Yahoo Finance

Dan, let me ask you when that when that Nvidia print hits tomorrow, crosses the wire, I'm just curious what what are the the data points, the metrics in that report you make a beeline for?

Yeah, I mean the first thing people are going to be looking at is how that margin integrity's holding up. That margin number has been a big part of what's driven it over three trillion dollars. I think another big part of it is going to be any sort of communication about the Blackwell ramp. It's it's not totally clear how much that shipping has kind of hit capacity and are there any constraints in their in their black well ramp because I think there's kind of this mixed belief. Some people are like it's it's fully shipping, it's fully ramped. Other people are hearing there's some heating issues. What's true? And then, of course, I think everybody here will agree, the guide. Um, over the last four quarters, each quarter the guide has been a little less aggressive. They've been great, they've been big beats, and they've been raises, but that raise from like quarter to quarter has gotten narrower and narrower making it even more necessary, Josh, that it's perfect. The guide is perfect. I think the whole I think the whole market is on stilts. And I think if Nvidia comes up anywhere short, even a mediocre guide, I'm really worried about what's going to happen for growth.

I mean, last time the stock didn't do much. The time before that it fell about 6%. Both of those times the S&P didn't do much anything. So, you know, we'll see. I mean, the other thing we have to look for, it's not going to be in the statement likely, but we'll be in the conference call, is discussion around DeepC, right? And we talked to analyst John Venn of KeyBank yesterday, who's confident that the chips that are going to Deep DeepC, which I think are the H20s, he expects an uptick in demand for those. But I imagine that's a lower margin product than Blackwell, right?

And there's still a lot of uncertainty around H20. There's still a lot of uncertainty around what Trump is going to do with chip controls, and there's still a lot of question marks around the entire DeepC uh proposition and what happened there. Um, what we did learn for sure from DeepC is that the more efficient model will actually lead to more GPU demand. That we are I think at less than like .1% of the consumption of AI. Now the infrastructure build, we'll keep talking about that is happening, but when we learn how to build more efficient models and we learn how to make the infrastructure work more efficiently, I think that scale up is going to happen really quickly. And so DeepC we did learn a lot from, Julie, but I do have a lot of concerns that we still really don't know what happened there and that the actual infrastructure that we were told versus reality, I'm not certain. And then, of course, all the Trump uh sort of politics going on between us and China and what's going to happen there, I mean, no wonder that the broader market is super uncomfortable. And this is the most uncomfortable I've seen the market broadly about AI since the sort of first big boom around the ChatGPT launch.