Connect your Claude, Codex or OpenClaw for full transcripts on millions of podcasts. Two week free trial, cancel anytime.

Start free trial →
DN Dan Nathan Dan Nathan On CNBC's "Fast Money"

“The data center guidance that we're getting, they may not get built, right? And we might end up having a lot of inventory. There's been a lot of talk of late about channel stuffing as it relates to GPUs.”

CNBC's "Fast Money" · October 2026

“The data center guidance that we're getting, they may not get built, right? And we might end up having a lot of inventory. There's been a lot of talk of late about channel stuffing as it relates to GPUs.” — Dan Nathan, CNBC's "Fast Money"

Context

Intel was falling on weak PC demand while talking up supply and demand out to 2030. Nathan argues long-range AI demand talk is shaky and that GPUs could pile up unused if planned data centers, many built on credit, don't get built.

Listen to the episode → More from Dan Nathan →

Transcript

CNBC's "Fast Money" Around 05:02 into the episode
Melissa Lee

…minutes in. CEO Lipu Tan saying third quarter demand, the environment has been more balanced than in previous quarters. So that's an interesting comment to say balanced without overheated or demand outstrips supply, you know, more balanced, implies we maybe reach this equilibrium. Yeah, I'll just

Dan Nathan

say this.

Speaker 4

…here, you know, I don't know if you guys looked at the calendar right now. I mean, we're getting to the end of 2026. I mean, that is a lifetime in a business that's very cyclical. So if you get to this point where, yes, we do see a pullback in demand. There has been at least enough of this hardware to be put in the servers that go into the data centers. And let me tell you something, people, the data center guidance that we're getting, they may not get built, right? And we might end up having a lot of inventory. There's been a lot of talk of late about channel stuffing as it relates to GPUs. If you buy these things, because they've been on tight allocation, but you can't build a data center if you're a NeoCloud, you know, and we know what's going on. They're being built on credit, right? So as you go down the stack, I mean, you have worse and worse companies, at least as far as their balance sheets are concerned, how to build this out. So I just think a lot of this now is getting to…

Melissa Lee

Mike, how do you see this quarter?

Michael Coe

Well, I mean, you know, it's interesting. I think this is kind of what Dan's talking about as well. I mean, looking out to 2030, anybody who has a rosy forecast, this is a business that didn't make any money at all last year. And it was losing almost 90 cents a share the year before that. And they…

Speaker 4

Intel inside. Everybody would rather have a Ryzen chip. So, you know, as far as I'm concerned, this thing was already trading more than 40 times forward before we got some news that isn't that great. It's not cheap. It wasn't cheap before these numbers came out. And even then at the discount we're…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

More from CNBC's "Fast Money"

“If we set aside Amazon stock and look at what Amazon said, this is probably the most bullish data point in AI over the past three or six months of all the data points that we get. … The fact that they went out two years, I can't remember when a company has given guidance out on a segment for two years. And I think that speaks to a confidence.” — Gene Munster, CNBC's "Fast Money" “They've been in the focal point of AI replacing them, but they are wired through all corporate systems. They're not a billboard. They're not a dashboard. They are integrated in every aspect of that corporation. … They're in the innards of the company, too hard to replace. I'd be a buyer.” — Steve Grasso, CNBC's "Fast Money"