WEBVTT - Nvidia CEO Jensen Huang Eases China Concerns, Touts AI Growth 0:00:02.560 --> 0:00:22.520 Bloomberg Audio Studios, podcasts, radio news. 0:00:17.520 --> 0:00:20.800 Live from San Francisco to our TV and radio audiences 0:00:21.040 --> 0:00:24.599 around the world. Welcome to a special edition of Bloomberg 0:00:24.680 --> 0:00:27.240 Technology imed Lovelo. In just a few moments in video, 0:00:27.320 --> 0:00:30.200 CEO Jensen Wang will join us for a live interview 0:00:30.200 --> 0:00:34.240 following the company's latest earnings report in video, posting revenue 0:00:34.240 --> 0:00:36.680 of forty four billion dollars for the fiscal first quarter, 0:00:37.159 --> 0:00:40.080 even as a slowdown in China weighed on results in 0:00:40.200 --> 0:00:44.159 video gave a positive outlook. I want to bring in 0:00:44.200 --> 0:00:47.599 Bloomberg's Ian King, who leads our coverage of semiconductors here 0:00:47.600 --> 0:00:50.400 at Bloomberg. And there is some math to be done, 0:00:50.520 --> 0:00:53.800 because the outlook for sales in the fiscal second quarter 0:00:53.920 --> 0:00:56.680 was forty five billion dollars plus or minus two percent. 0:00:57.160 --> 0:00:59.640 But the big disclosure was that there was eight billion 0:00:59.680 --> 0:01:03.720 dollars in revenue lost because of the restrictions on Age twenty, 0:01:04.160 --> 0:01:07.120 the China specific chip. What else do we need to 0:01:07.160 --> 0:01:08.120 know apart from that? 0:01:08.360 --> 0:01:10.160 Yeah, I mean the high level is here is. In 0:01:10.200 --> 0:01:13.280 some respects it was worse than people had feared, but 0:01:13.400 --> 0:01:16.520 in others it was much better than they had hoped. 0:01:17.319 --> 0:01:21.400 The China blow was arguably much worse than some analysts 0:01:21.400 --> 0:01:23.440 have projected. They were saying, you know, at the worst end, 0:01:23.480 --> 0:01:26.600 we're going to see seven billion dollars of revenue go away. 0:01:26.640 --> 0:01:29.400 Here in video said eight. But guess what still in 0:01:29.440 --> 0:01:32.720 line or that there or thereabouts with where consensus was, 0:01:32.800 --> 0:01:35.680 which means other parts of our business are doing well. 0:01:35.720 --> 0:01:37.640 There's black Well chip that they're able to ship to 0:01:37.720 --> 0:01:41.240 other places. Obviously they're getting supply, and that had, you know, 0:01:41.360 --> 0:01:44.400 kind of almost made up for some of the losses 0:01:44.400 --> 0:01:46.440 that we saw from China. 0:01:46.520 --> 0:01:50.320 There was a number fifty billion dollars of total addressable 0:01:50.360 --> 0:01:52.640 market in China loss. That's the way they've framed it. 0:01:52.680 --> 0:01:55.720 They've said that before. It was away from the numbers 0:01:55.800 --> 0:01:58.880 where the news was. I felt Jensen Wong opened the 0:01:58.920 --> 0:02:04.960 earnings call his portion addressing China head on that essentially 0:02:05.160 --> 0:02:08.160 that door is basically closed to them now in doing 0:02:08.160 --> 0:02:08.799 business there. 0:02:08.840 --> 0:02:10.600 What did he say, Yeah, I mean, this was in 0:02:10.639 --> 0:02:13.400 line with what he's been doing recently, where he's basically 0:02:13.440 --> 0:02:16.680 decided there's no point in messing around here, there's no 0:02:16.720 --> 0:02:19.640 point in working behind the scenes. I'm just going to 0:02:19.680 --> 0:02:23.519 go out there and lobby publicly. What he's telling the 0:02:23.880 --> 0:02:27.600 Trump administration in particular is, look, we if we lose China, 0:02:27.639 --> 0:02:29.760 if we're not allowed to play in China, we're going 0:02:29.840 --> 0:02:34.000 to be at risk competitively worldwide. It's not just a revenue, 0:02:34.240 --> 0:02:37.560 it's that because we're not in China, the Chinese companies 0:02:37.560 --> 0:02:39.480 are going to start to do it themselves. Thinking that 0:02:39.600 --> 0:02:43.040 China cannot do AI is wrong. You need to let 0:02:43.120 --> 0:02:45.960 us play there. That was his main message. 0:02:46.000 --> 0:02:48.880 He said that China will carry on with AI with 0:02:49.320 --> 0:02:52.720 or without Nvidia. In US newbergzine King, thank you very much. 0:02:52.760 --> 0:02:57.960 Let's get the investors reaction. Aoko Yoshioka, portfolio consulting director 0:02:58.400 --> 0:03:01.960 at Wealth Enhancement Group, joins us here on Bloomberg Technology. 0:03:02.560 --> 0:03:05.240 You listen to the call. Your main takeaways please. 0:03:06.480 --> 0:03:09.840 Sure, And you know, I think what Jensen said in 0:03:09.960 --> 0:03:14.040 terms of the AI era is here. I think that's 0:03:14.080 --> 0:03:17.400 the long term takeaway. You know, we didn't get answers 0:03:17.480 --> 0:03:21.840 on China. However, we know that demand is very strong 0:03:21.880 --> 0:03:24.280 and it's going to be strong for many years. There's 0:03:24.320 --> 0:03:27.840 still a lot of catalysts at Nvidia that's going to 0:03:27.840 --> 0:03:30.040 propel the stock for quite some time. 0:03:31.960 --> 0:03:37.720 I asked you on the show this morning. Well, the 0:03:37.720 --> 0:03:41.200 biggest question was we agreed it was China. Were all 0:03:41.240 --> 0:03:44.280 of your answers given and what you were looking for 0:03:44.440 --> 0:03:49.080 to understand the future of Nvidia's business in that market. 0:03:50.320 --> 0:03:53.880 No, I wish we had you know more details, but 0:03:54.080 --> 0:03:57.800 I think you know, Jensen and Collect gave as much 0:03:57.840 --> 0:04:01.720 as they could in regards to what is currently the 0:04:01.800 --> 0:04:05.280 situation with China, So there's not a whole lot we 0:04:05.320 --> 0:04:06.840 can do about it. I think we're all just going 0:04:06.880 --> 0:04:08.880 to have to wait and see whether or not they 0:04:08.920 --> 0:04:13.320 can come up with a cheaper Blackwell. Perhaps there's other 0:04:13.400 --> 0:04:16.520 ways that they can sell into the China market, but 0:04:16.600 --> 0:04:19.920 for now they're effectively closed to China, as they said 0:04:20.000 --> 0:04:22.560 during the call, So you know that's going to have 0:04:22.640 --> 0:04:27.240 to wait for more engineering and perhaps more news out 0:04:27.240 --> 0:04:28.080 of the administration. 0:04:28.920 --> 0:04:31.560 I want to bring Bloombergs in King back into the conversation. 0:04:31.600 --> 0:04:36.120 Ioko help us understand what Jensen Huang explained on the call. 0:04:36.200 --> 0:04:38.520 That they have a product or a category of products 0:04:38.560 --> 0:04:41.839 called Hopper, and as far as I understand the explanation, 0:04:43.040 --> 0:04:46.760 taking into account current US technology export controls, they have 0:04:46.960 --> 0:04:51.240 engineered that product, that chip to its lowest possible level 0:04:51.240 --> 0:04:56.040 of performance. What happened after that, Jensen essentially said it 0:04:56.080 --> 0:04:57.920 cannot get any less performant than that. 0:04:58.080 --> 0:05:00.800 Yeah, I mean the way to under standard is to 0:05:00.839 --> 0:05:04.159 go back. Hopper is a fundamental design that was the 0:05:04.200 --> 0:05:09.240 most important chip inarguably in this AI revolution. Rules came in, 0:05:09.279 --> 0:05:11.800 and these rules aren't like, you can't send this chip there. 0:05:11.880 --> 0:05:14.719 What they what they do is they set parameters, whether 0:05:14.800 --> 0:05:18.440 it's speed or memory or interface, all of these technical 0:05:18.480 --> 0:05:21.479 parameters get set in place and say you cannot have 0:05:21.560 --> 0:05:24.720 a chip that goes beyond these capabilities and send that 0:05:24.760 --> 0:05:27.760 to China. Page twenty was then a reaction to that, Hey, 0:05:27.800 --> 0:05:31.200 we'll make a chip that doesn't trigger these concerns, doesn't 0:05:31.200 --> 0:05:33.840 trigger these rules, and we'll send that. And then what's 0:05:33.839 --> 0:05:36.719 happened is basically the goalposts have been shifted again on 0:05:36.800 --> 0:05:39.039 in Vidia. So now they're at a point where we 0:05:39.240 --> 0:05:42.200 can't make this any worse and still have anything worth 0:05:42.279 --> 0:05:44.240 selling AOKO. 0:05:44.920 --> 0:05:47.680 The outlook for the fiscal second quarter is sales of 0:05:47.760 --> 0:05:51.560 forty five billion US dollars plus or minus two percent. 0:05:52.120 --> 0:05:54.880 But at the same time, in Nvidia told us that 0:05:54.920 --> 0:05:59.320 there was eight billion dollars in lost revenue opportunity relating 0:05:59.360 --> 0:06:04.080 to China, and to me, that seems like somewhere in 0:06:04.120 --> 0:06:08.760 their business in some geography, they found more demand and 0:06:08.800 --> 0:06:12.400 they sold more of something than they had initially anticipated. 0:06:12.440 --> 0:06:13.880 Have you been able to work that out? 0:06:15.600 --> 0:06:18.520 Absolutely? I mean, I think consensus numbers were closer to 0:06:18.640 --> 0:06:21.640 forty seven forty eight billion for two Q prior to 0:06:22.440 --> 0:06:25.440 the export van and so that the if you add 0:06:25.480 --> 0:06:29.000 back that eight billion that they said they lost for 0:06:29.200 --> 0:06:32.240 two Q, they'd be closer to fifty three billion. And 0:06:32.320 --> 0:06:36.000 so you know, the demand was very, very strong, and 0:06:36.040 --> 0:06:38.440 I think that continues, and I think we also got 0:06:38.520 --> 0:06:41.960 great information about margins in the second half. So really 0:06:42.000 --> 0:06:44.719 strong numbers and solid guidance out of Nvidia. 0:06:46.520 --> 0:06:48.560 Very shortly, Okay, we're going to speak to Jensen Wong 0:06:48.640 --> 0:06:49.800 live on this program. 0:06:50.160 --> 0:06:52.200 What's your question for Nvidia's CEO? 0:06:55.760 --> 0:07:00.440 Now, I think in terms of AI, just how things 0:07:00.560 --> 0:07:03.400 play out, especially on the enterprise. I think one of 0:07:03.400 --> 0:07:06.960 the things that he talked about is that enterprise AI, 0:07:07.760 --> 0:07:09.920 you know, is going to have to be back on prem. 0:07:10.240 --> 0:07:13.160 It's a little full circle for me when everything used 0:07:13.160 --> 0:07:15.640 to be on prem and we moved everything to the cloud, 0:07:15.920 --> 0:07:19.640 and now we're bringing back parts of it onto the 0:07:19.680 --> 0:07:24.560 premises as we need to protect those enterprise specific data. 0:07:24.880 --> 0:07:27.840 So I would love to understand how he views all 0:07:27.880 --> 0:07:29.480 of that playing out going forward. 0:07:30.640 --> 0:07:33.000 Okay, okay, a showcase, stay with us for just a moment. 0:07:33.080 --> 0:07:35.360 I'm looking at some of the after hours performance and 0:07:35.440 --> 0:07:39.040 video shares up more than four percent after ours highs. 0:07:39.320 --> 0:07:41.640 But then there are the other names, the other chip makers, 0:07:41.720 --> 0:07:46.880 those server assembly firms, super Micro, Dell. In the memory space, Micron, 0:07:47.000 --> 0:07:49.800 Qualcom is not as high as some of the others, Marvel, 0:07:49.840 --> 0:07:52.560 which got to shout out on the call from Nvidia, 0:07:52.880 --> 0:07:55.720 Blue Magazine King those after ours moves are interesting because 0:07:55.760 --> 0:07:59.280 basically the market is projecting a bit of confidence that 0:07:59.600 --> 0:08:02.000 all the picture stuff that Jensen's been talking about for 0:08:02.000 --> 0:08:05.200 some time appears intact, in other words, a commitment to 0:08:05.240 --> 0:08:08.760 continue investing in the infrastructure that supports AI. 0:08:09.160 --> 0:08:12.120 Yeah, I mean they want Uncle Jensen to tell everybody 0:08:12.120 --> 0:08:14.840 that everything's okay, and that's essentially what he tried to 0:08:14.880 --> 0:08:18.160 do today. But what is important to pick up on 0:08:18.240 --> 0:08:21.600 what our guest just said, Nvidia again said, guess what, 0:08:21.800 --> 0:08:25.000 these four or five big customers that are buying all 0:08:25.040 --> 0:08:28.720 of our gear, they're still fifty percent. So for the 0:08:28.760 --> 0:08:32.360 scenario that Jensen outlined to come true, which is that 0:08:32.640 --> 0:08:35.680 the enterprise is really using AI, that number has to 0:08:35.679 --> 0:08:39.559 go down. We have to see a large percentage go down. 0:08:39.920 --> 0:08:44.280 A Yoko Yoshoka, who's portfolio consulting director at Wealth Enhancement Group, 0:08:44.400 --> 0:08:48.040 we don't talk as often about Collect Cress, the CFO 0:08:48.200 --> 0:08:52.000 of Nvidia, but she spoke for the first twenty five 0:08:52.080 --> 0:08:55.319 minutes of that earning school, and throughout it she was 0:08:56.000 --> 0:08:59.800 going into great detail about all of Nvidia's customers away 0:08:59.800 --> 0:09:03.440 from the hyperscalers, all kinds of healthcare and industrial and 0:09:03.520 --> 0:09:09.040 robotics and consulting firms. But I don't quite understand that 0:09:09.160 --> 0:09:12.120 side of Nvidia's business, the software side, where they're not 0:09:12.240 --> 0:09:16.240 just selling GPUs to those that are building data centers. 0:09:15.880 --> 0:09:16.120 Do you. 0:09:17.920 --> 0:09:20.920 I think it's more the AI factories that they were 0:09:20.960 --> 0:09:25.120 talking about in conjunction with the software, you know. I 0:09:25.160 --> 0:09:28.559 think everybody needs to sort of pipe in and make 0:09:28.600 --> 0:09:31.520 sure they are getting AI in the manner in which 0:09:31.559 --> 0:09:34.040 they want to get it, whether it's through you know, 0:09:34.160 --> 0:09:38.040 the public cloud, which I think most enterprises are accessing 0:09:38.480 --> 0:09:41.440 some of that through that channel or other ways. And 0:09:41.480 --> 0:09:44.720 I think everybody's trying to explore how best to get 0:09:44.880 --> 0:09:47.319 the most out of AI from Nvidia. 0:09:48.960 --> 0:09:52.760 There was some discussion in King about recent announcements from 0:09:52.760 --> 0:09:55.480 the Middle East. Actually there wasn't a great lot of detail, 0:09:56.679 --> 0:09:59.640 particularly in how that relates to top line growth going forward. 0:10:00.080 --> 0:10:02.079 There was something that Jensen Wong said about what he's 0:10:02.120 --> 0:10:03.040 doing next week. 0:10:03.280 --> 0:10:04.240 He said, next. 0:10:04.040 --> 0:10:07.839 Week I will go to Europe, and in the context 0:10:07.880 --> 0:10:12.360 of further announcements, there's this expectation that Europe will start 0:10:12.400 --> 0:10:15.800 doing what others are doing, which is building out AI infrastructure. 0:10:16.280 --> 0:10:17.160 What did you make of. 0:10:17.120 --> 0:10:20.160 That and what was this sort of between the hints 0:10:20.200 --> 0:10:22.120 message I think that Jensen Wong was trying to share. 0:10:22.200 --> 0:10:24.319 Yeah, I mean the underlying message here is and this 0:10:24.360 --> 0:10:26.080 is a theme that he's been sort of forcing on 0:10:26.160 --> 0:10:28.280 us for a while, has been sovereign AI that a 0:10:28.280 --> 0:10:31.199 lot of companies, sorry, a lot of countries around the 0:10:31.240 --> 0:10:34.080 world want to keep their data within their borders and 0:10:34.120 --> 0:10:36.000 that to do the only way to do that is 0:10:36.040 --> 0:10:39.600 to build systems locally. We obviously saw the announcements in 0:10:39.600 --> 0:10:43.880 Saudi Arabia, in the United Arab Emirates, and so he 0:10:43.960 --> 0:10:46.880 got asked, Hey, when's this going to help You wouldn't 0:10:46.920 --> 0:10:49.240 answer that, but then he said, so will there be 0:10:49.240 --> 0:10:52.599 more announcements Jensen, and he was like smiling although we 0:10:52.600 --> 0:10:54.840 couldn't see him, and saying, I'm on my way to Europe. 0:10:54.960 --> 0:10:56.600 You can expect more announcements. 0:10:57.440 --> 0:10:59.880 Part of the question that the analysts pose to gen 0:11:00.080 --> 0:11:04.719 Some Wong was that analyst saying, historically, and remember, this 0:11:04.800 --> 0:11:08.040 company's been around a fair amount of time, there were 0:11:08.040 --> 0:11:11.760 moments on prior earnings calls where either Jensen Huang or 0:11:11.760 --> 0:11:17.079 collect Cress would give the streets something that was evidence 0:11:17.240 --> 0:11:21.120 that the big picture investment cycle was intact over a 0:11:21.240 --> 0:11:23.800 multi year horizon. Is how the analysts put it, that 0:11:23.880 --> 0:11:27.680 they could see what was happening would continue. What was 0:11:27.720 --> 0:11:29.760 your reaction to the response to that question? 0:11:31.559 --> 0:11:34.800 Sure, I mean, I think it was pretty positive in 0:11:34.800 --> 0:11:39.760 that their vision for how AI is going to impact 0:11:39.920 --> 0:11:45.760 every industry, every country is pretty spot on. It's been 0:11:45.800 --> 0:11:48.559 spot on. I think we're all trying to figure out 0:11:48.640 --> 0:11:52.559 how best to utilize it and what the parameters are. 0:11:52.600 --> 0:11:56.520 But it's impacting us in a true fundamental way, and 0:11:56.559 --> 0:11:59.560 I think that's something that NBDA has been talking about 0:11:59.600 --> 0:12:00.520 for quite some time. 0:12:02.440 --> 0:12:05.600 Jensen Wong talked about his relationship with President Trump, He 0:12:05.679 --> 0:12:08.480 said that he trusts the president, that the president has 0:12:08.520 --> 0:12:13.080 a plan and a vision. Whether you are on the 0:12:13.120 --> 0:12:15.760 cell side or the by side. When you listen to 0:12:15.800 --> 0:12:19.480 something like that, can you model for it the CEO 0:12:19.559 --> 0:12:23.040 of a technology company's relationship with the President of the 0:12:23.160 --> 0:12:25.760 United States and what that might mean for this company 0:12:25.960 --> 0:12:26.640 going forward. 0:12:28.360 --> 0:12:31.880 That's a really difficult thing to model for, especially when 0:12:32.320 --> 0:12:37.120 you know headlines can make changes pretty quickly. I think 0:12:37.160 --> 0:12:39.360 the one thing that you can rely on is at 0:12:39.440 --> 0:12:42.520 least you know, Jensen was talking about how manufacturing is 0:12:42.559 --> 0:12:47.240 going to be more high tech, more automated. We're not 0:12:47.360 --> 0:12:51.319 talking about, you know, low tech coming back, low manufacturing 0:12:51.400 --> 0:12:54.240 coming back to the United States. And I think that's 0:12:54.280 --> 0:12:56.880 the vision that we want to see going forward, especially 0:12:56.920 --> 0:12:57.840 in the United States. 0:12:59.800 --> 0:13:02.719 Just again the after hours trade and VideA higher four 0:13:02.800 --> 0:13:06.600 point three percent in late trading. Looking at other movers, 0:13:06.679 --> 0:13:11.079 super micro, Dell, some of the server assemblers also higher, 0:13:11.240 --> 0:13:13.920 in part because of that broader confidence of what this 0:13:14.040 --> 0:13:17.280 means for AI infrastructuring and King and video off. 0:13:17.320 --> 0:13:18.800 It's after ours highs though. 0:13:18.920 --> 0:13:20.600 You know, there was a moment in the earning school 0:13:20.720 --> 0:13:24.839 where people started to say, have we answered the China question. 0:13:24.920 --> 0:13:27.280 I think it's worth recapping the basics the outlook for 0:13:27.280 --> 0:13:30.320 the fiscal second quarter forty five billion dollars of revenue 0:13:30.360 --> 0:13:33.320 puss or minus two percent, but also this eight billion 0:13:33.400 --> 0:13:36.640 dollar figure of lost revenues because of the restrictions on 0:13:36.760 --> 0:13:38.800 h Chenney and China. What were the other numbers that 0:13:38.840 --> 0:13:39.640 we needed to look at. 0:13:39.760 --> 0:13:44.079 Yeah, I mean, as our guest mentioned, you know, sorry 0:13:44.559 --> 0:13:47.400 margins have been a concern, right, I mean, it's all 0:13:47.679 --> 0:13:50.880 a high class problem. But what the company was able 0:13:50.920 --> 0:13:53.440 to do today, which I'm sure help, was to say, look, 0:13:53.640 --> 0:13:55.880 despite all of the noise, despite all of the China, 0:13:55.960 --> 0:13:58.560 despite of everything else, we are still on course to 0:13:58.600 --> 0:14:03.040 get back to that five percent range later this year. 0:14:03.400 --> 0:14:05.880 And again, that is an astronomically high level for a 0:14:05.960 --> 0:14:10.120 company that is in this particular industry. And you know 0:14:10.280 --> 0:14:12.360 the fact that there were even concerns about it heading 0:14:12.360 --> 0:14:15.600 towards seventy percent is kind of arguably a joke anyway, 0:14:15.679 --> 0:14:19.320 But again, you know coll that Crest CFO was yet, Yeah, 0:14:19.320 --> 0:14:22.360 don't worry, We'll we'll get back there, and adding reassurance, 0:14:22.520 --> 0:14:26.200 adding back, you know, I think to that list of 0:14:26.400 --> 0:14:29.800 ticked off concerns that at Wall Street had about in Video. 0:14:30.240 --> 0:14:32.920 There is a longer term story about in Vidia's supply 0:14:33.040 --> 0:14:35.760 chain IOKO, and that's that it is like many others 0:14:35.800 --> 0:14:40.400 on shoring, they are working to assemble more of the 0:14:40.440 --> 0:14:43.880 server design, the complete package that goes into the data 0:14:43.920 --> 0:14:47.880 center here in America, insofar as in Video can control it. 0:14:48.080 --> 0:14:50.840 Jensen Wong spoke really positively about that. It's one of 0:14:50.840 --> 0:14:53.600 the good pieces of news. But the cost of doing 0:14:53.680 --> 0:14:56.640 business in this country is higher than other places in 0:14:56.680 --> 0:14:59.600 the world. How do you prepare for that as an investor? 0:15:00.080 --> 0:15:03.240 And when you think about videos quarterly earnings going forward? 0:15:05.080 --> 0:15:08.240 Sure, so, you know, I think he's especially called out 0:15:08.320 --> 0:15:12.800 a lot of partnerships too with TSMC, especially in Arizona 0:15:12.960 --> 0:15:15.440 and the build out that they're you know, doing out 0:15:15.440 --> 0:15:18.040 in Arizona. So I think you know that they're all 0:15:18.200 --> 0:15:20.880 talking about this and planning for this in a in 0:15:20.920 --> 0:15:25.280 a very methodical way, and you know, the costs are 0:15:25.280 --> 0:15:27.880 going to be the cost But I do think that again, 0:15:28.080 --> 0:15:31.320 we're bringing back higher tech, it's not going to be 0:15:31.840 --> 0:15:35.160 too low tech, and that is going to cost quite 0:15:35.160 --> 0:15:36.720 a bit and it's going to be difficult to do. 0:15:36.800 --> 0:15:40.320 But they're also generating just so much cash I mean 0:15:40.320 --> 0:15:44.280 to Ian's point. You know margins A's seventy five percent 0:15:44.520 --> 0:15:47.120 on You know, a company that's generating one hundred and 0:15:47.160 --> 0:15:51.360 twenty billion plus in revenues. You know, we're talking free 0:15:51.360 --> 0:15:55.480 cash flow, which is just astronomical for a company that 0:15:55.600 --> 0:15:58.760 was only generating twenty seven billion dollars in annual revenue 0:15:58.840 --> 0:15:59.720 just two years ago. 0:16:01.960 --> 0:16:06.400 Heyoka, Yoshioka, portfolio consulting director at Wealth Enhancement Group Bloomberg 0:16:06.480 --> 0:16:09.120 Z and King, who leads semiconductor coverage. That is the 0:16:09.240 --> 0:16:12.000 numbers you need to know and the reaction to Invideo's 0:16:12.000 --> 0:16:15.280 earnings report. Coming up, We'll continue our coverage of Video's 0:16:15.320 --> 0:16:19.000 earnings results and tune in for our conversation with CEO 0:16:19.320 --> 0:16:22.240 Jensen one that's coming up after a short break. This 0:16:22.280 --> 0:16:36.280 is Bloomberg Technology. Let's bring in in Video CEO Jensen one. 0:16:36.360 --> 0:16:40.200 Following the company's earnings and analysts call Jensen, thank you 0:16:40.240 --> 0:16:44.120 for your time here on Bloomberg Technology. You went into 0:16:44.120 --> 0:16:48.120 great detail about what's happening, particularly in China. I just 0:16:48.160 --> 0:16:50.280 wanted to go over something that I felt wasn't asked, 0:16:50.360 --> 0:16:54.480 which is fiscal second quarter sales forty five billion dollars 0:16:54.480 --> 0:16:58.400 plus or minus two percent eight billion dollars of lost 0:16:58.440 --> 0:17:03.120 revenue opportunity relating to China in age twenty. It seems 0:17:03.120 --> 0:17:06.119 as if you made up somewhere some demand for a 0:17:06.119 --> 0:17:08.840 different product, or from a different geography, or from a 0:17:08.880 --> 0:17:10.240 different corner of the market. 0:17:10.520 --> 0:17:11.200 What was that. 0:17:13.080 --> 0:17:13.760 Well, we got a. 0:17:13.720 --> 0:17:17.160 Whole bunch of engines firing right now. The biggest one, 0:17:17.160 --> 0:17:21.439 of course, is the reasoning AI inference. The demand is 0:17:21.520 --> 0:17:26.600 just off the charts. You see the popularity of all 0:17:26.640 --> 0:17:30.600 these AI services now, Chad, GBT, Gemini, you know, so 0:17:30.680 --> 0:17:32.680 on and so forth, GROC. I mean, they're just doing 0:17:32.680 --> 0:17:35.959 incredibly well across the board, and all of the APIs 0:17:36.000 --> 0:17:38.960 that they serve out and all the agentic AI services 0:17:39.000 --> 0:17:41.480 that are built on top of them, they're all doing 0:17:41.520 --> 0:17:44.760 incredibly well. I think that there's just a giant breakthrough 0:17:45.320 --> 0:17:51.080 in AI's capability, and inferencing has just become a giant workload. Second, 0:17:51.920 --> 0:17:54.360 people realize that Blackwell is just a home run Envy 0:17:54.440 --> 0:17:57.280 Link seventy two is a home run architecture. We designed 0:17:57.320 --> 0:18:00.399 it to be a thinking machine, a reasoning AIS system, 0:18:00.760 --> 0:18:05.000 and I think people now the confluence of the breakthrough 0:18:05.040 --> 0:18:08.879 and reasoning AI and the availability you know, now the 0:18:08.920 --> 0:18:12.520 emergence of Grace Blackwell and Velink seventy two perfect timing. 0:18:12.800 --> 0:18:16.120 I think that that's at the core a big part 0:18:16.119 --> 0:18:16.320 of it. 0:18:16.320 --> 0:18:18.159 And the second part of it is that our supply 0:18:18.240 --> 0:18:20.600 chain is growing and we're really ramping it up, and 0:18:20.640 --> 0:18:23.840 they're doing fantastically for us, and so all of these 0:18:23.880 --> 0:18:25.320 things are all coming together. 0:18:25.880 --> 0:18:27.720 And so it is fair to say that some of 0:18:27.760 --> 0:18:30.920 that additional supply on Blackwell and the demand for black 0:18:30.960 --> 0:18:34.520 Bell kind of made up for the opportunity lost in China, 0:18:34.560 --> 0:18:36.840 at least in the outlook for this current period. 0:18:38.560 --> 0:18:39.320 Yeah, I guess so. 0:18:39.560 --> 0:18:43.479 But you know, you just can't you can't underestimate the 0:18:43.520 --> 0:18:47.840 importance of the China market. This is the second largest 0:18:47.920 --> 0:18:52.080 AI market, This is the home of the world's largest 0:18:52.080 --> 0:18:56.159 population of AI researchers. And we want all of the 0:18:56.200 --> 0:18:58.960 world's AI researchers and all of the world's developers to 0:18:58.960 --> 0:19:02.920 be building on American stacks. And so, you know, irrespective 0:19:02.960 --> 0:19:07.359 of the nearer term revenue success that we have, we 0:19:07.400 --> 0:19:10.440 can't ignore the fact that the Chinese market is very important. 0:19:11.200 --> 0:19:14.760 You explained again in some detail that at least in 0:19:14.800 --> 0:19:18.520 the Hopper architecture, you have engineered down to the lowest 0:19:18.560 --> 0:19:22.639 spec possible. It's not possible to do anything different with 0:19:22.720 --> 0:19:27.119 Hopper for the Chinese market. But in your consideration for 0:19:27.400 --> 0:19:30.280 a different architecture or a chip for the Chinese market, 0:19:30.720 --> 0:19:33.159 is that what we're talking about, a new design or 0:19:33.200 --> 0:19:35.919 new class? And have you made that proposal to the 0:19:35.960 --> 0:19:39.200 administration just designing from the ground up a new chip. 0:19:40.080 --> 0:19:41.280 We're still thinking through that. 0:19:42.119 --> 0:19:47.719 They're just the limitations are quite quite stringent, quite limited, 0:19:47.760 --> 0:19:48.320 if you will. 0:19:49.200 --> 0:19:50.000 H twenty is. 0:19:52.080 --> 0:19:54.480 You know, as far down as we could take a Hopper, 0:19:54.920 --> 0:19:57.959 we don't know how to make it even less. And 0:19:58.000 --> 0:20:02.560 so that's really the limit. But so there aren't there aren't. 0:20:02.960 --> 0:20:05.840 You know, the limitations are quite stringent, so we have 0:20:05.960 --> 0:20:08.720 we have to really think through it. Whatever we make 0:20:09.080 --> 0:20:11.840 ultimately has to add value to the market. And so 0:20:11.920 --> 0:20:16.240 it's it's a really tight rope because because the Chinese 0:20:16.520 --> 0:20:20.200 competitors have evolved and advanced greatly over the last year, 0:20:20.760 --> 0:20:24.879 like everybody else, they're you know, doubling, quadrupling their capabilities 0:20:24.880 --> 0:20:28.600 every year, and the volume is increasing substantially. And remember 0:20:28.640 --> 0:20:30.919 these are data center chips. They don't have to be small. 0:20:31.280 --> 0:20:34.879 They could be quite large and and and you know, 0:20:34.920 --> 0:20:41.080 they without American technology, the availability of Chinese technology will 0:20:41.080 --> 0:20:44.040 fill the market. And so you know, we whatever we 0:20:44.119 --> 0:20:46.960 offer has to at least at least be competitive and 0:20:47.119 --> 0:20:48.439 has to add value to the market. 0:20:49.720 --> 0:20:50.120 Tensen. 0:20:50.200 --> 0:20:53.960 Does Huawei have an AI accelerator or a GPU that 0:20:54.119 --> 0:20:57.640 is performance is hah twenty or is performance of other 0:20:57.720 --> 0:20:59.320 classes of GPU that you. 0:20:59.320 --> 0:21:05.960 Make Huawei's technology based on based on our best understanding 0:21:06.040 --> 0:21:07.919 at the moment, uh, and we have a lot of 0:21:07.920 --> 0:21:12.240 ground truth there is probably comparable to an age two 0:21:12.400 --> 0:21:17.200 hundred and so they're they're they've been moving quite fast 0:21:17.240 --> 0:21:21.359 and and they've also offered uh this news system called 0:21:21.400 --> 0:21:25.679 cloud Matrix. UH that is that is scales up to 0:21:25.920 --> 0:21:29.520 even a larger system than our latest generation Grace Blackwell. 0:21:30.160 --> 0:21:32.760 And so so they're Huawei, as you know, is a 0:21:32.800 --> 0:21:35.800 formidable technology company, and they're not sitting still and they 0:21:36.160 --> 0:21:39.320 look for ways to look for ways to compete, and 0:21:40.160 --> 0:21:41.200 they're quite formidable. 0:21:43.280 --> 0:21:45.760 With that in mind, you've kind of explained the landscape 0:21:45.760 --> 0:21:48.480 now and Videa's ability to operate in China. I mean, 0:21:48.520 --> 0:21:50.600 you and I this year alone have already discussed the 0:21:50.640 --> 0:21:53.520 idea that they're they're fifty percent of the world's AI 0:21:53.640 --> 0:21:57.080 research is in or from China. But are you having 0:21:57.080 --> 0:22:00.640 a sense of the vacuum created that those big names 0:22:00.680 --> 0:22:03.159 I think about Tencent or Ali Baba or bayd that 0:22:03.200 --> 0:22:06.479 were buyers of age twenty that they've already pivoted and 0:22:06.560 --> 0:22:10.560 turned to the offering from Huawei because of the the 0:22:10.680 --> 0:22:11.879 policy that is in place. 0:22:13.600 --> 0:22:16.120 Yeah, they have no choice, but to you know, one 0:22:16.119 --> 0:22:18.959 of our one of the one of the challenges of 0:22:19.119 --> 0:22:23.440 the changing regulations is the ability for markets to trust 0:22:23.640 --> 0:22:29.040 the Nvidia and and ultimately American platforms. And so so 0:22:29.680 --> 0:22:34.359 it's prudent, I think for the Chinese customers UH to 0:22:34.520 --> 0:22:38.240 make sure that they they developed their stack on Huawei 0:22:38.880 --> 0:22:42.760 and because it's hard to rely on American technology at 0:22:42.760 --> 0:22:44.840 this point. And so that's one of the that's one 0:22:44.840 --> 0:22:51.360