WEBVTT - Nvidia’s Forecast Lifts AI Trade 0:00:02.560 --> 0:00:13.520 Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is alive 0:00:13.600 --> 0:00:17.400 from coast to coast with Caroline Hide in New York 0:00:17.680 --> 0:00:20.560 and v Lovelow in San Francisco. 0:00:22.720 --> 0:00:26.120 This is Bloomberg Tech coming up in Vidia's strong revenue forecast. 0:00:26.200 --> 0:00:30.440 SOUTHS investor concerns about an AI bubble. Our exclusive conversation 0:00:30.840 --> 0:00:33.040 with CEO Jensen one plus the. 0:00:32.960 --> 0:00:35.599 CEO of Palo Alto Networks joins us to discuss its 0:00:35.600 --> 0:00:39.160 own results and the latest acquisition to boost its AI offerings. 0:00:40.080 --> 0:00:43.440 An AI pioneer Yan Laccoon, will depart Meta at the 0:00:43.520 --> 0:00:45.440 end of the year to start a new company. 0:00:45.720 --> 0:00:46.920 Details later this hour. 0:00:47.440 --> 0:00:50.120 Now we focus in on the broader markets. We said 0:00:50.200 --> 0:00:53.760 yesterday so in video, so the market, So the market 0:00:53.760 --> 0:00:55.880 follows in Vidia higher, but we're at one point percent. 0:00:55.920 --> 0:00:58.440 Then that's that one hundred a breath of relief when 0:00:58.440 --> 0:01:01.360 it comes to AI anxiety instead, So it's a quelled, 0:01:01.360 --> 0:01:04.280 a bit ed, and not enough even to reshaking off 0:01:04.280 --> 0:01:06.320 some of the anxiety that comes with the jobs report 0:01:06.400 --> 0:01:09.800 that maybe maybe we could see a December rate cut. 0:01:09.920 --> 0:01:12.039 We built up a little bit the bond market pricing 0:01:12.040 --> 0:01:14.480 in now thirty five percent chance of a rate cup 0:01:14.520 --> 0:01:17.559 come December, but really all eyes on this AI trade 0:01:17.600 --> 0:01:19.160 today and all. 0:01:19.000 --> 0:01:19.720 Eyes on in video. 0:01:19.800 --> 0:01:22.679 The stock is up about three percent, had opened up 0:01:22.720 --> 0:01:25.600 about five percent and has paired some of that gain 0:01:25.680 --> 0:01:28.840 in the last half hour or so. Sixty five billion 0:01:28.880 --> 0:01:32.080 dollars of sales in the fiscal fourth quarter. Many see 0:01:32.080 --> 0:01:35.800 that as being healthily above consensus. There was a lot 0:01:35.840 --> 0:01:39.720 of fighting talk on the demand side, on the supply side, 0:01:40.080 --> 0:01:42.520 and how good in videous technology is. We spoke with 0:01:42.600 --> 0:01:46.880 Invidious CEO Jensen Wong in an exclusive interview post earnings. 0:01:47.120 --> 0:01:47.640 Take a listen. 0:01:49.560 --> 0:01:56.120 Sales are after charts for Blackwell and in video GPUs 0:01:56.120 --> 0:01:59.520 in the Cloud are sold out. We got plenty of 0:01:59.520 --> 0:02:03.080 Blackwells to sell you. We have lots of Blackwells coming, 0:02:03.840 --> 0:02:06.440 We're making a lot of Blackwells, and we have a 0:02:06.440 --> 0:02:11.919 bunch of Vera Rubens coming, and so business is very, 0:02:12.040 --> 0:02:16.440 very strong. But we've planned our supply chain incredibly well. 0:02:16.520 --> 0:02:19.280 We have the largest supply chain in the world. But 0:02:19.360 --> 0:02:21.040 we've got a bunch of black Wells to sell. 0:02:21.400 --> 0:02:24.560 Jensen I tried to go through what the CFO collect 0:02:24.560 --> 0:02:28.160 Cress said about China in the quarter gone. It seemed 0:02:28.200 --> 0:02:31.519 like there was not meaning for H twenty sales because 0:02:31.600 --> 0:02:34.360 the demand wasn't there even if you were permitted to 0:02:34.400 --> 0:02:37.000 sell EAH twenty and then in the current period and 0:02:37.040 --> 0:02:40.160 going forward in video, seems committed to working with both 0:02:40.200 --> 0:02:45.600 the United States and China to sell what COLLECT called 0:02:45.800 --> 0:02:49.760 more competitive compute. Where do we stand with that and 0:02:49.800 --> 0:02:53.080 could you just clarify what COLLECT was talking about in 0:02:53.120 --> 0:02:54.640 the current state of play for China. 0:02:56.040 --> 0:02:59.680 The most important thing she said is that we've said 0:02:59.720 --> 0:03:04.960 for time now our forecast for China is zero. All 0:03:05.000 --> 0:03:08.880 of our forecasts, that guidance that we showed zero, we 0:03:08.919 --> 0:03:11.600 should start. That's the most important thing that she said. 0:03:11.880 --> 0:03:14.960 She also said that effectively, China is a very important 0:03:14.960 --> 0:03:17.840 market to us. It's very important to the United States, 0:03:18.080 --> 0:03:21.760 It's very important to China. We would love the opportunity 0:03:22.120 --> 0:03:25.520 to be able to re engage the Chinese market with 0:03:25.639 --> 0:03:28.640 excellent products that we deliver and to be able to 0:03:28.680 --> 0:03:29.920 compete globally. 0:03:30.600 --> 0:03:34.239 During the call, the US Commerce Department issued a statement 0:03:34.440 --> 0:03:37.960 saying that you are now permitted to export up to 0:03:38.040 --> 0:03:41.720 thirty five thousand black Well chips each to both saudiast 0:03:41.840 --> 0:03:45.160 Humane and to the UAE through G forty two. But 0:03:45.240 --> 0:03:47.800 there are some requirements that the US has a view 0:03:48.560 --> 0:03:54.360 in particular around controls of preventing tech transfer to China 0:03:54.440 --> 0:03:56.400 through the Middle East. What can you tell us about 0:03:56.400 --> 0:03:59.200 your understanding of what the US government's asking of you there. 0:04:00.000 --> 0:04:04.440 That element has been around for a long time is 0:04:04.480 --> 0:04:08.360 to prevent diversion. Of course, over the years, people have 0:04:08.760 --> 0:04:14.360 speculated about diversion. We've chased down every single concern and 0:04:14.960 --> 0:04:20.480 we've repeatedly tested and sampled data centers around the world 0:04:20.880 --> 0:04:25.719 and found no diversion. And so this is an area 0:04:25.800 --> 0:04:29.839 that will continue to be rigorous on. And there's a 0:04:29.839 --> 0:04:34.080 lot of different ways to comply, and one of them, 0:04:34.240 --> 0:04:37.960 of course, is to have it be run by American cloud. 0:04:38.320 --> 0:04:41.000 Another way is just to make sure that we have 0:04:41.080 --> 0:04:45.120 measures put in place, whether technology or processes, to ensure 0:04:45.120 --> 0:04:46.239 that no diversion happens. 0:04:47.279 --> 0:04:50.960 In video CEO Jensen Huang speaking to us exclusively last night, 0:04:51.160 --> 0:04:53.880 let's dig in to that last bit from Jensen and 0:04:54.000 --> 0:04:56.960 dressing concerns around selling to them, at least selling into China. 0:04:57.240 --> 0:05:00.279 From Maxcina editor Mike Shepard can break down what is 0:05:00.320 --> 0:05:02.920 currently happening at the White House level. There's a pressure 0:05:03.000 --> 0:05:06.839 on Congress in particular to allow Jensen Wang and Nvidia 0:05:06.960 --> 0:05:09.960 to have greater access to some of these so called adversaries. 0:05:09.960 --> 0:05:13.080 What are you hearing, well, what we're hearing is that 0:05:13.120 --> 0:05:15.680 the White House would like Congress to hold off on 0:05:15.920 --> 0:05:21.160 legislation that would force Nvidia to put American customers first 0:05:21.520 --> 0:05:25.400 ahead of adversarial nations like China. And of course, Nvidia 0:05:25.480 --> 0:05:28.160 is proscribed from selling to China right now by US 0:05:28.240 --> 0:05:32.960 export controls and even by Beijing rejecting those aged twenty 0:05:33.120 --> 0:05:36.960 chips that we've talked about. That of one American authorization 0:05:37.160 --> 0:05:40.760 to be sold inside China, Nvidia has been looking for 0:05:40.839 --> 0:05:43.640 permission to do more, and that includes selling a deprecated 0:05:43.720 --> 0:05:48.520 version of its market leading Blackwell chip to China. Jensen 0:05:48.560 --> 0:05:51.920 Wang sees being able to compete in China as key 0:05:51.960 --> 0:05:54.320 to being able to compete around the world. But this 0:05:54.440 --> 0:05:59.320 measure is seen as really putting the handcuffs on Nvidia 0:05:59.360 --> 0:06:02.520 and on Ada as well, and both companies have moved 0:06:02.600 --> 0:06:03.080 against this. 0:06:03.440 --> 0:06:04.839 The bill is called the Gain Act. 0:06:04.880 --> 0:06:08.040 It would be attached to a piece of legislation that 0:06:08.279 --> 0:06:12.880 must pass. It governs defense policy, and it has bipartisan support. 0:06:13.080 --> 0:06:15.360 But now the White House is weighing in and seeking 0:06:15.400 --> 0:06:18.679 to get it pulled from the legislation, and it's unclear 0:06:18.760 --> 0:06:21.000 what would emerge in its place, because right now, Caro, 0:06:21.160 --> 0:06:27.720 there really is no uniform code governing export controls, especially 0:06:27.760 --> 0:06:30.240 as applied to China. And that's because if we go 0:06:30.360 --> 0:06:32.240 back all the way to the beginning of the year, 0:06:32.240 --> 0:06:35.200 and it seems so long ago, the Biden administration had 0:06:35.240 --> 0:06:40.280 tried to impose this AI diffusion rule, this framework that 0:06:40.360 --> 0:06:45.520 would prevent tech from leaking from allies like Saudi Arabia, 0:06:45.720 --> 0:06:49.080 like the UAE to China, and we don't really have 0:06:49.240 --> 0:06:51.719 anything here yet in place in Washington. So you see 0:06:51.760 --> 0:06:54.359 the China hawks on Capitol Hill and even inside the 0:06:54.360 --> 0:06:58.279 administration grasping for an alternative in this bill, which the 0:06:58.279 --> 0:06:59.880 White House is not trying to head off. 0:07:00.360 --> 0:07:02.000 Is at least an attempt at that. 0:07:03.440 --> 0:07:05.919 Shed last night was a little bit an issue of 0:07:05.960 --> 0:07:09.720 timing in chronology. So during the call the Analyst School, 0:07:09.920 --> 0:07:13.679 the Commerce Department put out a statement confirming that Nvidia 0:07:13.760 --> 0:07:16.400 is now allowed to export blackwell to the Gulf States. 0:07:16.920 --> 0:07:19.480 Is that just kind of rubber stamping something we knew 0:07:19.640 --> 0:07:20.000 was coming. 0:07:21.520 --> 0:07:24.360 We have widely expected this approval to come through, but 0:07:24.400 --> 0:07:27.840 it does take some time to work its way through 0:07:27.880 --> 0:07:30.800 the guts of the bureaucracy, but there is also a 0:07:30.880 --> 0:07:34.000 hurdle to overcome, and that is the concerns as we 0:07:34.240 --> 0:07:39.240 just discussed, that the technology does risk leaking to every 0:07:39.320 --> 0:07:42.520 serial nations, especially China, and that is because both the 0:07:42.600 --> 0:07:46.120 Saudis and the Amortis have had long standing business ties 0:07:46.200 --> 0:07:51.400 with China, and China's Huawei Technologies has a footprint in 0:07:51.440 --> 0:07:54.200 each of those two nations, and the concern is that 0:07:54.360 --> 0:07:58.920 the proximity of Huawei to American technology in a data 0:07:58.920 --> 0:08:03.000 center creates risk of some sort of leakage, either through 0:08:03.040 --> 0:08:05.560 software or something else, and that is something that the 0:08:05.680 --> 0:08:08.320 US government was really trying to seek assurances on. We 0:08:08.360 --> 0:08:13.400 don't have details on what was required of the emordis 0:08:13.440 --> 0:08:16.680 and include also of the companies, and you did your 0:08:16.680 --> 0:08:20.440 best in asking Jensen for specifics on that, but he 0:08:20.560 --> 0:08:23.960 really dug in on the whole broader question of diversion, 0:08:24.160 --> 0:08:27.000 and that is something that the company has been sensitive about. 0:08:27.280 --> 0:08:30.800 They have insisted that they know your customer programs have 0:08:31.000 --> 0:08:35.080 done enough to prevent diversion, and yet that hasn't assuaged 0:08:35.160 --> 0:08:38.000 all the concerns of China hawks here in Washington that 0:08:38.160 --> 0:08:41.960 advanced American technology could eventually leak back to China and 0:08:42.040 --> 0:08:46.720 somehow support its military and intelligence bureaucracies. 0:08:47.600 --> 0:08:50.320 Bloomberg's Mike Shepard, thank you very much. Let's get back 0:08:50.320 --> 0:08:53.640 to the Nvidia in the earning story. Ioko your Showcus 0:08:53.640 --> 0:08:56.959 portfolio consulting director at Wealth Enhancement Group one hundred and 0:08:57.000 --> 0:08:59.480 twenty four point five billion dollars in assets onder management 0:08:59.559 --> 0:09:02.520 and a pretty healthy chunk of Nvidia shares as well. 0:09:03.320 --> 0:09:04.000 I feel like. 0:09:04.280 --> 0:09:07.199 Gensen one was trying to give the trifecta a positive 0:09:07.240 --> 0:09:11.000 picture on demand, positive picture on supply, and even had 0:09:11.000 --> 0:09:13.120 the opportunity to kind of throw a bit of shade 0:09:13.120 --> 0:09:17.320 at A six or Google TPUs, saying that customers looked 0:09:17.360 --> 0:09:20.199 at alternatives and they came back to Nvidia. There was 0:09:20.240 --> 0:09:22.880 a lot there. What is the most important data point 0:09:22.920 --> 0:09:23.199 to you. 0:09:24.960 --> 0:09:28.240 Hied, Good to see you. Perhaps maybe the most important 0:09:28.240 --> 0:09:31.520 one was that Jensen wasn't wearing his signature leather jacket. 0:09:32.520 --> 0:09:35.480 But for the most part, for us, the when you 0:09:35.520 --> 0:09:38.520 look at the financials and when you look at the 0:09:38.559 --> 0:09:43.360 growth rate that Nvidia has posted and really delivering beyond 0:09:43.440 --> 0:09:48.200 expectations for where consensus estimates were, that was the most 0:09:48.240 --> 0:09:51.800 important metrics for us as we listen to the call. 0:09:51.960 --> 0:09:53.160 What I tried to do at the beginning of the 0:09:53.160 --> 0:09:55.720 conversation was get a sense of how supply constrained they 0:09:55.760 --> 0:09:59.080 are and how conservative that five hundred billion dollar forecast 0:09:59.120 --> 0:10:00.400 through the end of next year is. 0:10:00.679 --> 0:10:01.880 Let's start there. Listen to this. 0:10:03.320 --> 0:10:09.240 Sales are off the charts for Blackwell and and video 0:10:09.320 --> 0:10:13.240 GPUs in the cloud are sold out. We got plenty 0:10:13.240 --> 0:10:15.920 of black Walls to sell you. We have lots of 0:10:15.960 --> 0:10:19.960 Blackwells coming, We're making a lot of Blackwells, and we 0:10:20.000 --> 0:10:25.080 have a bunch of Vera Rubens coming, and so business 0:10:25.160 --> 0:10:26.400 is very very strong. 0:10:28.840 --> 0:10:31.760 Is that five hundred billion dollar figure which excludes China 0:10:32.080 --> 0:10:34.800 and is Blackwell Rubin through the end of calendar twenty 0:10:34.840 --> 0:10:36.600 six a bit conservative? 0:10:38.400 --> 0:10:42.480 Well, we heard analysts ask that question, right, and they 0:10:42.520 --> 0:10:46.840 have visibility into that five hundred billion dollar number, and 0:10:46.920 --> 0:10:49.240 so that doesn't include sort of the demand that will 0:10:49.280 --> 0:10:52.240 be coming forward in the next you know, fourteen to 0:10:52.360 --> 0:10:56.720 eighteen months, and so it perhaps is conservative. And I 0:10:56.720 --> 0:10:59.160 think that's the part that's so difficult for all of 0:10:59.200 --> 0:11:02.120 us to really wrap our heads around. These numbers are 0:11:02.160 --> 0:11:04.800 so large and they're growing at such a rate that 0:11:04.840 --> 0:11:08.040 we haven't seen before that it really becomes you know, 0:11:08.440 --> 0:11:11.280 for most of us, we're all skeptical that this is 0:11:11.320 --> 0:11:14.000 going to happen at the rate that it continues to 0:11:14.040 --> 0:11:14.880 happen at. 0:11:15.280 --> 0:11:19.160 What was interesting was the depreciation question was tackled. This 0:11:19.240 --> 0:11:21.240 is what's got Michael Burry and Itzzy. It's got what 0:11:21.360 --> 0:11:24.040 many have been seeing as a little bit of anxiety 0:11:24.120 --> 0:11:28.400 that ultimately the consumersm in video chips have been saying, look, 0:11:28.400 --> 0:11:31.880 these last longer than some had anticipated. But really it 0:11:32.000 --> 0:11:34.000 was made clear that the A one hundreds, the older 0:11:34.040 --> 0:11:36.240 type of chips are still working six years on. Ioka. 0:11:36.320 --> 0:11:38.200 Did that put that anxiety to bed? 0:11:39.559 --> 0:11:39.800 Yeah. 0:11:39.800 --> 0:11:42.000 I think it was a great way for collect to 0:11:42.120 --> 0:11:46.360 sort of clap back at that depreciation argument. You know 0:11:46.400 --> 0:11:49.720 that we are still using these older chips. They are 0:11:49.760 --> 0:11:53.960 still getting consumed simply because the demand is so great. 0:11:54.120 --> 0:11:56.040 So I do think there was a little bit of 0:11:56.040 --> 0:11:58.720 that that was put to rest and a relief I 0:11:58.760 --> 0:12:02.120 think from investors to an extent that they were still 0:12:02.160 --> 0:12:02.880 being utilized. 0:12:03.080 --> 0:12:06.160 To that extent, there's a relief today and we are 0:12:06.240 --> 0:12:09.560 up two percentage point Syoko. But I feel as though 0:12:10.360 --> 0:12:12.960 the entire market is a bit of an anxious boyfriend 0:12:13.000 --> 0:12:16.640 at the moment, trying to have soothing words currently given 0:12:16.679 --> 0:12:19.160 to it from whomever they can class and say, hey, 0:12:19.200 --> 0:12:21.360 I girlfriend in many ways, but what are you thinking 0:12:21.720 --> 0:12:23.599 of in terms of that this will soothe for the 0:12:23.640 --> 0:12:26.080 longer term? Are we're going to need weekly updates that 0:12:26.080 --> 0:12:27.280 we're not in some sort of bubble? 0:12:28.800 --> 0:12:31.360 You know, it is really difficult to sort of wrap 0:12:31.400 --> 0:12:33.679 our heads around. I think on one hand, we do 0:12:33.720 --> 0:12:36.520 see all of the demand continuing to grow at these 0:12:36.559 --> 0:12:39.480 exponential rates, But on the other hand, you see all 0:12:39.520 --> 0:12:43.199 this circularity and financing that I think concerns all of us, 0:12:43.240 --> 0:12:45.520 and then all of us who kind of grew up 0:12:45.600 --> 0:12:48.440 during the sort of tech bubble era. We've kind of 0:12:48.440 --> 0:12:50.760 seen this movie before, and we don't want it to 0:12:50.920 --> 0:12:53.240 end the way that it ended back in two thousand 0:12:53.720 --> 0:12:56.120 and So there's a lot of push and pull and 0:12:56.240 --> 0:12:59.840 a lot of monitoring on a daily, weekly, quarterly basis. 0:13:00.160 --> 0:13:02.720 And I think that's why in Video's earnings calls continue 0:13:02.760 --> 0:13:05.240 to be sort of the super Bowl of you know, 0:13:05.679 --> 0:13:09.439 equity investors sort of you know the calendars. 0:13:10.920 --> 0:13:15.040 What about in Video's valuation specifically, do you feel comfortable 0:13:15.040 --> 0:13:15.520 with that. 0:13:17.200 --> 0:13:21.040 You know, it's trading at twenty times next year's numbers, 0:13:21.080 --> 0:13:24.880 it's growing revenue at sixty percent, it's growing earnings at 0:13:24.920 --> 0:13:29.920 fifty plus percent. You know, it's there are reasonable valuations. 0:13:29.960 --> 0:13:33.400 They are not astronomical at like we saw during the 0:13:33.440 --> 0:13:34.040 tech bubble. 0:13:34.520 --> 0:13:35.480 However, you know. 0:13:35.720 --> 0:13:39.120 It's still relatively expensive. And it all hinges upon the 0:13:39.280 --> 0:13:43.320 duration of this growth rate. You know, does this continue 0:13:43.400 --> 0:13:47.520 at these levels for three five years or is it 0:13:47.679 --> 0:13:50.800 just this sort of near term twelve month growth rate 0:13:50.880 --> 0:13:54.319 that we're seeing. And I think that's this anxious investor 0:13:54.400 --> 0:13:57.600 base that's sort of watching this. When does the growth 0:13:57.679 --> 0:13:58.959 rates start to deteriorate? 0:13:59.240 --> 0:13:59.320 What? 0:13:59.360 --> 0:14:03.760 According to Blomberg analysis, if you're looking at what the 0:14:03.920 --> 0:14:07.240 general consensus is by twenty twenty eight, revenue growth is 0:14:07.280 --> 0:14:10.120 going to have slowed to twenty six percent. Just in 0:14:10.120 --> 0:14:12.480 twenty twenty seven, it will have slowed to forty five percent. 0:14:12.679 --> 0:14:14.760 But it's not just the law of ever greater numbers. 0:14:14.880 --> 0:14:17.319 I mean, we're already almost a five trillion dollar company. 0:14:17.360 --> 0:14:19.360 How can we expect it to keep growing at sixty 0:14:19.400 --> 0:14:21.280 percent out into twenty twenty eight and above? 0:14:22.760 --> 0:14:25.240 And I think that's the thing that's the skepticism that 0:14:25.320 --> 0:14:29.280 really sort of creeps into you know, the investor mindset 0:14:29.440 --> 0:14:32.680 is that how large can this get? And the numbers 0:14:33.160 --> 0:14:36.040 when you start to just grow at sixty percent plus 0:14:36.320 --> 0:14:40.360 year after year, it's just astronomical. And yet you know, 0:14:40.480 --> 0:14:44.040 Jensen's talking about three to four trillion dollars of annual 0:14:44.240 --> 0:14:49.120 spend for data center and AI infrastructure, and you know, 0:14:49.160 --> 0:14:51.280 if all of that or a good portion of that 0:14:51.400 --> 0:14:55.360 goes to Nvidia, the numbers continue to grow, and then 0:14:55.480 --> 0:14:57.880 you know the hyperscalers will grow as well because that 0:14:57.920 --> 0:15:01.080 benefits their cloud revenue. So it's just something that we 0:15:01.160 --> 0:15:04.640 continue to keep monitoring and we'll have to see each 0:15:04.720 --> 0:15:07.680 quarter not just from Nvidio, but from all of the 0:15:07.760 --> 0:15:13.040 tech ecosystem in terms of how large this AI platform 0:15:13.160 --> 0:15:14.040 is going to become. 0:15:14.680 --> 0:15:17.520 Okay, Yeshioka, it's always great to get your analysis from 0:15:17.560 --> 0:15:18.640 Welfare Enhancement Group. 0:15:18.640 --> 0:15:19.160 We thank you. 0:15:19.560 --> 0:15:23.760 Now, coming up AI pioneer Jan Lacun. He's leaving Meta 0:15:23.800 --> 0:15:26.480 to launch his own startup. What's next for him and 0:15:26.520 --> 0:15:29.520 what it says about Meta's own AI plans that's next? 0:15:29.920 --> 0:15:30.800 This is Bloomberg Tech. 0:15:43.640 --> 0:15:46.360 Jan Lukun, one of the so called godfathers of AI, 0:15:46.440 --> 0:15:49.200 is leaving his role at Meta to launch his own startup. 0:15:49.280 --> 0:15:51.880 Rumors of his departure have been swirling from more than 0:15:51.920 --> 0:15:54.120 a week, and it comes as the social media giant 0:15:54.440 --> 0:15:57.880 makes big changes to its AI efforts, including forming a 0:15:57.920 --> 0:16:01.440 new AI app lab earlier this summer and spending billions 0:16:01.560 --> 0:16:02.200 to high. 0:16:02.000 --> 0:16:05.280 New talent Bloom. As Kurt Wagner break the story I see. 0:16:05.280 --> 0:16:08.920 After we published our report, Jan Lkin posted on LinkedIn 0:16:09.080 --> 0:16:13.600 referring to rumors and clarified a few things. But I 0:16:13.640 --> 0:16:16.960 guess let's start with why he's leaving, what he's doing, 0:16:17.320 --> 0:16:18.080 where he's going. 0:16:19.160 --> 0:16:23.400 Yeah, So Jan Lucuhn is an AI researchature in sort 0:16:23.400 --> 0:16:26.000 of every sense of the term. He is looking at 0:16:26.000 --> 0:16:29.880 things on a very long time horizon. And as Meta 0:16:30.000 --> 0:16:32.680 has come into this sort of AI fight that we're 0:16:32.720 --> 0:16:36.440 seeing right now with open Ai and Google and Anthropic, 0:16:36.960 --> 0:16:39.760 they had really started to focus their AI efforts on 0:16:39.800 --> 0:16:42.640 the here and now, right on the chatbots, on the lms, 0:16:42.760 --> 0:16:47.080 the things that consumers are using today, and so Lacuhn's 0:16:47.120 --> 0:16:50.400 work is years of not decades, sort of down the line. 0:16:50.440 --> 0:16:52.680 And what we had heard in our reporting is that 0:16:53.000 --> 0:16:56.160 because of Meta's focus, he was starting to have trouble 0:16:56.160 --> 0:16:59.560 getting the resources that he felt he needed in order 0:16:59.600 --> 0:17:02.440 to do his research. The focus was certainly on things 0:17:02.440 --> 0:17:05.640 that were not what he was paying attention to, and 0:17:06.160 --> 0:17:08.639 ultimately Meta was going in a bit of a different direction, 0:17:08.840 --> 0:17:11.800 I think on AI than what he envisioned. And so, 0:17:12.200 --> 0:17:13.960 as you mentioned ed, this was something that had been 0:17:13.960 --> 0:17:16.520 discussed and sort of talked about really since this summer 0:17:17.160 --> 0:17:21.840 when they started pivoting towards the superintelligence effort, and you know, 0:17:22.240 --> 0:17:24.280 they did try to leave on good terms. Meta is 0:17:24.320 --> 0:17:26.560 going to be sort of involved in some capacity in 0:17:26.560 --> 0:17:28.520 his new startup, but it was time for him to 0:17:28.600 --> 0:17:30.560 move on and do his research elsewhere. 0:17:30.920 --> 0:17:35.080 Yeah, he describes as advanced machine intelligence, more sensory information, 0:17:35.200 --> 0:17:39.320 more physical AI. In many ways, the story articulates how 0:17:40.160 --> 0:17:43.320 the tide moved against him, and he certainly executives moved 0:17:43.320 --> 0:17:46.320 against him because Alexander Wang came in and sort of 0:17:46.320 --> 0:17:47.120 superseded him. 0:17:48.600 --> 0:17:48.760 Yeah. 0:17:48.840 --> 0:17:52.320 Well, lacun is a legend in the AI industry, or 0:17:52.400 --> 0:17:56.359 he's considered a quote godfather of this entire field. And 0:17:56.440 --> 0:17:59.040 yet when Meta decided to build this new lab the 0:17:59.080 --> 0:18:04.320 summer TV deal and go out and Acchoir essentially Alexander 0:18:04.320 --> 0:18:05.479 Wang from Scale Ai. 0:18:06.200 --> 0:18:08.680 He became Jan mccun's boss, right. 0:18:09.160 --> 0:18:12.800 I think he's more than thirty years younger, obviously very 0:18:12.840 --> 0:18:17.159 different type of expert in terms of AI, and so 0:18:17.240 --> 0:18:20.040 there was a feeling sort of in that moment where Okay, 0:18:20.320 --> 0:18:23.399 if they have this legendary AI researcher in the building 0:18:23.440 --> 0:18:25.960 and yet they're still going outside the company to bring 0:18:26.000 --> 0:18:28.600 in people to run their new AI effort, what is 0:18:28.680 --> 0:18:32.240 that signal right for John's future here? And I think 0:18:32.600 --> 0:18:35.119 again this was something that in some ways was a 0:18:35.119 --> 0:18:37.280 long time coming because of the fact that he is 0:18:37.320 --> 0:18:40.159 so forward and future looking with what he does, and 0:18:40.240 --> 0:18:42.920 Meta is fighting sort of for its AI life right 0:18:42.960 --> 0:18:47.640 here and now. But I think that layering of executives 0:18:47.720 --> 0:18:52.280 over the summer was probably another big reason that he's leaving. 0:18:52.000 --> 0:18:54.600 Today, and a big supporter of open sourcing that perhaps 0:18:54.640 --> 0:18:57.760 metas shifted slightly away from as well. Rumos Coo Wagner, 0:18:58.040 --> 0:19:00.679 great breakdown, Thank you very much. Indeed, while coming up, 0:19:00.800 --> 0:19:04.000 we'll talk about the Chinese university shaping the country's AI 0:19:04.119 --> 0:19:07.680 sector and its future startup leaders. This is blom Beg Tech, 0:19:16.119 --> 0:19:20.199 China's Singua University, but it's having an outsized impact on 0:19:20.200 --> 0:19:23.200 the country's AI industry and the school's professors and students. 0:19:23.400 --> 0:19:27.800 They received more patterns each year than MIT, Stanford, Princeton, 0:19:27.880 --> 0:19:31.600 Harvard combined. Let's go out to bluembg's executive tech editor 0:19:31.600 --> 0:19:34.440 Peter Elstrom for more on all this signifies about China's 0:19:34.440 --> 0:19:38.399 AI ambitions. They are wholly focused on making sure that 0:19:38.440 --> 0:19:42.840 they have many, many more experts and potential employees of 0:19:42.920 --> 0:19:43.400 the future. 0:19:46.000 --> 0:19:49.840 Yeah, Qingwa is a very prestigious university. For many years, 0:19:50.119 --> 0:19:53.399 it's educated the best in the brightest, especially in science 0:19:53.400 --> 0:19:56.639 and engineering in China. But what's really changing now is 0:19:56.720 --> 0:20:00.400 you're seeing this momentum in AI in particular. A couple 0:20:00.440 --> 0:20:02.960 things are feeding into that. First of all, Deep Seek. 0:20:03.040 --> 0:20:05.959 The success of Deep Deep Seek in its breakthrough model 0:20:06.280 --> 0:20:08.919 earlier this year really gave a lot of confidence to 0:20:09.000 --> 0:20:12.159 these young students that they too could build businesses that 0:20:12.200 --> 0:20:15.040 would be very, very successful. In addition to that, Chi 0:20:15.160 --> 0:20:18.240 Jinping and the Communist Party have been calling on the 0:20:18.320 --> 0:20:21.439 tech sector in China to be able to build in 0:20:21.480 --> 0:20:24.360 a number of very key areas for them, strategically important 0:20:24.400 --> 0:20:27.919 areas for them, especially AI. So you can get government funding, 0:20:27.960 --> 0:20:30.480 you can get tax subsidies, there's a lot of momentum 0:20:30.560 --> 0:20:33.320 to be able to build businesses, so the students there 0:20:33.600 --> 0:20:35.920 feel like they have a lot more opportunities in front 0:20:35.960 --> 0:20:38.360 of them. Venture capitalists are coming, They're funding a lot 0:20:38.359 --> 0:20:41.920 of these businesses. We're seeing quite impressive AI startups coming 0:20:41.960 --> 0:20:45.000 out of Chinwan the rest of China, Peter. 0:20:44.800 --> 0:20:46.800 As part of this Bloomberg Big Take, you look at 0:20:46.880 --> 0:20:50.840 the alumni that have gone away founded key startups, but 0:20:50.880 --> 0:20:54.680 also that those that have quite senior jobs at China's 0:20:54.720 --> 0:20:57.600 biggest technology companies. What are the best examples we need 0:20:57.640 --> 0:20:58.160 to know about. 0:21:00.480 --> 0:21:05.080 Well, Chinua has put their graduates in a number of 0:21:05.160 --> 0:21:08.200 different places, including leading companies like Ali Babo, which has 0:21:08.200 --> 0:21:11.680 done a lot of impressive work with the Gwen model 0:21:11.720 --> 0:21:14.840 in particular, bte Dance, the parent company of TikTok, and 0:21:15.359 --> 0:21:18.280 they founded startups like Jipu, which is another cutting edge 0:21:19.000 --> 0:21:23.240 company there. And really what you're seeing is this massive students. 0:21:23.840 --> 0:21:28.240 China graduates many more STEM graduates than the United States, 0:21:28.280 --> 0:21:30.800 and they have done that for many years. So you're 0:21:30.840 --> 0:21:34.639 seeing three and a half million students at last count, 0:21:34.680 --> 0:21:39.440 compared to eighty thousand in the United States. So it's 0:21:39.440 --> 0:21:42.199 a big magnitude difference for them in particular, and now 0:21:42.240 --> 0:21:44.159 they're able to get this kind of funding to be 0:21:44.160 --> 0:21:47.200 able to build up their tech sector, and you're seeing 0:21:47.200 --> 0:21:53.400 these impressive startups come out of the country's technology opportunities 0:21:53.600 --> 0:21:57.159 in the university city. Universities like Changhwa are also starting 0:21:57.200 --> 0:22:00.920 up their own incubators where students can get funding. They 0:22:01.000 --> 0:22:03.359 run a short tank like competition to be able to 0:22:03.359 --> 0:22:05.600 give these kids a chance to pitch their stories and 0:22:05.720 --> 0:22:07.240 possiblely get funding for them too. 0:22:08.400 --> 0:22:11.480 Bloomberg's Peter Elstrom, thank you very much on the latest 0:22:11.520 --> 0:22:14.600 big take. Actually, since we come to air on Bloomberg Tech, 0:22:14.640 --> 0:22:17.160 you see Nvidia kind of pairing some of its gains. 0:22:17.200 --> 0:22:20.120