Hi, I'm Tim Arcuri. I'm the semiconductor analyst here at UBS. Very pleased to have Qualcomm, and we have Akash Palkhiwala with us, and Akash is the CFO. I'm gonna turn it over to Akash. He's gonna make some comments, and then we can start Q&A.
Great. Thanks, thanks for having me here. It's, it's great to be here, and I was just telling you, just incredible conference and the, the location's excellent as well, so, happy to be here. So, so maybe, for people who are not as familiar with Qualcomm, the core asset of the company, from our perspective, is our technology portfolio. So if you think about what it takes to succeed in phones, you have to be good at a lot of connectivity technologies, Wi-Fi, Bluetooth, 5G, position, location, and you have to be very good at processing technologies, CPU, GPU, AI, audio, video camera. And our core strategy is: how do we take this technology portfolio that we've built and bring it to other industries, whether it's automotive, whether it's personal computers, tablets, IoT devices
And so we're in this fortunate place where the core asset of the company is very strong, and it is very transferable to all kinds of edge devices. And that's how we're approaching the market. Now, as we look forward, we think our competitive advantage and differentiation of this technology portfolio actually continues to get better. It's. We have our own custom CPU that we just released in a personal computer chip that I hope we talk about later in the conversation. We're gonna bring that CPU to all other devices. So it's gonna go into phones, it's gonna go into automotive, it's going to go into other edge devices. And then, of course, Generative AI is, we see it as a massive trend that's coming up on edge devices.
We've obviously well documented all the success Generative AI has had on the cloud side, but just like computing exists at both places, you do computing in the cloud, there's computing on the device, we're gonna see the same happen with Generative AI on the edge. And so that's gonna be a tremendous opportunity for us. We're in a very strong position because we have a neural processing engine that operates at extremely low power, and we think that's gonna be an advantage going forward. So, well positioned in various areas because of our technology portfolio and very focused on diversifying outside handsets into automotive and IoT devices. A real pause there.
Perfect. Thank you, Akash. So let's talk about handsets first.
Okay.
Smartphone. It's still 75% of your operating profit. And you did revise this year up a bit, versus your prior. For 2024, you guided 5G units up high single digit, well, between, you know, high single to low double digit, next year. Can you talk about what's driving your slightly more optimistic view next year versus this year?
Yeah. So as we, as you know well, through fiscal 2023 for us, we went through two kind of cycles within handsets. First is the demand was weaker than the previous year, and then second is there was significant inventory in the channel. And so if you think about handsets, it's a little further ahead than what you're seeing in industrial companies today. So exiting our fiscal 2023, which is a September year-end for us, two positive things happening in the handset industry as the base. The demand environment has stabilized, and so that's, that puts us in a positive frame of mind going into fiscal 2024. The second is, the channel inventory within Android has normalized as well, so that's not a drag to our demand anymore.
And so with those two things, as we go into fiscal 2024, we feel like we're in a good place. Now, within the handset market, of course, there is still the transition of 4G going to 5G happening, which is beneficial for us because really, our target market is the 5G market, more so than the overall handset market, and we expect it to grow high single-digit to low double-digit next year as well. So that's the framework for the handset business in my mind. There's two trends to keep in mind when you think about handset revenue versus units going forward. First is, the content continues to go up very significantly. And, and if you look at the last three years or four years, the content per device that we make has increased very substantially.
And then, when you look forward, we think that trend continues. And we already know the next two or three chips we're doing, and we expect this 10%-ish increase in content at the minimum as we look forward. The second thing to keep in mind is, we're seeing a mix shift move up, right? Because what is happening is, when people are buying new handsets, you're looking for a device that is more capable than the last one. It continues to be an extremely important personal device globally. It is the content consumption device in emerging markets. People don't watch TV on the screens anymore, they watch TV like this. And so it has become the content consumption device and which drives a lot of additional capability requirements as well. So those are kinda two positive things that are happening.
And then maybe to finish up, as we look at 2024, we're very optimistic about Generative AI use cases coming into handsets. That's just w e know the OEMs are extremely focused on it. Of course, the operating system partners are very focused on it. We are putting in the hardware that is required to enable it as well. And so as that comes in, there's a content advantage, there's opportunity to gain share, but there's also an impact on the overall market. We think it could be a driver for another replacement cycle.
Yeah, actually, I've been a little surprised, I mean, certainly at this conference so far, at how bullish companies are about, you know, next year in the handset market. Maybe we get some launches at Mobile World Congress that begin to have an AI wave in the, you know, smartphone market. Relative to that 10%, you know, like for like within a tier ASP growth, do you think that this causes that number to accelerate, or does that 10% already include the fact that you're gonna see AI in smartphones?
Yeah. So when you think about a content increase of 10%, it's really. There's in each generation of phones, there's a different driver for it, right? Sometimes it's driven by the CPU becoming more powerful, sometimes it's driven by AI coming in, sometimes it's driven by the GPU and the gaming performance becoming more important. And so rather than talk about specific numbers and specific years, we think of it as a longer-term trend. And next year, certainly, Gen AI is gonna be a very large portion of it, and there could be upside on top of that number. But to me, it's much better to talk about the long-term trend rather than talk about a specific year, because that obviously is gonna be transitory. To me, the long-term trend is very positive for us in the business.
And maybe one thing that question that I get all the time is just the risk of market cannibalization from Huawei and how, you know, you and most Western companies don't have any content there. You know, there are numbers that get thrown around like, you know, 100 million units next year. That sounds to me to be quite aggressive. The phone seems to be pretty old technology. I don't know that it really is like for like gonna cannibalize a lot of your market, but can you talk about that as a factor that you kind of think about next year?
Yeah. For us, as where I started this conversation, the most important thing is the technology leadership. So if you look at the chip that we just announced, we're in 4-nanometer going to 3. So we are gonna have significant leadership and process technology. As I mentioned, we have CPU, our customized CPU coming in, our Gen AI technology coming in. If you look at our forward roadmap with what we have, with what our customers are launching in their phones, the use cases they're deploying, we just think that we're in a great position long term from a roadmap perspective. In the end, we've competed a lot with Huawei, with other players in the industry, and what has held true all the way through is technology advantage always wins in the long term.
We're very comfortable with our technology advantage. Our partners are launching great devices. There's reason to be optimistic from my perspective.
Fiscal Q1, you've already guided, but on, you know, fiscal Q2, you gave some sense of sort of where you see it. You know, typically the seasonal bias is down, especially this year, I would think, given a, you know, little earlier launch from a, from a large OEM. So how do you think about that, sort of what a typical fiscal Q2, you know, calendar Q1 would be versus some of the puts and takes that we've seen in December?
Yeah. So, on the call, I addressed this question and I gave kind of a shape of the year. We expect the first and the fourth fiscal quarters to be the high points in the year, and third fiscal quarter is typically our low quarter. So the shape is gonna be consistent with that. We haven't obviously provided specific guidance for the March quarter at this point, but I think the overall trend for the year hopefully was instructive.
Great. And we talked about AI in phones, but what about in PC?
Yeah.
That's something that, I think everyone's a little bit sleeping on what the opportunity could be in PC. Can you talk just a bit about that? I know, you know, you have your new chip coming out, next year. Talk about sort of what you see as the success factors there.
Yeah. So maybe I'll try to address it in two parts. First, just talk generally about Gen AI and why we think we're in a great position. And it applies to our PC chip, but it also applies to a phone chip as well. So what we have is an architecture that where Gen AI use cases can run on the CPU, it can run on the GPU, but it can also run on this NPU, which is our neural processing unit, that does Gen AI at extremely low power. So the new chip that we announced for PCs, we can handle up to 13 billion parameter models at up to 30 frames per second. So very, very powerful capability on the device.
When you think about Gen AI at the device, the advantage is in addition to cost, because cost, once you've bought the device, it's free to run Gen AI on it, versus obviously pretty expensive to run it in the cloud. But in addition to cost, you have immediacy. Latency is very quick, so you can run several use cases that you can't with the cloud. There's privacy and security concerns that would get addressed. There's personalization, because now the use case is about you rather than a generic use case. Then we think all these advantages is gonna make it extremely interesting to run it on the device. The next step is, and this applies to how the next generation PC is gonna work, is AI is gonna run pervasively on the device.
So regardless of what you're doing, several of you are looking at a PC and typing right now, there's AI constantly running in the background. And think of it as a smart assistant. Think of it as something that is always saying, "This person is doing this activity. Is there a way I can help this person?" And whether you're in a Word document, whether you're in PowerPoint, whether you're doing a chat session, say, you send a message to someone saying, "Why don't we plan to meet when I'm in Arizona?" An AI assistant can give you a reminder and say, "Do you want to send an invite over to this person?" Or, "You're available from 2:30 to 5," where you could meet with this person.
That's just one example of things that can pervasively run in the background. The AI assistant is always thinking, "How can I help? How can I help? How can I help?" When you're doing that, you need to do it at extremely low power, and that's where our NPU is going to be a very significant advantage. For people in the PC, especially, if you're going to run use cases on the CPU, on the GPU, it, one, takes away the bandwidth of the CPU. Rather than doing something else, it has to run AI. Second is, it's not power efficient. What we'll be able to do is do it in a very power-efficient manner.
Now, going to your second question, your main question on the PC side, the chip that we announced about a month ago is really a very modern architecture, bringing the strength of our phone platform to PCs. And it is based on an ARM CPU core. It is our customized ARM CPU. What we have talked about in terms of performance is it would exceed the M2 Pro performance, and will do it at 30% lower power. We will also exceed i9 performance, and we'll do it at 70% lower power. Just tremendous advantage when you think about what that platform is gonna bring to the table. And in addition to that, it will also bring the Gen AI strength that I just talked about.
So the way I think about our PC market is there's today's PC, and there's tomorrow's PC, which is the AI PC. We have the opportunity to be the lead in AI PCs. Of course, a very optimistic goal there, but we really do feel we have the right product to be able to do that. And of course, it's a massive market, very large silicon TAM, and you can pick your assumption on what portion of the market we can get. We think it becomes very material to us financially quickly.
You feel like you already have the partnerships in place? I mean, these are customers that you don't typically, you know, deal with in the past. You feel like you have the partnerships with those customers? I mean, you know, deal with them, but they're not your, you know, your core, yeah, yeah, handset customers. But you feel like they know what's coming, and they're excited enough that they're willing to forego x86 to buy your-
Well, I think the PC industry is looking for an increase in the performance metrics, independent of Gen AI, right? They obviously have a competitor who is delivering a strong product, and the PC industry is looking to match and exceed the performance, and we have the ability to give the chip that allows them to do that. So there is a desire to use our chip to compete. The second is the Gen AI inflection point hit just at the right time for us, and the technology advantage that we bring to the table, especially for pervasively running Gen AI, is going to be tremendously important as well. So I think because our technology is good, that makes the partners very interested, and they're obviously very familiar with our platform at this point. They're looking at it very carefully.
We have several design wins in the pipeline, and so as we kind of go towards mid this year and, PC launches for back to school, we'll be talking a lot more about our partnership.
So, back to AI in the phones. So you gave the example in PC, but in phones, for, you know, use cases, I know you had Stable Diffusion running as a proof of concept earlier this year. Can you talk about some interesting use cases for the phone-
Yeah
that people don't think about?
Yeah, for sure. And then I'll tell you a few use cases, but the fundamental point here is that there'll be several more, hundreds more, maybe thousands more, and the entire ecosystem is going to come up with use cases. But just to get a sense of what is possible, Stable Diffusion, I'll start there since you mentioned that. We could run a Stable Diffusion model at in 12 seconds earlier this year. It takes 12 seconds to run it, cloud or somewhere in that range as well. We just said that with our new chip, you can run it in under 1 second, 0.6 seconds. And what Stable Diffusion is, is you can write in text and get an image as the output.
The text could be, show Mickey Mouse standing in front of Eiffel Tower, and it will - the Stable Diffusion model will create original art based on the instruction you gave within 1 second, in less than 1 second. The second use case is, think of it as, there's a model called ControlNet. What it does is, it takes an image, you give a text input, and it edits the image based on the text input. So this is removing people from pictures or changing the background. Go back to the Eiffel Tower example, you could take your photograph and say, "Show me in front of the Eiffel Tower," and it'll do that for you.
A third use case is, if you're trying to draw anything with a finger, and say you want to draw the Golden Gate Bridge, the model will help you in improving your drawing. So you can start with making a few strokes, and it'll just keep b ecause it knows what it should be drawing, what you should be drawing, it'll add art to it automatically so that it becomes. It looks great. Lots of use cases in camera, just tremendous number of use cases that you could think of. You could take a picture and say you wanted to zoom out in the picture, but you didn't have the image of what was around it. Well, an AI model can make that up for you.
It will look like you can just take a picture and quickly zoom out of it, which you didn't capture it that way in the first place. All the assistant use cases are obviously very, very applicable to phones, just as they are to PCs. One of the new kind of products that was recently launched, the Humane device, the AI Pin. I don't know if anyone saw it, but it's an ex-Apple team that has launched a new device. It's a pin that goes on your clothes, and think of it as a wearable device that goes on your clothes. It uses our Snapdragon chip in it. And it is just a great way of saying you just talk to it, and the entire UI is based on AI. There are no apps to deal with.
So you could think that the phone interface could change in a way where you, you just use the AI as a way to, interact with the phone, rather than our current app-based ecosystem. So there's just a lot that can happen. I'm not saying every single one of these things are gonna happen, but there is several use cases, and I think our job is to, enable the hardware, enable the software, bring it to the developers so that their imagination can take hold.
Just from a margin point of view there, so the NPU, there's already an NPU core on the chip today. You're just using software to basically enable that?
So the NPU core that we've had for the last 15, 20 years, was serving a very different function, and so the size and the scale of what it could do was different. Now we have, one, optimized it for neural models, and then second, supersized it, to make it extremely capable, and then we're building, a tools ecosystem on top so that it becomes easy for people to use the NPU, external folks, not just our internal software development team.
Great. We had a question that I think is a good one from the audience, and I think everyone sees maybe some potential headwinds in the smartphone business. I asked about Huawei. Obviously, you know, Samsung's trying to bring back their own modem unit. There's, you know, that's a potential headwind. And how do you sort of think about how all those balance out? Because there are some tailwinds. Obviously, you think that you know, units will be actually better next year, but there are some potential holes that you have to fill as well. So do you still see those, you know, netting out to you being able to grow your handset business next year?
So I think the best way to think about the handset business is, versus 2023 going to 2024, there's two significant advantages that I started with. The end market has stabilized, and you don't have the inventory hang, channel inventory hang. So that is fundamentally puts you at a better starting point than what we had last year. I think there is a lot of conversation about Samsung. What we've been very clear about is that we expect to have majority share in the new phone launches. And as I said before, when you look beyond 2024, our roadmap advantage, in our minds, accelerates. And so we feel very comfortable that as you go forward, we are in a great position with them. And this is not something. This is a new competition for us, right?
Like, we've competed with Exynos for a very long period of time, and what we've learned over the years is technology always wins. So if you have the best technology, you're gonna be in a great place. That's how we're executing now.
Great. Then there was this second question, whether AI Compute will get separated and ring-fenced within the device in China, such that Chinese OEMs can pick and choose the accelerator alongside of the SoC from Qualcomm or MediaTek?
So if you think about the way we make our technology available to our OEMs, and this is globally, is we supply the hardware, we give the software tools so that they can make it do what they want it to do. And so we're enabling technology. We're obviously not doing the final use case. We don't make the final device. And so I think the enabling technology is still gonna be very relevant, and the OEMs are gonna decide how to use it. We don't think of and this is the history in phones. You don't have separate chips to do different things.
Mm-hmm.
In phones, it's always been one single SoC.
Mm-hmm.
That is the architecture that is almost not optional, given the things you're trying to do in a device, the size constraints you have, the power constraints, form factor constraints you have. So we're, we're very much a believer in, as you look forward, you're gonna end up with a single SoC model.
Great. Let's talk about another adjacency that I think is pretty exciting, which is autos. I think you have one of the best stories out there. You have a very customizable platform. OEMs want that. They don't want a, you know, black box, as some of your competitors have, you know, tried to push. Can you talk about key milestones there, and sort of, how big do you think that business can be over time?
Yeah. So we've given a pretty specific revenue forecast in the long term. So I'd kind of leave it at that. I think this is a business that can be at very large scale. And the good part about it, about the business is, as we look forward, we think we kind of started with components. We had the connectivity set of chips, then we had Digital Cockpit chips, then we had ADAS chips, we added ADAS software to it. But what is happening is it's increasingly becoming a platform. Because what OEMs want is quick time to market, especially the modern OEMs have a very different way of launching new cars.
When you go to one silicon and software provider, and you can develop a platform that goes from high-tier cars to low-tier cars, that goes from connectivity to compute to ADAS, and you can reuse your software across all these ecosystems, it just becomes a very interesting value proposition. Now, we've added a lot of capabilities to it. As an example, we have a cloud product that allows you to—the car can talk to the cloud, and then it can interface into, as an example, Salesforce system. So auto OEMs are very interested in having an ecosystem of applications that they use on a day-to-day basis, and we are enabling that for them as well.
So I just feel like this is a very strong business position for us. It is getting stronger. Component business is translating into a platform business, and so we're pretty excited. Now, AI, we—I know we talked about phones and PC, but you should think of it as something that applies to automotive in a big way as well, and especially when you think about use cases inside the car everything you could do on a phone or a PC is gonna happen inside the car as well. And then it's also something that applies to IoT devices, because we see a lot of edge devices in IoT are going to have AI capability as well. And so it's an advantage that translates to all these end markets we are pursuing.
We've talked mostly about your component business, but on the licensing side, do some of these developments potentially help your licensing business as well? I guess, in as much as they drive more content, that's gonna help your licensing revenue. Is that the answer?
Yeah, I think that's, that's the way to think about it. It also strengthens our IP portfolio, right? And so, obviously, having a stronger IP portfolio is extremely important for us from a licensing perspective. But we think of our licensing business as tied to wireless and wireless technologies. And then the other patents that we have around processing AI will become an important part of the overall portfolio as we go forward.
You really haven't, you know, you used to give us a royalty rate, you don't do that anymore. It's a bit more of a black box. Is it fair to say, though, that in these adjacent markets, that the early discussions in terms of the model, is it a similar licensing model to what it was in the handset markets?
You should think of as the framework of the licensing model really carries over to anything that's connected with 5G or 4G, any wireless technology. So, when we think about automotive, the same kind of framework of licensing applies.
So it's a rate times a, times a price?
Yeah, that's right. And then there are certain If you're doing a module, there is a fixed royalty rate around it as well. So this, again, same framework, we just applied to a new set of devices.
Got it. And then, can you talk a little bit about the European Commission is looking at, you know, standard essential patents, and is that a risk that we should think about over the longer term?
So there's always a debate on this topic, and it's gone around for several years. We are obviously tracking it very closely, but overall, I think if anything, there is a trend globally to protect IP for people who are doing innovation. Make sure that the IP has protection around it, and I think that's gonna continue to apply. That's a fundamental premise that we are seeing all countries embrace, and that's the thing that's most important to us in mind.
So, I don't ask about stocks too much in, you know, sessions like this, but yours has been a laggard this year. And when you meet with investors, what are the pieces to the story that you think are being underappreciated as you sort of look at your stock price, and it hasn't gone up as much as some of your peers have? What is being missed? I mean, obviously, we've talked, you know, most of the time about your very powerful IP portfolio. What's the aspect of the story that people miss when you look at your stock price?
Yeah, so I, I think, diversification is a priority for us, and it's a priority for the investors. And, and as o ne of the things that obviously investors would look for is proof points in, in success. And so I think we're on this journey. We've done a lot in automotive already. We are in a great place in XR metaverse devices when that market scales up. Personal computers, I think, you're gonna see us give more proof points on that. I think the broader IoT industrial market, especially with AI coming in, is gonna become interesting as well. So there's a lot of opportunity for us, and, and again, as you said, it's the relevance of the technology portfolio to those areas. And I think investors are looking for proof points, and we're, we're executing on it.
So maybe as a final question, let's actually talk about IoT. IoT is now a $6 billion business for you this year, so it's a big, big business. December guidance implies down about 20% year-over-year, which is really still quite a bit better than a lot of other IoT businesses out there. You're still, you know, $400 million-$500 million higher than you were before COVID. Can you sort of talk about why you're doing better than some of your peers are in IoT? Why your IoT business is, you know, so much bigger now than pre-COVID?
Yeah. So, the starting point is the same as what we've discussed. It's the relevance of the technology portfolio. But when you think about our IoT business, you almost have to think about it in parts, because there's, like, a lot inside it, right? So if you think of consumer devices, it's tablets, PCs, metaverse devices, and then wearable devices. And all these, all these devices, they're kinda like a phone, right? Like, they do a lot of the same functions, they need the same processing, they have displays as well. And so we are able to take our technology and apply it to all of them, and that's kind of the first part of IoT. We think, especially with PCs, we have a massive inflection point coming up, which could be very significant.
Of course, we have to go and make it happen, but we have the right things in place for it. The second part of the IoT business is industrial, and that's where we are at the very front end of the curve. We have a relatively small business. There are obviously other companies that have much larger position. But what is happening to those devices is they're all getting connected to the cloud. They are all needing incremental processing, and now all of them are going to end up doing GenAI. That puts us at a massive advantage, I think, because a lot of the other companies don't have those assets. We do. And so the technology assets put us in a great place there. And then the final part of IoT is edge networking.
For us, edge networking is the Wi-Fi access point business, where we are number one already, and then 5G as a broadband technology at operators, where we are also number one. So we're in a very strong position in my mind. There are a couple markets that are very large, especially, industrial and PCs, where we are at the front end of the curve, and we have the opportunity to make big businesses out of it.
Great! Well, we're out of time, but thank you, Akash.
My pleasure. Thank you.