Snapdragon Summit 2026: Qualcomm’s AI Vision Is Bigger Than Your Phone

Personal AI, for me, for what I'm hearing from the folks at Qualcomm, they have a number of different devices from earbuds to pins to, of course, smartwatches and phones.

And really, it's about the technology they're enabling in each of those devices.

It depends, you know, like the user may have a preference for whether they be a watch and phone person or a pen and earbud person, you know, whatever your... specific mode of enabling a personal AI agent, they're working with OEM partners to enable that.

And then they also talked quite a bit about establishing really what we refer to, and I think what the industry is beginning to evolve around is this knowledge graph of the user.

So, your devices learn about you, they have the technology to be, you know, contextually aware, the environment you're in, the things you're doing, and then actually react.

You know, we're talking about agentic AI, right? What does it really mean? Well, agentic AI is an AI assistant that's going to not just, you know, answer your questions, but answer a question, react to your needs, and then go a step further, schedule appointments, book it in your calendar, get you the Uber, but really just go the full, I don't know, what was the AI that Tony Stark had? Jarvis.

Go the full Jarvis for you.

Trademarked.

Yeah.

Yeah, nothing to do with Qualcomm.

But yeah, I mean, that's what's interesting to me is that's what we're beginning to, we're learning about the user and then taking the steps to become fully agentic on your behalf.

Well, and you mentioned that knowledge map.

Marco, I wanted to, we heard from a couple companies this week that are trying to perfect that sharing of that knowledge map, or at least sharing information so you have the same experience across all these devices.

We heard from Aware, and we also heard from Liquid AI, which is coming out with their own kind of product, subproduct, called Liquid Context, to be able to share that information.

So what do you think about the possibility? And is this a solution, or do we still need kind of a MCP-type standard to actually get this done?

Tough call.

So the software is really going to be where the magic happens, right? We tend to focus on chips and accelerators and who's running what workload.

But to tie that all together, we need the right software stack.

We need the right tools.

We need the right connectivity to get that knowledge graph sort of transferred across all our devices seamlessly without having to think about it.

We don't want to press buttons.

We don't want to go through 10 different applications.

We want to work on the device we're working on.

Go to our, you know, if you're on your phone, get to your PC, everything's just there on the PC.

If you have a workload too heavy for that PC, it goes out to the cloud automatically, everything just gets done, right? So that magic's all gonna happen in software and connectivity.