On this episode of The Six Five – Insider Edition, hosts Daniel Newman and Patrick Moorhead welcome Qualcomm’s CEO, Cristiano Amon, during Snapdragon Summit for an inside look at Qualcomm’s vision for AI, and what it means for individuals and enterprises, as well as how the company is approaching developing Generative AI at the edge.
Their discussion covers:
- Qualcomm’s vision for AI, with its potential to change our lives, including how we learn, work, and play and what it will mean for each of us – as individuals, as enterprises
- How Qualcomm is approaching developing generative AI at the edge and why on device so critical
- How generative AI impacts how we engage with our vehicles
- Qualcomm’s R&D history in AI and their AI processing capabilities on-device for several edge applications including smartphones and PCs
- How Qualcomm positioned to maintain a leadership role in the AI space
Be sure to subscribe to The Six Five Webcast, so you never miss an episode.
You can watch the full video here:
Or Listen to the full audio here:
Disclaimer: The Six Five webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.
Patrick Moorhead: The Six Five is live here in Hawaii. We are at the Snapdragon Summit 2023. Gosh, Dan, this might be our fourth or fifth year. When we’re having this much fun, I just lose track. And we’re getting what we would’ve expected, a lot of conversation on new devices, new products, new platforms, but our favorite topic over the past nine or 10 months, AI.
Daniel Newman: Edge Computing. Oh. Well, we’ll get to that though, right?
Patrick Moorhead: Yes.
Daniel Newman: But, in serious, Pat, I’m going to go out on a limb, and I’m going to actually say, I think we got more than we expected. And, in this conversation, we’re going to dig into it, but we’ve gotten into this routine of going to events, and everyone being like, “Oh, my gosh. You’re going to be so excited.” And we’re like, “We go to 100 events a year.” Probably not, but I have to say, I tweeted a little bit… I X’ed a little bit more than normal, my LinkedIn feed has been blown up, I’m hearing a lot from the press, and I think this particular Snapdragon Summit may have been the most successful that I’ve been to, and that’s something to say.
Patrick Moorhead: Yeah, it was big. It was market expansion. It was doing something in markets that Qualcomm traditionally hadn’t done, and that was in PCs. But, hey, let’s get onto our discussion with our guest. Cristiano Amon, great to see you. Welcome back to The Six Five.
Cristiano Amon: Oh, I’m very happy to be here. I’m very happy to see you both here at the Snapdragon Summit. Hope to see you every year.
Patrick Moorhead: Yes.
Cristiano Amon: And I agree with the comment that was just made. I think this is a very important Snapdragon Summit for Qualcomm. If anything, when I say that Qualcomm is changing from a communications company to a connected processing company, I hope we deliver on that statement during this summit.
Daniel Newman: Yeah, I always come in with a bit of skepticism. I think, as Analysts, that’s our job. But I am always surprised and pleased when the message is better, bigger, more exciting, gives us something to talk about, keeps our phones ringing. I’ve been to the recent IAA event, and I saw some really great things in what you’re doing with automotive, and we’ll hit that today.
But Pat alluded to it, AI, I want to start there, because we’ve been talking about it for a while. We had you on at MWC. You really started talking about Edge and AI with us, showing us off the stable diffusion last year, but you’ve come so far in such a short period of time. Talk a little bit about that vision, how that vision’s evolving, and what you shared here today on your story around AI.
Cristiano Amon: No, very good. Look, this is actually one of the main topics of the conversation at Snapdragon Summit. We’re working very hard to get to the point that you want to have a Gen AI phone and a Gen AI mobile computing platform, and Gen AI is fundamentally changing user experience. Look, I am going to be a little bit like Qualcomm style and I’m going to say, we started talking about this before it was popular.
Daniel Newman: Before it was cool.
Cristiano Amon: If you remember, when we talk about 5G era, we said 5G and AI are going to come alongside and it’s going to blur the lines between the Cloud and the devices, and that’s what we started to see and were able to demonstrate, especially as you think about the new use cases, but I’m going to take it from the top.
What you have right now is the next big change in computing and it’s called accelerated computing. It’s the ability to do AI pervasively, to do that in a way that are like kernels running on a CPU. You’re going to have to that AI engine running and it’s going to start changing how we interact with devices. And one of the things we wanted to do at the summit is to spend time talking about what exactly happen on device. I think we all see and we’re incredibly excited about what’s happened on the Cloud, and AI started on the Cloud and it’s developing and evolving in the Cloud, but, at the same time, the AI is developing on device on its own way. And the two are meeting in the middle, when we talk about hybrid AI, where the device and the Cloud are working as one.
And it’s changing not only the user experience, but even a bold statement is changing the role of apps, especially when we think about phones and our computers, we think about an app-centric environment, but now the AI can really bridge the Cloud and the apps in a way that is completely transparent to you because it’s just your copilot, your system. And that’s one of the main things we talk about at this show. And I’ll say what is really cool is the things that we always knew about mobile devices, and now think about Qualcomm talking mobile devices, that includes the PC because we are in the PC now with some of the things we did.
But mobile devices, they have something which is unique. They have context, they have your favorites, they know about you, they know where you are, they know what’s relevant to you at that time, and that’s a more accurate input to a Gen AI model if you go from that device to the Cloud and give the Cloud a head start, or things that are just unique to you that is going to run on a device. And I think that’s the whole thing about we’re entering the Gen AI era, and the devices have an important role to play, and we’re working hard to enable that.
Patrick Moorhead: Cristiano, one thing I think you nailed today, which has been this question on Cloud versus on-device, but I have to ask you, because you’re investing a ton of resources into figuring out and lighting up experiences, can you talk about what this might mean to an enterprise user or even a consumer for on-device Generative AI?
Cristiano Amon: Yes. Why don’t we start with the consumer and we’ll try to show some of the examples as we’re trying to put this into what you’re going to visualize. Let’s go talk about a consumer and talk about a consumer on their phone. Things that you do every day, like you do every day, you text people. You and I, we talk on WhatsApp.
Patrick Moorhead: Too much. Yeah.
Cristiano Amon: And as you’re talking, you’re providing input. Think about this, Pervasive AI that is running all the time, as you’re talking, as you’re messaging, you’re providing input, and that input has some relevance. Think about now as we describe, this AI engine is running out of time and it has one job, it’s to try to predict what you are going to do as a human that interfaces with the machine. Yes, you text and you tell me, “Cristiano, I just met Dan here in Hawaii.” And the machine know that you took a picture with Dan, so as soon as you say that, you’re going to have a little tab for your gallery app with the picture already selected. And it’s going to ask “Pat, do you want to share that picture with Cristiano?” Then you say, “Look, we need to talk about those things. Why don’t we get together?” And I said, “I may come to San Diego in the first week of November.” When you say that, you get a little prompt that said, “Those are available times in your calendar. Do you want to send a meeting invite or Teams invite?”
Those are things that you do every day and that’s just going to get assisted. But now let’s change the conversation to photography, for example. You take a photo and you you change the background, think a model like ControlNet, image text, image, multimodal, change, share the photo. You do that many times a day. And then the last thing I’m going to give an example is phone calls. We do phone calls. You make a phone call, get the call summarized for you and tell you if you have any action that you agree. So those things are going to be things that are going to be running on device with the real context in real time.
And then on the other part of your question, which is commercial applications. Look, I go back to Microsoft, which has been an incredible partner and we’ve been on this journey together. Microsoft talk about $30 per month, per employee. I’ll say that is inexpensive. If you think about the productivity gains, if you’re looking about what is an hourly wage, that is-
Patrick Moorhead: A reaction. Exactly.
Cristiano Amon: … very inexpensive. And think about things like finding a file, think about getting a draft of your email, get a preliminary response, all the things you see with the copilot and now with the real context of your PC for the things that are relevant to you. So I think we’re just the beginning. As you said, this is moving very fast. What is fascinating, Pat, is if we had this conversation about 12 months ago, I’ll give you one example. We have hundreds of examples right now and it’s going to go to thousands of examples. So it’s very exciting.
Patrick Moorhead: Yeah. I’m feeling confident that we’re jelling in on some applications that almost everybody can relate to. And the irony about this agent, your first example, we were talking about that 30 years ago, but we’ve been unable to actually do it. And I believe the technology is here now to be able to sort this out. It’s not perfect yet, but it’s getting there where we can deliver those experiences. Exciting stuff.
Daniel Newman: Do you think it’ll do the calendar well enough? Now inside joke, right? But Pat’s very specific about how his calendar’s run. So it would learn, right? And eventually it’ll do it exactly the way you like. And I still won’t put it on my calendar.
Patrick Moorhead: I think your agent will “automagically” put it on your calendar.
Daniel Newman: I actually need it more than he does. But this stuff is really compelling and I think it’s a cross between reading your mind, knowing where you need to go, and then of course being able to keep your entire network engaged. I mean collaboration for all the technology and all the advancements we’ve had, there’s such an opportunity to make it better. And what’s on those devices that intelligence really is what’s been lacking, whether it’s your VA or an Assistant or it’s someone on your team, really kind of not only knowing what’s on the calendar, but what you’re thinking, who have you been talking to, who have you maybe got a scheduled thing with. This really works. But here’s the thing, Qualcomm is really well understood, Cristiano, for its success in phones, but I’ve spent some time, I know you have too, Pat, really looking at the business beyond the automotive and where you’re moving into PC. Generative AI can’t only be a phone thing for Qualcomm. Can you talk a little bit about how you see this expanding with the portfolio expansion?
Cristiano Amon: Absolutely. It is not a phone thing only. I think Gen AI is like this now general purpose technology that is going to be a part of every single device and that’s what we’re doing. We developed this engine to accelerate Gen AI and every device and we can see with clarity how this is going to make a difference. So we give examples of phones, we give examples on PCs, but I’m incredibly excited about what’s happening on the cars right now. So I’m going to make two comments on the car. The first comment is, when you’re behind the wheel, natural language communication in visual communication is incredibly natural. Think about when you think about large visual models, how many cameras now are in the car looking at things in real time. We have the information about you and you’re behind the wheel and you’re going to have a conversation with the car.
The second part is the car is really changing its role. And this, I know it’s a bold statement of what we’re going to say, but the car is becoming a new living space. You have the work. You have work, you have home, you have a car. And as the car becomes like this digital living space in your digital cockpit connected to the Cloud and the car has more autonomy, there are other things you’re going to be able to do in the car, and then Gen AI is perfect for that. I’m going to go from the conversation we just had about daily activities you do on a phone. And I’m going to transport that to some of the things we’re doing with our partners in automotive.
Think about something that simple. You leave your office, you’re going to go home. So there is a couple information that the car knows, how to get you home, it knows what time it is. It knows what you usually do. So for example, if you usually leave the office and you call your wife. So the car is going to say, “Dan, time to call your wife. I’m going to call your wife for you. Then you have to pick up groceries.” It already planned the stop. If you listen to the conversation and say, “We need to order,” it may even order something for you. So even something as simple as going from A to B with the knowledge where you are and the ability to have this system running is going to change the experience. And I think we’re just the beginning of this change. And what I like about this is across phones, across computers, across cars, it kind of blurs the difference between the Cloud device.
You may do things based on an app you have installed, you can go straight to the Cloud, it almost doesn’t matter, it’s completely transparent for you. But the key thing is you have this AI engine that is running. And in the case of what I like about it, I know that’s a long answer, but this is very, very interesting. You may have all of those constructs, right? You’ll have computing OSS applications, you have all those things, but then eventually you, as a human, you are seeing, you hear everything, you look at a screen. So the AI is looking at all those things, even though it’s your data’s private, the cameras in the car are looking at everything around it and then it can have access to everything. So it’s a completely new paradigm shift when you think about the role of computing and apps and how we are going to interact with those devices.
Patrick Moorhead: So Cristiano, we talked about smartphones, PCs, cars, and we even saw some hearables that are using more and more advanced AI, and then we have the Cloud. There’s a lot of players who want to win there and who were saying they’re going to win, and I’m curious, how do you view your position in all of this AI Ecosystem for on-device AI when it seems like everybody’s gunning for this? Maybe it’s a question about differentiation or winning your unfair share of those markets. And this isn’t inexpensive. I mean the work you did up there takes a lot of resources as well.
Cristiano Amon: Yes. I’ll say this on-device chain AI and the capability to do in a way that is going to deliver on all of those things we just talk about and all those possibilities and do it with the Cloud, that’s a Qualcomm size problem.
Patrick Moorhead: I heard that on 5G, which was, we really do really the best at the hardest problems.
Cristiano Amon: Yes. And it’s a combination of having the ability to invest at scale in some breakthrough technologies, plus the ability to push the boundaries of what’s possible with the ability to do that in a way that still works in a battery powered device. A great example of that is how we thought about the computing hardware required to make that as a reality. And that’s not only having the best CPU, we’ll talk about that in a little bit, but the GPU, but also a dedicated NPU. And what is unique about it is, the CPU could do many things and it’s going to do AI, the GPU can do many things, it can do AI, but the NPU has one role. It’s almost like when people said, “Why don’t you do this?” I remember when we had these conversations, when you started to do video, you could do video decode on the CPU. It’s just not a great idea. You end up building a video decoder because you need specialized hardware that’s all that it’s going to do at a very efficient power.
So the NPU is there that anything you’re going to run pervasive and it’s going to run all the time, and I think that’s a very unique differentiation of Qualcomm, and I’m going to give you the proof point. We used to show in February, we’re running an NPU stable diffusion. It’s a very common model, a large visualization model at 15 seconds. Some of our competitors were saying, “Oh, we can also do it in like 45 seconds. We showed today .6 seconds.” So I think that tells you that we have a very differentiated platform and it’s not a coincidence that you see some of the greatest companies in the world really partner with Qualcomm because if you are going to be a Gen AI leader, you can’t ignore the device. And if you look at a number of models that we have now natively running on the NPU and we show over there, I think Snapdragon is really becoming a platform of choice for you to bring your model to on-device AI.
Daniel Newman: It’s definitely an end, right? I mean we’ve got the Cloud to Edge and it’s going to be additive, Cristiano, it’s not-
Cristiano Amon: That’s right.
Daniel Newman: The Cloud has a role, but the Edge and the device has a really substantial role and we’re going to see this big training to inference shift. Right now a lot of the AI dollars have gone to training, but eventually it’s going to shift to inference, and inference does very well on the device. It’s very efficient. And so that’s where the story becomes very compelling. Now you teed this up, so I’m going to go ahead and wait. You don’t knock it out of the park right when you go, okay?
Cristiano Amon: Yeah.
Daniel Newman: But I’m going to hit this one long. So the CPU has been something really since the Nuvia acquisition, everybody’s been really excited and you’ve been doing this for a while. Both Patrick and I have come out. I think just yesterday I was on CNBC and I actually mentioned, I think Qualcomm’s got a pretty substantial lead on its arm variance because you’ve been at it six years plus seven now. But you made big announcements, you made big claims and of course Pat and I, we are Analysts, so we believe you, but we always trust but verify. I’d like to kind of hear, because these were, I mean outperforming Apple, outperforming x86 in really big ways, and you got on stage and we like to say we do the victory lap. There were a couple victory laps run up there. Talk a little bit about the advancements, talk about how this came to be and how do you maintain that because you know it’s a constant back and forth in this space.
Cristiano Amon: Yes. Look, one, so there’s a lot to unpack here, but let’s just go straight to the CPU. I think there’s an incredibly, I think, proud moment for the company, especially because we’ve been saying that as the company go into all of those other markets and we find new end markets for our technology, which really has always been about communications computing and now Artificial Intelligence. And we set ourselves on a mission and that came in particular when we made the decision that we fundamentally believe in the conversions of mobile and compute and we’re going to enter the PC space. And we’ve been on this journey with Microsoft that we knew would take several steps to transition the Windows that is consistent, to ARM, and we said we need to have the leadership position in this space because that’s the Qualcomm DNA. That’s why you see us doing mobile with Snapdragon every year. As we go into this space, we need to have the leadership position.
And we set ourselves to design what will be from the ground up, the most efficient and powerful CPU for any better power device. And that’s what we did. I think then we’re so happy with the acquisition of Nuvia. That’s an incredible team. I think we have some of the best CPO Designers in the world. I know that’s a big claim to make, but in the period that we’ve been to develop this and what we show, and I want you guys to go run the benchmark school, go run it. Some of the things we show over there, we did it after running like a hundred times.
And I’ll be honest, we put ourselves into it, we are going to create the Windows SoC that will be competing with the M-series. And I think the team exceeded our expectations. I’m not ashamed to say that, you show what we did, we exceeded the single threaded performance of the M2 Max. So we have the world’s fastest Arm CPU, not designed by ARM, and we have in the process become the first CPU. It was an i9-13980HX that you use in a gaming device. And we didn’t lose the power advantage, but that’s a typical Qualcomm thing. We have to reinvent ourselves every year. We have to have a new Snapdragon every year. So I think that’s just the first step in being in this market, pushing the boundaries of performance. And as I said, I want you two to make your travel plans for Snapdragon Summit 2024, because there’s going to be more CPU news coming at that time.
Patrick Moorhead: No, it’s super exciting and it turned out better than I expected and I’m excited to get a hold of devices. But I wanted to ask, I know our time is running out here, but I wanted to talk to you about the PC market and the way that you’re getting into it. We talked about the product and in many cases you have a 2x to a 10x advantage on certain parameters, which again, you’re not new to PCs, but this is like an absolute step up. It’s really a quantum leap increase here. How are you addressing the market differently with Oryon and the Platform X Elite versus what you did before? Maybe talk about just your aspirations for the PC market.
Cristiano Amon: Yeah, so I won’t make a share commitment, but I’ll-
Patrick Moorhead: Definitely not asking.
Cristiano Amon: No, I understand you’re not asking.
Patrick Moorhead: There’s no way you can answer that.
Cristiano Amon: But I’ll give you an answer.
Patrick Moorhead: You take one. Yeah.
Cristiano Amon: I’ll give you an answer. But I need to break this down into pieces. First of all, everything we have been doing in PC today, and we’ve been very careful also of not setting expectations in the past, because it was leading up to this moment, because it was not an easy journey for both us and Microsoft of an ecosystem switch from x86 to R. And I think for example, Windows 11, you now have the ability for the Microsoft emulator to run every application, we needed to get a number of new capabilities. We have a lot of native applications. So everything to date is in preparation for this moment where I think we now have the capability, the maturity of the ecosystem and the platform that can be very meaningful for Qualcomm. The second data point is, we’re super excited about the announcement of some of our other semiconductor companies saying they’re going to build ARM compatible CPU. It validates that the new SAM, that’s the new SAM for Qualcomm, which is really the mobile compute PC ecosystem, is available now with an ARM compatible instruction set.
And then that leads to the final part of the answer. What are we doing now and what’s different? So on consumer, it’s kind of a little bit easier to understand. I think it provides Microsoft with an Apple compete solution. It’s thin and light, it’s meeting the demands of higher performance at the same time with higher mobility, more battery life and a lot of different capabilities that you can do if you have this much processing capability on the CPU and the GPU. You can have a thin and light laptop that you can play games, entertainment, you can do creative. You look at all of the different things you can do for productivity, for creativity, et cetera, as a consumer. On the enterprise, there’s one interesting thing that changes, and I think that’s not because of the CPU, but because of AI. One thing that changes in the enterprise is the enterprise or the commercial naturally will move slower to the new ecosystem.
So we expect that it’s going to have a lag for enterprises to adopt this next generation PC in the same way that enterprises, for example, didn’t adopt Windows 11 right away. But there’s one thing that is very different right now, which is Windows AI. So if you look about the copilot, if you look about the amount of value that exists for Microsoft with the copilot, there’s an incredible incentive to accelerate that transition, and we’ll be able to ride the wave with all of the investments we did earlier. So I’ll say now without making any share projections, because we’re going to be careful about it, but I think as we get to the second half of calendar 24, if those devices are going to launch with any Windows, that’s going to start to be meaningful and we’re really excited about that.
Daniel Newman: Tell you what, I want to take one for a test drive.
Patrick Moorhead: No, I mean I might slip one in my bag on the way home, but hopefully nobody will notice.
Daniel Newman: Can’t confirm or deny that I haven’t already done that. Speaking of a test drive, I also want to ride in that i7 out front. You brought that beautiful BMW out here. You are going to take the Six Five guys for a spin casino?
Cristiano Amon: I don’t know. Absolutely. Let’s do it.
Patrick Moorhead: There we go.
Daniel Newman: I love it. Well, I want to just go ahead and say thanks so much for taking the time here. We know it’s a very busy schedule. It’s a busy couple of weeks with all the stuff going on, all the travel, and of course I think the earnings aren’t coming up too far away. I don’t want to ask him about that. But I’ll say something to you. But congratulations on all the announcements here at Snapdragon Summit. We can’t wait to check out those benchmarks and of course, watch the continued evolution of your products both on the handsets and the PCs.
Cristiano Amon: Well thank you so much. Great pleasure. I always love talking to both of you, and stay tuned for more.
Patrick Moorhead: We’ll see you in 2024. And sooner, of course.
Cristiano Amon: Yes.
Daniel Newman: All right, everybody. Hit that subscribe button. Stay with us here on the Six Five. We appreciate you tuning in. But for Patrick, for myself, for Hawaii, it’s time to say goodbye. We’ll see you all soon.