WEBVTT - Nvidia Gets Into the PC Market With New Chip 0:00:02.520 --> 0:00:13.520 Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is alive 0:00:13.560 --> 0:00:17.360 from coast to coast with Caroline Hyde in New York 0:00:17.680 --> 0:00:19.599 and ever though in sent Francisco. 0:00:22.560 --> 0:00:25.239 This is Bloomberg Tech coming up in videos, Next target 0:00:25.520 --> 0:00:27.760 the PC and u AI chip sends and video and 0:00:27.800 --> 0:00:32.200 Friends soaring while rival slide after Jensen Hung's Computech's. 0:00:31.680 --> 0:00:34.960 Keynote, Plus All Eyes on SpaceX will dive into how 0:00:35.000 --> 0:00:38.080 the upcoming Mega Ibo is already reshaping. 0:00:37.600 --> 0:00:39.520 Wall Street and New York. 0:00:39.560 --> 0:00:41.680 Tech Week kicks off today, will be joined by Tech 0:00:41.840 --> 0:00:44.720 NYC CEO Judy Samuels on what to expect. 0:00:44.840 --> 0:00:47.080 First, we check in on how the market is trying 0:00:47.080 --> 0:00:49.960 to digest once again concerns about the Middle East conflict, 0:00:49.960 --> 0:00:53.040 concerns that sees far is getting further away to have 0:00:53.159 --> 0:00:55.800 any sort of long term perspective with the US and 0:00:55.880 --> 0:00:58.360 Iranian talk seeming to break down at this current moment. 0:00:58.560 --> 0:01:00.880 Look we're looking at we're two tens of percent higher 0:01:00.880 --> 0:01:03.120 on the last that one hundred. Big tech is trying 0:01:03.120 --> 0:01:05.920 to shake off the geopolitical risk ed. More broadly, take 0:01:05.959 --> 0:01:06.319 a look. 0:01:06.200 --> 0:01:11.720 At it, yep, Nvidia taking on the PC market for decades. 0:01:11.920 --> 0:01:15.160 You find in Nvidia GPUs in the PC all about 0:01:15.240 --> 0:01:16.000 gaming now. 0:01:16.160 --> 0:01:17.120 It is the CPU. 0:01:17.200 --> 0:01:20.640 It's based on ARM architecture, ARMS up sixteen percent. It 0:01:20.720 --> 0:01:24.399 takes on the traditional market of Intel and AMD. They 0:01:24.440 --> 0:01:27.959 are down significantly in Vidia outline the specs here they. 0:01:27.840 --> 0:01:34.920 Are introducing RTX Spark. Everything we've learned over thirty three 0:01:35.000 --> 0:01:42.040 years distilled into one show Blackwell RTXGPU with six thousand, 0:01:42.280 --> 0:01:45.600 one hundred and forty four couter course, one petaflop of 0:01:45.720 --> 0:01:50.760 AI performance, a custom twenty core Grace CPU built in 0:01:50.840 --> 0:01:56.680 partnership with Media Tech, fused by m vlage one hundred 0:01:56.720 --> 0:01:59.520 and twenty eight gigabytes of unified. 0:01:59.120 --> 0:02:03.639 Member Bloomberg, Zy and King with us leads our courage 0:02:03.640 --> 0:02:07.640 of semiconductors that was classically in video typically Jensen and 0:02:07.680 --> 0:02:11.160 its presentation. What do we need to know about RTX 0:02:11.160 --> 0:02:12.079 Spark super chip. 0:02:12.320 --> 0:02:14.440 Yeah, I mean this has been go rumored for a 0:02:14.480 --> 0:02:16.800 long time that in video would directly get into this 0:02:16.919 --> 0:02:19.960 market with a CPU, and now we've finally seen it. 0:02:20.280 --> 0:02:22.480 And the way he's packaged it is he's saying, look, 0:02:22.480 --> 0:02:24.520 if AI is coming to the laptop, if we're going 0:02:24.560 --> 0:02:26.520 to start talking to our laptops, if we're going to 0:02:26.520 --> 0:02:29.520 start expecting them to think, we'll need a different kind 0:02:29.520 --> 0:02:31.760 of chip. And here's what I prepared earlier. 0:02:32.560 --> 0:02:33.200 And why is. 0:02:33.200 --> 0:02:35.600 It so significant that it's better than XAVY six? Why 0:02:35.639 --> 0:02:37.880 is this such a competitive threat as the market seems 0:02:37.919 --> 0:02:39.639 to think, to the likes of Intel, to the likes 0:02:39.680 --> 0:02:40.680 even of Qualcom. 0:02:41.720 --> 0:02:44.920 Well, we asked them repeatedly, give us a comparison, how 0:02:45.360 --> 0:02:48.320 is it better tell us exactly what it does that 0:02:48.360 --> 0:02:51.320 the others can't. And what they said was wait and see, 0:02:51.720 --> 0:02:54.280 will show you when these devices come out. But the 0:02:54.320 --> 0:02:56.440 market is taking this as Look, there's been lots of 0:02:56.520 --> 0:02:59.799 attempts to get our architecture into the PC market. None 0:02:59.840 --> 0:03:03.600 of them have gone particularly well, apart from arguably Apples. 0:03:03.960 --> 0:03:05.920 But guess what this is in Vidia. This is a 0:03:05.960 --> 0:03:08.800 company with all the resources it needs, with an established 0:03:08.800 --> 0:03:12.239 brand in PCs, and with a you know, a real target, 0:03:12.800 --> 0:03:16.080 and a technology that can differentiate itself with. So that's 0:03:16.120 --> 0:03:18.679 why we're seeing Intel and AMD trading off. 0:03:19.280 --> 0:03:19.960 In Vidia is. 0:03:20.080 --> 0:03:22.280 Holding its gains at four percent. I find that a 0:03:22.320 --> 0:03:25.120 really interesting reaction. If you remember in and I know 0:03:25.120 --> 0:03:27.120 it as a while ago, didn't you and I spend 0:03:27.120 --> 0:03:31.040 a whole month testing aipcs. They're out there, they exists. 0:03:31.080 --> 0:03:34.239 In that case, it was quite commonside. You know, what 0:03:34.560 --> 0:03:36.960 do you make of that? You know, Jensen's kind of 0:03:36.960 --> 0:03:39.440 introducing this as a brand new category, but I thought 0:03:39.480 --> 0:03:40.600 it it is really there. 0:03:40.680 --> 0:03:43.240 No, I mean it isn't. ARM in the PC is 0:03:43.280 --> 0:03:45.800 not there. In VideA itself tried it more than a 0:03:45.800 --> 0:03:48.440 decade ago. It didn't work out. So this isn't a 0:03:48.480 --> 0:03:53.000 revolutionary idea at a fundamental level. But again, this is 0:03:53.040 --> 0:03:56.520 in Vidia, this is the biggest company in the world, and. 0:03:56.440 --> 0:03:58.600 This is media tech, and this is ARM and all 0:03:58.600 --> 0:04:00.360 the other companies are getting a base bit in the 0:04:00.440 --> 0:04:02.680 glory as well. Ian King, is so great to have 0:04:02.720 --> 0:04:04.320 you on the show to start us off. Look and 0:04:04.400 --> 0:04:07.360 video clearly remains at the very center of the AI trade, 0:04:07.360 --> 0:04:11.240 but as the technology matures, investors have been diversifying their bets. 0:04:11.560 --> 0:04:14.400 It's all of the future opportunities. We say, Matt witness 0:04:14.440 --> 0:04:17.560 hither with us all Spring Global Investments portfolio manager. I 0:04:17.600 --> 0:04:20.600 mean we had seen a bit of a well plateauing 0:04:20.800 --> 0:04:23.400 of in videos rise of late, but now it manages 0:04:23.440 --> 0:04:25.800 to rekick start. How much do you still need to 0:04:25.839 --> 0:04:28.920 be in the very infrastructure the bowels of the AI trade. 0:04:29.480 --> 0:04:32.440 Well, it's a great important question right now, and you 0:04:32.520 --> 0:04:35.040 talked a little bit about some of the uncertainty in 0:04:35.080 --> 0:04:37.919 the market. I think it's really forcing investors to remain 0:04:37.960 --> 0:04:40.760 in some of those safe haven names. Ours cap take 0:04:41.000 --> 0:04:44.320 tech makes sense, strong visibility with cash flows and their positioning, 0:04:44.360 --> 0:04:47.280 But we think diversification is going to be important going forward. 0:04:47.880 --> 0:04:51.520 Concentration risk is real, so we're taking a holistic approach 0:04:51.600 --> 0:04:56.479 to tech in the AI ecosystem. As those CAPEX dollars 0:04:56.480 --> 0:05:00.279 started to spread across the AI value chain, so big 0:05:00.320 --> 0:05:02.159 cap tech names like the Mac seven are going to 0:05:02.160 --> 0:05:04.800 play an important part with the evolution of AI, but 0:05:04.800 --> 0:05:07.320 it's it's important to start allocating some of those investment 0:05:07.400 --> 0:05:08.920 dollars to other areas of the market. 0:05:09.839 --> 0:05:13.560 Matt, how seriously are you taking this nvideo entry into 0:05:13.960 --> 0:05:16.000 the PC market with a CPU. 0:05:17.560 --> 0:05:18.719 Well, it's important. 0:05:18.760 --> 0:05:22.839 It's it's gonna that edge kind of compute as as 0:05:23.400 --> 0:05:27.920 that that insatiable demand for more performance kind of extends 0:05:27.960 --> 0:05:29.200 to those end products. 0:05:29.400 --> 0:05:31.000 They're going to have to be more capable. 0:05:31.040 --> 0:05:34.200 And so as we move to inference and AGENTIC AI 0:05:34.279 --> 0:05:36.520 and physical AI, we're going to see different parts of 0:05:36.520 --> 0:05:39.479 the market with different requirements and eventually new winners. So 0:05:39.760 --> 0:05:43.120 I think it's important to have opportunities in different solutions 0:05:43.120 --> 0:05:45.720 sets to handle that. And I think in Vidia coming 0:05:45.720 --> 0:05:48.080 into the market there's going to play an important role. 0:05:48.640 --> 0:05:52.720 We've seen entire adjacent industries take off that are at 0:05:52.760 --> 0:05:54.599 the heart of the bottlenecks. We think of the energy 0:05:54.640 --> 0:05:57.839 trade that has picked up, even nuclear companies. We've seen 0:05:58.120 --> 0:06:00.720 the focus on new types of when it comes to 0:06:00.760 --> 0:06:04.039 photonics map. But where are the opportunities that haven't already 0:06:04.080 --> 0:06:05.080 been ridden in this wave? 0:06:06.160 --> 0:06:08.600 Yeah, and you know, we're focusing a lot of our 0:06:08.640 --> 0:06:11.960 time on the application layer right now. Look, we kind 0:06:11.960 --> 0:06:15.080 of look and view as that the AI value chain 0:06:15.160 --> 0:06:18.000 is having kind of five important layers. You've got energy 0:06:18.000 --> 0:06:19.880 and power like you talked about you talked about the 0:06:20.440 --> 0:06:24.680 infrastructure with Chips, computing infrastructure. You've got the hyperscalers, You've 0:06:24.680 --> 0:06:26.640 got those AI models, and then on top of that 0:06:26.720 --> 0:06:31.160 is the application layers. Those are those those adopters and 0:06:31.200 --> 0:06:33.920 the abilities for industries where they're going to be able 0:06:33.920 --> 0:06:37.120 to drive automation, decision support and kind of productivity gains. 0:06:37.240 --> 0:06:39.080 They're going to need software to do a lot of 0:06:39.080 --> 0:06:41.240 that for them, and that's where we've been spending a 0:06:41.240 --> 0:06:44.479 lot of our time and focus from an investment perspective. 0:06:44.520 --> 0:06:46.080 Oh, I don't want to front run and where ED 0:06:46.200 --> 0:06:48.320 is going to take this conversation just after this map. 0:06:48.360 --> 0:06:51.800 But software has been getting a bid today, So are 0:06:51.839 --> 0:06:53.920 you a believer that there are going to be ultimate 0:06:53.960 --> 0:06:56.160 winners out of us going to get more detailed on 0:06:56.200 --> 0:06:57.400 the application there for us? 0:06:58.120 --> 0:06:59.240 Well? I think so. 0:06:59.400 --> 0:07:02.240 I mean, I think companies are starting to focus on 0:07:02.400 --> 0:07:05.600 how important it is to have that AI is going 0:07:05.640 --> 0:07:07.479 to be an important driver of what they're going to 0:07:07.480 --> 0:07:09.680 be able to do, and they're going to need software 0:07:09.720 --> 0:07:12.000 to come along with them along that journey. And so 0:07:12.360 --> 0:07:16.040 those companies that have robust access to that data, that 0:07:16.160 --> 0:07:19.640 have the operational scale and are fully embedded within their 0:07:20.040 --> 0:07:22.520 customer suites, I think are going to be important to 0:07:22.520 --> 0:07:25.040 have that ecosystem help get them to where they're going 0:07:25.080 --> 0:07:27.800 to be going when it comes to productivity, enhancement or 0:07:27.800 --> 0:07:29.040 whatever they're going to use AI for. 0:07:31.160 --> 0:07:33.600 It's a very difficult morning to make sense of what's 0:07:33.600 --> 0:07:35.600 going on in the world. I've got some d ram 0:07:35.640 --> 0:07:38.640 and nan flash data we kind of breather in May 0:07:39.000 --> 0:07:42.560 pricing slowed down flat on land. There is still an 0:07:42.560 --> 0:07:46.120 ongoing situation in the Middle East. IDC has this forecast 0:07:46.120 --> 0:07:48.280 for the PC market that the PC market overall is 0:07:48.280 --> 0:07:52.320 going to decline eleven percent in calendar twenty six. When 0:07:52.360 --> 0:07:55.440 you have a big moment like computechs and Jensen Wong, 0:07:55.560 --> 0:07:57.720 probably the most important person in the world of technology 0:07:57.800 --> 0:07:59.920 is on stage and the market reacts like it does, 0:08:00.640 --> 0:08:02.920 do you actually learn anything useful from that map? 0:08:03.920 --> 0:08:06.640 Well, I think it just gets back to the diversification 0:08:06.960 --> 0:08:09.720 points that we want to make sure that our clients 0:08:09.880 --> 0:08:13.600 are paying attention to. You need to have a holistic 0:08:13.640 --> 0:08:16.200 approach to whatever you're doing from an investment perspective, whether 0:08:16.200 --> 0:08:18.480 it be across all ten sectors. But I also think 0:08:18.520 --> 0:08:21.960 it's important just within the AI ecosystem itself to make 0:08:22.000 --> 0:08:24.640 sure you're spreading your bets across to all those layers, 0:08:24.880 --> 0:08:26.760 to make sure when there are big moves, when there 0:08:26.760 --> 0:08:29.840 are those bottlenecks that are going to force pricing up 0:08:30.080 --> 0:08:33.360 and really force the agenda, I think you need to 0:08:33.400 --> 0:08:36.520 make sure that you have exposure to all five layers 0:08:36.679 --> 0:08:38.360 to make sure that you don't lose out on any 0:08:38.360 --> 0:08:39.760 opportunities that are out there. 0:08:39.960 --> 0:08:42.640 And Matt, we become very myopic sat in the United 0:08:42.679 --> 0:08:45.679 States with the biggest winners being US names. But this 0:08:45.760 --> 0:08:48.400 is a global supply chain and with some huge wins 0:08:48.400 --> 0:08:50.600 in South Korea, over in Japan and Europe. Where are 0:08:50.640 --> 0:08:53.040 you thinking geographically, Well. 0:08:53.040 --> 0:08:55.120 I think it starts here, primarily here in the US, 0:08:55.160 --> 0:08:57.480 but I think there are some names outside of the 0:08:57.600 --> 0:09:00.200 US ASML I think is a great example of that. 0:09:00.200 --> 0:09:02.640 That's going to play an important part of the build 0:09:02.640 --> 0:09:05.920 out and their ability to help provide with these these 0:09:05.960 --> 0:09:09.319 big customers need to drive those product sets in the 0:09:09.400 --> 0:09:09.960 right direction. 0:09:12.040 --> 0:09:13.360 Matt, just really really quick. 0:09:13.559 --> 0:09:16.280 You don't stay up at night worried about memory bottlenecks 0:09:16.400 --> 0:09:17.880 or you do lose a lot of sleep because of that. 0:09:18.720 --> 0:09:20.360 I do, I absolutely do. 0:09:20.679 --> 0:09:23.120 I think it's you know, these cycles are probably the 0:09:23.200 --> 0:09:27.000 duration of these cycles are going to continue to evolve 0:09:27.040 --> 0:09:29.520 and move out further down the timeline, and I think 0:09:29.520 --> 0:09:31.000 a lot of that has to do with the fact 0:09:31.000 --> 0:09:34.640 that these clean rooms are are are are There's only 0:09:34.720 --> 0:09:37.120 a few of them out there, and they're they're. 0:09:36.960 --> 0:09:37.800 Taking up time. 0:09:38.120 --> 0:09:39.880 These green fields are going to take two to three 0:09:39.920 --> 0:09:42.640 years before they come online. So I think the memory 0:09:42.720 --> 0:09:45.640 names are going to continue to to to drive pricing 0:09:45.640 --> 0:09:47.400 in the direction that they needed to do for the 0:09:47.679 --> 0:09:48.480 stocks to work. 0:09:49.280 --> 0:09:50.920 Matt Weimer of all Spring. Great to have you back 0:09:50.960 --> 0:09:52.679 on the show, Thank you very much. Coming up, we're 0:09:52.679 --> 0:09:55.240 going to discuss how the SpaceX I p O is 0:09:55.280 --> 0:09:57.480 already changing the world of finance. 0:09:57.559 --> 0:09:57.959 Details. 0:09:58.040 --> 0:10:00.680 Next, the other big story out of copy techs and 0:10:00.760 --> 0:10:03.360 in the markets is software. Look at the gains for 0:10:03.400 --> 0:10:06.400 example on service now in Adobe very simply put Gensen 0:10:06.400 --> 0:10:09.600 one arguing that the biggest users of software will be 0:10:09.640 --> 0:10:10.280 AI agents. 0:10:10.320 --> 0:10:12.559 Later in the show, will dig in. This is Bloomberg Tech. 0:10:22.320 --> 0:10:26.280 Even before it begins trading. SpaceX is reshaping Wall Street. 0:10:26.280 --> 0:10:28.880 Elon Musk is pitching investors on a company that combines 0:10:28.960 --> 0:10:32.559 rockets and AI with a claimed market opportunity twenty eight 0:10:32.600 --> 0:10:36.280 and a half trillion dollars. That is today's big number. 0:10:36.559 --> 0:10:39.920 That scale is challenging long held assumptions about how companies 0:10:39.920 --> 0:10:43.760 are valued, who ultimately drives stock market demand. Bloomberg's tech 0:10:43.800 --> 0:10:46.520 equity reporter Isabeli is one of the team on today's 0:10:46.520 --> 0:10:50.760 big take all about how already SpaceX is just so large, 0:10:50.760 --> 0:10:53.920 so consequential that benchmark's having to tear up the rule 0:10:53.920 --> 0:10:56.000 books as to how they let them start trading. 0:10:56.160 --> 0:10:57.880 I couldn't have put it better tearing up the rule 0:10:57.920 --> 0:10:59.960 book or read buysing if you want to be more diplomat. 0:11:00.440 --> 0:11:03.000 So the companies behind the most familiar names you can 0:11:03.040 --> 0:11:05.199 think of, like Nasak one hundred and foot Sea Russell, 0:11:05.360 --> 0:11:08.040 they've already changed the rule book from Nasak the seasoning 0:11:08.080 --> 0:11:10.960 period or the call it the waiting period when the 0:11:11.000 --> 0:11:14.520 company debuts they go public, they have to sit there 0:11:14.559 --> 0:11:17.320 for a while before three months before they'll consider you 0:11:17.360 --> 0:11:19.400 to be part of the Nasak one hundred, they've narrowed 0:11:19.400 --> 0:11:21.960 that down to fifteen days. For the foot Seat it's 0:11:22.240 --> 0:11:25.320 five days. The SMPS and consultation from twelve months to 0:11:25.440 --> 0:11:28.080 six months. So of course you have proponents saying, yes, 0:11:28.160 --> 0:11:30.400 this is what we want because it will reflect the market. 0:11:30.640 --> 0:11:33.400 But you have critics saying, but wait, when companies go public, 0:11:33.400 --> 0:11:35.320 they're usually volatile, and don't you want to wait and 0:11:35.360 --> 0:11:37.720 see first, you know, have a normal price discovery. So 0:11:38.000 --> 0:11:40.720 of course, just like with anything, there's critics and there 0:11:40.720 --> 0:11:42.040 are also supporters. 0:11:41.600 --> 0:11:44.080 And the critics a worried perhaps about protections for retail 0:11:44.080 --> 0:11:47.120 investors or new entrants. Meanwhile, SpaceX is all about the 0:11:47.120 --> 0:11:50.600 retail investor. I mean a thirty percent allocation towards them. 0:11:50.960 --> 0:11:54.360 How are we seeing this managing to go into sometimes 0:11:54.559 --> 0:11:58.120 Elo Musk's criticism himself of passive investments exactly. 0:11:58.160 --> 0:12:01.200 So in the tracking fund estimated to buy nearly twenty 0:12:01.240 --> 0:12:03.520 billion dollars worth of SpaceX. That's massive. I don't think 0:12:03.520 --> 0:12:06.240 we've seen that before. And going with retail traders. I 0:12:06.280 --> 0:12:08.680 went on Reddit, which is the most reliable source on Earth. 0:12:08.760 --> 0:12:11.920 That's obviously a joke. I'm smiling if people are just 0:12:11.960 --> 0:12:15.000 listening and some people were like, why will we even 0:12:15.040 --> 0:12:17.960 buy this? It will be in our brokerage, in the 0:12:18.000 --> 0:12:19.920 index funds in just a few months. So there is 0:12:19.960 --> 0:12:23.600 that thinking of like, how will this reshape passive investing? 0:12:23.640 --> 0:12:26.400 Will traditional money managers have to reallocate because they don't 0:12:26.440 --> 0:12:28.679 want to be too overweight on one thing? I mean, 0:12:28.880 --> 0:12:31.760 where is SpaceX going to be categorized under even industrials 0:12:31.920 --> 0:12:35.800 tech or communication? So lots of unknown but definitely consequential. 0:12:39.040 --> 0:12:41.840 As well. Tough decisions need to be made. 0:12:42.040 --> 0:12:45.560 We spent a lot of last week talking about Tesla 0:12:45.880 --> 0:12:49.120 and the idea that SpaceX merges with Tesla in the future. 0:12:49.400 --> 0:12:51.080 You just showed a quote from Scott Sogny. 0:12:51.360 --> 0:12:54.640 One point he makes is that actually Tesla and Bitcoin 0:12:54.760 --> 0:12:56.720 is examples. You know, and I thought about the bitcoin holdings 0:12:56.720 --> 0:12:57.840 at SpaceX last week. 0:12:58.280 --> 0:12:59.959 They might actually get hit by this. 0:13:00.360 --> 0:13:04.079 Why because a play on Tesla and a play on 0:13:04.120 --> 0:13:06.920 SpaceX is really just a belief on Elon. And if 0:13:06.920 --> 0:13:09.920 you're invested in Tesla and SpaceX, are you too over 0:13:10.000 --> 0:13:11.040 concentrated on an Elon? 0:13:11.120 --> 0:13:11.240 Now? 0:13:11.280 --> 0:13:13.520 I know that these are two separate companies, and within 0:13:13.559 --> 0:13:16.520 those two companies are multiple companies, it feels like, and 0:13:17.000 --> 0:13:20.120 they have different valuations, different technicals, but it's all really 0:13:20.120 --> 0:13:22.080 Elon at the end of the day. So that's what 0:13:22.240 --> 0:13:25.040 one analyst was telling me. Also, like, if you're into Tesla, 0:13:25.040 --> 0:13:26.880 would you sell that because you're going to be in 0:13:26.880 --> 0:13:29.560 SpaceX and SpaceX of course it's also in the index. 0:13:29.600 --> 0:13:31.880 So really lots of over concentrated risk. And it's not 0:13:31.960 --> 0:13:35.640 just Tesla. I mean, soon we'll see OpenAI, we'll see Anthropic, 0:13:35.679 --> 0:13:37.840 which the show has covered a lot. What's that going 0:13:37.840 --> 0:13:40.880 to mean for our for your brokerage as an ordinary American? 0:13:40.920 --> 0:13:43.000 I mean, you're going to have such a huge concentration 0:13:43.200 --> 0:13:45.120 of chech and do you want that for your portfolio? 0:13:45.720 --> 0:13:49.280 Blimberg ysabarly tough stuff, Thank you very much. Now, when 0:13:49.280 --> 0:13:51.840 the Apple Watch went on sale in twenty fifteen, it's 0:13:52.120 --> 0:13:55.040 on not just the other high tech wearables but also 0:13:55.120 --> 0:13:58.120 the wider mechanical watch market, and as you can see, 0:13:58.240 --> 0:14:01.480 that meant hit to sales of traditional in that space. Now, 0:14:01.520 --> 0:14:04.880 Apple's aiming to use that playbook in the eyewear market. 0:14:04.920 --> 0:14:08.200 Bloombost Mark German leads our coverage of Apple and that's 0:14:08.240 --> 0:14:09.880 the subject of this weekend's power On. 0:14:10.559 --> 0:14:11.440 We've talked about. 0:14:11.200 --> 0:14:14.720 This before, you know why how Apple and the ewear market. 0:14:14.720 --> 0:14:17.560 But that historic playbook, you broke it down. What do 0:14:17.559 --> 0:14:18.040 we need to know? 0:14:19.240 --> 0:14:22.200 Yeah, it's interesting when the Apple Watch launch, you see 0:14:22.200 --> 0:14:25.120 what you saw. Within a few years the entire mid 0:14:25.200 --> 0:14:28.600 tier mechanical watch industry get completely upended. 0:14:28.720 --> 0:14:28.920 Right. 0:14:29.200 --> 0:14:32.240 There was Movado, there was Fossil, there was Swatch, and 0:14:32.520 --> 0:14:35.840 those companies are still around, but their revenues are considerably 0:14:35.920 --> 0:14:37.920 lower than they were. 0:14:37.840 --> 0:14:38.760 A decade ago. 0:14:39.080 --> 0:14:41.880 The high end of the watch market, companies like Pateech, 0:14:42.120 --> 0:14:46.360 rolex ap all the hot ones right now and the 0:14:46.400 --> 0:14:49.520 Covid luxury boom they did fine and what you're going 0:14:49.560 --> 0:14:51.680 to see is the same thing play out in glasses 0:14:52.000 --> 0:14:55.920 I believe where Apple and therefore other smart glasses makers 0:14:55.960 --> 0:14:59.960 are going to completely disrupt that mid tier market brand. 0:15:00.080 --> 0:15:04.400 It's like ray Ban, Warby Parker, etc. Now you've seen 0:15:04.480 --> 0:15:07.840 ray Ban, Warby Parker Pivot too Smart. So really the 0:15:07.920 --> 0:15:09.840 non smart glasses I think are going to. 0:15:09.760 --> 0:15:10.560 Get wiped out. 0:15:10.880 --> 0:15:14.320 But the very high end Maison Bonet Cardier, et cetera. 0:15:14.720 --> 0:15:15.800 Those guys I think are going. 0:15:15.800 --> 0:15:16.240 To be okay. 0:15:16.600 --> 0:15:19.080 I mean, we really do see how RayBan has positioned 0:15:19.160 --> 0:15:23.480 himself like Zotoca more broadly alongside Meta. But where do 0:15:23.560 --> 0:15:26.720 you think has also managed to start to get in 0:15:26.960 --> 0:15:29.280 ahead of the curve from a digital tech perspective, not 0:15:29.320 --> 0:15:31.400 having to just depend on we'd tie ups like we 0:15:31.440 --> 0:15:34.240 see with APN Swatch for example, which does prove pretty successful. 0:15:35.280 --> 0:15:35.480 Yeah. 0:15:35.600 --> 0:15:37.560 No, I mean Meta has done a great job. Obviously, 0:15:37.600 --> 0:15:39.960 they've pioneered the space there. They're going to be launching 0:15:39.960 --> 0:15:42.840 new smart glasses. I guess it's June first already, so 0:15:42.960 --> 0:15:46.600 later this month. They have partnerships with Exertal Exotica. I 0:15:46.640 --> 0:15:49.000 would expect them to eventually roll out new designs that 0:15:49.040 --> 0:15:52.600 are in house form factors. So certainly it's going to 0:15:52.640 --> 0:15:55.240 be a pretty booming space, and Apple coming into the 0:15:55.280 --> 0:15:58.720 space with an iPhone exclusive pair of smart glasses is 0:15:58.720 --> 0:16:00.800 going to drive interest for the rest the industry, and 0:16:00.840 --> 0:16:03.280 I think Meta has a strong chance for the android 0:16:03.360 --> 0:16:04.000 side of the world. 0:16:05.560 --> 0:16:07.920 Very very quick, Mark, is there anything that we learn 0:16:08.040 --> 0:16:11.600 from from the vision pro story about how they approach glasses? 0:16:13.360 --> 0:16:16.040 You know, I think that there's really it's really two 0:16:16.040 --> 0:16:19.880 different stories, you know, the vision Pro obviously thirty five dollars, 0:16:20.920 --> 0:16:26.120 very heavy, very tech tour de force on the glasses 0:16:26.240 --> 0:16:29.880 something quite lighter and a lot cheaper. So I think 0:16:29.880 --> 0:16:31.920 they're two distinct categories. 0:16:32.640 --> 0:16:34.800 Really is again one of the most read things over 0:16:34.840 --> 0:16:37.160 the course of the weekend, Bloomberg's Mark Gum and thank 0:16:37.200 --> 0:16:39.480 you very much, all things power on. Next coming up 0:16:39.560 --> 0:16:43.960 from video generation to robotics, Luma Ai CEO and Jane's 0:16:43.960 --> 0:16:46.080 going to be joining us on the next frontier in 0:16:46.200 --> 0:16:48.800 artificial intelligence. This is blombag Tech. 0:17:01.840 --> 0:17:05.280 Another Nvidio message at computext. The next AI race is 0:17:05.320 --> 0:17:08.160 in the physical world. That brings us to Lumerai, which 0:17:08.160 --> 0:17:12.400 is launching an open research lab focused on solving generalization 0:17:12.720 --> 0:17:16.480 in robotics. Joining us now CEO and it Jane. We've 0:17:16.480 --> 0:17:19.520 talked quite a bit on this program about the challenge. 0:17:19.560 --> 0:17:22.919 The challenge is that in the physical world, models need 0:17:22.960 --> 0:17:27.119 to be multimodal. But I think let's start by defining 0:17:27.160 --> 0:17:30.480 the problem generalization. That might be a newer term. What 0:17:30.600 --> 0:17:31.800 are you what are you getting at there? 0:17:31.960 --> 0:17:32.120 Oh? 0:17:32.160 --> 0:17:32.959 Thanks for having me. 0:17:33.000 --> 0:17:37.439 First of all, currently, pretty much all robots are trained 0:17:37.440 --> 0:17:41.040 by showing a few examples of specific tasks, So the 0:17:41.200 --> 0:17:44.480 entire field basically trains one task at a time. If 0:17:44.480 --> 0:17:46.680 you compare that to the world of language models, where 0:17:46.720 --> 0:17:48.800 you know, you train a large model and then it 0:17:48.840 --> 0:17:51.240 is able to handle a wide variety of tasks unseen 0:17:51.280 --> 0:17:54.120 problems as well. So that is an example of generalization. 0:17:54.480 --> 0:17:57.600 And in robotics we are we have this critical gap where, 0:17:57.680 --> 0:17:59.440 like you know, we are just stuck in the value 0:17:59.480 --> 0:18:03.679 of specific tasks in order for robotics to be impactful 0:18:03.680 --> 0:18:05.439 in the world and for us to be able to 0:18:05.480 --> 0:18:07.040 just talk to them and ask them like, hey, okay, 0:18:07.359 --> 0:18:09.800 do this, then like you know, when you're done with that, 0:18:10.040 --> 0:18:11.200 go go take care of that thing. 0:18:11.640 --> 0:18:14.720 Maybe a new scenario being presented for the first time. 0:18:14.520 --> 0:18:18.080 For the first time, So generalization is this problem of 0:18:18.240 --> 0:18:21.440 how do we allow robots to solve generally any. 0:18:21.240 --> 0:18:24.359 Task, even in a world where you have access to 0:18:24.359 --> 0:18:27.440 a lot of synthetic or virtual data. Having a grounding 0:18:27.480 --> 0:18:29.879 in physics the real physics of the world is a 0:18:29.920 --> 0:18:32.480 principle challenge. So why is it that having an open 0:18:32.480 --> 0:18:34.320 science phys quai that have is a solution? 0:18:34.720 --> 0:18:36.320 It sounds somewhat abstract with. 0:18:36.200 --> 0:18:39.320 Respect, right, So I mean there's two aspects to this. 0:18:39.640 --> 0:18:42.199 One that it is open and the second is the 0:18:42.200 --> 0:18:43.720 world that we are specifically doing. 0:18:43.520 --> 0:18:45.679 Just open being open source, you mean open. 0:18:45.440 --> 0:18:48.719 Being open science, open source, both of those aspects. So, 0:18:49.760 --> 0:18:51.960 first of all, what the solution actually looks like from 0:18:52.000 --> 0:18:57.080 our perspective. Currently, the best solution is brute forcing this approach, 0:18:57.080 --> 0:19:00.120 which is gather data on every single task you can 0:19:00.119 --> 0:19:03.080 imagining combination of task everything humans do from picking up 0:19:03.119 --> 0:19:05.479 cups to you know, digging like you know, minds all 0:19:05.480 --> 0:19:09.080 of these things and gather data one piece at a time. 0:19:09.320 --> 0:19:12.879 That's a practically impossible solution. On the other hand, the 0:19:12.920 --> 0:19:14.760 teams at LOOMA, the work we have been doing for 0:19:14.800 --> 0:19:17.639 the past four years is in building out these general 0:19:17.640 --> 0:19:21.040 systems out of multimodeal data Internet scale multimodel data and 0:19:21.160 --> 0:19:25.760 extracting signals from that. That allows control, that allows simulation 0:19:25.800 --> 0:19:28.560