WEBVTT - The AI Genie Could Deliver on the Hype 0:00:02.360 --> 0:00:06.880 You're listening to Asia Centric from Bloomberg Intelligence, the podcast 0:00:06.920 --> 0:00:10.160 that pulls back the curtain non global business so you 0:00:10.200 --> 0:00:14.320 can invest better across the Pacific realm. I'm Tom Corbett 0:00:14.360 --> 0:00:15.480 in Hong Kong. 0:00:15.400 --> 0:00:19.680 And I'm John Lee. Artificial intelligence, its rapid emergence has 0:00:19.720 --> 0:00:22.919 shaken up the tech world with eye popping potential to 0:00:23.040 --> 0:00:25.520 change the way we work, live, and do business. 0:00:25.720 --> 0:00:29.640 From chips to the cloud, and from data centers to chatbots. 0:00:29.720 --> 0:00:34.320 AI is already plowing ahead, opening up a world of possibilities, 0:00:34.320 --> 0:00:37.800 but also prompting some to waive the caution flag. 0:00:38.120 --> 0:00:42.480 Some estimates value AI's business opportunity in the billions and trillions, 0:00:42.840 --> 0:00:45.640 But what are the risks and can the aigenie deliver 0:00:45.760 --> 0:00:46.040 on the. 0:00:46.040 --> 0:00:50.240 Hype As we start moving more towards AI processing worst 0:00:50.240 --> 0:00:54.440 case scenario by twenty thirty, if things aren't done correctly 0:00:54.680 --> 0:00:59.000 and there aren't any energy efficiency bottles that are implemented, 0:00:59.360 --> 0:01:02.720 it could count ten percent of all of global energy production. 0:01:03.240 --> 0:01:08.040 Let's bring in Woojin Ho, senior tech analyst with Bloomberg Intelligence. 0:01:08.120 --> 0:01:11.440 Woojin Welcome, Hey Tom, thanks for having me on, Hi, John, 0:01:11.440 --> 0:01:12.399 how are you? Hi? 0:01:12.440 --> 0:01:16.240 With Jin which in chat GPT sparked off huge excitement 0:01:16.280 --> 0:01:20.320 on artificial intelligence. Is this hype real and how big 0:01:20.360 --> 0:01:21.560 is this market opportunity? 0:01:21.959 --> 0:01:24.680 I do think the hype is real. I think this 0:01:24.760 --> 0:01:28.440 is very much unlike the metaverse. This is something that 0:01:28.920 --> 0:01:32.320 can increase the productivity for people, and this is why 0:01:32.360 --> 0:01:35.920 we're starting to see a lot of real investment going 0:01:35.959 --> 0:01:40.440 in by the cloud providers like Microsoft, Google, Amazon and 0:01:40.760 --> 0:01:42.080 Meta on the US side. 0:01:42.440 --> 0:01:44.600 And how big is a market opportunity do you believe? 0:01:45.000 --> 0:01:48.240 It's still unclear right because we can look at it 0:01:48.280 --> 0:01:51.520 in multiple different facets. There's an infrastructure side of the 0:01:51.600 --> 0:01:54.600 ledger and there's also a services side the ledger. We 0:01:54.640 --> 0:01:57.639 do know that hundreds of millions of dollars have already 0:01:57.640 --> 0:02:01.240 been invested in it right now not seeing much revenue. 0:02:01.600 --> 0:02:04.520 But if they can start producing revenue or make it 0:02:04.520 --> 0:02:08.080 into a revenue generating opportunity from a service aside, we're 0:02:08.120 --> 0:02:11.360 talking about the scopes of hundreds of billions and possibly 0:02:11.440 --> 0:02:12.160 into trillions. 0:02:12.400 --> 0:02:15.120 Wou jin Ho, you talked about how you believe the 0:02:15.200 --> 0:02:18.640 hype around AI is real. Can we dive down a 0:02:18.680 --> 0:02:21.600 little deeper into that. What are we overlooking or what 0:02:21.639 --> 0:02:25.360 are we underestimating in terms of the potential, the power, 0:02:25.480 --> 0:02:26.120 and the risk. 0:02:27.040 --> 0:02:30.680 Sure, So think about how we're using AI today. It's 0:02:30.680 --> 0:02:33.240 more for efficiency gains, and I think we just started 0:02:33.280 --> 0:02:37.320 to tap into what we can do with AI. Think 0:02:37.320 --> 0:02:40.400 about what AI can do, what chat, GPT and other 0:02:40.480 --> 0:02:43.399 AI engines can do like it. It can tap into 0:02:43.560 --> 0:02:48.480 multiple different databases and come out with an output that 0:02:49.080 --> 0:02:52.200 would probably happen and split seconds that could have happened 0:02:52.200 --> 0:02:55.640 in hours, days or months. Right, So we're talking about 0:02:55.639 --> 0:02:58.840 some massive productivity gains just from the use of AI, 0:02:58.960 --> 0:03:00.440 if it's done and use properly. 0:03:00.760 --> 0:03:04.200 Is it an overstatement to say that AI represents a 0:03:04.440 --> 0:03:07.120 quantum leap in the evolution of tech? 0:03:07.680 --> 0:03:09.880 Still unclear if that's going to be the case, right, 0:03:10.200 --> 0:03:12.600 But it does seem as if it's a step function 0:03:12.760 --> 0:03:15.639 change in terms of tech, much like what we saw 0:03:15.639 --> 0:03:19.320 on the Internet back in the early nineties. And once 0:03:19.320 --> 0:03:24.279 people learn how to use AI and the AI tools effectively, 0:03:24.680 --> 0:03:28.400 you'll see those productivity games potentially in. 0:03:28.360 --> 0:03:31.280 Which companies do you believe stand to benefit from this trend? 0:03:32.040 --> 0:03:34.920 Well, we're already starting to see some of the early beneficiaries. 0:03:34.920 --> 0:03:38.320 It's right now, it's mainly hardware and video is the 0:03:38.360 --> 0:03:39.680 first company that comes to mind. 0:03:39.760 --> 0:03:41.720 Clearly they manufacture or. 0:03:41.600 --> 0:03:45.320 Produce the GPUs, but there's also the compute processing side, 0:03:45.760 --> 0:03:50.040 which Intel as well as a MD and then you 0:03:50.080 --> 0:03:53.440 have some of the networking players like Arrested Networks and 0:03:53.480 --> 0:03:57.440 networking chip guys like Broadcom that produce chips to help 0:03:57.440 --> 0:03:59.160 support the AI side. 0:03:59.640 --> 0:04:03.000 And how much computer power does AI require? For example, 0:04:03.040 --> 0:04:06.640 if we ask a question to Google search a traditional 0:04:06.680 --> 0:04:10.240 search versus a say, an AI powered search like CHATGBT, 0:04:10.760 --> 0:04:12.880 can you give us some sense of how much data 0:04:13.440 --> 0:04:14.600 and searching is required. 0:04:15.320 --> 0:04:17.839 So if we think about the compute power that's needed 0:04:17.920 --> 0:04:22.440 right now, you need the highest available processor and processor speed. 0:04:22.880 --> 0:04:25.880 Now that there are two levels of compute that we 0:04:25.920 --> 0:04:29.120 need to think about. First is actual on the compute 0:04:29.120 --> 0:04:32.039 processing side, and that's where Intel and AMD come in, 0:04:32.440 --> 0:04:35.159 and think about it in the scope of ten to 0:04:35.440 --> 0:04:40.200 one hundred times in compute power to process all of 0:04:40.200 --> 0:04:43.320 the information that's coming through those servers. Right now, we're 0:04:43.320 --> 0:04:48.000 talking about the GPUs which Nvidia offers. The cloud guys 0:04:48.080 --> 0:04:51.120 just can't get their hands or enough GPUs to support 0:04:51.200 --> 0:04:53.720 what they want to do on AI. We're talking about 0:04:53.800 --> 0:04:56.960 thousands and tens of thousands of GPU chips and GP 0:04:57.040 --> 0:05:00.000 aligned cars that they want to support. So we're talking 0:05:00.120 --> 0:05:03.920 about a lot of processing power is necessary just to 0:05:03.920 --> 0:05:07.559 do some high end AI functions relative to general cloud 0:05:07.560 --> 0:05:09.000 computing functions. 0:05:09.240 --> 0:05:13.839 And woujin ho. What constraints do those power concerns impose 0:05:14.160 --> 0:05:16.440 upon AI and its potential. 0:05:17.160 --> 0:05:20.880 Well, that's a challenge here right when we're starting about 0:05:20.960 --> 0:05:25.320 high power compute, we're talking about a lot of energy 0:05:25.320 --> 0:05:28.479 that's being used an dissipated coming out of these servers 0:05:28.480 --> 0:05:30.960 and these data centers. Let me give you a stat 0:05:31.839 --> 0:05:35.640 all the data centers owned by the clouds they generate 0:05:35.880 --> 0:05:39.240 or use up roughly consumer about three percent of all 0:05:39.320 --> 0:05:43.320 of the energy that's being produced globally. And as we 0:05:43.400 --> 0:05:49.159 start moving more towards AI processing, because they are and 0:05:49.440 --> 0:05:54.520 energy suck, that could potentially continue to increase the energy consumption. 0:05:54.960 --> 0:05:59.280 Worst case scenario by twenty thirty, if things aren't done 0:05:59.360 --> 0:06:05.520 correctly and there aren't any energy efficiency models that are implemented, 0:06:06.040 --> 0:06:08.680 it could account for ten percent of all of global 0:06:08.760 --> 0:06:09.440 energy production. 0:06:09.640 --> 0:06:11.640 We're talking about the cloud guys. 0:06:11.640 --> 0:06:15.320 Which think this sounds like a potential environmental nightmare. 0:06:15.640 --> 0:06:18.480 Well, for some of the Bloomberg listeners who are very 0:06:18.800 --> 0:06:22.840 ESG conscious it is really a potential nightmare for the 0:06:22.839 --> 0:06:23.680 cloud guys. 0:06:24.120 --> 0:06:27.240 Could you draw analogies to say cryptocurrency. I read this 0:06:27.320 --> 0:06:30.719 stat that bitcoin usage consumes more power than some mid 0:06:30.760 --> 0:06:31.880 sized countries in the world. 0:06:32.400 --> 0:06:35.440 Yeah, if you think about the bitcoin ecosystem, it's still 0:06:35.440 --> 0:06:36.200 a fairly. 0:06:36.120 --> 0:06:38.520 Concentrated ecosystem if you think about it. 0:06:38.920 --> 0:06:43.120 There are some bitcoin farms that try to build out 0:06:43.200 --> 0:06:47.799 their own servers, but now we're talking about larger cloud 0:06:47.839 --> 0:06:51.479 guys that are trying to build AI farms at a 0:06:51.560 --> 0:06:54.799 larger and grander scale than than a bigcoin server farm. 0:06:55.120 --> 0:06:58.600 We're not talking about thousands of people that are trying 0:06:58.600 --> 0:07:03.080 to mine bitcoin. We're talking about millions and hundreds of 0:07:03.120 --> 0:07:06.920 millions of people to billions of people trying to access 0:07:07.240 --> 0:07:10.239 AI data coming out of the cloud, and the cloud 0:07:10.280 --> 0:07:13.680 guys are trying to support that. I do believe that 0:07:13.720 --> 0:07:17.480 the hyperscale cloud providers are cognizant of that, and they 0:07:17.520 --> 0:07:21.840 are trying to implement new methodologies to reduce the power consumption. 0:07:22.000 --> 0:07:23.320 But that's going to come over time. 0:07:23.760 --> 0:07:27.320 And wou Jinho a lot of those questions about AI 0:07:27.600 --> 0:07:31.400 and chat GPT and the power that they require, the 0:07:31.560 --> 0:07:34.200 US is not the only country that wants to pursue 0:07:34.200 --> 0:07:37.880 a piece of this technology. China is also very interested, 0:07:38.200 --> 0:07:43.040 and you have enough power constraints in China right now 0:07:43.320 --> 0:07:46.480 where that could be an issue for them if they 0:07:46.480 --> 0:07:47.360 go down this road. 0:07:48.000 --> 0:07:51.800 And China has two challenges. One for us is the 0:07:51.840 --> 0:07:55.240 access to the latest generation of chips. And one of 0:07:55.280 --> 0:07:59.280 the advantages of getting the latest generation of chips is 0:07:59.760 --> 0:08:04.400 less power consumption because the smaller geometry chips doesn't consume. 0:08:04.120 --> 0:08:05.800 As much power and it's much more efficient. 0:08:06.360 --> 0:08:09.680 And two you need fewer of those boxes as the 0:08:09.720 --> 0:08:13.080 AI engine becomes more efficient. The issue is is that 0:08:13.200 --> 0:08:16.200 China will not have access to the latest generations of chips. 0:08:16.320 --> 0:08:21.040 There actually, and Video offers a China version of their GPUs, 0:08:22.040 --> 0:08:25.200 whether it's as an Apple eight hundred or the new 0:08:25.440 --> 0:08:29.520 hs In Henry eight hundred, which is the China version 0:08:29.560 --> 0:08:31.920 of those chips, They're just not going to be as 0:08:32.200 --> 0:08:37.000 efficient and power efficient as China. So to compensate, what 0:08:37.120 --> 0:08:40.280 China may need to do is to build more of 0:08:40.360 --> 0:08:43.800 the server and these AI clusters, and that's just going 0:08:43.840 --> 0:08:46.360 to consume much more power. The power needs are just 0:08:46.400 --> 0:08:49.520 going to increase and grow. And to your point, Tom, 0:08:49.760 --> 0:08:52.920 you know China has been a power starved We've seen 0:08:52.920 --> 0:08:56.480 the blackouts during twenty one and twenty two, and if 0:08:56.520 --> 0:08:58.720 that's going to be an issue, the lights might go 0:08:58.840 --> 0:09:01.640 out because they really want to push the AI engine. 0:09:01.640 --> 0:09:04.679 You can see a world of clashing imperatives emerging because 0:09:04.920 --> 0:09:08.840 upwards of two thirds of China's power is generated today 0:09:09.360 --> 0:09:12.360 by coal, which is of course a fossil fuel, so 0:09:12.400 --> 0:09:15.640 you have that factor coming in as well. 0:09:15.040 --> 0:09:17.880 Right, And that's exactly the point, right, It's like you 0:09:18.000 --> 0:09:20.720 have to pick the better of two evils. Do you 0:09:20.760 --> 0:09:23.400 want to really start pushing the AI engine just to 0:09:23.400 --> 0:09:25.720 compete with the rest of the world or the US 0:09:25.720 --> 0:09:28.560 in this case, or do you want to manage and 0:09:28.600 --> 0:09:32.360 conserve energy to support factories on a day to day basis. 0:09:32.600 --> 0:09:35.320 Would you just continue from what you were saying. US 0:09:35.360 --> 0:09:40.000 regulations are prohibiting certain chips to be exported to China, correct. 0:09:40.080 --> 0:09:40.840 That is correct. 0:09:40.920 --> 0:09:44.640 So there are some sanctions that the US imposed on 0:09:45.040 --> 0:09:48.640 US technologies that they cannot ship to China, and Nvidia 0:09:48.800 --> 0:09:51.840 is one of those companies and they actually line itemed 0:09:52.240 --> 0:09:56.000 a couple of Nvidia chips, Nvidia AI chips, the A 0:09:56.160 --> 0:10:00.439 one hundred and above. So what the US is essentially 0:10:00.440 --> 0:10:03.920 trying to do is slow down the progress of China's 0:10:03.920 --> 0:10:05.160 AI development. 0:10:05.080 --> 0:10:08.959 So China can still use certain invidio chips. I think 0:10:09.000 --> 0:10:12.079 you mentioned A eight hundred. Can you compare the eight 0:10:12.559 --> 0:10:14.679 eight hundred to say the most advanced of the A 0:10:14.840 --> 0:10:16.240 one hundred or the H one hundred. 0:10:16.920 --> 0:10:19.480 Sure, So some of the notes that I looked at 0:10:19.760 --> 0:10:21.920 in preparation for this, and it just seems as if 0:10:22.280 --> 0:10:25.440 a lot of it is the interconnect communications between the 0:10:25.480 --> 0:10:30.440 graphics cards, right, that's one. So speed and latency is 0:10:30.559 --> 0:10:35.080 very key for FURAI and also with the processing power 0:10:35.280 --> 0:10:38.040 of those chips as well, so it's not going to 0:10:38.040 --> 0:10:40.319 be as powerful, and that's one of the reasons why 0:10:40.840 --> 0:10:46.199 China may need to have more servers and more GPUs 0:10:46.240 --> 0:10:48.760 to compensate for the less processing power. 0:10:49.080 --> 0:10:52.200 So how many years of generations behind would China be 0:10:52.320 --> 0:10:54.920 in terms of, you know, the usage of these Nvidia chips. 0:10:55.480 --> 0:10:58.520 Well, if you want to add capacity, they'll try to 0:10:58.600 --> 0:11:02.000 keep up as close as possible my senses, and if 0:11:02.040 --> 0:11:04.679 we think about the power dynamic as well, the energy 0:11:04.679 --> 0:11:07.679 consumption dynamic as well, they could be anywhere between three 0:11:07.760 --> 0:11:11.360 to five years behind in terms of the chips itself. 0:11:12.559 --> 0:11:16.480 You're listening to Asia Centric from Bloomberg Intelligence. By the way, 0:11:16.600 --> 0:11:19.200 if you like what you hear, and we hope you do, 0:11:19.600 --> 0:11:24.520 please rate us on Apple Podcasts, Spotify, Google Podcasts, or 0:11:24.559 --> 0:11:27.520 wherever you may be listening to us. Of course, more 0:11:27.520 --> 0:11:31.360 stars are better. Your feedback matters, and we love hearing 0:11:31.360 --> 0:11:36.800 from our listeners. Our guest is wu Jin Ho, senior 0:11:36.880 --> 0:11:41.280 tech analyst with Bloomberg Intelligence. Wujin The Financial Times ran 0:11:41.400 --> 0:11:45.720 an opinion piece in early May which rather boldly declared 0:11:45.800 --> 0:11:50.400 that Apple is a in their words, Chinese company. They 0:11:50.520 --> 0:11:53.280 argue that Apple is so deeply embedded in China that 0:11:53.400 --> 0:11:57.160 rapid decoupling becomes a rather tall order. How do you 0:11:57.240 --> 0:12:00.440 see that unfolding? Is China going to let Apple just 0:12:00.600 --> 0:12:04.000 go quietly? What are the risks to Apple of moving 0:12:04.120 --> 0:12:08.359 out of China and moving their manufacturing to India and elsewhere. 0:12:08.800 --> 0:12:11.880 That's a fantastic question, tom So, if we think about 0:12:11.960 --> 0:12:16.120 what's happening with Apple in particular, as well as some 0:12:16.160 --> 0:12:18.800 of the geopolitical pressure that the US has started to 0:12:18.840 --> 0:12:22.560 press US manufacturers, it seems as if Apple has. 0:12:22.480 --> 0:12:24.280 Been moving along quite rapidly. 0:12:24.800 --> 0:12:27.160 We actually did a lot of work on the Chinese 0:12:27.160 --> 0:12:30.360 supply chain prior to the ft, and initially we thought 0:12:30.360 --> 0:12:34.280 it was about over ninety nine percent of Apple smartphones 0:12:34.280 --> 0:12:37.040 with manufacturing in China. Now I'm going to give a 0:12:37.120 --> 0:12:39.880 nod to one of my colleagues, Stephen Saying, based out 0:12:39.880 --> 0:12:43.559 of Taiwan. We did another layer of analysis on the 0:12:43.640 --> 0:12:46.840 China supply chain and it seems as if the Chinese 0:12:46.880 --> 0:12:50.559 manufacturers and the assemblers have really started to move out 0:12:50.559 --> 0:12:54.600 of their operations to de risk their geographical exposure. 0:12:54.720 --> 0:12:57.800 It's just as if it's more like close to thirty. 0:12:57.480 --> 0:13:01.720 Five percent of the factories are in China now and 0:13:02.240 --> 0:13:05.839 they've started to move out to the Asian right, whether 0:13:05.880 --> 0:13:09.320 it's Vietnam and Malaysia and as well as a little 0:13:09.360 --> 0:13:12.280 bit more in India. Now, China is still going to 0:13:12.280 --> 0:13:15.320 be a mainstay. I still argue that there's still a 0:13:15.320 --> 0:13:19.080 lot of manufacturing expectise in China because of their multiple 0:13:19.160 --> 0:13:23.760 years of smartphone development and smartphone component development. But at 0:13:23.800 --> 0:13:26.560 the end of the day, the assembly part of it 0:13:26.880 --> 0:13:29.040 and some of the component part of it, not only 0:13:29.080 --> 0:13:33.959 by the US manufacturers, but the Chinese manufacturers have started 0:13:34.000 --> 0:13:36.880 to move things out as well because they recognize the 0:13:36.920 --> 0:13:39.439 geopolitical risk that some of the U S sanctions may 0:13:39.480 --> 0:13:40.800 potentially hold going forward. 0:13:40.920 --> 0:13:45.400 So, currently ninety eight percent of iPhones are assembled in China. 0:13:45.520 --> 0:13:48.440 What's your expectations for say, like five years out. 0:13:48.200 --> 0:13:51.880 Well, ninety eight percent of the smartphones have actually shifted more, 0:13:52.320 --> 0:13:56.040 has accelerated a lot faster, and it could be closer 0:13:56.080 --> 0:14:00.520 to forty percent, you know, in twenty thirty or even 0:14:00.559 --> 0:14:05.400 thirty percent in twenty thirty in China as these assembly 0:14:05.760 --> 0:14:11.560 manufacturing sites move into Vietnam, Malaysia, in India. Now, I 0:14:11.600 --> 0:14:14.920 know India has been a hot button topic recently because 0:14:15.200 --> 0:14:19.160 Apple has been propping up some manufacturing facilities out there 0:14:19.200 --> 0:14:25.000 as well, and India has its advantages, but they still 0:14:25.000 --> 0:14:27.120 have to ramp up in terms of the infrastructure. 0:14:27.880 --> 0:14:30.920 Is India You're ready for that kind of a challenge technologically? 0:14:31.560 --> 0:14:37.160 Technologically there is engineering expertise. The one thing that India 0:14:37.160 --> 0:14:39.480 still needs to build up that that China has and 0:14:39.560 --> 0:14:43.000 India lacks is an ecosystem, right and that will take 0:14:43.040 --> 0:14:46.560 multiple years to build up. The other thing is the infrastructure. 0:14:47.080 --> 0:14:50.920 Where they lag in infrastructure is. 0:14:50.880 --> 0:14:52.280 On the communication side. 0:14:52.960 --> 0:14:56.080 If I look at the speed test, dot net or 0:14:56.200 --> 0:14:59.640 UKULA data, they rank eighty eighth in the world in 0:14:59.720 --> 0:15:04.200 terms terms of high speed fiber optic communications. But if 0:15:04.240 --> 0:15:06.640 we compare it to where China and the US, they're 0:15:06.680 --> 0:15:09.400 fourth and eighth in the world. And I want to 0:15:09.440 --> 0:15:11.640 go back in history where how did China get to 0:15:11.680 --> 0:15:15.040 where it is today as a major manufacturing hub. One 0:15:15.080 --> 0:15:17.640 of the things that China did starting in twenty twenty 0:15:18.240 --> 0:15:21.120 was to lay out the fiber optic networks, and they 0:15:21.240 --> 0:15:25.400 represent roughly one third of all fiber optic investments globally, 0:15:25.440 --> 0:15:29.520 one third. And once you get the communication networks now down, 0:15:30.200 --> 0:15:34.880 then you can start the digital transformation necessary for everything 0:15:34.920 --> 0:15:39.400 else that is touched technologically, whether it's a manufacturing site, 0:15:39.520 --> 0:15:42.160 whether it's a consumer site, whether it's the Internet, and 0:15:42.200 --> 0:15:44.920 so on and so forth. India has yet to be there, 0:15:45.320 --> 0:15:48.160 and India will not get there until twenty twenty five, 0:15:48.240 --> 0:15:49.000 twenty twenty six. 0:15:49.160 --> 0:15:51.840 So the fiber optic network's being laid down, that's the 0:15:51.960 --> 0:15:54.720 tipping point for where China stands today. Would you say 0:15:55.000 --> 0:15:56.760 that's the case or is that an overstatement? 0:15:57.240 --> 0:15:57.320 No? 0:15:57.680 --> 0:16:00.920 I think that China made the playbook. I think Korea 0:16:01.000 --> 0:16:04.320 made the playboat as well. India started to recognize that 0:16:04.720 --> 0:16:08.000 and I would even say that US lags from a 0:16:08.000 --> 0:16:11.680 fiber rollout, and they're really started to lay out fiber 0:16:11.720 --> 0:16:14.040 as well over the next three to five years to 0:16:14.680 --> 0:16:16.560 really build up their digital infrastructure. 0:16:16.840 --> 0:16:19.120 And I do view fiber. 0:16:19.360 --> 0:16:22.400 I know that five G is touted as the communication 0:16:22.520 --> 0:16:25.480 but you cannot have wireless unless you have a wire 0:16:26.080 --> 0:16:29.440 to connect it back to a network. So fiber is 0:16:29.960 --> 0:16:33.080 the key tipping point for digital transformation success. 0:16:33.600 --> 0:16:37.240 Well, jin can Apple effectively detangle from the China supply chain. 0:16:37.560 --> 0:16:40.600 Is the certain parts of the manufacturing supply chain that 0:16:40.640 --> 0:16:42.480 you just cannot get away from China. 0:16:42.600 --> 0:16:48.480 In terms of manufacturing, well, there are several parts. Quite frankly, 0:16:48.600 --> 0:16:51.280 if you look at some of the components where China 0:16:51.400 --> 0:16:56.120 leads in, they lead in LCD panel, screen panel. We 0:16:56.240 --> 0:16:59.120 start to see that ship from Samsung and LG going into. 0:16:58.920 --> 0:17:01.600 BOE and it's mainly manufactured in China. 0:17:02.480 --> 0:17:05.480 If you see the display glass that's manufactured in China, 0:17:05.640 --> 0:17:10.280 some of the acoustics components manufactured in China. And one 0:17:10.320 --> 0:17:13.160 of the things that China smartly did over the last 0:17:13.240 --> 0:17:17.560 couple of years is to double down on legacy, more 0:17:17.680 --> 0:17:22.840 mundane chips, whether it's power management chips, if they can't 0:17:22.840 --> 0:17:27.160 get the advanced node chips that TSMC offers, they're going 0:17:27.240 --> 0:17:30.000 to be the market winners for some of these legacy 0:17:30.040 --> 0:17:33.160 node chips, and for every advanced NOE chip, you will 0:17:33.200 --> 0:17:37.360 need quite a bit of legacy node content inside the iPhones, 0:17:37.359 --> 0:17:39.000 and Apples just can't get away from that either. 0:17:39.760 --> 0:17:44.320 So the idea of Apple decoupling completely from China, you're saying, 0:17:44.359 --> 0:17:46.000 that's probably not going to happen. 0:17:46.840 --> 0:17:50.040 I don't think that's not only an Apple statement. I 0:17:50.080 --> 0:17:55.080 think that's a global electronics statement. I think the aspirations 0:17:55.119 --> 0:17:58.960 are there, but I do think that China is going 0:17:59.000 --> 0:18:02.800 to be a last thing manufacturing partner over the long term. 0:18:03.280 --> 0:18:06.000 Was you if I could circle back to artificial intelligence 0:18:06.080 --> 0:18:10.639 or AI, which Asian companies stand to benefit from this trend? 0:18:11.240 --> 0:18:13.840 Well, let's start from the bottom up. It's going to 0:18:13.880 --> 0:18:18.280 be the server companies predominantly. They will be getting the 0:18:19.400 --> 0:18:22.919 large GPU line cards and chips from Nvidia and Intel 0:18:22.920 --> 0:18:25.040 an AMD. But it will be a companies someone like 0:18:25.400 --> 0:18:29.000 an Insper or a Lenovo or an H three C 0:18:29.640 --> 0:18:31.919 those other companies. But if we start moving up the 0:18:32.040 --> 0:18:34.760 chain on the services side, it's going to be someone 0:18:34.840 --> 0:18:38.080 like a buy Do or Tencent as they build out 0:18:38.119 --> 0:18:41.280 those AI services and try to offer it to the 0:18:41.320 --> 0:18:43.199 local economies and local regions. 0:18:43.520 --> 0:18:46.120 And what about on the hardware side, Well, on. 0:18:46.080 --> 0:18:50.440 The hardware side, it's going to be inspur Lenovo H 0:18:50.560 --> 0:18:54.600 three C, as much as I want to say someone 0:18:54.640 --> 0:18:55.400 like a Huawei. 0:18:55.640 --> 0:18:57.840 On the networking side. 0:18:58.080 --> 0:19:03.520 There's somewhat negatively impacted because of the US sanctions as well. 0:19:03.560 --> 0:19:06.679 They just don't have the access to the technologies on 0:19:06.720 --> 0:19:10.359 the networking side as they hope. So if anyone's going 0:19:10.440 --> 0:19:12.520 to benefit from the networking side, it's probably going to 0:19:12.560 --> 0:19:13.159 be H three C. 0:19:14.080 --> 0:19:17.600 And what about the Asian cheap makers like Taiwan Semi, 0:19:17.640 --> 0:19:20.440 TSMC or Samson Electronics or esk Heinex. 0:19:20.840 --> 0:19:23.399 So if we think about it from the chip maker front, 0:19:23.680 --> 0:19:27.280 AI needs a lot of advanced process notes and TSMC 0:19:27.359 --> 0:19:30.840 is going to be the leading beneficiary right now because 0:19:31.000 --> 0:19:35.480 they make the five nanometers seven animeter and in line 0:19:35.520 --> 0:19:39.359 for the threes and the two animeters, and for AI workloads, 0:19:39.440 --> 0:19:43.000 you actually ead a lot, a lot, a lot of memory. 0:19:43.320 --> 0:19:46.919 And if we think about ske Heinex and Samsung not 0:19:46.960 --> 0:19:49.800 only the DRAM side, but also on the nan side, 0:19:50.280 --> 0:19:53.600 they are going to be the biggest beneficiary among the 0:19:53.640 --> 0:19:55.440 Asian technology. 0:19:54.880 --> 0:19:59.879 Providers, wujin Hole. Could it be that we don't know 0:20:00.440 --> 0:20:04.600 what we don't know about AI and chat GPT and 0:20:04.640 --> 0:20:08.400 its potential and how would you compare its emergence today 0:20:08.600 --> 0:20:11.560 with the way the Internet emerged in the early to 0:20:11.600 --> 0:20:12.680 mid nineteen nineties. 0:20:13.480 --> 0:20:16.520 So there's a lot that we don't know, to your point, 0:20:16.600 --> 0:20:19.600 that we don't know going out twenty thirty years from now, 0:20:20.080 --> 0:20:23.680 sure in terms of what chet chipee can do, how 0:20:23.720 --> 0:20:27.000 it'll evolve. If we think about what the early Internet 0:20:27.240 --> 0:20:32.000 was in the nineteen nineties, it was basic email, right. 0:20:32.160 --> 0:20:33.960 And then it evolved. 0:20:33.520 --> 0:20:36.960 To very simple graphics to virtual. 0:20:36.600 --> 0:20:40.480 Realities, a clunkier version then of what it is today. 0:20:40.600 --> 0:20:42.359 Exactly much clunkiler version. 0:20:42.400 --> 0:20:45.280 But if you think about what the Internet actually provided 0:20:45.400 --> 0:20:48.440 back in the early nineteen nineties, it created a whole 0:20:48.840 --> 0:20:53.040 vast of news services. We could imagine a service like 0:20:53.080 --> 0:20:58.280 an Amazon or an Uber, or food delivery services like delivery. 0:20:58.440 --> 0:20:59.720 Let's just say back. 0:20:59.520 --> 0:21:01.879 Then, it's not just the technology, it's the way it's 0:21:01.920 --> 0:21:03.440 applied exactly. 0:21:03.600 --> 0:21:06.720 I don't know how people are going to apply AI 0:21:06.920 --> 0:21:10.480 going forward. There are some scary aspects of it in 0:21:10.560 --> 0:21:14.359 terms of the productivity gains of AI will lead to 0:21:14.440 --> 0:21:19.239 potential job losses, but it might be productivity gains that 0:21:19.320 --> 0:21:22.600 we can really leverage and there might be more jobs 0:21:22.640 --> 0:21:24.320 created as a result of that. 0:21:24.880 --> 0:21:26.400 So it's just I don't know. 0:21:26.359 --> 0:21:29.480 What's going to happen in the next five, ten, twenty 0:21:29.560 --> 0:21:32.320 thirty years from now from the evolution of AI. 0:21:32.960 --> 0:21:36.800 But those productivity gains could also bring with them investment 0:21:36.920 --> 0:21:40.879 opportunity exactly, economic growth, and all the things that go 0:21:40.960 --> 0:21:42.400 on with that exactly. 0:21:42.480 --> 0:21:46.800 It can actually be seeds to new industry. It could 0:21:46.840 --> 0:21:50.159 be seeds to new technologies that AI was able to 0:21:50.160 --> 0:21:54.640 discover in the speed of years which would have been decades. 0:21:55.200 --> 0:21:59.720 So it depends on how companies implement AI and how 0:21:59.720 --> 0:22:04.040 they leverage the technology to really drive growth going forward. 0:22:04.040 --> 0:22:07.040 And I think every company. Look, I understand that the 0:22:07.040 --> 0:22:09.680 cloud guys, whether it's either the US or the Chinese 0:22:10.400 --> 0:22:13.600 on the China side, have gotten a lot of attention, 0:22:14.160 --> 0:22:16.600 but I know that a lot of corporates are trying 0:22:16.600 --> 0:22:19.600 to figure out new ways or ways to leverage AI 0:22:20.080 --> 0:22:22.200 for their own benefit and we'll see what. 0:22:22.160 --> 0:22:22.639 They do with it. 0:22:23.080 --> 0:22:24.760 Which there seems to be a lot of press on 0:22:24.880 --> 0:22:29.280 how AI can help with the search functions like Google Microsoft. 0:22:29.880 --> 0:22:32.320 Do you see a point where people actually pay to 0:22:32.600 --> 0:22:34.840 have subscription to these services. 0:22:35.800 --> 0:22:38.200 Yes, but not for a search function. 0:22:38.600 --> 0:22:43.000 I do think that again, going back to how AI evolves, 0:22:43.200 --> 0:22:45.720