00:00Let's start just with the outlook for the fiscal fourth quarter, around $65 billion.
00:05Depending on where you stood going into that, that is meaningfully above consensus to some.
00:12How do you react to that data point?
00:15Yeah, the data point is that the slope of the spend is continuing to move higher.
00:22And I think that that was a concern going into the print, especially after Jensen sort of let the cat out of the bag on how strong demand has been going with the $500 billion of backlogs that he talked about at GTC back in October.
00:39And I think expectations because of that had increased going into expectations for the results for the October quarter as well as the guidance.
00:48And so, you know, with the acceleration of revenue in both the October quarter versus July and then further acceleration into the January quarter, I think that took away some of the concerns.
01:02And then, you know, the commentary on the call, which I think was very important, was that they're obviously taking more orders.
01:08And so you just talked about this, which is that there could be, you know, not could be, there will be further upside to that $500 billion backlog as we think about 26.
01:18You know, one of the things, in addition to the slope increasing, is that the runway here is increasing.
01:24So we get a lot of questions from investors around sustainability of the spend.
01:29Obviously, the hyperscalers have been increasing CapEx on a quarterly basis for quite a while right now.
01:34But, you know, these types of data points would suggest that the runway on the spend and the demand for AI compute continues to, you know, lengthen.
01:44And so I think those are very strong data points for the AI trade, you know, looking forward.
01:51Ruben, just as you were speaking, NVIDIA's gain is now just nine tenths of a percent in a session.
01:56I'm not saying that's your fault.
01:57That's just curious timing.
02:00The base case assumption is zero revenues from China.
02:04You know, Jensen could not have been more clear about that.
02:06But does that not then give quite a lot of opportunity for some upside if they can get the China situation resolved?
02:14Yeah, I think that's right.
02:16I think, you know, at some point it is likely that we'll see, you know, companies like NVIDIA and potentially AMD and others figure out ways to work with both governments.
02:27The company talked about this on the call last night.
02:29But, you know, to your point, even without China and, you know, this came up on the last quarter, I mean, they had a beat and another raise without, you know, any H20 assumptions in the guidance.
02:42But, you know, even without that, you do have what I mentioned, which is the hyperscale of CapEx going up.
02:49And I think something that is still not fully appreciated is the rate at which sovereign governments are spending on AI.
02:56So, you know, we talk about the five major hyperscalers in the United States and the CapEx budgets that are growing there.
03:04But NVIDIA is doing, you know, quite a bit of growth in sovereign nation AI deployments.
03:10And I think that that's something that's going to continue as well.
03:12So even without China, obviously there's a big demand backdrop here.
03:16And like I said, that backdrop is extending into, you know, sort of a multi-year investment cycle.
03:22Jensen and the company more broadly really spelled out, again, they own 90 percent of the AI accelerator market or the data center market.
03:30Is that going to change? You just referenced AMD.
03:33Yeah. You know, I think, you know, at some point you will see, you know, just if for nothing else, you know, capacity that AMD, that's, I'm sorry, NVIDIA can bring to the table, you know,
03:45could enable others to participate here.
03:48The hyperscalers obviously have their own ASIC platforms that they're trying to bring to the market.
03:55But I think, you know, the way to think about this is that the overall TAM assumptions continue to rise, right?
04:01So a year ago or so, we were talking about multiple hundreds of billions of dollars in potential AI compute TAM.
04:07And now we're talking about trillions of dollars, right?
04:09So NVIDIA's talking about one to three trillion dollars.
04:11AMD had an analyst meeting just a few weeks ago where they talked about a million dollar, I'm sorry, a trillion dollar TAM as you get out to the end of the decade.
04:19And as that TAM increases, even if NVIDIA's share comes down, which it likely will, I think, you know, this concept of all boats rise, you know,
04:28they're going to continue to grow revenue in our view, you know, just on the overall TAM assumption increasing.
Be the first to comment