Sources · 3 min · Oct 8, 2026

Qualcomm sees energy efficiency as its data center opening

CA

Cristiano Amon

CEO, Qualcomm

Alex HeathHost · Founder & host, Sources
Left The Verge to start Sources in 2025; it has made over $1M since, mostly from sponsors.

About this segment

Amon traces Qualcomm’s move from mobile through PCs and cars to data centers. He says compute demand and energy constraints create room for efficient CPUs and accelerators.

From “Why Qualcomm is betting on AI data centers”, Oct 8, 2026. Watch the full episode

“our DNA is building chips, assuming there's a battery on the other side. That's our DNA.”

CACristiano Amonat 33:36 · Sources33:36

Takeaways

Qualcomm expanded into data centers by building on capabilities from other markets.
AI inference is splitting into more specialized workloads.
Energy efficiency is central to Qualcomm’s hardware strategy.

Watch 3:21 from 31:02

In this episode

31:02Qualcomm sees energy efficiency as its data center opening3:21
34:54Qualcomm wants an open software platform for AI chips4:19Up next
18:20AI smartphones will rely on agents and personal context3:29

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

31:09
a mobile dna we started in mobile we've been the most concentrated company in mobile uh we understand that we'll continue to do that and but there were several steps as we started to diversify the company we thought that the technology we had could be relevant and disrupt in other markets and we started to build computing assets and we went started to make the computer bigger there was a process you cannot burn any step you have to go to that process and we went into pcs and cars and it's interesting when you think about car some of their chips they run autonomy from both level 2++, level 3, level 4, and we have the silicon running this.
31:55
This is like a server. It's a server class processor for you to do those things. So we had to develop the capability. So it became a natural thing for us to expand the next step of the company to go to the data center. The second part of the answer is This data center thing is moving at a breakneck speed. And everything that people believe was true, it changed really fast. And I'll give an example.
32:27
In the very beginning, it was all about NVIDIA GPU. And even people want to talk about inferencing, a new hardware for inference, the answer was, you train on NVIDIA GPU when you're not doing training, you do inference on NVIDIA GPU. Only NVIDIA GPU. then inference start against scale. People said, oh, there's different solutions to do inference. There's more efficient. There are more better performance, lower TCO.
32:55
And if that wasn't true, NVIDIA will have nothing to do with Grok, okay? But now you keep going, and you said the compute demand is very high. I don't have a lot of energy. I need to find a specialized solution. So the inference started to break down in pre-fill and decode, and even the code is breaking down. So we look at this scenario, we say, the first pace, memory technology is gonna change, compute technology is gonna change, very pace.
33:24
In our business, it doesn't matter when you enter, if you have something innovative and disruptive, there's room for you. So that's the second tier. So we said, if we can do something unique, we're gonna have a market. Let's just go focus on doing something unique And then get to the layer number three, which is our DNA is building chips, assuming there's a battery on the other side. That's our DNA.
33:50
When I look at a data center, this is the amount of compute. This is the availability of energy. Yeah. So we see, oh, we can do something here. And that's kind of where we are. We focus on can we build energy efficient, high performance CPUs, high performance energy efficient accelerators. that's our space and it's just a natural evolution of Qualcomm. So it sounds like designing for the battery was a key insight for why you're doing this.