Snapdragon Architecture Deep Dive 2025 (with Kedar Kondap)

neXt Curve had the privilege of being invited to the first Snapdragon Architecture Day, which took place in San Diego, CA, unveiling the second-generation Snapdragon X2 Elite and Snapdragon X2 Extreme AI PC processors.

Leonard Lee of neXt Curve had an opportunity to have a casual chat with Kedar Kondap, SVP & GM of Compute & Gaming, at the Qualcomm headquarters about the microarchitecture and core IP innovations of his latest SoCs, a new generation of Oryon CPU cores, and the power-efficiency that changed the PC game as we knew it.

Please subscribe to our podcast which will be featured on the neXt Curve YouTube Channel. Check out the audio version on BuzzSprout or find us on your favorite Podcast platform.

Also, subscribe to the neXt Curve research portal at www.next-curve.com and our Substack (https://substack.com/@nextcurve) for the tech and industry insights that matter.

NOTE: The transcript is AI-generated and will contain errors.

Transcript

Leonard Lee: 0:00

Hey Kadar, how's it going? Good. How are you Lyric? I'm really good everyone. This is Leonard Lee, executive analyst at Ncur, and I'm here in San Diego. The best city in the world, right? Wouldn't you agree? A hundred percent. Yeah. There's no debate. Yeah, there's absolutely no debate. But, this is really great because this is the first architecture deep dive, that's great day for Snapdragon. And this is really exciting because. We're talking about X two, right. Which you guys announced. Snapd Dragon X two Extreme. Yeah. SOCs were announced at Snapdragon Summit. That's right. So yeah, this is great to have an opportunity to talk to you about. These, pretty impressive chips that you guys have announced. you've, invited press and analysts from around the world and have opened up the kimono a bit, given us, much deeper insight into how you guys have engineered, these SOCs and getting some really. Impressive. Jen, over Jen, Right. Improvements in performance. So I say this on behalf of a lot of the analysts, fellow analysts who are tenants here. We really appreciate the opportunity to do these deep dives and get under the hood to understand how you guys are achieving. these generational improvements. And so I wanted to give you an opportunity to maybe give. The next curve audience, some insights on how you guys made that happen. the CPU, the GP, and the MPU, but maybe if you can, from a foundational perspective, help us understand how you guys are making these step changes.

Kedar Kondap: 1:38

Yeah.

Leonard Lee: 1:38

cause they

Kedar Kondap: 1:39

Leonard, first of all, thanks for joining us. Thanks for joining us at Summit. Thanks for joining us here at Architecture Day. Yeah, obviously for us it's an exciting time. We, started with this mission in, entering the PC space with, the objective to drive, innovation And bring excitement back into the PC space. Our intent is very simple. If you start with, the philosophy of the fact that we want to delight customers, the fact that we want to bring innovation, we feel like the PC just hasn't seen innovation in a very long time. we've been able to bring excitement in phones and we wanna bring that same level of excitement in windows. For us, I think the core, fundamental part of our design and philosophy is that. When you think of these devices as something that you carry around with you all the time, something that you wear all the time, power is obviously the most important thing. And so the way we bring innovation in is by focusing and starting our design philosophy with focusing on driving the best performance. At the lowest power. And I think what you probably heard, during the course of the architecture day through all of the engineers is there are decisions that are made in our entire platform. It's not just the core SOC, but the entire platform That are made, that are driven towards driving that experience. So as long as we're consumer centric, focused mm-hmm. And delivering the best experience mm-hmm. And you know that you want to drive that best experience, we'll always focus and bias ourselves towards technology. So whether it's the CP. Whether it's the graphic score, whether it's our audio core, the NPO, you name it. The core tenets of our design philosophy starts with, yeah. Best performance of the lowest power.

Leonard Lee: 3:13

Yeah. And one of the things I've noticed, this is just like a takeaway from today's sessions. Was from a micro architecture perspective, especially from a memory standpoint, you guys have made some significant improvements.

Kedar Kondap: 3:26

back to, you know what we want the end objective to be, right? So we focused, as you know, on three big vectors. The best performance, incredible battery life. And third was AI And when you think of AI and where the use cases are evolving, they're all very much tied to memory. Yeah. You obviously want to be close to, the platform. You wanna be able to go, run, time before token. Super important in terms of what matters for a good AI experience. Obviously DR bandwidth. So the reason why we announced the extreme is we had more DR bandwidth available'cause we know. A lot of the AI use cases that everybody's running tend to read more and more DDR R memory bandwidth. So a lot of, tenants go towards driving those experiences that are needed to get the best AI outcome. So we moved micro architecture back and you back into understanding what exactly is needed to go make this happen.

Leonard Lee: 4:15

You and I talked earlier before we hit the record button about the heterogeneity of. Computing supporting AI applications, right? Correct. I think right now there's such a fixation on large language models and that singularly being what is ai, right? But when you look at the entirety of how, AI applications are built, there's a lot of other stuff underneath that require and run more optimally on different ip. It's not necessarily one or the other. And that's one of the things that, you can. Really appreciating the designs and the design philosophy you guys have brought, right, to the, the PC space. Honestly, you guys. you know, people say game changing all the time, right? And sometimes they do it in a gratuitous way. You guys, honestly, with the X series, have changed the game, right? Thank you. You made the industry, especially the PC industry, rethink, what the PC experience, Should be. And have. Pretty much delivered on it. You know, AI for the PC seems to be much more than just these simple concepts that we associate with ai.

Kedar Kondap: 5:25

You know, I think it'll evolve very quickly.'cause AI is such a ubiquitous thing. It's gonna evolve for every use case. It's gonna evolve for every consumer differently. Yeah. It's gonna get deployed in commercial differently. And so our intent, again, the core philosophy is we wanna drive disruption with technology similar to what we've done in folks. And, you know, whether it's the camera, whether it's, the audio. The proximity of these features between a phone and a pc. It's incredible, Like we're still on video calls. You're still using the camera, you're still talking You wanna put display. so it brings a very natural progression in terms of, use cases on a phone versus the pc.

Leonard Lee: 5:58

Yeah. And you know what, it's also, interesting'cause we're also observing how some of the principles of mobile computing and architectures and design are making their way up to the data center. That's right. you need power efficiency, you know? Correct. that's a big deal. Correct. Right now. And you guys bring a lot of that DNA, across. different categories of computing and applications It's interesting to see how you guys are scaling everything, you know? I know that like when you introduced Orion, those are one of the things that you guys made a big point about is how it's scalable. It is correct. Especially as you make this move toward, data center, it'll be interesting for us to observe. Correct. How you guys continue to scale it across different, uh, correct. You know. Categories. Good. So yeah. Hey, thanks a lot. It was great. Thanks Leonard. I really appreciate you, uh, having thanks for joining us, having me, and it's an absolute pleasure, being here and to our next curve audience. thanks for, listening in. We'll see you next time. Thank you.