WEBVTT - Nvidia Sees 70% Growth as AI Boom Accelerates 0:00:02.520 --> 0:00:04.210 Bloomberg Audio Studios. 0:00:04.690 --> 0:00:05.370 Podcasts. 0:00:05.690 --> 0:00:06.210 Radio. 0:00:06.650 --> 0:00:07.050 News. 0:00:13.480 --> 0:00:17.000 Bloomberg Tech is live from the heart of Silicon Valley 0:00:17.540 --> 0:00:19.820 with Ed Ludlow in San Francisco. 0:00:22.730 --> 0:00:26.450 This is Bloomberg Tech. Coming up, NVIDIA sees 70% revenue 0:00:26.470 --> 0:00:30.380 growth next fiscal year, far exceeding analyst estimates. We'll break 0:00:30.420 --> 0:00:34.960 down the chipmaker's astonishing outlook. Plus, Salesforce jumps after giving 0:00:35.000 --> 0:00:38.339 an outlook for strong revenue expansion and deepening its ties 0:00:38.620 --> 0:00:42.669 with Anthropic. And Hugging Face comes out with its newest robot, 0:00:42.810 --> 0:00:46.850 a duck that can walk and talk. And even Rollerskate. 0:00:46.870 --> 0:00:48.890 We're going to be joined by Hugging Face co-founder and 0:00:48.909 --> 0:00:50.950 chief science officer, Thomas Crosby. 0:00:51.479 --> 0:00:51.800 Wolf. 0:00:52.380 --> 0:00:55.910 Nvidia is the story. Nvidia is the driver of markets. 0:00:55.970 --> 0:00:59.570 The stock is up now more than 8% accelerating and 0:00:59.630 --> 0:01:03.360 on track for its biggest jump since April of 2025. 0:01:03.350 --> 0:01:07.429 In very quick succession, the stock has changed direction and 0:01:07.450 --> 0:01:10.880 the story has changed. That is driving different indices higher. 0:01:11.380 --> 0:01:11.800 The news? 0:01:12.900 --> 0:01:15.800 Top line growth, 70% in fiscal 28. Let's bring in 0:01:15.819 --> 0:01:20.370 Bloomberg's Ian King. NVIDIA does not give full year fiscal 0:01:20.390 --> 0:01:24.510 year outlooks, but they did. And it's a big one. Yeah, no, absolutely. 0:01:24.610 --> 0:01:27.449 Everything changed with that one phrase from Colette on the 0:01:27.490 --> 0:01:31.770 call yesterday. And just for context, Wall Street was expecting 0:01:31.810 --> 0:01:34.929 a brilliant year next year. It was expecting 45% growth. 0:01:34.970 --> 0:01:37.009 And then NVIDIA comes in and says, no, it's going 0:01:37.030 --> 0:01:39.110 to be more like 70. And guess what? If we 0:01:39.130 --> 0:01:41.560 had enough supply, if we get more supply, it could double. 0:01:41.980 --> 0:01:43.600 That's the amount of orders we've got. 0:01:44.620 --> 0:01:47.250 I want to bring up something that Colette Crest, the CFO, 0:01:47.280 --> 0:01:49.470 told you and I on the phone, which is basically 0:01:50.530 --> 0:01:53.490 the revenue forecast would be greater were it not for 0:01:54.110 --> 0:01:58.730 supply constraints. The supply constraint is still severe. Explain that 0:01:58.850 --> 0:02:00.870 number and what Colette Crest was trying to get at. 0:02:01.270 --> 0:02:03.270 Yeah, I mean, she said to us, you'll remember, I 0:02:03.620 --> 0:02:06.760 am making a point here. I am telling everybody, whether 0:02:06.800 --> 0:02:10.889 it's the memory chip makers, whether it's the made-to-order silicon makers, 0:02:11.030 --> 0:02:13.710 everybody in the supply chain. Hey, demand is here. 0:02:14.370 --> 0:02:15.090 Get working. 0:02:15.410 --> 0:02:18.200 Get putting more production in place. We can do this. 0:02:18.320 --> 0:02:20.160 There's more business to be done. 0:02:21.560 --> 0:02:25.960 There is not a perfect set of numbers because ultimately 0:02:25.980 --> 0:02:29.600 there is some margin impact from higher memory prices. Nvidia 0:02:29.620 --> 0:02:32.410 will raise prices of its technology at the start of 0:02:32.470 --> 0:02:37.130 next year. Explain that, that sort of bottom line situation 0:02:37.169 --> 0:02:37.480 for us. 0:02:37.730 --> 0:02:40.239 Yeah, I mean, we kind of laughed with Colette on 0:02:40.260 --> 0:02:43.620 the call saying how many CFOs go on a call, 0:02:44.480 --> 0:02:49.120 lower their margin by three percentage points. and still get 0:02:49.470 --> 0:02:52.570 a stock rise, and that's not right. So what's going 0:02:52.610 --> 0:02:54.690 on there is memory has just got out of hand. 0:02:54.830 --> 0:02:57.970 Memory has to be packaged with NVIDIA's products. There just 0:02:58.030 --> 0:03:01.089 isn't enough of it. There's demand everywhere, and NVIDIA's had 0:03:01.130 --> 0:03:03.630 to pass that along. And of course, that's having an 0:03:03.690 --> 0:03:04.570 impact on margins. 0:03:04.889 --> 0:03:07.430 All right, Bloomberg's Ian King, thank you very much indeed. 0:03:07.490 --> 0:03:10.570 It was a very interesting set of numbers call. And 0:03:10.610 --> 0:03:13.470 then the reaction. Our next guest says the real surprise 0:03:13.990 --> 0:03:17.130 is in NVIDIA's earnings, the durability of that growth outlook. 0:03:17.510 --> 0:03:21.000 Doug Huber is Deputy Chief Investment Officer at Wealth Enhancement Group, 0:03:21.060 --> 0:03:25.380 which across funds has about $ 1. 9 billion worth of NVIDIA stock. 0:03:25.860 --> 0:03:28.299 What do you mean by durability? What was the evidence 0:03:28.360 --> 0:03:32.680 you saw that this goes beyond, I suppose, fiscal 2018? 0:03:32.680 --> 0:03:35.840 Yeah, I think, you know, frankly, it really did shift 0:03:35.880 --> 0:03:39.800 the dynamic, to me at least, in that massive outlook 0:03:39.820 --> 0:03:43.020 they put forward, right? I think it shifted the thought 0:03:43.040 --> 0:03:46.600 process of is demand slowing and moving this to is 0:03:46.620 --> 0:03:50.530 the industry physically able to build enough compute to satisfy it, right? 0:03:50.540 --> 0:03:52.550 I think that was that shocking number that Ian was 0:03:52.570 --> 0:03:55.590 just talking about with 70% growth next year, but that's 0:03:55.630 --> 0:03:57.730 supply constraint. I mean, it could be 100, 120, right? 0:03:57.770 --> 0:04:01.450 And so I think Everybody thought this cycle was kind 0:04:01.490 --> 0:04:03.490 of nearing, and they were pretty clear that there was 0:04:03.510 --> 0:04:06.430 a lot of demand, not only through hyperscalers, but also 0:04:06.450 --> 0:04:10.120 through the ASI channels. And so I think that's really compelling. 0:04:10.870 --> 0:04:12.990 So, Doug, let's go there. When Ian and I were 0:04:13.030 --> 0:04:15.530 on the phone with the CFO at Cress, there was 0:04:15.570 --> 0:04:19.270 a heavy emphasis on the hyperscalers still surging. They've already 0:04:19.330 --> 0:04:22.070 put their orders in for next year. If there was 0:04:22.110 --> 0:04:24.450 a set of numbers under the hood, it was the 0:04:24.510 --> 0:04:28.010 concentration of business still in the hyperscalers. What do you 0:04:28.070 --> 0:04:30.760 make of that now and the idea that NVIDIA feels 0:04:30.800 --> 0:04:34.260 that broadens beyond fiscal 2018? 0:04:34.260 --> 0:04:37.560 Yeah, I think she pointed out on the call that 0:04:37.580 --> 0:04:40.580 they are buyers of chips, essentially, where some of the 0:04:40.700 --> 0:04:43.460 other channels, it's much more of a total infrastructure, right? 0:04:43.480 --> 0:04:45.380 And I think that build-out takes longer. I think the 0:04:45.420 --> 0:04:48.560 integrations take longer. And so, yes, I think we expected 0:04:48.580 --> 0:04:51.370 to see better numbers or at least potentially a little 0:04:51.390 --> 0:04:53.730 bit better numbers from those other channels. But it does 0:04:53.790 --> 0:04:55.729 sound to me, and I think they alluded to the 0:04:55.750 --> 0:04:58.650 fact that that is coming through, that's pulling through. And 0:04:58.690 --> 0:05:01.029 you can see those as being really large channels, whether 0:05:01.089 --> 0:05:04.310 it's the sovereigns, the neoclouds, the enterprise solutions. I mean, 0:05:04.330 --> 0:05:07.200 I think that will be an interesting channel for them. 0:05:07.220 --> 0:05:09.320 The question I have, and I don't think they've really 0:05:09.360 --> 0:05:11.080 touched on it to date, is did they get the 0:05:11.120 --> 0:05:13.260 same margin out of that, given that it sounds like 0:05:13.300 --> 0:05:14.500 it's a higher lift to implement? 0:05:15.760 --> 0:05:19.830 I find that interesting, you know. There's a difference, I suppose, 0:05:20.190 --> 0:05:24.380 in selling Vera Rubin systems to the hyperscalers at scale 0:05:24.740 --> 0:05:28.180 and doing lots of little deals with what our enterprise 0:05:28.240 --> 0:05:31.289 is essentially. Do you want to see NVIDIA shift away 0:05:31.320 --> 0:05:33.430 its reliance from the hyperscaler? Is that something that you 0:05:33.450 --> 0:05:34.670 would demand of the company? 0:05:36.910 --> 0:05:38.310 I don't know if I'd demand it of the company. 0:05:38.330 --> 0:05:40.370 They're clearly doing a pretty good job. But I do 0:05:40.410 --> 0:05:42.800 think it's interesting to see that. Sensible point of view. 0:05:44.860 --> 0:05:47.630 But I think they're interestingly moving to more of that 0:05:48.190 --> 0:05:51.870 integrated AI platform, right? So whether it's the GPUs, the CPUs, 0:05:51.930 --> 0:05:54.890 the networking, the software, the inference, and how they can 0:05:54.930 --> 0:05:57.010 apply that, I think that has set them apart a 0:05:57.029 --> 0:05:59.270 little bit. And I think it protects them, especially as 0:05:59.290 --> 0:06:02.380 some of these hyperscalers are developing in their own custom silicon. 0:06:04.020 --> 0:06:05.840 Doug, I'm going to go back to the shares. Actually, 0:06:05.860 --> 0:06:08.240 they've just pulled back a touch below an 8% gain, 0:06:08.260 --> 0:06:11.279 but it's pretty astonishing, right? The change in the storage. 0:06:12.620 --> 0:06:15.120 on track for its best day since April of 2025 0:06:15.120 --> 0:06:18.290 at some point. What is it specifically that the market 0:06:18.350 --> 0:06:21.210 is cheering so much here? How unusual it is to 0:06:21.230 --> 0:06:24.510 have a full fiscal year guide, something more than that? 0:06:25.940 --> 0:06:28.820 It seems to be really anchored to that guidance, right? 0:06:28.839 --> 0:06:30.960 I think if you saw when it first came out, 0:06:31.000 --> 0:06:33.540 it actually trended down right after the announcement. I think 0:06:33.580 --> 0:06:35.800 people looked at that margin number and looked at the 0:06:35.839 --> 0:06:38.290 forward guidance that the margin actually might decrease next year. 0:06:38.710 --> 0:06:40.970 But then on the call, as they really hammered this 0:06:41.029 --> 0:06:43.910 potential future growth opportunity, I think that really woke folks 0:06:43.990 --> 0:06:46.330 up to the fact that There is still a lot 0:06:46.550 --> 0:06:49.270 of demand and we are actually trying to figure out 0:06:49.310 --> 0:06:50.910 how do we supply it all. And I think that 0:06:50.950 --> 0:06:54.070 was a real defining point in that call that you 0:06:54.089 --> 0:06:56.550 actually saw the stock move. The price action moved kind 0:06:56.570 --> 0:06:58.229 of right around when that kind of came up. 0:07:00.670 --> 0:07:04.800 So the argument is top line growth for that fiscal 0:07:04.810 --> 0:07:08.500 year would be greater were it not for supply constraints. 0:07:09.020 --> 0:07:13.020 So demand is doubling. It's at 100%. Do you take 0:07:13.080 --> 0:07:14.280 that at full face value? 0:07:18.180 --> 0:07:20.390 That's a great question. I think that's what everybody's trying 0:07:20.410 --> 0:07:21.480 to figure out. I mean. 0:07:23.150 --> 0:07:25.630 I think more data points as we go through here 0:07:26.130 --> 0:07:28.930 will be interesting to see outside of just NVIDIA. Are 0:07:28.950 --> 0:07:31.630 we seeing that CapEx continue to go up? Are we 0:07:31.670 --> 0:07:34.550 seeing the folks, the neoclouds and others continue to press 0:07:34.590 --> 0:07:36.810 that spend? And how are they guiding? I think that 0:07:36.850 --> 0:07:40.290 will be certainly confirmatory to what they're saying. But it 0:07:40.550 --> 0:07:44.290 does appear that they are quite bullish on their calls. 0:07:44.330 --> 0:07:46.390 We know that. But I think this one felt and 0:07:46.430 --> 0:07:48.970 had evidence points, to me at least, that seem to 0:07:49.030 --> 0:07:52.570 indicate that this is a really sustainable and quite a 0:07:52.950 --> 0:07:58.000 real possibility that that growth is there. Now, whether it's 70, 100, 60, 0:07:58.000 --> 0:08:00.040 who knows? And obviously, the market is going to price 0:08:00.060 --> 0:08:02.900 this on what they execute on. But it certainly seems 0:08:02.920 --> 0:08:05.880 like there's a lot of tailwind, maybe more tailwind than 0:08:05.900 --> 0:08:07.560 I think anybody might have appreciated here. 0:08:08.800 --> 0:08:12.450 Doug, in prior quarters, The question was always, is the 0:08:12.530 --> 0:08:16.880 bigger picture story intact? Like AI's full momentum. And that 0:08:16.920 --> 0:08:19.940 chart I was just showing was adoption on different metrics 0:08:19.960 --> 0:08:23.840 against the frontier labs. What Colette Cress in particular argued 0:08:23.880 --> 0:08:27.070 was like, look, we have full confidence that those frontier 0:08:27.110 --> 0:08:29.850 labs are going to go turn into mature companies, not 0:08:29.890 --> 0:08:33.190 just revenue. They're going to be great businesses. No one 0:08:33.210 --> 0:08:34.979 seems to be lingering on that this morning. 0:08:35.030 --> 0:08:35.250 Why? 0:08:36.850 --> 0:08:38.470 It's a great question. I mean, it does come up 0:08:38.530 --> 0:08:40.490 a lot. You know, how are they using their balance sheet? 0:08:40.510 --> 0:08:43.000 I think she said they've invested 50 billion dollars or 0:08:43.040 --> 0:08:45.420 plan to invest 50 billion dollars in the labs. And 0:08:45.440 --> 0:08:48.079 I think it is a great breeding ground. Right. If 0:08:48.140 --> 0:08:50.849 you can be intimately involved in the growth of these 0:08:50.920 --> 0:08:51.990 so-called farm teams. 0:08:52.050 --> 0:08:52.370 Right. 0:08:52.450 --> 0:08:55.329 And have your product integrated early, have the system set 0:08:55.390 --> 0:08:59.780 up so it's it's. utilizing your infrastructure from the start 0:09:00.360 --> 0:09:03.340 if they catch traction and if they become the next 0:09:03.380 --> 0:09:06.560 big player. Being in early is unbelievable, and it could 0:09:06.620 --> 0:09:09.600 really lead to a lot of upside for NVIDIA and 0:09:09.640 --> 0:09:12.610 for their ability to push that product through. I don't 0:09:12.650 --> 0:09:16.350 know if people are just not quite discounting that yet 0:09:16.410 --> 0:09:19.209 and saying, listen, we'll evaluate that as time goes on. 0:09:19.250 --> 0:09:23.010 We're more focused on the actual numbers. But I thought 0:09:23.030 --> 0:09:24.450 that was a really interesting piece of the phone call. 0:09:26.030 --> 0:09:28.210 Doug Huber of Wealth Enhancement, thank you so much for 0:09:28.230 --> 0:09:32.650 joining us here on Bloomberg Tech. Let's take a look 0:09:32.670 --> 0:09:36.449 at today's big number, $ 45 billion. That's how much Anthropic 0:09:36.550 --> 0:09:39.830 is thinking of spending to rent AI cloud computing capacity 0:09:39.850 --> 0:09:43.929 from NScale's flagship data center in West Virginia. That's according 0:09:43.980 --> 0:09:47.240 to sources who say the commitment's over six years and 0:09:47.280 --> 0:09:52.710 represents about 460 megawatts of power. Let's stick with Anthropic 0:09:52.730 --> 0:09:55.400 and add Salesforce into the mix as the two companies 0:09:55.740 --> 0:10:01.120 are deepening their partnership with Anthropic, integrating Salesforce's products. within Claude. 0:10:01.420 --> 0:10:03.800 The news was revealed as Salesforce also gave an outlook 0:10:03.840 --> 0:10:08.230 for strong revenue expansion, sending shares jumping beyond 21% the 0:10:08.270 --> 0:10:10.829 most in six years. Yeah, the most in six years. 0:10:16.190 --> 0:10:21.310 That stock move, the anthropic effect, or something more related 0:10:21.350 --> 0:10:22.050 to anthropic? 0:10:22.470 --> 0:10:24.620 It feels like the anthropic effect. I mean, the numbers 0:10:24.650 --> 0:10:27.500 were good. They weren't amazing. Of course, Salesforce and the 0:10:27.520 --> 0:10:30.720 whole sector has been dealing with this whole SaaSpocalypse fear 0:10:30.760 --> 0:10:33.599 that the model layer is going to eat the application layer. 0:10:34.179 --> 0:10:37.459 And I think seeing Dario and Mark Benioff together in 0:10:37.520 --> 0:10:40.690 a TV interview, that showed people, hey, there's a bit 0:10:40.710 --> 0:10:43.590 of a non-aggression treaty here. They're going to work together. 0:10:43.670 --> 0:10:46.860 The application layer is here to stay. at least for 0:10:46.880 --> 0:10:48.110 the foreseeable future. 0:10:48.929 --> 0:10:51.870 What was the earnings story for Salesforce just beyond the numbers? 0:10:51.950 --> 0:10:55.240 Like how is CRM doing and all the things that 0:10:55.300 --> 0:10:58.580 Salesforce is known for? Or is this them doing real 0:10:58.720 --> 0:10:59.400 AI stuff? 0:11:00.520 --> 0:11:04.000 Effectively, their story is that you sell AI products kind 0:11:04.020 --> 0:11:06.680 of at the margins, right? You get people to upgrade 0:11:06.700 --> 0:11:09.380 their subscriptions, you get them to buy some tokens to 0:11:09.450 --> 0:11:13.430 weave AI into their existing platform. And what's really important 0:11:13.470 --> 0:11:17.390 is they show that customers are not attritioning off the platform. 0:11:17.770 --> 0:11:19.679 That's been the big fear that customers are going to 0:11:19.700 --> 0:11:22.719 go grab an LLM and build their own CRM. But 0:11:22.820 --> 0:11:25.360 at least for now, it is not happening, which is 0:11:25.420 --> 0:11:27.330 giving comfort a lot to investors. 0:11:27.429 --> 0:11:30.630 This is a big stock move. Sales for WhatsApp, 21% 0:11:30.630 --> 0:11:33.390 on track for its best day since August 26th, 2020. 0:11:33.390 --> 0:11:35.880 and the stock now trading at a January high. So 0:11:35.900 --> 0:11:39.320 the situation has changed. Bloomberg's Brady Ford, thank you very much. Indeed, 0:11:39.360 --> 0:11:43.040 Bloomberg's learned that SpaceX CEO Elon Musk and former White 0:11:43.080 --> 0:11:45.940 House AI czar David Sachs are scheduled to speak at 0:11:45.980 --> 0:11:50.120 a tech-focused Group of 20 ministerial meeting in North Carolina 0:11:50.179 --> 0:11:54.540 next week, along with NVIDIA's Jensen Wang, OpenAI's Sam Altman, 0:11:54.840 --> 0:11:58.400 and Meta president Dina Powell McCormick. That's all according to... 0:11:58.700 --> 0:12:01.720 To sources, Bloomberg senior tech editor Mike Shepard joins us 0:12:01.760 --> 0:12:04.470 now from Washington with more. It seems like an important 0:12:04.500 --> 0:12:07.750 meeting and it's certainly a top tier list of participants. 0:12:07.770 --> 0:12:08.610 What's the point of it? 0:12:10.010 --> 0:12:12.950 Well, the point really, Ed, is at this ministerial meeting, 0:12:12.970 --> 0:12:16.610 they're trying to lay the agenda out for the leaders 0:12:16.809 --> 0:12:20.130 ultimately when they gather in Florida at President Donald Trump's 0:12:20.170 --> 0:12:23.480 golf club in Doral. you'll have leaders from the 20 0:12:23.480 --> 0:12:28.060 largest economies gathered there in December. But beforehand, you really 0:12:28.100 --> 0:12:30.370 have to make sure all the ducks are lined up 0:12:30.660 --> 0:12:33.770 and everybody is on roughly the same page when it 0:12:33.830 --> 0:12:37.030 comes to key points of the global economic agenda. And 0:12:37.070 --> 0:12:41.980 certainly these days, artificial intelligence and advanced technologies are really 0:12:42.130 --> 0:12:45.840 at center stage. And it is a top concern for U.S. officials. 0:12:45.880 --> 0:12:49.060 They want to make sure that There's general agreement among 0:12:49.300 --> 0:12:51.640 these nations that are going to take part, and they 0:12:51.679 --> 0:12:57.000 include China. They include other participants like India and Brazil 0:12:57.120 --> 0:13:00.420 and a number of EU countries as well. They want 0:13:00.460 --> 0:13:02.880 to make sure that there is agreement on how to 0:13:02.960 --> 0:13:06.760 approach regulation of AI in particular. And one of the 0:13:06.800 --> 0:13:08.730 concerns that they have is that there could be too 0:13:08.780 --> 0:13:10.970 heavy a touch. So what they're hoping for is at 0:13:11.030 --> 0:13:13.170 this meeting next week to agree on a set of 0:13:13.230 --> 0:13:17.200 principles known as the Carolina Principles, that would lay out 0:13:17.400 --> 0:13:21.579 a lighter touch for regulation, push for more foundational research, 0:13:21.940 --> 0:13:26.640 call for more private public sector cooperation, and overall try 0:13:26.720 --> 0:13:30.000 to set the stage for a common ground between governments 0:13:30.020 --> 0:13:32.640 so that you don't see heavy regulation in one country, 0:13:32.820 --> 0:13:37.280 a lighter approach in another, and overall disagreement and imbalances. Now, 0:13:37.480 --> 0:13:38.979 this will be a bit of a heavy lift, and 0:13:39.020 --> 0:13:42.750 that's why they're bringing in these blue link names, these 0:13:43.210 --> 0:13:47.120 high profile folks to try to help their argument there. 0:13:48.860 --> 0:13:51.900 Let's go to Indiana. Take me to Indiana, Shep. We 0:13:51.920 --> 0:13:55.290 broke a story that SK Hynix, the Korean memory name 0:13:55.309 --> 0:13:58.869 that we've been tracking so closely, has broken ground on 0:13:58.890 --> 0:14:01.670 something pretty significant in that state. What is the project? 0:14:02.050 --> 0:14:03.170 What is the dollar commitment? 0:14:04.530 --> 0:14:08.260 Well, this is a long-awaited packaging plant that has been 0:14:08.340 --> 0:14:13.000 planned for Indiana, for West Lafayette, for a couple of 0:14:13.059 --> 0:14:15.260 years now, and they are finally kicking this off. It 0:14:15.320 --> 0:14:18.900 is getting some government financing through the Chips and Science 0:14:19.000 --> 0:14:22.060 Act that was signed by President Joe Biden in 2022. 0:14:22.060 --> 0:14:25.380 It'll be a total of a $ 4 billion investment by 0:14:25.400 --> 0:14:27.600 the company, and the idea is that it will create 0:14:28.400 --> 0:14:31.900 as many as 1,000 jobs there. It will also address 0:14:32.000 --> 0:14:35.030 one of the key concerns here in U.S. chipmaking, is 0:14:35.130 --> 0:14:40.130 that there is not enough packaging capacity to go around. 0:14:40.600 --> 0:14:43.510 And packaging, as you know, Ed, is so crucial, especially 0:14:43.550 --> 0:14:45.470 in this memory area. They really need to be able 0:14:45.490 --> 0:14:47.490 to stack those chips to be able to make them 0:14:47.710 --> 0:14:51.200 more efficient. And it is a huge step by SK Hynix, 0:14:51.270 --> 0:14:54.940 which just a few months ago had its high-profile debut 0:14:54.980 --> 0:14:59.320 in U.S. markets, financial markets. Now we're seeing its expansion 0:14:59.420 --> 0:15:01.170 physically here with that new plant. 0:15:02.360 --> 0:15:03.660 Later in the show, we're going to speak to the 0:15:03.700 --> 0:15:06.580 LAMS research CEO, another company that's saying we've got to 0:15:06.660 --> 0:15:10.230 focus on advanced packaging. Bloomberg's Mike Shepard, thank you so much. Now, 0:15:10.260 --> 0:15:14.190 coming up, Meta settled a landmark social media case yesterday, 0:15:14.590 --> 0:15:18.190 which could cost the company up to $ 18 billion. University 0:15:18.210 --> 0:15:21.570 of Virginia Law School professor Danielle Citron joins us next. 0:15:22.190 --> 0:15:33.240 This is Bloomberg Tech. As part of Meta's landmark settlement, 0:15:33.420 --> 0:15:36.710 Meta's rolling out new safeguards to Instagram and Facebook that 0:15:36.750 --> 0:15:40.410 are currently limited to US users. But risks remain, even 0:15:40.450 --> 0:15:45.170 with guardrails, starting with properly vetting children with age verification tools. 0:15:45.570 --> 0:15:48.790 Bloomberg's Alex Levine joins us now with the details. You've 0:15:48.830 --> 0:15:53.300 put out some really interesting work overnight since the settlement broke. 0:15:54.520 --> 0:15:58.620 Let's start with the basics of the mechanisms, age verification 0:15:58.950 --> 0:16:02.190 and the limits that Meta's put in place, and it's U.S. only. 0:16:03.330 --> 0:16:05.970 Yeah, I mean, I think that age verification is really 0:16:06.050 --> 0:16:09.210 central to this. And as we heard in several of 0:16:09.250 --> 0:16:11.070 the trials that have played out this year, including the 0:16:11.110 --> 0:16:14.270 most recent one with the state AGs in Meta, age assurance, 0:16:14.360 --> 0:16:16.660 age verification is something that not only Meta, but the 0:16:16.700 --> 0:16:19.040 entire tech industry has struggled with for years. 0:16:18.800 --> 0:16:19.340 At this point. 0:16:20.020 --> 0:16:23.440 And given that several of the safety features that Meta 0:16:23.460 --> 0:16:28.800 is using committing to are really hinging almost entirely on 0:16:28.840 --> 0:16:31.640 their ability to correctly determine a user's age. I think 0:16:31.660 --> 0:16:34.970 there's a lot of open questions about how effectively all 0:16:34.990 --> 0:16:37.290 of this is going to work if this indeed is 0:16:37.350 --> 0:16:40.450 something that Meta and its competitors have been challenged with. 0:16:41.360 --> 0:16:43.480 You and Riley Griffin put out the story where the 0:16:43.520 --> 0:16:46.720 headline is, kids will be kids. And that's the risk 0:16:47.140 --> 0:16:50.040 with MetaSafeguards, the opening. How does a company know for 0:16:50.100 --> 0:16:52.860 sure if a kid is a kid on the internet? 0:16:52.880 --> 0:16:57.250 There's actually like two ideas there. Verification that kids will 0:16:57.270 --> 0:16:59.910 want to push the boat on how many hours they 0:16:59.930 --> 0:17:00.950 can use on these platforms. 0:17:01.590 --> 0:17:02.070 Of course. 0:17:02.250 --> 0:17:04.810 And even just overnight and this morning, if you look 0:17:04.869 --> 0:17:07.470 on social media and you start to see kids' reactions 0:17:07.550 --> 0:17:10.399 to some of this news, you see that some of 0:17:10.440 --> 0:17:12.280 them don't agree with it. Some of them think it'll 0:17:12.300 --> 0:17:16.820 be a good thing. Some of them, just regularly, some 0:17:16.880 --> 0:17:19.740 of them discuss how they have very easy time signing 0:17:19.800 --> 0:17:22.540 up for all of these platforms with very little pushback 0:17:22.859 --> 0:17:26.190 without using their real birthdays. So I really do think 0:17:26.230 --> 0:17:29.270 that there's sort of a tension between what the companies 0:17:29.310 --> 0:17:32.490 are putting on the parents, the kids, the families, and 0:17:32.530 --> 0:17:35.240 then what the parents, kids, and families are actually going 0:17:35.270 --> 0:17:38.670 to do. But it remains to be seen whether any 0:17:38.690 --> 0:17:41.669 of Meta's efforts to really strengthen its age assurance technology 0:17:41.750 --> 0:17:42.510 are going to pan out. 0:17:43.450 --> 0:17:46.429 Bloomberg's Alex Levine. Thank you very much. Let's stay on Meta. 0:17:46.450 --> 0:17:49.990 Joining us now is Danielle Seatrum, professor of law at 0:17:50.020 --> 0:17:52.879 the University of Virginia Law School, School of Law, who 0:17:52.920 --> 0:17:57.480 believes the outcome is, quote, a pittance for Meta. Let's 0:17:57.520 --> 0:17:59.640 start there. What do you mean by that, professor? 0:18:01.250 --> 0:18:05.290 You know, when Meta settled with the FTC for $ 5 0:18:04.859 --> 0:18:09.420 billion over the Cambridge Analytica affair, that was, I think, 0:18:09.560 --> 0:18:12.670 something like a couple of days' worth of their advertising income. 0:18:13.609 --> 0:18:18.270 So let's just put in perspective the profits of $ 46 billion, 0:18:18.320 --> 0:18:21.780 I think, from last quarter or this part of this year. 0:18:22.800 --> 0:18:25.220 The first half of this year was something like $ 50 billion. 0:18:26.280 --> 0:18:31.020 So $ 17. 5 billion over 10 years is a small amount 0:18:31.560 --> 0:18:34.980 of their income. It's a small hit they were happy 0:18:35.020 --> 0:18:35.380 to take. 0:18:36.480 --> 0:18:39.740 So you are talking about the financial settlement. But for 0:18:39.780 --> 0:18:44.140 lots of people, the changes that Meta has agreed to, 0:18:44.550 --> 0:18:47.660 to the Facebook and Instagram platforms, are the news, right? 0:18:47.700 --> 0:18:51.990 These are significant policy changes and operational changes. So in 0:18:52.130 --> 0:18:55.409 one example, two-hour daily limits, and it is for the 0:18:55.490 --> 0:18:59.430 parents to authorize any additional time beyond that. That brings 0:18:59.470 --> 0:19:01.889 the role of the parent into this. Is that a 0:19:01.930 --> 0:19:04.330 sufficient mechanism for enforcement? 0:19:06.369 --> 0:19:09.640 As a parent myself... So I don't mean to demean parents. 0:19:09.780 --> 0:19:14.940 Parents are overwhelmed. They're overtaxed. And to put the burden 0:19:14.980 --> 0:19:18.160 on parents, you know, kids are very persuasive. So if 0:19:18.240 --> 0:19:20.790 they want to override and insist to their parents, like, 0:19:20.810 --> 0:19:24.850 you can trust me, you know, allow me to override 0:19:24.910 --> 0:19:27.929 this two hour limit, they're going to do it. And 0:19:27.950 --> 0:19:31.359 I think relying on parents is a mistake. that both 0:19:31.520 --> 0:19:35.260 law and the children's online privacy protection act and sort 0:19:35.300 --> 0:19:39.740 of our current lawmakers approach and, you know, uh, parental 0:19:39.780 --> 0:19:43.430 control is the approach of the FTC. It's the current 0:19:43.490 --> 0:19:46.050 emphasis on current law and legal proposals. 0:19:46.070 --> 0:19:47.510 And I think it's a mistake. 0:19:48.090 --> 0:19:53.380 Parents are overburdened. Um, they often have their kids' best 0:19:53.460 --> 0:19:58.820 interests in mind. But kids are persuasive with kids. And 0:19:58.880 --> 0:20:01.800 for some kids, it's dangerous. So if you put parents 0:20:01.840 --> 0:20:06.330 in the driver's seat and kids have emerging sexual orientation, 0:20:06.350 --> 0:20:10.450 they're experimenting, their gender identity... it can be quite dangerous 0:20:10.490 --> 0:20:13.170 for parents to be in total control of their social media. 0:20:14.270 --> 0:20:17.600 I've studied the court filing for the agreement. I see 0:20:17.710 --> 0:20:23.439 no legal requirement imposed on parents to monitor, enroll in 0:20:23.460 --> 0:20:27.980 the supervision. I see an agreement that regulates meta. Just 0:20:28.800 --> 0:20:32.250 that mechanism, explain why it will or will not work 0:20:32.310 --> 0:20:32.790 in your view. 0:20:34.880 --> 0:20:38.670 So parents, so under the Children's Online Privacy Protection Act, 0:20:38.910 --> 0:20:42.270 which is one of the series of claims in the lawsuit, 0:20:42.930 --> 0:20:47.290 it requires that parents give verified parental consent for the collection, use, 0:20:47.350 --> 0:20:53.050 and sharing of children's data, so data of kids under 13. 0:20:53.050 --> 0:20:56.730 And so in that regard, we are putting parents, law 0:20:56.910 --> 0:21:01.139 is asking parents to give permission. But as your prior, 0:21:01.180 --> 0:21:06.420 you know, kids often, you know, they get around that 0:21:06.520 --> 0:21:10.240 requirement by putting an age for their birth date that 0:21:10.300 --> 0:21:13.820 makes them over 13. So we get around parental consent 0:21:14.200 --> 0:21:17.240 in that first way. You know, the second is parents 0:21:17.260 --> 0:21:19.639 will often give that consent, even if kids, you know, 0:21:19.740 --> 0:21:23.540 are honest about their birth dates. And putting parents in 0:21:23.580 --> 0:21:27.420 charge of that. So the default is under this settlement agreement, 0:21:27.880 --> 0:21:31.970 That kids will have only two hours, so teens, and 0:21:31.990 --> 0:21:35.729 that parents can override that default. You can imagine that 0:21:35.810 --> 0:21:41.060 that default will be easily overridden. Kids talking to their 0:21:41.100 --> 0:21:43.260 parents or simply doing it themselves. Hey, mom, can I 0:21:43.260 --> 0:21:47.030 have your iPad? Can I have your laptop? That is, 0:21:47.290 --> 0:21:51.850 kids are pretty persuasive. They're really smart. They carve out privacy. 0:21:52.470 --> 0:21:55.230 We know this from social science that they do it themselves. 0:21:55.390 --> 0:21:58.980 I'm not convinced that the architectural promises made in this 0:21:59.040 --> 0:22:04.360 settlement will do very much. And unfortunately, the settlement doesn't 0:22:04.420 --> 0:22:08.869 prevent the amassing, the use of the sharing internally of 0:22:08.920 --> 0:22:11.510