Things and so forth, and that’s what we’ll be executing. So very different than being a license only and legal battle company, it’s a whole different world right now.
Now we will get into all of those topics, and I think it’s very interesting, the broad diversity strategy, but to tie this conversation together, you talk about being a semiconductor company. I think one of the honestly more bizarre episodes of my career analyzing the tech industry is Qualcomm did have a semi business and then they shut it down, and started shipping Snapdragon processors that were basically just Arm cores that were taken off the shelf. We won’t get into the Arm business, as far as the ongoing legal bits, but just big picture, Qualcomm abandoned their own designs. Then obviously a few years later you bought Nuvia and you are all back in. What happened there? Because I agree with your strategy, I’m going to give you more time to fully articulate it, but why would have Qualcomm ever thought it wouldn’t be a semiconductor company?
CA: All right, so let’s unpack that. We have been a semiconductor company since we designed our first chip for CDMA phones, and never stopped being a semiconductor company. I think what you ask is, within your chips, within your semiconductor products, you use to develop your own CPU, then you stopped developing your CPU, started using Arm and now you’re back developing your own CPU post-acquisition of Nuvia.
Yes, very good clarification.
CA: Let me start with that because it’s super logical, super easy to understand. So why did we hire, build, put together, made some acquisitions and put together a CPU team that was actually based at that point in time, in RTP — why did we put a CPU team together? It was very simple, I’ll go back to what I just told you. When we realized the potential of 4G technology as an evolution from 3G to 4G, we realized that the phone needed to become a computer. So when we look at what’s available at that time within the Arm ecosystem, because we use Arm cores for the CPU portion or for chips future phones.
I think my audience understands the distinction between the Arm instruction set versus a core from Arm. And if they don’t, it’s okay, they’ll figure it out. So go ahead.
CA: All right. So at a time, we look into the ecosystem of what are the Arm processors at the time, and there is nothing that could have the performance of what you will see for example, in a personal computer in a mobile phone. So we realized the smartphone SOC would not be possible if we did not develop a CPU. So we developed our own CPU and when we launched it, Scorpion was the fastest CPU with an Arm structure set running on a battery powered device, and we followed with a second generation, great. Eventually as we got into future generations, Arm caught up and Arm started to offer their own CPU cores. They were comparable to what we could develop ourselves, so we were just very pragmatic and focused our engineering resources in other areas that are beyond CPU. We focused on unique differentiation workloads such as graphical processor unit, getting more powerful GPUs for rendering high definition gaming, et cetera.
We invested in other areas, started licensing CPUs from Arm, but then we had exactly the same situation happening again. We started on this journey with Microsoft, to build the next generation PC, we are the Arm ISA partner of Microsoft, we’re using the Windows on Snapdragon, and we realized that we needed a CPU that wasn’t available in the ecosystem, that could be the leader in performance in the segment. That’s when we acquired Nuvia, and I think that Nuvia has been integrated into Qualcomm and we have been developing our own custom CPU called Oryon, and it’s going to be applied across our entire portfolio of products. The first instance is on PCs and it’s history repeating itself. We needed to go develop something because we believe the level of performance is going to be required for leadership is not available in the ecosystem, and we’re very bullish about what we can do with our new SOC, including our Oryon CPU. Sorry for the long answer.
I’ll be totally honest, I think you answered your question about being the company man versus throwing previous management under the bus, because it seems like to have your own CPU team, to disband it, and then have to rebuild it seems like a mistake from my perspective, but we can leave it there.
Qualcomm’s Diversification Strategy
You talked a bit about this diversification and I think obviously Qualcomm — one of the founding companies of mobile, and it didn’t just have a leadership position in mobile, but mobile was then just exploding as a market, and you’re along for the ride and driving the bus in part, for sure, and now you get to today and mobile is obviously a massive market, but it’s not really a growing market anymore. To that end you’ve been talking a lot since taking over about the importance of diversification. What’s the high level pitch on this, about where Qualcomm is going to find growth in a world where mobile is just going to be at a steady state?
CA: Here’s how I will characterize the Qualcomm strategy. Mobile is not a growth market but it is a massive market, so I think Qualcomm is always going to be in mobile. We are going to be the company that’s going to be setting the pace of innovation in mobile, you’re not going to see Qualcomm moving away from mobile. However, we want to grow the company and therefore, we need to find other end markets for our technology that will continue to grow in the same way that mobile grew.
Mobile is a very large market. I think every person on earth has a phone because I think there are more phones than people these days, and it is a fully penetrated market. It’s a massive scale market, but it’s a market that is growing at single digits at best and it moves up and down with macroeconomics, it’s a replacement market. Consumer confidence, disposable income dictates, “Oh, I’m going to buy a new phone now, I’m going to buy a new phone later.” But still, it’s a very big market.
However, the technology that we develop for mobile is transforming many industries because of the scale of mobile. Think about mobile as the largest mankind development platform and I’ll give you some very easy examples to understand this. People use, and this is exactly how I started thinking about this actually, it actually was this simple, I used to go to people’s cars and people will buy a holder and they put their smartphone right there, and that holder in front of their dashboard. Why are they doing this? Because the phone can provide more information that is useful to them than the systems in the car can. So we figured out we can create the Snapdragon Digital Cockpit, where we can leverage our technologies in mobile and scale that to automotive.
Then you think about something else, think about the PC. The PC we see now a full conversion between mobile and PC. If you look at what happened in the pandemic, the PC is now a communication device as much as a mobile phone’s a communication device. I’m sure you’re talking to me right now through your PC and that’s what you’re doing. We’re communicating, we’re doing video telephony, so we see this opportunity of conversions, our technology can go to the PC market and actually make a difference, actually create a leadership position when you think about ultra mobility and work from anywhere. So what we did is we create a mechanism that the technology we develop for mobile we can actually scale to all of those end markets outside the data center. The car is the ultimate mobile device, the conversion of mobile and PC.
We look at the limitations of the screen size and we said, “Why not making glasses and make the bridge between physical and digital spaces”? We’re looking into the broader digital transformation of enterprises, a smart camera, a robot, and we said those technologies we have for mobile are scalable to those end markets and it’s accretive to margins because most of the R&D to support this is from mobile. So I think that’s in essence, the strategy of the company and it’s working. We have said that now the non-mobile business accounts to about approximately 30%, which we’re very happy with it, it was in like 27% a couple earnings calls ago. Mobile’s very big but I think we are in the right trajectory and over time, I think we expect those things to be a much larger percentage of total Qualcomm revenue.
Yeah, it makes total sense. It’s like a classic cash cow strategy, but in this case, the cash cow isn’t just generating cash, it’s also generating, to your point, the underlying R&D and technology that can be applied, and this is the beauty of chips.
One of the questions about the chip bit is there’s a weird balance here where Apple gets to manifest their chip design differentiation by virtue of them only being available in their devices; however you have this situation where you want to sell to everyone in Android, and obviously you compete with MediaTek in those regards and you’re doing very well in the premium end of the market in particular, but also those individual OEMs want to differentiate their phones, and so there’s this tension about who appreciates and captures that differentiation.
You go back to the 90s and you had the Intel Inside consumer campaign where they wanted customers to actually care about their chip despite the fact that’s something that they didn’t choose directly. Is that something that you think about? There are Qualcomm advertisements, but is that something you see the need to push on that you need consumers to care that it’s a Snapdragon chip, or is it sufficient that phone companies and OEMs know, look, if you want the best performance, you have to go with us?
CA: Very good. Look, I have to tell you, that’s my favorite question so far of this podcast. I’m so happy you asked this question because I want to tell you a fascinating story. Qualcomm has never been a B2C company, it’s been a B2B company. So historically, you have not seen Qualcomm traditionally investing in the brand, especially the Qualcomm brand. However, the conversation we just had, we have a very mature smartphone user base and consumers are now more and more interested in the technology and what are the capabilities of their phone, what is the next phone they’re going to buy? So I’ll give you a data point, which is incredible. In China, awareness of Snapdragon is 80% in the China population.
The China market, I can confirm, it’s very attuned to the feeds and speed specifications of what goes in their phones for sure.
CA: And that’s propagating to other markets. You look for example, European customers saying, “Finally we got phones now with Samsung brand with Snapdragon, that’s what we wanted” — you see that happening. The interesting thing is this happened in a different way than the PC. I think consumers look at their phone as their inseparable device, their most important consumer electronics purchase decision, and they’re interested about the technology to make the phone work. The second part of the answer to the question is remember, in the history of computing — because you said about the two different models — in the history of computing and it’s kind of a recent history you have the PCs, you have now phones, eventually they’re going to be glasses or something like that.
But you always have the competition between two models, the vertical model and the horizontal model, and as technology goes through its cycles, you see that the two models change. Our strategy with Snapdragon is to enable the horizontal model, and that doesn’t prevent individual OEMs from having their own differentiation. As a matter of fact, if you think of Snapdragon as a development platform for the OEMs, and we have seen incredible differentiation they have done in areas such as camera, into their gaming performance, in audio, in the broader multimedia between themselves leveraging the same platform, and I think we do invest a lot to make sure that OEMs can differentiate, and we continue to demonstrate what’s possible to do with mobile, putting Snapdragon as exactly the flagship platform that is driving the pace of technology innovation.
Yeah, it’s interesting. I think the way that I think about your branding and customers being aware of you and a positive outcome for Qualcomm is not that consumers buy phones because it has Qualcomm like it’s a top level differentiator, but rather they won’t accept a phone that isn’t Qualcomm so it’s like a baseline and then it’s on the OEMs to build above that, and to the extent you can establish that expectation. I remember with Stratechery, the first early years, people would buy it because it was like, “Oh, this gives me unique insights and a differentiator” and then there was some parts of Silicon Valley where you felt compelled to buy it. On one hand, you have grumpier consumers, on the other hand, that’s actually a very effective sales model where people feel like they feel compelled to do it and if Qualcomm could get to that position where “I’m not even going to consider it unless it has Qualcomm and then I’ll look at the differentiating factors”, that seems like a very good place to be.
CA: I agree with that, then also I think Snapdragon continues to grow. For example, we have a Snapdragon Insider’s Community of enthusiasts. They’re really wanting to know what’s coming and want to know, what are the capabilities of the chip is. It’s now crossed over to 10 million users and remember, for a B2B company to have 10 million users on the insiders enthusiasts community on Snapdragon, we’re very happy with that result.
So you mentioned the glasses and XR, XR the being virtual reality and augmented reality together. Why does it make sense for Meta to partner with you instead of making their own chip?
CA: This is pretty obvious, at least to me.
Hey, I got to give you some softballs here.
CA: Thank you very much, I really appreciate it. So if you look at the core competence of Qualcomm it really is the ability to actually get — think about the challenge of a glass, it has even more design restrictions than you find on a phone, it has to be very light, if it doesn’t look like this [Amon points to his glasses], people are not going to use it. You have to have a lot of computing power because you’re going to have to display those things to you and create a mixed reality, a fully immersive virtual reality and it has to be light, it cannot be hot, and you have to have battery life, and you have to make it work across the cloud in other devices, so this is what we do.
And if you look at Meta, especially as if you look at all their social network apps, you can easily conclude that the future of social could also be fully immersive like you and I are having right now, we’re doing the podcast and you can see my image, I can see your video, we’re talking about it but that could be a hologram in the future. You can think about what’s the power of a hologram, for example, if you have it in a social network.
So what you do? You combine those two assets, and it’s fair to imagine that we’re probably going to be a better option for them than having to come up with all the innovations on the chip side versus concentrating what they are really good at, and that’s why we think about those strategic partnerships that we have with them and many other companies. I think it is naturally how companies are going to become successful. It’s exactly the same reason when people ask me, “Why don’t you have a foundry? Why don’t you manufacture your own chips? You’re the largest fabless company.” My answer is pretty simple, I don’t have any core competencies manufacturing chips, I have core competence in designing chips and that’s why we partner with foundries.
Yeah and as large as you may be, you are not large enough to be filling a foundry. Intel had to face that reality and needless to say they’re a large chip maker as well.
Snapdragon Digital Chassis
Let’s talk about autos, one of your big focus areas. It’s kind of interesting because you talk about this shift of “We’re not going to be a mobile company, we’re going to be this diversified strategy and we’re going to have autos and Internet of Things” as your two categories. Qualcomm I think people don’t realize has had a strong position in infotainment systems for a very long time. What makes your approach today worth calling out as this is a distinct shift in strategy compared to what you did before?
CA: At the end of the day, I think auto is now one of the brightest spots in our diversification strategy. Look, we have in a short period of time, over $30 billion of pipeline, and that is going to translate into revenue over the next few years as those new car models launch.
But what’s unique about this is the approach we took, actually one clarification if I may. I think Qualcomm always had a position in connectivity in the car, especially with telematics. We started back when OnStar started with General Motors, but this whole concept of a digital cockpit is recent, it’s just a few years, but the important thing is to step back and say “What was our strategy in auto and why are we being so successful right now?” If you are an investor looking into a car company today, you’re going to ask them two questions: are you going electrical and are you digital?
The digital is a very difficult question to answer because the car companies will need a strong partner. So what we did, instead of thinking about different components that go into the car, we said can we help the car companies create a platform for innovation, a whole digital platform that can scale up and down across all the different brands and tiers, and that what’s the Snapdragon Digital Chassis. Snapdragon Digital Chassis is a combination of multiple solutions working together, it’s the connectivity to the car and to the cloud, and is a framework for you to have a lot of connectivity services into the car; it’s the entire digital cockpit experience, which is way bigger than infotainment; it’s the dashboard, the infotainment, the rear side entertainment, the smart mirrors, the heads up display and including some aspects of monitoring of the driver, the cameras they’re monitoring the driver, which integrates with another piece of the Digital Chassis, which is assisted driving and autonomy. I think that platform really resonated well with the industry, I think it’s no secret we’re virtually working with every brand out there across a multitude of tiers of cars, and we really excited about the digital transformation of the automotive industry.
I think that that’s a good articulation of just so many things becoming digital. At the same time, you made the point before that people would get that little mount in their car because the phone is still the center of everything. One of your partners is GM, they’re in the news because they’re going to stop supporting CarPlay in their electric vehicles. Do you think it’s realistic to have an experience independent of the phone, or is it probably better to incorporate the phone in addition to all these things that are car only, the heads up display, the smart mirrors, all those sorts of things?
CA: I need to use this opportunity to make a technical clarification, because it’s actually an important data point. When you look at CarPlay or Android Auto, in the absolute majority of the cases, it’s a projection, a technology projection from your phone to your dash, to your infotainment system. However, if you look at everything else that the car is doing, it needs to do, that is just a tiny piece of the puzzle. So at the end of the day, the platform and the car is significantly more powerful and more complex, especially because there are things that are important for you.
To give an example, navigation, let’s talk about navigation. The existing navigation system you see in your phone is to tell you how to go from one place to the other place, and I tell you how you can go there, what street you take and what’s the traffic like. The navigation for the car is very different. The car, especially when you think about assisted driving, the car is paying attention to potholes, paying attention to accidents, paying attention to traffic signs, it’s a whole different navigation. So I think the car will need significantly more, it’s completely unrealistic to think that you’re going to run the digital cockpit experience projecting from a phone, especially because the car computer needs to be fault-tolerant. You may crash your multimedia or your game that you’re playing, you can’t crash the dashboard, that’s not possible.
So at the end of the day, the answer is really what GM is doing is developing a platform of the car that is so powerful that actually you don’t need your phone. Everything you’re going to get on the phone is going to be there and that’s the expertise we contribute to the industry coming from the phone side.
So you just acquired Autotalks this week, very timely and it happened like twelve hours ago, it was announced. Tell me about that.
CA: Look, we have been talking about this technology called CV2X for a very long time. CV2X is basically car-to-car, car-to-infrastructure, car-to-bicycles, car-to-motorcycles, car-to-pedestrian and I think we as an industry, especially as we think about the car as this connected computer on wheels and part of a network, I think we can build an intelligent transportation system with zero accidents as a goal, and there’s a number of technologies you do need in order to reduce accident improved safety, for example, assisted driving and autonomous systems. But also you have to leverage the telecommunications network in this technology and CV2X is externally powerful.
Think about the use cases. You are in your car and you’re driving into an avenue, let’s just say at 45, 50 miles per hour, and on the sidewalk there is a pedestrian and it’s running towards the sidewalk, the same direction you’re going. That’s a very safe environment. However, the pedestrian coming towards you is not a safe environment. The ability for the car to sense what’s around it, to look at the position and the other cars, to position the pedestrian, the bicycles, and feed that information into the decision system for the assisted driving or autonomy, it’s very important. You can get to an intersection and the car will know what’s happening at the intersection because the traffic light’s telling the car, and see the position of the other cars, so this is going to be a future of intelligent transportation. This company [Autotalks] has a very good solution, it’s complementary to what we’re doing, and I think it’s going to be very well-received by the automakers. It’s about facilitating get this technology in as many cars as possible.
Is it fair to say your approach relative to say Nvidia — Nvidia I’ve always compared them a bit to Apple, they tend towards the vertical model in some respects — but in this regard you are building from the ground up, working with the automakers, you acquired the sensor company, you acquired another company last year in this space and you want to organically come up with the automakers as they become more computerized as opposed to saying, “Look, we have this top-down system for autonomous driving that you’re just going to drop in and abide by”. Is that how you think about the differentiation of how you are going to market versus Nvidia?
CA: I’m incredibly happy the way you describe it, because I have to tell you this is one thing that defines Qualcomm and you can go back in our history, and I know about our history and you’re going to see that, you’re going to see that everywhere. Qualcomm creates ecosystems, we don’t believe it’s just one company that’s going to be innovating, and let’s go back to how we establish our licensing business. So the way we think about this is can we create a platform and that can help the existing auto ecosystem to become very successful companies as they go through the transition to digital, and how do we create an environment that every company can contribute to? At the end we think that’s what drives the biggest innovation, because I’m actually adding the R&D capabilities of all those companies togethers, versus one single company and one single R&D is responsible for innovation in an entire industry. So that’s kind of what we do in the automotive business and that’s the reason I think the core companies have been partnered with Qualcomm, because they know it’s a long-term partnership and is a partnership where not only we have the right to innovate, but they have the right to innovate as well.
China
You have a very strong position with the Chinese automakers, particularly electric cars. Is there a core proposition of the Qualcomm growth story in auto that Chinese autos start being exported in a major way, sort of like the Japanese and the Koreans before them? Is that something that you’re thinking about as you think about this opportunity and is there a risk factor that maybe the reception to Chines autos abroad — particularly as the environment changes — may not be as large as it might have been five years ago?
CA: Look, I think it’s important, maybe there are different things worth talking about. First one is the progress that the Chinese automotive industry did with EV is really incredible. If you look of the number of models, the frequency in the launch now models and how that’s performing in the market, I think it’s very different than used to be a few years ago, so that’s the first part. The second part, there are a number of markets that the Chinese auto industry has been starting to participate and usually you see markets such as Latin America, Southeast Asia, Middle Eastern, Africa and Eastern Europe. As those markets change to EV, it’s very possible that the Chinese companies will have a strong position in the market as well as the existing car companies. From a Qualcomm standpoint, I think it will benefit from continuing to support our customers with our platform across all those different models.
More broadly, two-thirds of your revenue does come from China. Now obviously that includes a ton of smartphones including a prominent, how do you refer to in your earnings call, a “modem-only customer,” which of course is Apple who manufactures in China, so that’s why the revenue is accounted as being from there even if it’s actually worldwide revenue. Do you see the increase in tensions between the US and China — is that a meaningful risk factor for your business?
CA: Look, you said two-thirds of revenue, it’s a little bit high, but I think we have a big business in China as you would expect. If any company that has a leadership position, if you just look at this size of the GDP, you’re going to have a big business in the United States, going to have a big business in China. I think we are somewhat in a privileged position because what we do right now, it’s important for China, is important for the United States, and I think what Qualcomm does hopefully is an example what a successful China cooperation is, I can list them to you. It has elements of respect for intellectual property, has export of semiconductors and it has growth in technology areas. If you look at what’s happening with the current tensions, and I think we all hope that things improve, but I think there has been statements made publicly that the focus of the attention is not impacting chips for phones, for cars, for computers. To date, we have not been impacted by any restrictions. It’s hard to know what the future is going to be, we just hope that things improve, but we expect our business in China to continue to expand, and as we diversify our company, we’ve seen that. We now have more customers that we didn’t used to have in China. We used to have a lot of customers on the phone side, now we have a lot of customers on auto as well in IoT.
Internet of Things
With regards to IoT, Qualcomm has a very expansive definition of the category. For example, your PC chip efforts that you talked about before are considered a part of Internet of Things. What is your definition? Is it basically everything else?
CA: So to be fair, I think the IoT for us is very, very broad.
To be fair, nobody knows what IoT means, so you get to make whatever definition you want.
CA: (laughing) It’s funny, I talked to a lot of people that in the past, they will say that what corporations will do is everything that they don’t know what category it is, they say this is IoT.
That’s right, that’s right, exactly.
CA: We took a different approach. The reason we put that in the IoT category is because we’re doing all of those different things, and we needed to put them in one category, but they are not mobile, because at Qualcomm it is very easy to say this is mobile, we see that. So the way we think about IoT, and I agree with you, not all IoT is created equal, but the way we think about IoT is we’ve divided three categories, new products that are going to be available for consumers and it doesn’t mean it’s not going to go to the enterprise, but it’s just kind of how we think about it.
You’ve seen in our IoT consumer segment, you see the transition of PCs to Windows on Snapdragon, we see the products for mixed reality, virtual reality, augmented reality, and you see products such as wearables, smart watches, earbuds, those things.
Then in the IoT networking segment, or we call as networking, what we have is our WiFi access point business, WiFi networking business. As you know, we’ve been the number one in retail in the United States, we have been the number one in the enterprise, that business is now growing because the enterprises are all upgrading their communication systems to support collaboration. We have in the future the transition that has already happened from WiFi 6 to WiFi 7. The other thing we have there is 5G as a broadband technology we call Fixed Wireless Access. For example India, that is starting to deploy 5G, a lot of the broadband in addition to fiber, is 5G will be used to connect people to the Internet, so that’s in the IoT networking bucket.
And then you have the IoT industrial bucket and that’s very, very broad. That ranges from retail, energy, oil and gas, warehousing, manufacturing, when we actually provide a number of different devices, whether you have sensors, processing connectivity to the cloud, helping companies on their journey of digital transformation.
Generative AI
You mentioned earlier, you mentioned IOT, you mentioned cars, and then generative AI. There is a critique that could be levied with this inclusion which is that of course everyone’s talking about generative AI these days, how could you not talk about generative AI? Why is it actually not just a talking point for Qualcomm? Why do you see this as a meaningful and significant driver of your business going forward?
CA: Yes, very good. Actually we’ve been working on this or preparing for this, I should say, before it was popular. Let me explain the reason I’m saying this. This is the unique thing about Qualcomm, to try to actually invest in things significantly ahead of where the market opportunity is.
So we have this view, which I think you’re going to agree with me, that computing — think about the story of the CPU. Computing started in the mainframe, and then it went to the personal computer, went to the palm of your hands, computing evolved into the edge and became distributed computing. So we thought, the same thing’s going to happen with the computing of AI, or inference. So we’ve been developing what is the best possible hardware architecture that you can put on the phone, put it on a PC, put it on a car, and have the ability to run high-performance AI at the edge.
Then, one thing happened, all those large language models start to become very popular and we showed at MWC this year, we showed Stable Diffusion running on a phone in just a few seconds. You’re going to see now in the summer, it’s going to get significantly higher and we are going to demonstrate phones running models in excess of 10 billion parameters, we’re going to show PCs running models in the order of 20 billion parameters.
So what is the opportunity for Qualcomm, which is very transformative actually — you should look at this as a confidence building moment for all of this diversification we’re doing in all those areas. I’m going to talk about the use cases. First of all, if you are going to run those large language models only in the cloud, I think you probably saw a number of different articles that came out and saying the cost of something as simple as a search will go significantly higher in the data center because the computational power needed to put in the data center. What if you can run those models into the device? What if the device can actually give the cloud a head start, or build this hybrid AR architecture on top of it? You don’t have to have every model in the category of 200 billion plus parameters. You look at the number of startup companies that are developing all sort of different applications for audio, for text, for video, for photos, and the ability for you to run those things on the device.
An enterprise for example is thinking — if you read about it, you’re going to see people saying, I won’t say the company name, but was in the news last week. Some employee got fired because their source code was used on a large language model, which is open source, and the source code went to the open source. So enterprises could say, “I want those things to run locally”, especially when you think about large language models for code generation or code verification.
I’ll give another example. I won’t say the name of the company, but a very large retailer said, “Look, I have an incredible application, I can do a smart shopping cart that you can come in and you type in, ‘I have a family of four, I have fifty bucks, what should be my grocery list?’ and then you run those large language models, ‘What do you have in the store? Tell them options what should they buy and actually direct them how to pick them.’” There’s a multitude of applications. For us, it creates now the opportunity to have the ability to provide a platform that can actually run those large language models in the palm of your hands, in your car, and in the computer. The car one is the one that I find it fascinating. When you are in — I don’t know, look, I may show my age, but I don’t know if you watched it when you were younger, Knight Rider and there was that car called KITT.
Yep.
CA: That’s very possible to do with large language models running in a car, and the ability — you’re going to be talking to your car, and a voice is a very good user interface when you’re behind a wheel, and whether you’re going to do searches on different things on a map, when it will get information. So I think we’re in the beginning of something very exciting and we look at generative AI as an opportunity to create a new upgrade cycle on phones, very big differentiation on PCs. What if we can run instances of ChatGPT locally on the Office Autopilot? And the opportunity on cars.
I think the point you made that is a very important one, which is, if inference can be run locally, it is effectively free, as far as the consumer goes, the cost is their own battery life. That’s obviously a compelling opportunity for Apple and by extension, to your point, because you’re the horizontal alternative, it’s definitely a a big opportunity for you. I’m definitely with you on that.
One of the things that I think is going to be important to making this work though is that there be a healthy and thriving open source ecosystem. You mentioned Stable Diffusion and the way that so much of the innovation was just getting it to run locally, which was really compelling. You’ve seen a similar thing happen with LLaMa, which is not open source per se, but is available and you’ve seen the same sort of optimizations. Do you see this also as really critical for Qualcomm, that there be the development of an open source ecosystem, becoming sort of the commoditized complement to your chips? Is that something you feel the need or burden that you should contribute to and make sure comes to be?
CA: I think the answer is absolutely yes. But the way I will characterize the answer is, it’s incredible what’s happening in the open source community. Those models are evolving very fast. Look, we are still in May of 2023, you’re going to see in a year from now how much more those models going to evolve, and a lot of things going to be available and the open source community is very active. This is all an opportunity. It’s really an opportunity to create a number of different things that are very useful to the user. It’s going to change a bit how you think about apps in the OS and how you interact with some of those models and those are going to happen in parallel to what we do today, and I think that Qualcomm’s going to be more active in this space, especially with the open source community, because there’s going to be a lot of innovation.
Sentiment and Supply Chains
How concerned are you about your supply chain? You mentioned before, you don’t make your own chips. When it comes to, especially with TSMC becoming so dominant, Apple’s top dog there. MediaTek is right down the road. How do you make sure you get enough supply? You had some nice pieces in the Financial Times and Wall Street Journal about solving chip sourcing problems when you first took over. But as you go forward, is this a concern? Is this why you’ve sort of been talking about, or Intel loves to talk about, the fact that you’ve explored a partnership with them, at least. Is that something that is a concern looking forward, that you’re not a year or two behind just because of capacity reasons?
CA: Let’s talk about the scorecard, right? So today, I think we’re the second-largest customer at TSMC, I believe, I think you know who number one is, we’re close number two. We’re the largest customer of Samsung, we’re the largest customer of GlobalFoundries and the largest customer at SMIC, we’re the largest fabless company in the world.
Our view on this is actually because we don’t have any manufacturing, our view is actually this very same view of many governments. We want semiconductors to have a very resilient supply chain because we need that supply chain, but we’re not alone, I think the world needs that supply chain, semiconductors are in everything, so there are a couple things we’re doing. One of the things that we do is, we actually have a very well put together, efficient ability to multi-source the products. For example, when you think about products in 4nm technology or 5nm technology, we had both Samsung, what we call 4LPX, as well as TSMC 4nm for Snapdragon 800. We do balance our supply chain, but the reality is, that’s not enough.
So the way we see this is, if you step back within the next five to ten years, you can argue somewhere between the five to ten years window, we’re going to have to double the total manufacturing capabilities of the globe and semis, especially as you look at the future of the economy is digital — you look at all of those industries in the digital transformation, smart energy, smart everything and that requires semiconductors. So if you’re going to do that, why not do it in a geographically diversified way? If anything, just don’t think about geopolitics, just natural disaster alone. Our perspective is all of those investments that are being made, for example, fabs in the United States, I think hopefully fabs in Europe, we’re willing to work with our supplier actually to have some of our capacity built in those sites, and that’s what we’re doing right now.
I’ll end by talking about Intel. We have said in the very beginning of this supply chain crisis, we want more foundries. So if Intel wants to be a foundry and that’s what they indicated, we encourage that, we support that, and if they have a competitive product, we’ll be interested in using it. Right now we don’t have any product with Intel. I think they’re still on this journey to become a foundry, to have a technology that is competitive with TSMC and Samsung and when they do, we’ll be interested in considering them as well as we consider TSMC and Samsung, for example, for the leading node.
I think the trick is making sure customers care about what is in their phones, down to the name on the processor, but ideally not to the manufacturer of the processor. Differentiating, “Oh, that’s a TSMC chip, that’s a Samsung chip.”
One final question, maybe just to wrap up why you’re on with me, I think it probably sort of ties into this. One of the big issues with Qualcomm obviously has always been the relationship with Apple and it’s sort of this double-edged sword where on one hand it’s a huge customer, and on the other hand, everyone’s waiting for the guillotine to drop when they’re finally going to build their own modem, they’re going to move on. I think you’ve been very upfront, “We’re going to project as if they’re not there and it’s a pleasant surprise if they stay there, that’s sort of a good thing”. But big picture, do you think that Qualcomm — in contrast to Intel, everyone is waiting for good news, “Please, give us good news, we want to hear it” — it kind of feels like Qualcomm, everyone’s waiting for bad news, “When’s the bad thing going to happen?”. Does that feel like an accurate estimate of the way people think about Qualcomm and is that frustrating? Is that why you’re out here talking, wanting to turn around the narrative? “We’re not just a mobile company, we have these huge growth opportunities,” et cetera.
CA: I think that’s not a fair way to characterize it. I think what we did with the Apple relationship, we have taken every possible step to basically tell our investors that our planning assumption does not include Apple, so everything is upside. If you look, that’s actually the assumption that we make, and I think the industry is making. Talking about that relationship, look, we have a great relationship. I’m sure we’re a great supplier and we have, through our technology, have enabled our partner Apple to launch incredible products, and I believe from a supply chain standpoint, as well as our technology, I think we’re best in class supplier of modem technologies. At the end of the day, I think they had indicated the ability to go design their own modem. If they do, that’s not in our planning assumptions, if they continue to work with Qualcomm, we always going to be there to supply to Apple. That’s one part of the conversation.
The second part of the conversation, that’s the broader issue, and I think that’s why we’re being very active. It’s what we just talked about, everybody recognizes the Qualcomm position in mobile; they recognize the Qualcomm position on the technology standpoint, our contribution to standards. You should expect that Qualcomm’s going to be very big in 6G, when that happens, everybody recognizes that, but the market doesn’t grow. So because it doesn’t grow, our multiples are associated with the mobile market, and we are very active finding new home for technology, which is very efficient, very accretive to margins, so that we can create growth opportunities for investors and that’s the reason we’re spending so much time on this.
Well, I appreciate you spending the time with me. We actually went a little long, but I thought it was a very interesting conversation. And yeah, I agree with you, I think your position is quite compelling, I think the horizontal/vertical distinction is the accurate one and semiconductors — they’re very highly leverageable, so it makes a lot of sense.
CA: Very good. Pleasure talking to you, my friend.
This Daily Update Interview is also available as a podcast. To receive it in your podcast player, visit Stratechery.
The Daily Update is intended for a single recipient, but occasional forwarding is totally fine! If you would like to order multiple subscriptions for your team with a group discount (minimum 5), please contact me directly.
Thanks for being a supporter, and have a great day!