WEBVTT - Chip War Author on Global Fight for Tech Supremacy 0:00:01.000 --> 0:00:05.280 Semiconductors are at the heart of America's geopolitical rivalry with China. 0:00:05.960 --> 0:00:09.640 They're vital to products we use every day, including smartphones 0:00:09.680 --> 0:00:14.880 and computers, and for advanced technologies like AI, electric vehicles 0:00:14.920 --> 0:00:18.800 and defense systems. China's made in twenty twenty five policy 0:00:19.320 --> 0:00:23.600 a decade ago called for aggressive manufacturing and self sufficiency 0:00:23.640 --> 0:00:24.239 in this space. 0:00:25.040 --> 0:00:28.680 That's been worrying Washington ever since, with both the Biden 0:00:28.720 --> 0:00:32.720 and Trump administrations adding tougher rules like cutting access to 0:00:32.760 --> 0:00:36.800 American technology and certain semiconductors, and it's been met with 0:00:36.880 --> 0:00:40.480 limited success. Chinese firms are still finding ways of getting 0:00:40.520 --> 0:00:44.519 the hands on US and Taiwanese advanced semiconductors, and the 0:00:44.600 --> 0:00:48.200 Chinese government continues to double down on its own industry. 0:00:49.200 --> 0:00:53.320 So where are we in a great chip war? Who's winning? 0:00:53.520 --> 0:00:56.840 If anyone? Could we see more thawing in the relationship 0:00:57.000 --> 0:00:59.800 as China and the US try to negotiate a trade deal, 0:01:00.480 --> 0:01:03.760 and how can regions like Taiwan, South Korea and Japan 0:01:04.240 --> 0:01:08.240 benefit from this rivalry. You're listening to Asia Centric from 0:01:08.240 --> 0:01:11.240 Bloomberg Intelligence. I'm Katydmitreva in Hong Kong. 0:01:11.760 --> 0:01:13.360 I'm John Lee also in Hong Kong. 0:01:13.920 --> 0:01:16.720 And today we have Chris Miller, who wrote the book 0:01:16.959 --> 0:01:20.520 on Justice. He's a professor at Toft University and the 0:01:20.560 --> 0:01:23.280 author of Chip War, The Fight for the World's Most 0:01:23.319 --> 0:01:28.160 Critical Technology, often seen as required reading for understanding this space. 0:01:28.680 --> 0:01:31.399 He joins us in Boston. Thank you for joining us, Chris, 0:01:32.040 --> 0:01:33.000 Thank you for having me. 0:01:34.120 --> 0:01:38.440 Chris in your own words, why are semiconductors so important 0:01:38.480 --> 0:01:39.520 to the great powers? 0:01:40.280 --> 0:01:43.679 Well, I think there are two reasons why. The first 0:01:43.720 --> 0:01:46.880 is that chips are just important for the global economy 0:01:47.640 --> 0:01:51.440 writ large. If you look at the world's biggest companies 0:01:51.520 --> 0:01:56.520 by market capitalization, they're all either designers of chips, producers 0:01:56.560 --> 0:01:59.240 of chips, or companies that really couldn't function without chips. 0:02:00.000 --> 0:02:01.040 Look at global trade. 0:02:01.080 --> 0:02:04.320 The largest single flow in all of global trade is 0:02:04.360 --> 0:02:06.920 the flow of chips into China each year, from Taiwan, 0:02:07.000 --> 0:02:11.399 from Korea, from Southeast Asia. And so you can't understand 0:02:11.760 --> 0:02:14.840 modern economies without chips. That's one reason. The second is 0:02:14.840 --> 0:02:18.399 that when you look forward towards the application of artificial 0:02:18.400 --> 0:02:21.959 intelligence to every aspect of the economy, what you find 0:02:22.120 --> 0:02:26.000 here too, chips are central the advances that we've seen 0:02:26.040 --> 0:02:28.760 in AI over the past couple of years have been 0:02:28.840 --> 0:02:31.920 driven by better and better semiconductors, which are both used 0:02:31.919 --> 0:02:35.520 to train these ultra capable AI systems but also in 0:02:35.639 --> 0:02:39.400 deploying them at scale. And so, just like all of 0:02:39.400 --> 0:02:42.120 the world's governments, tech CEOs too are betting that if 0:02:42.160 --> 0:02:44.560 you want to be able to build and deploy the 0:02:44.600 --> 0:02:46.640 best AI systems, you've got to be able to access 0:02:46.680 --> 0:02:50.480 to large numbers of high quality AI chips. And so 0:02:50.560 --> 0:02:54.720 these two factors, the macroeconomic centrality and the importance to AI, 0:02:55.040 --> 0:02:57.520 have put chips not just on the radar screens of 0:02:58.200 --> 0:03:02.320 Silicon Valley leaders, but also of the world's top political officials. 0:03:03.000 --> 0:03:06.040 And just sort of a historical point for those who 0:03:06.040 --> 0:03:10.000 haven't read the book yet, when did this all begin? 0:03:10.480 --> 0:03:14.079 In terms of this importance of chips, Well. 0:03:13.960 --> 0:03:15.799 In some ways it dates back to the origins of 0:03:15.840 --> 0:03:18.760 the chip industry. The first ships were invented in the 0:03:18.840 --> 0:03:23.280 late nineteen fifties. The aim was to miniatureize computing powers. 0:03:23.400 --> 0:03:26.000 You could fit a computer in the nose cone of 0:03:26.000 --> 0:03:28.640 a missile and guide it more accurately towards its target, 0:03:28.919 --> 0:03:32.280 and this was critical for the Cold War competition between 0:03:32.280 --> 0:03:35.800 the United States and the Soviet Union. But it really 0:03:35.920 --> 0:03:38.160 kind of kicked up a notch about ten years ago 0:03:38.520 --> 0:03:42.280 when the United States was the dominant player in the 0:03:42.320 --> 0:03:44.960 chip supply chain, not the dominant manufacturer of chips, but 0:03:45.000 --> 0:03:48.040 the dominant producer of many of the key inputs into chips, 0:03:48.080 --> 0:03:51.240 the ip the design know how, many of the key tools, 0:03:51.840 --> 0:03:54.800 and China realized that it's economy was becoming more and 0:03:54.920 --> 0:03:57.840 more dependent on ships that at the time it was 0:03:58.200 --> 0:04:01.839 almost completely an importer of. It produce many at all 0:04:01.840 --> 0:04:04.320 domestically in the chips that it did produce were all 0:04:04.880 --> 0:04:07.640 low end, and so in twenty fourteen, China launched a 0:04:07.800 --> 0:04:13.440 major industrial policy effort trying to develop more self sufficiency 0:04:13.480 --> 0:04:17.800 in semiconductors. It set up a very large central government 0:04:17.960 --> 0:04:20.359 investment fund, and each of the provinces set up their 0:04:20.400 --> 0:04:23.279 own investment funds to try to build a self sufficient industry. 0:04:23.960 --> 0:04:26.039 And this was seen not just by the US but 0:04:26.080 --> 0:04:29.520 also by other key players Taiwan, Korea, Japan is a 0:04:29.960 --> 0:04:33.520 economic and potentially down the road, a security threat. 0:04:34.200 --> 0:04:37.359 And where is China now in its aims to become 0:04:37.440 --> 0:04:40.719 self sufficient? How would you grade the efforts so far? 0:04:42.520 --> 0:04:45.440 Well, the chip industry is large and complex, and you 0:04:45.480 --> 0:04:47.719 get a different answer depending on which corner of the 0:04:47.760 --> 0:04:49.040 hip industry. 0:04:48.640 --> 0:04:51.000 You look at. So there's different types of chips. 0:04:51.040 --> 0:04:53.920 There's processor chips, there's memory ships, and there's also different 0:04:53.960 --> 0:04:58.640 segments of the supply chain, the design process, the manufacturing itself, 0:04:58.839 --> 0:05:01.080 the production of the tool rules that are used to 0:05:01.080 --> 0:05:03.080 make chips, and you need to kind of look at 0:05:03.080 --> 0:05:07.760 each of these segments individually to really understand how China 0:05:07.839 --> 0:05:11.040 stacks up against other competitors. At a high level, I 0:05:11.040 --> 0:05:14.240 would say this, On the one hand, China's import of 0:05:14.360 --> 0:05:16.920 chips from the rest of the world, which means largely 0:05:16.920 --> 0:05:20.520 from Taiwan and Korea and their assembly facilities in Southeast Asia, 0:05:21.000 --> 0:05:24.039 generally hits a new high each year it's grown for 0:05:24.120 --> 0:05:26.880 most of the last decades. China is importing larger numbers 0:05:27.000 --> 0:05:30.560 at least larger dollar values, of chips from abroad. But 0:05:30.640 --> 0:05:33.760 China is also producing more at home as well, and 0:05:33.800 --> 0:05:36.760 so when it comes to lower range and mid range ships, 0:05:36.960 --> 0:05:40.400 China's building a lot of self sufficiency in the types 0:05:40.400 --> 0:05:43.480 of ships that go into cars and to refrigerators and 0:05:43.880 --> 0:05:46.919 industrial equipment, but the high end ships are getting even 0:05:47.080 --> 0:05:50.080 more important. Both the chips that go in phones and PCs, 0:05:50.160 --> 0:05:54.000 but also critically the processors and the memory ships that 0:05:54.080 --> 0:05:56.800 make AI possible in data centers, and for these cutting 0:05:56.880 --> 0:06:00.640 edge ships, China is still mostly importing the chips that 0:06:00.680 --> 0:06:04.240 its tech firms need to deploy advanced computing at scale. 0:06:05.240 --> 0:06:09.719 So how far behind is China behind global leaders in 0:06:09.800 --> 0:06:11.599 terms of being self sufficient? 0:06:13.440 --> 0:06:16.760 Well, it really depends again on what exactly one wants 0:06:16.800 --> 0:06:22.000 to measure. If you ask how far behind is China 0:06:22.040 --> 0:06:24.960 when it comes to designing smartphone processors, I would say 0:06:24.960 --> 0:06:28.200 it's hardly behind at all. The smartphone processors that Huawei's 0:06:28.279 --> 0:06:31.360 high silicon arm designed are by a lot of metrics, 0:06:31.400 --> 0:06:33.560 almost as good or just as good as what Apple 0:06:33.760 --> 0:06:39.200 or Qualcom would design, And for every different category of 0:06:39.279 --> 0:06:41.680 chip you get a different rough metric. But there's a 0:06:41.680 --> 0:06:46.279 couple places where China remains meaningfully behind, and it's been 0:06:46.360 --> 0:06:49.839 struggling to replicate the capabilities of industry leaders. And I'll 0:06:49.880 --> 0:06:52.520 flag two of them which are the most critical. The 0:06:52.560 --> 0:06:56.239 first is in the production of the machines that make chips. 0:06:56.680 --> 0:06:58.960 If you go inside a chip making facility in China today, 0:07:00.120 --> 0:07:03.280 to the complex machines, especially at the high end, are 0:07:03.279 --> 0:07:08.520 imported imported from Japan, from the Netherlands and from US firms, 0:07:08.560 --> 0:07:11.120 and China has been trying to build out its own 0:07:11.160 --> 0:07:13.240 suite of machines that can make chips. But this is 0:07:13.280 --> 0:07:17.240 extraordinarily complex equipment, and it's been struggling to make a 0:07:17.240 --> 0:07:21.000 lot of progress in replacing foreign chip making tools with 0:07:21.040 --> 0:07:24.920 domestic variants. And because of that, China's also struggled to 0:07:25.040 --> 0:07:28.760 ramp up cutting edge capacity to manufacture chips itself. And 0:07:28.800 --> 0:07:32.840 so you can look, for example, at the leading Chinese firm, SMIC, 0:07:33.040 --> 0:07:35.120 which is a chip maker, and compare it to the 0:07:35.160 --> 0:07:38.280 leading Taiwanese firm, which is the world's leader, t SMC, 0:07:39.040 --> 0:07:42.480 and they have similar manufacturing trajectories. And so you can 0:07:42.560 --> 0:07:45.640 look at what SMIC does this year and look at 0:07:45.680 --> 0:07:48.800 what TSMC did several years ago and measure the gap 0:07:48.880 --> 0:07:52.720 between them as a way of assessing where China stacks up. 0:07:52.760 --> 0:07:56.760 And here you find that SMICK right now in Shanghai 0:07:56.840 --> 0:08:00.320 has got a seven nanimeter process that it has just 0:08:00.480 --> 0:08:04.120 ramped at scale. It's producing ai chips, producing smartphone processors. 0:08:04.320 --> 0:08:06.880 But TSMC and Taiwan did that six years ago. So 0:08:06.920 --> 0:08:10.160 there's a five or six year gap between what Smith 0:08:10.240 --> 0:08:13.280 can produce at scale and what TSMC can produce a scale. 0:08:13.320 --> 0:08:15.680 So those are two segments where it's clear there is 0:08:16.320 --> 0:08:19.960 still a significant gap between China's capabilities and the leading 0:08:20.000 --> 0:08:24.400 Taiwanese US Japanese firms. But those are only two ways 0:08:24.400 --> 0:08:27.400 you could measure the gap. And in other segments, especially design, 0:08:28.240 --> 0:08:31.320 China's got very impressive capabilities, and I would argue it 0:08:31.320 --> 0:08:33.880 has been closing the gap in recent years. 0:08:36.000 --> 0:08:40.559 Maybe a technical point, but you mentioned that China kind 0:08:40.559 --> 0:08:45.320 of legs behind in the equipment that makes chips. So 0:08:45.800 --> 0:08:47.920 could you describe just a bit why these machines are 0:08:47.920 --> 0:08:50.720 so complex and why China is perhaps behind in it. 0:08:50.720 --> 0:08:53.600 Because when you think of China as a manufacturer, I 0:08:53.640 --> 0:08:56.360 mean the world's factory floor, it's really got the supply 0:08:56.480 --> 0:08:58.839 chains down. So what is it about the equipment that 0:08:58.880 --> 0:09:00.240 they're not able to do now? 0:09:00.920 --> 0:09:02.160 Yeah, it's a great question. 0:09:02.240 --> 0:09:05.360 When I first myself began studying the industry, it was 0:09:05.440 --> 0:09:07.160 kind of a puzzle, why could Chiny to make everything 0:09:07.200 --> 0:09:10.040 except for these tools? And the answer, it turns out, 0:09:10.080 --> 0:09:12.480 is that these are the most complex and precise tools 0:09:12.520 --> 0:09:15.960 humans have literally ever produced. So if you take, for example, 0:09:16.600 --> 0:09:20.679 the most complicated and expensive tool called an extreme ultraviolet 0:09:20.800 --> 0:09:24.040 photo lithography tool. These are tools produced by just a 0:09:24.080 --> 0:09:27.280 single company in the world, ASML of the Netherlands. They 0:09:27.280 --> 0:09:31.440 have hundreds of thousands of components, including the flattest mirrors 0:09:31.520 --> 0:09:34.240 humans have ever made, the most powerful lasers ever deployed 0:09:34.240 --> 0:09:38.480 in the commercial device, the capable of manufacturing at nanometer 0:09:38.640 --> 0:09:42.640 scale that's billionths of a meter with close to perfect accuracy. 0:09:43.080 --> 0:09:44.599 And that's just one of the tools you need to 0:09:44.640 --> 0:09:47.880 make chips. There are many others that are only scarcely 0:09:48.120 --> 0:09:50.320 less complex, and so when you start to look at 0:09:50.360 --> 0:09:53.360 these tools, you realize that there's nothing that comes close 0:09:53.400 --> 0:09:56.800 in terms of complexity and really any other industry. And 0:09:56.840 --> 0:09:59.560 so it's not a surprise that China hasn't caught up 0:09:59.600 --> 0:10:02.920 because most of these types of tools, there's just one 0:10:03.000 --> 0:10:05.960 or two companies in the entire world that can undertake 0:10:06.080 --> 0:10:09.400 this activity. And people have been ask why hasn't China 0:10:09.440 --> 0:10:12.319 cut up and the production of extreme multi violet photo 0:10:12.320 --> 0:10:14.520 lithography tools, Well, no one else has caught up either. 0:10:14.600 --> 0:10:16.760 There's just one company that can do it. Others have 0:10:16.880 --> 0:10:21.240 tried and failed because the engineering is so complicated and 0:10:21.280 --> 0:10:24.360 the business model is so hard to make work, and 0:10:24.400 --> 0:10:26.520 so it really is the fact that we're at the 0:10:26.559 --> 0:10:30.920 frontiers of engineering the frontiers of physics that explain why 0:10:31.400 --> 0:10:33.520 it's not easy for China to replicate this, even though 0:10:33.600 --> 0:10:36.600 China's a course, very good at manufacturing many other types 0:10:36.600 --> 0:10:37.160 of machinery. 0:10:38.360 --> 0:10:41.679 Asia Centric is produced by Bloomberg Intelligence. We're more than 0:10:41.720 --> 0:10:45.560 five hundred experienced analysts and strategists work around the clock 0:10:45.640 --> 0:10:49.120 to bring you timely, world class research. Our coverage spans 0:10:49.160 --> 0:10:53.480 two hundred market industries, currencies, commodities, and industries, as well 0:10:53.480 --> 0:10:56.520 as over two thousand equities and credits. To learn more 0:10:56.520 --> 0:11:00.000 about Bloomberg Intelligence, visit b I go on the blue 0:11:00.080 --> 0:11:03.040 the terminal. If you like what you hear, don't forget 0:11:03.120 --> 0:11:06.760 to subscribe and chair Chris, Look, I read your book, 0:11:06.760 --> 0:11:08.880 But for the uninitiated who didn't read your book, can 0:11:08.920 --> 0:11:13.400 you explain why like this industry lends itself to monopolies 0:11:13.520 --> 0:11:15.880 like you mentioned that ASML pretty much like no one 0:11:15.880 --> 0:11:22.640 else makes these you know, lithic of methography on machines, 0:11:24.360 --> 0:11:26.160 and I think you said no one else makes these 0:11:26.280 --> 0:11:31.120 TSMC also manufactures over ninety percent of the advanced AI chips. 0:11:31.480 --> 0:11:34.160 Why is this the case? That there are monopolies formed 0:11:34.240 --> 0:11:35.120 in the supply chain. 0:11:35.720 --> 0:11:35.920 Yeah. 0:11:35.960 --> 0:11:38.960 I think there are two factors that lend the chip 0:11:39.000 --> 0:11:43.560 industry to these sort of oligopolistic rin monopolistic markets. The 0:11:43.559 --> 0:11:48.800 first is the capital expenditures are often tremendous. So for TSMC, 0:11:49.400 --> 0:11:53.320 every new factory they build is the world's most expensive 0:11:53.320 --> 0:11:56.040 factory because each factory is more expensive and breaks the 0:11:56.080 --> 0:11:59.079 prior record. We're talking twenty or twenty five billion dollars 0:11:59.400 --> 0:12:03.240 per FACI, and so this scope for a new startup 0:12:03.320 --> 0:12:08.240 firm to compete is approximately zero. That has really kept 0:12:08.760 --> 0:12:12.000 competition down because the number of players that could even 0:12:12.120 --> 0:12:16.560 conceive of taking on TSMC is very small, and the 0:12:16.559 --> 0:12:19.800 same dynamic is true in other segments of the supply chain. 0:12:19.800 --> 0:12:22.120 The dollar values are lower, but not much lower. 0:12:22.160 --> 0:12:24.840 Even if it's only three billion dollars for a new 0:12:24.880 --> 0:12:28.720 factory that makes the ultra purified a silicon wafers, for example, 0:12:28.760 --> 0:12:32.520 that's still a monumental capital expense for most companies, and 0:12:32.559 --> 0:12:36.800 it deters new entrants from even trying. The second factor 0:12:36.840 --> 0:12:41.200 is that the R and D is really extraordinarily highly specialized, 0:12:41.800 --> 0:12:44.520 and for a lot of the process steps and making chips, 0:12:44.760 --> 0:12:48.840 the only way you learn about how to undertake that 0:12:48.880 --> 0:12:52.080 specific step is by having already been in the industry 0:12:52.120 --> 0:12:55.000 and learning alongside your suppliers and your customers. 0:12:55.440 --> 0:12:56.240 And so, if you think of. 0:12:56.240 --> 0:12:59.080 The way the chip industry's R and D works, every year, 0:12:59.520 --> 0:13:02.640 the entire industry works together to bring a new process 0:13:02.679 --> 0:13:06.520 that's more capable than the prior years, often dramatically more 0:13:06.520 --> 0:13:08.679 capable in terms of the computing power it produces. And 0:13:08.760 --> 0:13:12.640 so you've got really tight alignment between the existing chemicals producers, 0:13:12.760 --> 0:13:17.600 existing machinery producers, existing chip manufacturers themselves to make sure 0:13:17.640 --> 0:13:20.760 that these processes arrive on time and that all the 0:13:20.800 --> 0:13:24.040 necessary components are together. There's a lot of shared learning 0:13:24.160 --> 0:13:26.960 inside of that small circle. But if you're an outsider 0:13:27.160 --> 0:13:30.000 who's never been part of that learning process, there's no 0:13:30.040 --> 0:13:32.080 way to do it on your own. You can't just 0:13:32.320 --> 0:13:35.960 test on your own very easily whether your new chemical 0:13:36.000 --> 0:13:40.360 will work with a existing extreme ultraviolet photogthography machine. If 0:13:40.400 --> 0:13:41.800 you want to buy one of those machines yourself, it 0:13:41.920 --> 0:13:44.840 costs you one hundred and fifty million dollars. And moreover, 0:13:44.920 --> 0:13:47.000 companies aren't going to let you run a lot of 0:13:47.040 --> 0:13:49.679 tests in their fab because they just spend twenty billion 0:13:49.720 --> 0:13:50.920 dollars building a fab they. 0:13:50.800 --> 0:13:52.280 Need to run that at full capacity. 0:13:52.320 --> 0:13:54.720 So it's really hard to break in from an R 0:13:54.720 --> 0:13:56.760 and D perspective, And when you combine that with the 0:13:56.760 --> 0:14:00.240 capital intensivity, it means that many of the firms that 0:14:00.440 --> 0:14:03.560 have key roles in the supply chain have had them 0:14:03.600 --> 0:14:05.800 for the last decade or in some cases the last 0:14:06.080 --> 0:14:09.400 three or four decades, and that has kept competition down 0:14:09.440 --> 0:14:10.280 in the hip industry. 0:14:10.920 --> 0:14:13.120 So you wrote your book in twenty twenty two, that 0:14:13.280 --> 0:14:16.439 was under the Biden administration. We're now under the Trump 0:14:16.440 --> 0:14:21.120 administration two point zero. What are the big differences you've 0:14:21.200 --> 0:14:25.320 seen in how both administrations have handled this issue? Are 0:14:25.360 --> 0:14:28.080 there some similarities that come up, but also some of 0:14:28.080 --> 0:14:30.640 the big big differences, because of course under Biden we 0:14:30.760 --> 0:14:35.440 had the Chips IRA Act, But under both administrations it 0:14:35.440 --> 0:14:37.000 seems like a bipart is an issue. 0:14:38.400 --> 0:14:39.640 Yeah, I think it is, and I think that the 0:14:39.680 --> 0:14:45.480 similarities are to me more substantial than the differences. So first, 0:14:45.480 --> 0:14:48.000 there's a focus on domestic manufacturing in the United States, 0:14:48.080 --> 0:14:50.520 an effort to reduce reliance both on the lower end 0:14:50.520 --> 0:14:53.760 ship produced in China, but also dependence on high end 0:14:53.800 --> 0:14:57.920 chips produced in Taiwan the Chips Act. Although Biden claimed 0:14:57.960 --> 0:15:00.800 credit for it, it actually emerged under the trumpetministration. That was 0:15:00.800 --> 0:15:03.840 when the Congress first started discussing the bill, and people 0:15:03.880 --> 0:15:06.320 like Secretary of State Mike Pompeo played a major role 0:15:06.440 --> 0:15:09.600 in some of the early investments that the Taiwanese from 0:15:09.640 --> 0:15:12.480 TSMC announced the United States. So there's been a through 0:15:12.560 --> 0:15:15.920 line of support for manufacturing chips in the United States 0:15:15.960 --> 0:15:17.640 that has been bipartisan. 0:15:17.960 --> 0:15:19.760 I think that's a point of continuity. 0:15:20.120 --> 0:15:22.120 I think the second point in continuity has been a desire 0:15:22.200 --> 0:15:27.000 to stop China from accessing the most advanced ship making technologies. 0:15:27.480 --> 0:15:31.520 The export controls on ASML tools, for example, started under 0:15:31.520 --> 0:15:35.440 the first administration, were expanded under Biden, and we'll see 0:15:35.480 --> 0:15:37.560 they could well be expanded further at some point under 0:15:37.560 --> 0:15:40.360 the current TRUP administration. So that's also a through line. 0:15:40.800 --> 0:15:44.080 And I think the third is concern about Chinese industrial 0:15:44.120 --> 0:15:47.320 subsidies in the chip industry and implications for trade policy. 0:15:47.760 --> 0:15:52.320 The Trumpdministration in round one imposed higher tariffs and Chinese 0:15:52.320 --> 0:15:55.920 ships imported the United States. Biden expanded those tariffs, and 0:15:56.440 --> 0:15:59.680 of course President Trump right now is considering what he's 0:15:59.680 --> 0:16:00.120 going to do. 0:16:01.000 --> 0:16:01.920 On chip tariffs. 0:16:01.960 --> 0:16:04.200 But there's widespread expectation that there will be another round 0:16:04.200 --> 0:16:06.960 of chip tariffs coming at some point later this year, 0:16:07.000 --> 0:16:09.560 and so that's also a through line of concern about 0:16:09.640 --> 0:16:12.520 Chinese subsidies. I think there are differences. Obviously, you know, 0:16:12.520 --> 0:16:16.320 we're going to see where exactly Trump's chip tariffs land, 0:16:16.320 --> 0:16:18.160 and they might end up being much broader on a 0:16:18.160 --> 0:16:21.040 whole range of different countries, not just China focused. 0:16:21.080 --> 0:16:22.920 If so, that would be a difference. 0:16:22.960 --> 0:16:25.160 But I think when you zoom out, you actually end 0:16:25.240 --> 0:16:30.480 up seeing a pheromounta continuity, despite some nuances that are different. 0:16:31.280 --> 0:16:31.440 Chris. 0:16:31.880 --> 0:16:34.480 President Trump seems to be dialing back on some of 0:16:34.480 --> 0:16:38.560 the policies now. Under the previous administration, they banned the 0:16:38.680 --> 0:16:42.280 export of high end in Nvidia chips into China, but 0:16:42.960 --> 0:16:47.720 Donald Trump recently suggested that they could allow the export 0:16:47.800 --> 0:16:51.600 of certain Age twenty chips back into the country. What 0:16:51.640 --> 0:16:54.080 does this mean, Well, I. 0:16:54.000 --> 0:16:56.280 Think it's too soon to have a confident judgment. I 0:16:56.320 --> 0:16:59.440 would say when the Biden administration left office, in Videau's 0:16:59.520 --> 0:17:03.360 H twenty were still legal to sell China. Then in 0:17:03.520 --> 0:17:06.680 month four of the trumpdministration, the President banned the export 0:17:06.720 --> 0:17:09.400 of Age twenty chips, and now there's suggestions they might 0:17:09.440 --> 0:17:12.040 be allowed again. As of today, we're still at a 0:17:12.160 --> 0:17:17.520 tighter restrictive policy than when the Biden administration ended. 0:17:17.520 --> 0:17:18.879 When the Trump administration started. 0:17:18.880 --> 0:17:21.920 We'll see exactly how this plays out, but I think 0:17:21.960 --> 0:17:25.959 it speaks again to a continuity, which is that both 0:17:26.000 --> 0:17:29.679 Biden and Trump feel like they need to balance action 0:17:29.960 --> 0:17:32.520 on chip issues with the broader. 0:17:32.320 --> 0:17:33.480 US China relationship. 0:17:33.800 --> 0:17:36.879 So into the Biden administration, it was regularly the case 0:17:37.000 --> 0:17:40.119 that the Commerce Department would have a new regulation ready 0:17:40.119 --> 0:17:42.960 that would limit this or that aspect of the chip trade, 0:17:43.240 --> 0:17:45.200 and then it would wait for three months or six 0:17:45.240 --> 0:17:48.119 months or nine months while the State Department found what 0:17:48.119 --> 0:17:50.360 they thought was a quote unquote good window in which 0:17:50.400 --> 0:17:53.280 to roll out a new restriction that the Chinese wouldn't like. 0:17:53.359 --> 0:17:55.840 And cour we're seeing the trumpministration as something similar. 0:17:55.920 --> 0:17:56.320 Right now. 0:17:56.480 --> 0:17:59.919 The President wants to negotiate a trade deal with China. 0:18:00.400 --> 0:18:03.160 Their White House is trying to get rare earth magnets 0:18:03.160 --> 0:18:05.520 flowing again, and so there's concern in the White House 0:18:05.560 --> 0:18:09.640 that if they were to take punitive steps on chips 0:18:09.760 --> 0:18:12.800 or on chip making tools, that this would disrupt the 0:18:12.880 --> 0:18:16.119 US China relationship. One can debate whether that's the right approach, 0:18:16.160 --> 0:18:18.760 but I think it is a striking similarity to a 0:18:18.760 --> 0:18:22.040 lot of the delays and the kind of half hearted 0:18:22.080 --> 0:18:25.120 decision making that you saw in the Biden administration when 0:18:25.119 --> 0:18:27.680 it came to whether or not to restrict certain types 0:18:27.720 --> 0:18:29.280 of tools or chips. 0:18:30.040 --> 0:18:34.199 So, after the Trump administration recently allowed the export of 0:18:34.480 --> 0:18:39.200 H twenty chips, very closely afterwards, China called in Nvidia 0:18:39.280 --> 0:18:43.320 staff to have a discussion with them behind that is 0:18:43.920 --> 0:18:48.960 at least partly the Trump administration suggesting to track these chips. 0:18:49.320 --> 0:18:52.399 And this is all happening within the context of a 0:18:52.880 --> 0:18:57.199 trade war and trade negotiations. So one of the questions 0:18:57.240 --> 0:19:00.280 we've been thinking about is, you know, why are the 0:19:00.440 --> 0:19:03.760 US and China right now fighting over what are essentially 0:19:03.840 --> 0:19:07.560 kind of third rate chips. I mean, these aren't like 0:19:07.640 --> 0:19:10.920 the most advanced chips out there. As John ass like, 0:19:10.960 --> 0:19:12.919 how should we read this kind of back and forth. 0:19:13.960 --> 0:19:16.399 Well, I wouldn't say it's right to consider the H 0:19:16.480 --> 0:19:19.439 twenties third rate chips. I think if you're a Chinese 0:19:19.440 --> 0:19:21.119 tech company, these are first rate. These are the US 0:19:21.160 --> 0:19:23.919 you can get at scale can you smuggle in small 0:19:24.000 --> 0:19:27.160 volumes of the higher quality H one hundred yes. Can 0:19:27.200 --> 0:19:31.280 you access small volumes of Huawei ASSN chips Yes? But 0:19:31.359 --> 0:19:34.240 can you access either by the tens or hundreds of thousands, No, 0:19:34.440 --> 0:19:37.560 both because Huawei can't produce that efficient scale and because 0:19:37.600 --> 0:19:39.720 smuggling you can do a medium scale, but not at 0:19:39.760 --> 0:19:42.919 the scale you need. So you know, judging by the 0:19:42.960 --> 0:19:46.760 purchasing patterns of byte Edance and Ali Baba Intense, these 0:19:46.760 --> 0:19:49.080 are great chips, which is why they really want to 0:19:49.080 --> 0:19:51.440 spend hundreds of millions or billions of. 0:19:51.400 --> 0:19:52.240 Dollars on them. 0:19:52.800 --> 0:19:54.280 And I think that that is why this is such 0:19:54.280 --> 0:19:57.399 a critical issue, both for the US and for Chinese 0:19:57.440 --> 0:19:57.920 tech firms. 0:19:58.000 --> 0:19:59.640 If they really were third rate, we wouldn't be talking 0:19:59.640 --> 0:20:00.000 about them. 0:20:00.440 --> 0:20:03.960 Chris like and Vidia CEO Jensen Huang was very critical 0:20:04.119 --> 0:20:08.200 about the US ban on chips to China. He said 0:20:08.240 --> 0:20:11.360 it hurts US businesses, and he also insinuated that it's 0:20:11.359 --> 0:20:16.480 going to accelerate China's domestic semiconductive development. Do you agree 0:20:16.480 --> 0:20:18.199 with him? Do you think these mergers are in some 0:20:18.320 --> 0:20:19.400 way self defeating? 0:20:20.520 --> 0:20:22.960 Well, I think there's no doubt that in the short 0:20:23.040 --> 0:20:25.720 run it reduces in video sales. That's certainly clear. So 0:20:26.000 --> 0:20:29.720 does it hurt that business? I think the answer is yes. 0:20:30.640 --> 0:20:33.280 One of the dilemmas that any sort of export control 0:20:33.359 --> 0:20:35.760 regime faces is that if you restrict the sale of 0:20:35.800 --> 0:20:38.679 a certain tool, you hurt the tool makers, but you 0:20:39.040 --> 0:20:42.560 benefit the domestic companies that use that tool, who have 0:20:42.640 --> 0:20:45.560 less foreign competition. So, for example, if you restrict the 0:20:45.560 --> 0:20:48.160 sale of chip making tools, it's bad for the companies 0:20:48.160 --> 0:20:51.280 that make chip making tools, but it's actually better for 0:20:51.480 --> 0:20:54.679 chip makers in the West, including in Taiwan and South Korea, 0:20:55.119 --> 0:20:57.680 because they don't have strong competition in China at the 0:20:57.680 --> 0:21:00.440 cutting edge. And if you thought about the counterfactuals pose 0:21:00.840 --> 0:21:03.800 the Trudministration had not leaned on the Netherlands in twenty 0:21:03.880 --> 0:21:07.160 nineteen and said the Netherlands don't sell UV tools to China, 0:21:07.600 --> 0:21:10.840 China would be operating today a fleet of many UV tools, 0:21:10.880 --> 0:21:14.520 possibly the second largest behind Taiwan in the world, and 0:21:14.560 --> 0:21:17.560 that'd be a totally different landscape for the world's chip makers. 0:21:18.000 --> 0:21:20.520 It'd be better for sellers of chip maker tools, but 0:21:20.560 --> 0:21:22.359 worse for chip makers. And we're in a similar dynamic 0:21:22.440 --> 0:21:24.960 right now with chips. If you sell more chips it's 0:21:24.960 --> 0:21:27.520 good for chip makers, but it's probably worse for the 0:21:27.520 --> 0:21:30.720 companies that use chips, which are the big tech companies, 0:21:30.760 --> 0:21:32.280 cloud computing companies. 0:21:31.880 --> 0:21:34.960 And AI labs. Why is it that. 0:21:35.280 --> 0:21:40.320 We haven't seen large deployments of AI chips buy Chinese 0:21:40.359 --> 0:21:43.280