00:00Yeah, I mean, straight to the outlook. And by the way, I would have done the same thing. Go right
00:04there. Revenue in the fiscal first quarter, $78 billion, plus or minus 2%. Consensus was really high. But going into
00:13this, it's a very simple equation.
00:15There were many names from across sell side and buy side that said they wanted to see NVIDIA beat revenues
00:22overall in the current period by the multiple billions of dollars against consensus. And so one thing I would say
00:29to confirm something is that the fiscal first quarter, which is the April quarter, does not assume any compute revenue
00:39from China in the outlook, does not factor in any compute revenue of China.
00:43That, I guess, is something to check off the list. It was one of the questions we had.
00:47I want to bring in Jay Goldberg from Seaport Research Partner. He's senior analyst for semiconductors and electronics. He joins
00:54us also from San Francisco.
00:56Jay, anything in here that confirms your thesis or does it sort of go against the sell rating that you
01:03have on NVIDIA?
01:04Yeah, I mean, it goes, I mean, it's a good quarter. It's a good quarter and a good guide. It's
01:08above consensus. It's above my expectations.
01:11But good for them. We're still waiting to see a lot more detail.
01:15What do you want to see?
01:16I want to see. I want to hear more. I have to talk about gross margin for next quarter. I
01:19think that's going to be in the CFO commentary.
01:22Yeah.
01:23Oh, Ed, jump in. Go ahead.
01:25No, I don't have it yet, but it will be in the CFO commentary.
01:28But 75.2% versus 75% expectations. I mean, can Ed, can you use that as sort of a
01:34guide for how the company could see first quarter?
01:37I mean, on a basis point, basis, yeah. Look, like margins are impacted by a number of things.
01:44It has been a sell side question on prior calls, the percentage of a server that NVIDIA contributes to.
01:51And again, what's changed in the period is NVIDIA starting to sell, for example, CPUs as standalone product.
02:00And that was a part of the meta deal that was included.
02:03They talked about CoreWeave being an early user of CPU as a standalone product.
02:08You know, how does that impact the margin profile going forward?
02:12But yeah, I mean, what they do is they own more of the server.
02:16And that has been one of the ways in which that they've got margins to where they are.
02:21And clearly being 0.2% above, you know, where the hopes were is not a bad thing.
02:27Yeah, I'm looking at that CFO commentary.
02:30As we know, revenue expected to be $78 billion, plus or minus 2%.
02:34Ed, as you correctly pointed out, not assuming any data center compute revenue from China in their outlook.
02:40Gap and non-gap gross margins expected to be 74.9% and 75% respectively, plus or minus 50
02:47basis point.
02:49Inclusive of a one-tenth of 1% impact from stock-based compensation expense.
02:54Just looking at what else we have here.
02:57Operating expense of gap and non-gap operating expenses to be approximately $7.7 billion, $7.5 billion respectively, including.
03:05That's inclusive of $1.9 billion of stock-based compensation expense.
03:09A headline getting my attention is NVIDIA says that hyperscalers were just over 50% of 4Q data center revenue.
03:16Jay Goldberg, I want to repeat that headline for you.
03:19Hyperscalers were just over 50% of 4Q data center revenue.
03:22Is that a sign that NVIDIA is diversifying revenue?
03:26Who made up the other half of data center revenue?
03:31Is it automakers?
03:32Like, who's buying?
03:33Well, I think that's one of the most interesting things that's taking place with NVIDIA is they're not just reshaping
03:38the semiconductor industry.
03:39They're trying to reshape the cloud computing industry as well.
03:42And that's why we see all these NeoClouds getting investment from or support from NVIDIA.
03:47NVIDIA would much rather have 100 NeoCloud customers than three hyperscale customers.
03:51Of course.
03:52And they're doing a really good job of creating that, willing that into existence.
03:57So those other customers, Ed, come on in here.
03:59Those other customers are like the CoreWeaves and the other NeoClouds?
04:03Yeah, I mean, what is still important and is fact is that obviously the vast majority has been the hyperscalers
04:12plus meta, right?
04:14And so they've wanted to kind of switch between a story, what they used to call an AI factory.
04:21At first, an AI factory was an on-prem data center that enterprises and maybe some of the NeoClouds would
04:27operate themselves.
04:29Now an AI factory is what you define any data center that's running AI workloads in.
04:35So they have chopped and changed between that story.
04:37But its origin was that they wanted to go beyond the hyperscalers to have people own their own infrastructure for
04:44AI workloads if they were, say, a software company or an enterprise of a slightly different size.
04:49So, let's go ahead.
Comments