Welcome to another Cheerious Cast. I'm here in wonderful Maui — eat your hearts out — at the Qualcomm Snapdragon Summit, and a lot of announcements this week. And I have the honor to be here with a friend, Ziad Asghar, who is the Senior Vice President and General Manager of XR, Wearables, and Personal AI. Now, first off, that means you're in charge of all the cool stuff.
So, I'm very excited that you took over this role.
Thank you. Thank you. [laughter] No, this has kind of been building up the right way, right? I mean, I've worked on AI for a long time at Qualcomm, and these are the devices that are going to carry AI to the consumer. So, it's like a perfect fit. So, I jumped at the chance of being able to do this.
Exactly. And, you know, we've been talking — we've talked to investors, we've talked to the industry, we talked to... on and on and on over the last couple of years. But you, you know, and we've seen AI evolve.
But it's not really about the technology anymore. It's about the experience. And I think that's, you know, agentic started bringing us into that AI experience, right? So, let's start off with that. What do you think this personal AI experience looks like?
Yeah. I mean, I usually say it this way, but I actually mean it, which is that these devices — these new what I call AI-native devices and all — should give you superpowers. What I mean by that is they better do use cases that the existing devices cannot do.
So, for example, I'll give this example. I was in China. I'm talking to one of the customers. They gave me glasses that are based on our technology, Snapdragon AR1. And basically the whole discussion is going on in Mandarin. And I can see the full translation coming up on my displays on my glasses, saying, 'Hey, speaker one, speaker two.' It's separating out who's saying what. And I mean, that's just an example of what you can do with this thing. We have another partner in San Diego, actually — the company's called Arya, I believe. And what they have done is they've taken the input from, for example, the Meta Ray-Ban glasses, and for a visually impaired person, they actually go and help them get out of a difficult situation if they're having trouble navigating out of a place, for example. Right? So the applications are really amazing things that you cannot do without these glasses or what we can do with the advent of these things. And the other angle that I think about always is that, look, always the new form factors have emerged with a new human-machine interface, and with the advent of agentic AI, if that's the new human-machine interface, what the device looks like — and that is what these new devices are. They are the ones that fit perfectly with the agentic AI sort of experience.
And you know that I'm very passionate about this. I keep telling people I don't want to be pulling out a phone, this little brick, in 10 years. I want it all connected. I want everything working. Matter of fact, you know, it's a five-and-a-half-hour flight over here from Phoenix, and I spent the entire time on my phone actually researching stuff and learning stuff, basically just interacting with ChatGPT.
Oh, wow. [laughter] Interesting.
So, it was like, 'Didn't you watch movies?' No, I actually had conversations going on — like multiple conversations going on.
Same here. I didn't watch a movie either. Yeah.
Yeah. So, it's like, and they're like blown away by that. But I am doing so much with that. But still my main interface is a PC and that smartphone. What do we have to do to change that? Because I want that. I want the smart glasses. I want the smartwatch. I want the devices around me to be the experience.
I think the way I'd put it, look, you said, you know, you still use the PC.
In the same way as this change happens and we move to these AI-native devices, I don't think the smartphone goes away, but the use cases might transfer from one device to another, just like we did with the PC space, right? We used to write a lot of emails on the PC; I send a lot of emails over my smartphone at this point in time. Now, what is that next step where more of these use cases move from the smartphone to these new devices? And I think that will happen. The other thing that I think is going to be very critical with these devices is it actually gives you context. So what these devices give you, they give you context, and that context is critical for agentic AI to give you that amazing experience. Which means now, if I were wearing my smart glasses that are right over there, they would know that, hey, I'm in a kind of like an office setting. I am, you know, having a video session with my friend over here. It can actually integrate that into the responses that it provides. My smartwatch knows my heart rate. It knows my stress level. It knows my activity level. Am I sitting? Am I walking? Am I running? All that context is a critical ingredient to be able to do that agentic experience in the best way possible. And I think that's what's missing today, right? And that's what we plan to bring. Each of this — and you know this for a long time — each of our solutions has a sensing hub inside it.
Right, it's just an example of only one of the technologies, but there are myriad such technologies. But the really amazing thing is, I also call this always a Qualcomm-sized problem. Why? Because now you've got to do all these use cases, but you've got to do it with a device that's one-twentieth the battery of a smartphone. And that's a problem that we love to solve here at Qualcomm. Well, and that context is a very unique aspect of this because it's not just what's around me, but it's knowing what I'm interested in. So, if something changes, you know, in the markets or, you know, in politics or whatever, it's going to let me know so I don't have to go searching for it.
That's right. So, you know, that creates, you know, and this is a question I get a lot, is, okay, well, if these devices are always on and always listening, isn't there a privacy issue? It's absolutely a valid concern from a lot of the folks. I think I usually talk about it this way: that with privacy, we want to be able to give all the power to the consumer to use exactly, or to give as much access as they want to give. So here is some of the technologies that we're working on to actually make this very seamless. To give credit to our current partners, they've done a good job, right? You take a picture, the light comes on, tells you that somebody's taking your picture. If somebody's recording, you have a blinking light that's telling them, hey, there's a recording that's being made. You may want to step aside. What I want to do is, as you know, within our sensing hubs, we also have this technology that we call always sensing camera. What we can do is, without you — before you store that image, can I just recognize that there is a face? Not identify, recognize, and I obfuscate it. Why I say that is because to be able to do exactly what we just talked about, which is these agentic experiences in the context, I don't need the faces. I need the context around the face. So if I can just do that, I can still get the full power of agentic AI, contextualized and personalized to my needs, but not have the privacy concern that people have, right? And I think that's exactly what we're driving towards with the technologies that we're building into our products.
That sounds very much like the HIPAA laws, where you can share information as long as it's not the personal information about the person.
Exactly. So we want to make sure that it's purely, you know, generic sort of information. It's not specific to any one person. And, you know, I mean, you know this also for the longest time: as we are able to do more of this work on the device, we're actually able to reduce that concern even further. So, you know, what I showed at the summit was our ability — pretty much the first platforms in the world that can actually take one-bit multimodal AI models and run them on the device. One bit.
Right? So that's what allows us to be able to bring these experiences to the device. You can just do the simple math, and now it takes a fourth of the space in memory. These devices are very memory-constrained, and that was a limitation. So we work with our partners to be able to have that use case, and it runs perfectly.
Well, and that's a key part of agentic AI, is being able to run multiple models so you don't have to have the big foundational model.
