This is the Daily Tech News for Monday, October 27th, 2025.

We tell you what you need to know. We follow up on context of stories and we help each other understand.

Today it seems like everybody's coming for NVIDIA.

Qualcomm enters the AI chip competition.

Ooh, the battle is heating up. I'm Tom Merritt.

And I'm Rob Dunwood.

Let's start with what you need to know with that big story.

Qualcomm announced it has developed a data center chip based on its hexagon neural processing unit.

That's the one it uses in its mobile chip.

So it's not like they haven't done an MPU before, but they are taking that work

and turning it into a data center chip designed for inferencing. Inferencing is when you execute a model.

That's what happens when you ask chat GPT something. It goes to the data center for inferencing

versus training, which is what they do to teach the model how to answer those questions.

Like systems from Nvidia and AMD that use GPUs, Qualcomm's accelerators will come in a system that uses a full liquid-cooled server rack.

So if you want to just get the card, put it in your data center, you can do that.

Qualcomm says it will also sell its chips separately for clients who wish to design their

own racks.

You may hear this called hyperscalers.

That's just talking about cloud providers mostly like AWS and stuff.

So if you want to build your own system and use some Qualcomm parts, you can do that too.

It also suggested Nvidia and AMD might become customers for some of its parts here.

Because, you know, why not? They'll sell to anybody.

...