WEBVTT - Kopi Time E108 - Chris Miller on Chip War 0:00:06.179 --> 0:00:09.720 Welcome to Kobe Time, a podcast series on Markets and 0:00:09.728 --> 0:00:13.239 Economies from Devious Group Research. I'm ba chief economist, welcoming 0:00:13.250 --> 0:00:15.670 you to our 108th episode. 0:00:16.388 --> 0:00:20.038 If any book can make general audiences understand the Silicon 0:00:20.047 --> 0:00:23.358 age and finally recognize how it rivals the atomic age 0:00:23.368 --> 0:00:27.159 for drama and import chip war is it? That's what 0:00:27.169 --> 0:00:29.589 Virginia Hefferman wrote at the end of her New York 0:00:29.599 --> 0:00:30.518 Times book review. 0:00:30.829 --> 0:00:33.759 I'm very glad to have the Chip Wars author on 0:00:33.770 --> 0:00:38.020 Coy Time. Chris Miller is associated professor of International History 0:00:38.119 --> 0:00:41.759 at the Fletcher School at Tufts University. He published Chip 0:00:41.770 --> 0:00:44.720 War last year and it has been a major success 0:00:44.729 --> 0:00:48.319 winning the Financial Times best business book award for 2022. 0:00:48.330 --> 0:00:50.659 Chris Miller, a warm welcome to Cope time. 0:00:51.720 --> 0:00:54.330 Thank you for having me. So great to have you. 0:00:54.340 --> 0:00:57.889 Uh Chris, I'll begin by playing the devil's advocate. Uh 0:00:57.900 --> 0:01:01.689 the nuclear era quantum mechanics, DNS double he legs man 0:01:01.700 --> 0:01:04.410 on the moon. Even in my world, the futures and 0:01:04.419 --> 0:01:08.679 options in finance. All of these civilization, transforming advances took 0:01:08.690 --> 0:01:11.449 place in the 20th century with little or no shape, 0:01:11.459 --> 0:01:12.830 processing power behind them, 0:01:13.290 --> 0:01:18.300 modern cities, industrial agriculture, amazing bridges and tunnels, aviation. All 0:01:18.309 --> 0:01:21.330 these predate the computer age. And yet in your book, 0:01:21.339 --> 0:01:25.699 you state that today's military, economic and geopolitical power are 0:01:25.709 --> 0:01:29.589 built on a foundation of computer chips. Are you overseeing 0:01:29.599 --> 0:01:30.080 the case? 0:01:33.760 --> 0:01:36.379 Well, I, I think you're right that humanity has had 0:01:36.389 --> 0:01:40.199 lots of advances before chips were invented. Um, but it's 0:01:40.209 --> 0:01:43.519 also the case today that it's hard to go about 0:01:43.529 --> 0:01:47.400 your daily life without interacting with not just dozens but 0:01:47.410 --> 0:01:52.440 hundreds or thousands of semiconductors in your phone. There are 0:01:52.449 --> 0:01:53.000 dozens of semi 0:01:53.480 --> 0:01:57.319 devices in your computer in the data centers that process 0:01:57.330 --> 0:01:59.540 and store your data in your car. If it's a 0:01:59.550 --> 0:02:04.080 new car, it will have on average 1000 semiconductors inside 0:02:04.089 --> 0:02:08.418 in your refrigerator in your dishwasher in your microwave. Uh 0:02:08.429 --> 0:02:11.729 And almost every device you touch, there are lots and 0:02:11.740 --> 0:02:12.960 lots of semiconductors and 0:02:13.309 --> 0:02:17.169 they are so ubiquitous around us that we don't even 0:02:17.179 --> 0:02:20.250 think about them. Uh But in fact, as we learned 0:02:20.258 --> 0:02:24.220 uh during the pandemic era, shortages, modern economy simply can't 0:02:24.229 --> 0:02:25.960 function without them. 0:02:26.470 --> 0:02:29.168 And so in that sense, I think they're the most 0:02:29.179 --> 0:02:33.619 economically critical technology because during the pandemic, the largest dollar 0:02:33.630 --> 0:02:37.869 value uh disruption to manufacturing uh in terms of component 0:02:37.880 --> 0:02:41.440 shortages came from uh semiconductor shortages. And so I think 0:02:41.449 --> 0:02:46.199 that was one very brief and relatively small illustration of 0:02:46.210 --> 0:02:49.978 just how critically economically important semiconductors have become 0:02:51.580 --> 0:02:56.960 economic criticality. Aside. Do you also see security and conflict 0:02:56.970 --> 0:03:01.070 propensity coming in with the availability or lack there of chips. 0:03:04.309 --> 0:03:07.330 Well, certainly militaries are just as reliant on ships as 0:03:07.339 --> 0:03:11.600 the rest of us. Um militaries have reoriented the way 0:03:11.610 --> 0:03:15.720 they devise systems to take advantage of the sensing, the communications, 0:03:15.729 --> 0:03:19.279 the computing power that semiconductors enable. And just like today, 0:03:19.288 --> 0:03:22.119 it's impossible to buy a car without many, many chips 0:03:22.130 --> 0:03:22.339 and stuff. 0:03:22.505 --> 0:03:25.384 So too, it's impossible to feel a modern tank or 0:03:25.395 --> 0:03:29.755 a plane or a drone without lots of chips as well. 0:03:29.764 --> 0:03:33.535 And just like in cars or consumer devices, the more 0:03:33.544 --> 0:03:36.544 advanced the communication, the more advanced the sensing, the more 0:03:36.554 --> 0:03:39.085 advanced the computing, the more chips are inside the same 0:03:39.095 --> 0:03:40.434 is true of defense systems 0:03:40.750 --> 0:03:45.279 as well. Which is why today every major military is 0:03:45.289 --> 0:03:48.360 thinking harder than ever about its access to semiconductors, both 0:03:48.369 --> 0:03:51.350 in terms of assuring that it can access the chips 0:03:51.360 --> 0:03:54.279 that it needs. But also trying to find new ways 0:03:54.289 --> 0:03:57.600 to apply the capabilities that chips enable the defense systems. 0:03:57.610 --> 0:03:59.520 If you talk to defense ministries, you'll hear a lot 0:03:59.529 --> 0:04:03.779 today about autonomy about applying A I to defense systems 0:04:03.789 --> 0:04:07.580 and intelligence capabilities. And all of that means using more 0:04:07.589 --> 0:04:10.419 and more semiconductors for military purposes. 0:04:11.259 --> 0:04:12.820 Right? And in your book also, I think there are 0:04:12.830 --> 0:04:17.209 these fascinating chapters about the Soviet Union's attempts to get 0:04:17.220 --> 0:04:20.750 some degree of uh security on ship making and how 0:04:20.760 --> 0:04:24.950 spectacularly fi you know, it all ended in largely failure. 0:04:24.959 --> 0:04:26.899 And you point out that by the early nineties, the 0:04:26.910 --> 0:04:28.730 gap was so insurmountable that even 0:04:28.988 --> 0:04:31.488 Soviet defense specialists would say one of the major reasons 0:04:31.500 --> 0:04:34.719 for their decline was the tech related uh gap with 0:04:34.730 --> 0:04:38.959 the US and its allies. Um Chris, your book has 0:04:38.970 --> 0:04:42.518 just fantastic first. So 67 chapters on the history of 0:04:42.529 --> 0:04:44.880 chip making. And we go all the way back to 0:04:44.890 --> 0:04:48.720 the forties and fifties, the uh the noise and Shockley 0:04:48.730 --> 0:04:49.859 of the world. And 0:04:49.945 --> 0:04:52.975 by the way, I did my phd at Urbana Champagne, Illinois. 0:04:52.984 --> 0:04:55.565 I did not know about John Barin till I read 0:04:55.575 --> 0:04:57.674 your book. And now I know why there's a Barine 0:04:57.684 --> 0:05:03.315 quad at Illinois. Um So looking at at that uh and, 0:05:03.325 --> 0:05:06.815 and the incredible success that Silicon Valley has had over 0:05:06.825 --> 0:05:10.844 the last 50 60 years in terms of pushing uh innovation. 0:05:11.024 --> 0:05:14.974 So help us sort of walk through the critical ingredients 0:05:14.984 --> 0:05:17.875 behind this continuous series of innovation in the area of 0:05:17.885 --> 0:05:18.734 computer chips. 0:05:21.339 --> 0:05:22.390 Well, the the the 0:05:22.399 --> 0:05:24.649 key to the rate of innovation in the chip industry 0:05:24.660 --> 0:05:28.019 is to understand Moore's law. Um Gordon Moore was one 0:05:28.029 --> 0:05:31.160 of the co-founders of Intel even before he founded Intel, 0:05:31.170 --> 0:05:33.029 though he was involved in the chip industry. And he 0:05:33.040 --> 0:05:37.190 realized very, very early on that the number of transistors 0:05:37.200 --> 0:05:39.790 per chip was doubling every year or two. 0:05:40.079 --> 0:05:43.079 Um And at the time a cutting edge ship would 0:05:43.089 --> 0:05:46.169 have a very small number of transistors on it. Today, 0:05:46.178 --> 0:05:50.089 your phone has billions of transistors, but just uh six 0:05:50.100 --> 0:05:53.410 years after the first ship was invented, that he saw 0:05:53.420 --> 0:05:56.238 a pattern of doubling every year or two and predicted 0:05:56.250 --> 0:06:00.269 that that rate of doubling would continue for some time. 0:06:00.279 --> 0:06:01.049 He thought 0:06:01.890 --> 0:06:03.920 continue for at least a decade. It turns out it 0:06:03.928 --> 0:06:06.880 continued all the way up basically to the present. And 0:06:06.890 --> 0:06:09.799 what that's meant is that the chip industry has improved 0:06:09.809 --> 0:06:14.678 its capabilities more rapidly over a longer time horizon than 0:06:14.690 --> 0:06:17.399 any other type of technology in all of human history. 0:06:17.410 --> 0:06:21.558 Nothing else comes remotely close to that rate of technological improvement, 0:06:21.570 --> 0:06:24.920 exponential growth for uh more than half a century. It's 0:06:24.928 --> 0:06:26.980 something that the rest of the economy can only dream 0:06:26.988 --> 0:06:27.380 of 0:06:27.709 --> 0:06:30.429 in most companies if sales grow by 5% or if 0:06:30.440 --> 0:06:33.339 technology improves by 10% it's a great success. And in 0:06:33.350 --> 0:06:35.829 the chip industry, that's an extraordinary failure. And so III 0:06:35.859 --> 0:06:39.700 I like to compare that to other types of veggies, aviation, 0:06:39.709 --> 0:06:43.579 for example. Um you know what if planes flew twice 0:06:43.589 --> 0:06:46.549 as fast every year while also being reduced in cost 0:06:46.559 --> 0:06:48.399 on a regular basis? Will they be flying 0:06:48.678 --> 0:06:51.339 uh many times faster than the speed of light at 0:06:51.350 --> 0:06:54.450 this point? If that rate of growth were sustained over 0:06:54.459 --> 0:06:56.928 half a century. And so it's it's physically impossible for 0:06:56.940 --> 0:07:00.019 that to happen. But the chip industry has done something 0:07:00.209 --> 0:07:04.269 similar and and when you look at the proliferation of 0:07:04.279 --> 0:07:08.230 chips across society and across economy, that's been made possible 0:07:08.238 --> 0:07:11.369 because Moore's law has driven down the cost of computing. 0:07:11.470 --> 0:07:13.549 You want a data point that the first chips that 0:07:13.559 --> 0:07:16.630 were available commercially had just four transistors on them 0:07:16.910 --> 0:07:18.959 today. You can go to Amazon and buy a, a 0:07:18.970 --> 0:07:21.980 thumb drive with a billion transistors or 10 billion transistors 0:07:21.989 --> 0:07:25.920 for about $10. Which means that ballpark, the cost of 0:07:25.929 --> 0:07:31.290 computing has fallen by roughly 1 billion fold uh since 0:07:31.299 --> 0:07:34.100 uh since the 19 fifties. And that's all possible. Thanks 0:07:34.109 --> 0:07:34.760 to Moore's Law, 0:07:35.559 --> 0:07:40.279 right? I was looking up the Apple iphone 12 had 0:07:40.290 --> 0:07:43.970 the A 14 processor, something like 10 billion I CS. 0:07:44.320 --> 0:07:47.920 And the iphone 14 pro max has the A 16 0:07:47.929 --> 0:07:51.489 processor with 16 billion. There's a 60% increase in the 0:07:51.500 --> 0:07:54.239 I CS just in the latest ones. Um I wanna 0:07:54.250 --> 0:07:57.100 talk to you later about the likelihood that this will 0:07:57.109 --> 0:07:59.510 not go on forever and what that means. But uh 0:07:59.589 --> 0:08:01.510 before we go in that direction, 0:08:01.750 --> 0:08:03.459 uh Chris, you spent a lot of time in your 0:08:03.470 --> 0:08:06.489 book talking about the immense complexity and the cost associated 0:08:06.500 --> 0:08:11.230 with producing chips uh share with us some idea of how, 0:08:11.239 --> 0:08:13.019 how complex and how coffee that is. 0:08:14.679 --> 0:08:17.970 Well, to make an advanced chip, you need to make 0:08:17.980 --> 0:08:20.359 a chip with billions of transistors, each of which is 0:08:20.369 --> 0:08:23.190 roughly the size of a Coronavirus. So it's, it's the 0:08:23.200 --> 0:08:26.859 most complex manufacturing humans have ever done. You think in 0:08:26.869 --> 0:08:31.429 most manufactured goods, they have bolt tolerances measured in millimeters 0:08:31.440 --> 0:08:35.530 or 10th of millimeters. But in the chip industry, misplacing 0:08:35.539 --> 0:08:40.689 a single atom can at times cause catastrophic failure to 0:08:40.719 --> 0:08:42.309 the way the semiconductor works. 0:08:42.405 --> 0:08:46.255 So there's no other industry that combines the scale manufacturing 0:08:46.265 --> 0:08:50.695 transistors by the billions with the level of precision required. 0:08:50.705 --> 0:08:53.405 And and doing this is it's, it's simply the most 0:08:53.414 --> 0:08:55.875 complex uh manufacturing that we do. And as a result, 0:08:55.885 --> 0:08:58.195 it requires the most complex supply chain 0:08:58.590 --> 0:09:02.530 that exists. And so today, it's fundamentally impossible, simply impossible 0:09:02.539 --> 0:09:06.340 to create an advantage ship without drawing on machine tools, 0:09:06.349 --> 0:09:12.010 software materials, um software from multiple different uh companies in 0:09:12.020 --> 0:09:15.260 multiple different countries. And there are several that stand out. 0:09:15.270 --> 0:09:20.200 Uh the US Japan, the Netherlands, uh Taiwan and Korea 0:09:20.210 --> 0:09:21.718 are the five big players 0:09:22.099 --> 0:09:25.460 uh in the chip industry. Uh They're the only five 0:09:25.469 --> 0:09:29.358 countries that have um uh a substantial share of the 0:09:29.369 --> 0:09:32.349 chip industry's revenue in double digit terms. Uh And you 0:09:32.359 --> 0:09:34.950 can't produce a cutting edge uh chip without drawing on 0:09:34.960 --> 0:09:37.369 expertise from uh each of those countries. 0:09:38.109 --> 0:09:41.049 And if I am not just taken like a lithography, 0:09:41.059 --> 0:09:44.440 the extreme ultraviolet lithography machines cost like $300 million to 0:09:44.450 --> 0:09:45.239 make or something like that. 0:09:46.109 --> 0:09:48.869 That, that's right. Yeah, the, the, the next generation lithography 0:09:48.880 --> 0:09:52.439 machines will cost 300 billion or so dollars and like 0:09:52.450 --> 0:09:54.679 a cutting edge ship making facility as a result of 0:09:54.690 --> 0:09:56.770 all the machinery you need inside of it uh can 0:09:56.780 --> 0:10:01.478 cost $20 billion or so the most expensive factories in 0:10:01.489 --> 0:10:04.349 human history and 70% or so. The cost of a 0:10:04.359 --> 0:10:07.169 new chip fab fab is what we call a manufacturing 0:10:07.179 --> 0:10:10.239 facility is just to pay for the machine tools. So these, 0:10:10.369 --> 0:10:10.809 these tool 0:10:10.903 --> 0:10:13.482 tools that can lay down thin films and materials, just 0:10:13.492 --> 0:10:17.552 four atoms thick with basically complete deformity or carve canyons 0:10:17.562 --> 0:10:20.002 into silicon, just a couple of atoms wide. This is 0:10:20.013 --> 0:10:23.093 the most complex machinery humans have ever made that requires 0:10:23.102 --> 0:10:25.513 the supply chain that stretches from the Netherlands to the 0:10:25.523 --> 0:10:28.852 US to Japan. And you can't make advanced ships without 0:10:28.862 --> 0:10:30.932 these tools. And these tools themselves 0:10:31.015 --> 0:10:34.976 are are not just made exclusively in these three countries. 0:10:34.995 --> 0:10:37.655 They're also made by a tiny number of firms. Uh 0:10:37.666 --> 0:10:39.296 And in the case of some of the tools, there's 0:10:39.306 --> 0:10:41.036 just one firm that knows how to make them. And 0:10:41.046 --> 0:10:43.556 so you mentioned the lithography tools when it comes to 0:10:43.565 --> 0:10:47.125 advanced lithography, a SML and the Netherlands has 100% market 0:10:47.135 --> 0:10:50.296 share uh of the tools that are necessary to make 0:10:50.306 --> 0:10:51.036 cutting edge ships. 0:10:51.830 --> 0:10:53.659 I think there's one part in your book, you have 0:10:53.669 --> 0:10:56.409 this amazing analogy. As you say that the Carl Zeiss 0:10:56.609 --> 0:11:00.440 lens or mirror that is required to reflect the laser 0:11:00.580 --> 0:11:03.770 in the in the uh fat process for Eu V. 0:11:03.919 --> 0:11:06.289 It's like uh the, the perfection in that is like, 0:11:06.299 --> 0:11:08.780 you know, trying to hit a golf ball on the 0:11:08.789 --> 0:11:10.789 moon from here like the laser has to be that 0:11:10.799 --> 0:11:12.320 accurate or the pure 0:11:12.419 --> 0:11:15.010 on the class has to be like if there's one 0:11:15.020 --> 0:11:18.429 millimeter of impurity, it, it's like the size of Germany, 0:11:18.440 --> 0:11:20.949 that's the size of the mirror and that's a tolerance level, 0:11:20.960 --> 0:11:23.210 which is absolutely mind boggling. But I, I really like 0:11:23.219 --> 0:11:25.250 the way you portray it because I think we sort 0:11:25.260 --> 0:11:29.489 of understand the word complexity, we're really understanding it. And 0:11:29.500 --> 0:11:31.369 I think that sort of analogy really, really helped. Uh 0:11:31.380 --> 0:11:32.929 definitely it helped me. 0:11:33.200 --> 0:11:37.780 Um OK, so it's a truly globalized process, all sorts 0:11:37.789 --> 0:11:41.549 of companies have very specific and world dominating roles in it. 0:11:41.729 --> 0:11:43.799 So I take it you don't believe the Chinese can 0:11:43.809 --> 0:11:44.789 do it all by themselves. 0:11:47.049 --> 0:11:48.919 Well, I think it depends on, on what we mean. If, 0:11:48.929 --> 0:11:53.460 if we mean can China produce 2023 levels of technology 0:11:54.020 --> 0:11:56.679 self sufficiently at some point in the future? I think 0:11:56.690 --> 0:11:58.239 the answer is yes, they will be able to, the 0:11:58.250 --> 0:12:00.978 question is how far in the future will that be possible? 0:12:01.630 --> 0:12:06.159 And because the industry races forward very rapidly. If any 0:12:06.169 --> 0:12:09.000 country catches up to the cutting edge in 20 years time, 0:12:09.010 --> 0:12:11.659 it's a meaningless accomplishment because the cutting edge will be 0:12:11.669 --> 0:12:15.099 so far uh so far advanced. So I I think 0:12:15.109 --> 0:12:19.250 answering your question requires specifying the time horizon. Do I 0:12:19.260 --> 0:12:24.419 think China can indigenously develop cutting edge capabilities soon using 0:12:24.429 --> 0:12:27.640 solely domestically produced technology? I doubt it. Uh just because 0:12:27.650 --> 0:12:29.419 I don't think any country can do it uh 0:12:30.164 --> 0:12:32.864 soon. Uh And you know, if I look at the 0:12:32.875 --> 0:12:37.094 largest producers of uh in the semi hunter industry, the US, 0:12:37.104 --> 0:12:41.723 uh uh Taiwan, Japan, Korea, none of them have any 0:12:41.734 --> 0:12:44.664 sort of remotely plausible pathway over the next 10 years 0:12:44.674 --> 0:12:47.304 to be completely self sufficient ship making. And so it 0:12:47.315 --> 0:12:49.674 seems highly implausible to me that China which is today 0:12:49.684 --> 0:12:52.825 still a small player in revenue terms, um, has a 0:12:52.835 --> 0:12:54.164 plausible pathway of its own. 0:12:54.820 --> 0:12:56.869 I mean, this may be an unfair question because this 0:12:56.880 --> 0:12:59.669 happened just like last week. So the made 60 phone 0:12:59.679 --> 0:13:01.739 that Huawei came up with and just a couple of 0:13:01.750 --> 0:13:03.809 days ago, I saw that they're also increasing some foldable 0:13:03.820 --> 0:13:06.799 phone and these seem to have nine nanometer chips and 0:13:06.809 --> 0:13:10.699 then made by Ken uh or a K nine process 0:13:10.710 --> 0:13:14.109 or something like that. Um Is it still like a 0:13:14.119 --> 0:13:16.349 made in China phone or there is still a huge 0:13:16.359 --> 0:13:19.429 amount of design and other elements that are coming from 0:13:19.440 --> 0:13:21.130 a globalized supply chain process? 0:13:22.820 --> 0:13:25.289 Well, we're still learning more about the the chip inside 0:13:25.299 --> 0:13:27.460 of the Huawei phone. I think what we can pretty 0:13:27.469 --> 0:13:30.439 safely assume thus far is that uh it's clear that 0:13:30.450 --> 0:13:33.469 the pro the manufacturing process used to manufacture the chip 0:13:33.479 --> 0:13:35.460 is what's called a seven nanometer process. So this is 0:13:35.469 --> 0:13:39.130 a process that um TS MC first brought online 0:13:39.400 --> 0:13:43.960 uh about five years ago uh in Taiwan. Um It's 0:13:43.969 --> 0:13:47.229 a process that is now used in in mid and 0:13:47.239 --> 0:13:51.049 low range smartphones, newer smartphones will have uh tsm CS 0:13:51.059 --> 0:13:53.950 most advanced process, uh five nanometers 0:13:54.419 --> 0:13:58.640 uh in them. Um We don't know about the efficiency 0:13:58.650 --> 0:14:02.699 of the manufacturing yet. So Huawei's manufacturing partner, Smi um 0:14:02.710 --> 0:14:04.590 hasn't released any data nor do I think they will 0:14:04.599 --> 0:14:07.478 any time soon. So uh whether this chip is being 0:14:07.489 --> 0:14:11.939 produced at huge inefficiency or relatively efficient, we're still going 0:14:11.950 --> 0:14:14.179 to find out um what we do know is that it, 0:14:14.190 --> 0:14:18.020 it was uh produced largely using imported tools. Um because 0:14:18.030 --> 0:14:20.960 until recently, SMI was able to buy the most 0:14:21.039 --> 0:14:23.260 advanced as well as the sort of second generation most 0:14:23.270 --> 0:14:26.690 advanced tools from the US, from Japan uh and from 0:14:26.700 --> 0:14:29.570 the Netherlands. Uh And so it had purchased a large 0:14:29.580 --> 0:14:31.330 quantity of those tools and is now still able to 0:14:31.340 --> 0:14:34.320 use uh at least some of those. Uh and this 0:14:34.330 --> 0:14:36.890 ship was made largely using those tools. So, you know, 0:14:36.900 --> 0:14:40.010 in some ways, it's, it's certainly uh evidence of incremental 0:14:40.020 --> 0:14:42.669 progress in the ship industry. It's probably not evidence of 0:14:42.679 --> 0:14:45.750 catching up because TS MC is also moving forward. And 0:14:45.760 --> 0:14:47.750 so the gap between TS MC and 0:14:47.989 --> 0:14:52.630 China's capabilities, uh it doesn't appear to be decreasing. Um 0:14:53.080 --> 0:14:55.780 Is it e evidence of self sufficiency for China? Not 0:14:55.789 --> 0:14:59.599 really because again, the machines that were producing these chips 0:14:59.609 --> 0:15:04.340 were uh largely imported uh machines. So it's, it's, it's 0:15:04.349 --> 0:15:06.400 progress of a sort, but I, I wouldn't call it 0:15:06.409 --> 0:15:06.900 a breakthrough. 0:15:07.659 --> 0:15:09.440 Tell me a little bit about the term that you 0:15:09.450 --> 0:15:12.669 use inefficiency or efficiency. So I think there the yield 0:15:12.679 --> 0:15:14.909 of semiconductor is like a very important part of the 0:15:15.169 --> 0:15:17.780 bottom line of any manufacturing process. So give us a 0:15:17.789 --> 0:15:20.150 sense of, you know what kind of yield is needed 0:15:20.159 --> 0:15:22.840 and how complex it is the process of getting decently 0:15:22.849 --> 0:15:25.010 efficient yield out of a semiconductor production. 0:15:26.070 --> 0:15:27.989 Yeah, you yield is the key variable in it. That 0:15:28.000 --> 0:15:29.840 that just means the share of chips you produce that 0:15:29.849 --> 0:15:34.090 actually work. Um And because chips are so complex at 0:15:34.099 --> 0:15:37.539 the cutting edge, uh it is very, very, very easy 0:15:37.549 --> 0:15:40.000 to make errors in the manufacturing process that cause a 0:15:40.010 --> 0:15:42.760 chip not to work. And there are ways you can 0:15:42.770 --> 0:15:45.109 try to make your chip more or less resilient to 0:15:45.119 --> 0:15:48.619 having errors. But whenever any company brings online, a new 0:15:48.695 --> 0:15:52.645 manufacturing process more advanced with smaller transistors, there's always a 0:15:52.655 --> 0:15:56.044 very uh significant learning curve that has to be overcome 0:15:56.174 --> 0:15:59.304 and yields always start low and then they're always intended 0:15:59.315 --> 0:16:03.385 to end uh very high in the 90% level um 0:16:03.395 --> 0:16:08.565 uh as close to 100 as possible. Um And the, the, 0:16:08.575 --> 0:16:11.905 the pro the process of improving yields is therefore critical 0:16:11.914 --> 0:16:15.895 to financial viability for companies. And so a key question 0:16:15.905 --> 0:16:17.065 in your ability to 0:16:17.320 --> 0:16:20.130 price uh effectively, if you're a chip maker is what 0:16:20.140 --> 0:16:23.210 yield you could produce at. Because if you're producing a 50% yield. 0:16:23.609 --> 0:16:26.729 In other words, you gotta spend um amount of, you know, you, 0:16:26.739 --> 0:16:29.400 you need to, you need to, to fund the production 0:16:29.409 --> 0:16:31.059 of twice as many chips as you actually get. In 0:16:31.070 --> 0:16:33.080 terms of working chips, you're gonna be a lot less 0:16:33.090 --> 0:16:36.710 competitive than a company with 90% yield. Uh And so we, 0:16:36.719 --> 0:16:39.409 we don't have good data uh about yields from companies. 0:16:39.419 --> 0:16:42.349 They're very secretive about their yield data. Um But we 0:16:42.359 --> 0:16:44.190 know that the the leading chip makers 0:16:44.284 --> 0:16:47.065 uh end up with very high yields uh in the 90% 0:16:47.075 --> 0:16:51.054 range um for their, their, their mature processes. And so 0:16:51.065 --> 0:16:53.534 if you look at the new Huawei phone, you know, 0:16:53.544 --> 0:16:56.205 a fundamental question is, what's the yield? And I don't 0:16:56.215 --> 0:16:57.484 think we're gonna know the answer to that question any 0:16:57.494 --> 0:17:00.994 time soon. But any assessment of how viable a processes 0:17:01.005 --> 0:17:03.494 is or not has to uh has to try to 0:17:03.505 --> 0:17:04.954 get us a sense of the yield. 0:17:05.680 --> 0:17:08.930 So Chris, you mentioned that the Huawei phone uh could have, 0:17:08.939 --> 0:17:12.609 you know, homegrown processors, but the machine tools, the critical 0:17:12.619 --> 0:17:17.400 component of preparing that semiconductor are all imported. Um So 0:17:17.410 --> 0:17:21.310 give us a sense of state of sanctions in play 0:17:21.319 --> 0:17:25.119 right now. Are we saying that China is restricted from 0:17:25.130 --> 0:17:30.199 importing any lithography machines going forward or is the idea that, 0:17:30.209 --> 0:17:32.150 you know, they'll just get stuck in whatever they were 0:17:32.160 --> 0:17:33.889 allowed to import and 0:17:34.400 --> 0:17:37.010 they can keep on importing that but nothing else after that. 0:17:38.689 --> 0:17:41.489 So over the past uh year, the US, Japan and 0:17:41.500 --> 0:17:44.449 the Netherlands have each released a set of controls which 0:17:44.459 --> 0:17:49.000 have uh basically tried to stop China from buying new 0:17:49.010 --> 0:17:52.699 tools that could be used at sub 16 or 14 0:17:52.709 --> 0:17:55.959 nanometer production. So China has been able in the pass 0:17:55.969 --> 0:17:59.119 to buy lots of tools and has bought huge volumes 0:17:59.130 --> 0:18:02.479 of tools that used for 14 nanometers, 10 nanometers and 0:18:02.489 --> 0:18:04.489 seven is the the note after that which is the 0:18:04.500 --> 0:18:07.319 note that the Huawei phone was uh produced on 0:18:07.680 --> 0:18:10.540 and so until recently and actually the the Dutch regulations 0:18:10.550 --> 0:18:12.410 don't come into force until the end of this year. 0:18:12.420 --> 0:18:15.849 So it's still the case that um that Chinese buyers 0:18:15.859 --> 0:18:18.800 can buy uh certain tools from the Netherlands. But once 0:18:18.810 --> 0:18:21.979 all these restrictions are enforced, the uh the the net 0:18:21.989 --> 0:18:26.079 effect will be to limit China's purchase of Chinese firms 0:18:26.089 --> 0:18:28.849 purchase of any new tools that can be used primarily 0:18:28.859 --> 0:18:30.670 for sub 14 nanometer 0:18:31.099 --> 0:18:36.479 uh production. Now most chips by volume are actually produced 0:18:36.489 --> 0:18:40.198 at larger geometries. In other words, older manufacturing technologies at 0:18:40.209 --> 0:18:44.130 40 nanometers or 100 and 30 nanometers. But in dollar terms, 0:18:44.140 --> 0:18:46.339 the most expensive chips are the ones that are produced 0:18:46.349 --> 0:18:49.729 at more advanced technologies. And so inside your smart, 0:18:49.800 --> 0:18:53.119 the main application processor is produced at something pretty close 0:18:53.130 --> 0:18:56.659 to cutting edge inside your PC. Inside data centers, the 0:18:56.670 --> 0:18:59.609 key chips are uh are are are also pretty close 0:18:59.619 --> 0:19:00.719 to the cutting edge. And so for a lot of 0:19:00.729 --> 0:19:04.859 key technology areas, there are huge performance gains to be 0:19:04.869 --> 0:19:07.589 had by producing on a cutting edge technology or something 0:19:07.599 --> 0:19:08.430 very close to it. 0:19:09.619 --> 0:19:13.310 So the implications of this could be rather dire for 0:19:13.319 --> 0:19:16.449 China's medium term outlook. So for the time being, they 0:19:16.459 --> 0:19:19.489 can produce some chips and get on with it. But 0:19:19.500 --> 0:19:20.630 in terms of pushing 0:19:21.439 --> 0:19:24.310 gps that you need to train A I models or 0:19:24.319 --> 0:19:26.150 if you want to run the next generation of data 0:19:26.160 --> 0:19:30.188 centers or develop the next generation of weapons, um they're 0:19:30.199 --> 0:19:32.409 basically doomed to failure. 0:19:34.319 --> 0:19:36.790 Well, II I think I would, I would differentiate first 0:19:36.800 --> 0:19:40.510 between the the restrictions on tools used to produce ships 0:19:40.520 --> 0:19:42.159 below 14 nanometers 0:19:42.579 --> 0:19:45.750 um and the restrictions on specific chips themselves. So the US, 0:19:45.760 --> 0:19:48.849 Japan and the Netherlands have restricted tools used for sub 0:19:48.869 --> 0:19:52.739 14 nanometers, but they haven't restricted the transfer of chips 0:19:52.750 --> 0:19:56.020 to China, sub 14 nanometers. And so today, China imports 0:19:56.030 --> 0:19:59.229 almost all of its sub 14 nanometer chips from Taiwan, 0:19:59.239 --> 0:20:02.949 from Korea, from the US uh from from elsewhere, which 0:20:02.959 --> 0:20:05.560 is the reason that China spends as much money each 0:20:05.569 --> 0:20:07.369 year importing chips, it spends importing oil, 0:20:07.449 --> 0:20:11.159 it's largely smartphone processors, PC processors. And so it's actually 0:20:11.170 --> 0:20:15.060 not much of a change to China's chip buying patterns 0:20:15.150 --> 0:20:17.800 to have it keep buying chips from abroad. Um In 0:20:17.810 --> 0:20:21.050 so far as the regulations uh by limiting China's domestic 0:20:21.060 --> 0:20:24.540 production force China to buy chips uh abroad. There are 0:20:24.550 --> 0:20:28.169 also as you mentioned, specific US rules that ban the 0:20:28.180 --> 0:20:31.089 transfer of GP US or chips used to train A 0:20:31.099 --> 0:20:32.239 I systems to China. 0:20:32.560 --> 0:20:35.750 Uh but these are very, very specific rules governing just 0:20:35.760 --> 0:20:38.409 one type of chip. And so the US has via 0:20:38.420 --> 0:20:40.989 its rules kept wide open the transfer of any chip 0:20:41.000 --> 0:20:44.479 for smartphones, any chip for autos, any chip for P CS. 0:20:44.930 --> 0:20:48.270 Um almost all types of chips except GP US. So 0:20:48.280 --> 0:20:52.089 China can still buy very advanced ships from abroad for smartphones. 0:20:52.270 --> 0:20:54.910 The only restriction is when it comes to the chips 0:20:54.920 --> 0:20:57.140 that are used to train A I systems. And those 0:20:57.150 --> 0:20:59.229 are the chips called GP US that are largely produced 0:20:59.239 --> 0:20:59.949 by NVIDIA. 0:21:00.680 --> 0:21:03.310 Would the state of sanctions remain where they are or 0:21:03.319 --> 0:21:05.329 the net is about to get tighter. 0:21:06.890 --> 0:21:10.079 Well, I think a lot depends on um first off 0:21:10.089 --> 0:21:12.520 uh the direction of of us politics US has an 0:21:12.530 --> 0:21:16.599 election uh next year which could bring policy change. Um 0:21:16.609 --> 0:21:21.020 but second Chinese policy um as well, I think, you know, 0:21:21.030 --> 0:21:23.760 one of the key questions, the key uncertainties is 0:21:24.000 --> 0:21:27.560 what cost China is willing to bear to pursue its 0:21:27.569 --> 0:21:31.510 self-sufficiency drive. And we've seen the Chinese government spend tens 0:21:31.520 --> 0:21:34.959 of billions of dollars a year from 2014 to present 0:21:34.989 --> 0:21:39.819 towards self sufficiency. I think uh that rate will certainly 0:21:39.829 --> 0:21:42.229 continue it. But I wonder if it will even intensify 0:21:42.239 --> 0:21:44.000 and if China is willing to spend more money on 0:21:44.010 --> 0:21:47.810 self sufficiency. In other words, produce more chips at a 0:21:47.819 --> 0:21:51.089 very inefficient manner domestically, you could end up buying fewer 0:21:51.099 --> 0:21:52.109 chips from abroad. 0:21:52.400 --> 0:21:54.399 And if China starts buying fewer chips from abroad, it 0:21:54.410 --> 0:21:56.890 also loses some of its leverage because right now, one 0:21:56.900 --> 0:22:00.750 of the factors limiting foreign restrictions on China is that 0:22:00.760 --> 0:22:02.770 foreign and chip firms still sell a lot of chips 0:22:02.780 --> 0:22:05.688 to China. But if they begin to be restricted from 0:22:05.699 --> 0:22:09.510 market access by China's self sufficiency drive, then you'll have 0:22:09.520 --> 0:22:14.089 less lobbying in Washington, in Tokyo uh in European capital 0:22:14.099 --> 0:22:16.900 saying be careful with your restrictions, you're gonna hurt our 0:22:16.910 --> 0:22:20.040 market because if that market's already declining, that'll become much 0:22:20.050 --> 0:22:20.869 less powerful 0:22:21.229 --> 0:22:23.800 argument. And that's I think a very tricky balance for China. 0:22:23.810 --> 0:22:26.260 It wants self sufficiency for sure, but it also values 0:22:26.270 --> 0:22:29.030 the fact that because its market is valuable, it gives 0:22:29.040 --> 0:22:32.109 it some bargaining power in its relations with the West. 0:22:32.459 --> 0:22:35.780 It's, it's really fascinating, right? So if I don't have 0:22:35.790 --> 0:22:40.319 an advanced domestic industry, interdependence remains uh but the moment 0:22:40.329 --> 0:22:43.109 I progress interdependence, we can and maybe there is sanctions 0:22:43.119 --> 0:22:45.938 at that point, but it's too late because I am independent. 0:22:46.140 --> 0:22:51.389 Um Chris, your book showcases uh TSM CS singular role 0:22:51.400 --> 0:22:55.188 in the manufacturing of the world's most advanced chips. Um 0:22:55.199 --> 0:22:57.530 And you mentioned earlier as well that there are other 0:22:57.540 --> 0:23:00.709 single nodes of vulnerability. A smo being another one. So 0:23:00.719 --> 0:23:04.140 how risky it is. Is it for like one company 0:23:04.150 --> 0:23:07.079 at this day and age being responsible for 100% of 0:23:07.089 --> 0:23:09.290 something that the world needs critically? 0:23:11.400 --> 0:23:13.939 Well, I don't think anyone designing a supply chain would 0:23:13.949 --> 0:23:18.750 have opted in uh to a system of complete vulnerability, 0:23:18.760 --> 0:23:21.089 complete reliance on a single firm. Like we're all reliant, 0:23:21.099 --> 0:23:24.489 for example, on a S MLS lithography machines, I think 0:23:24.500 --> 0:23:27.219 everyone would prefer a competitive market. Certainly the buyers of 0:23:27.229 --> 0:23:27.750 lithography 0:23:27.829 --> 0:23:31.400 machines would prefer a competitive market right now. They're price takers. 0:23:31.430 --> 0:23:36.069 Um But I think there's particular concern around uh TS 0:23:36.079 --> 0:23:40.379 MC because of the risks to Taiwan. Uh that I 0:23:40.390 --> 0:23:44.050 think most people assess have grown somewhat in, in recent years. 0:23:44.500 --> 0:23:48.910 The, the, the Taiwan question I think is, is important, 0:23:48.920 --> 0:23:52.380 not just because of TSM CS dynamic as being the 0:23:52.390 --> 0:23:55.679 central player of in producing advanced ships, but also because 0:23:55.729 --> 0:23:59.390 it has become this critical producer of advanced ships precisely 0:23:59.400 --> 0:24:04.050 as the us ability to credibly defend Taiwan has deteriorated. 0:24:04.780 --> 0:24:07.290 I think that's the second half dynamic. It's not just 0:24:07.300 --> 0:24:09.380 a chip story, it's also a story of the military 0:24:09.390 --> 0:24:12.400 balance shifting in Taiwan's favor. And so for most of 0:24:12.410 --> 0:24:17.760 the last 50 years, uh tech executives, investors, um electronics 0:24:17.770 --> 0:24:19.849 firms didn't have to think that hard about the military 0:24:19.859 --> 0:24:21.930 balance because it was obvious if there were a war, 0:24:21.939 --> 0:24:24.780 everyone knew who would win, everyone knew which would win. 0:24:24.790 --> 0:24:26.859 Beijing and Washington and Taipei everywhere. 0:24:27.150 --> 0:24:28.780 Today. If there were a war, there's a lot of 0:24:28.790 --> 0:24:31.219 uncertainty about what the outcome would be. And I think 0:24:31.229 --> 0:24:36.520 that has inspired uh an increased sense of uncertainty about 0:24:36.530 --> 0:24:39.790 whether China might be willing to escalate. And if so 0:24:39.800 --> 0:24:41.739 what the ramifications would be for the supply chain. And 0:24:41.750 --> 0:24:44.510 so that's really the reason why we're talking about uh 0:24:44.520 --> 0:24:49.119 Taiwan risk is that the military balance has shifted dramatically 0:24:49.130 --> 0:24:49.938 in China's favor. 0:24:51.040 --> 0:24:53.319 Right. Yeah, I don't see anybody forcing a SML to 0:24:53.329 --> 0:24:57.369 start making lithography missions elsewhere because is not seen as 0:24:57.410 --> 0:24:59.790 a note of vulnerability out. There are these, you know, 0:24:59.800 --> 0:25:02.099 terrific chapters in your book where you sort of talk 0:25:02.109 --> 0:25:05.790 about the war game scenarios around which China Taiwan, uh 0:25:05.800 --> 0:25:08.209 the situation can unravel. And I think that was also 0:25:08.219 --> 0:25:11.339 very instructive. Ok. I want to come back to something 0:25:11.349 --> 0:25:14.469 we talked about earlier, Gordon Moore and his incredible Christian 0:25:14.479 --> 0:25:17.290 call 55 years ago that, you know, 0:25:17.550 --> 0:25:21.500 chip speed processing speed can double 18 to 24 months. 0:25:21.599 --> 0:25:23.319 How long can that last Chris? 0:25:26.020 --> 0:25:28.310 Well, I, nobody knows the answer to that question. Even 0:25:28.319 --> 0:25:30.949 the smartest people in the industry are just speculating. I 0:25:30.959 --> 0:25:32.489 think we can say a couple of things. 0:25:33.290 --> 0:25:38.160 One is that since the 19 eighties, Moore's law's death 0:25:38.170 --> 0:25:40.800 has been regularly predicted and it's always been wrong. 0:25:41.609 --> 0:25:44.479 In the 19 eighties. There were very distinguished computer uh 0:25:44.489 --> 0:25:48.209 engineers who were highly confident that Morris Law would come 0:25:48.219 --> 0:25:50.319 to an end that they were completely wrong in the 0:25:50.329 --> 0:25:53.650 early two thousands. Gordon Moore himself said that he couldn't 0:25:53.660 --> 0:25:57.219 fathom ways that the industry could keep Mori's law alive 0:25:57.229 --> 0:25:58.218 for more than 0:25:58.560 --> 0:26:01.060 and here we are two decades later. Uh and it's 0:26:01.069 --> 0:26:04.139 still alive. So I think we should be very cognizant 0:26:04.150 --> 0:26:06.979 of the fact that past predictions of Moore's Law's failure 0:26:06.989 --> 0:26:10.760 have always been wrong. Uh And that should I think 0:26:10.770 --> 0:26:12.698 sensitize us to the fact that there are lots of 0:26:12.709 --> 0:26:14.859 smart people trying to find ways to keep it alive. 0:26:15.280 --> 0:26:17.400 I think also it's the case that we, we have 0:26:17.410 --> 0:26:19.099 a pretty clear line of sight. 0:26:19.319 --> 0:26:21.369 If you look at the road maps uh that have 0:26:21.380 --> 0:26:24.290 been put out by TS MC or Intel uh over 0:26:24.300 --> 0:26:26.770 the next decade, how we're gonna get at least another 0:26:26.780 --> 0:26:31.150 decade of transistor shrinkage beyond that point, nobody knows. But 0:26:31.160 --> 0:26:33.770 it's always been the case that more than a decade out, 0:26:33.780 --> 0:26:36.760 nobody knows. It's just too far out to, to, to 0:26:36.770 --> 0:26:37.869 realistically plan, 0:26:38.510 --> 0:26:41.650 right? Certainly will be the place. No, no, please go ahead. 0:26:42.560 --> 0:26:44.849 Well, II I it, it certainly will be the case 0:26:44.859 --> 0:26:47.469 that at some point you can't string transistors any further. 0:26:47.479 --> 0:26:50.468 We're already measuring them in terms of nanometers. Uh Soon 0:26:50.479 --> 0:26:53.729 we'll be in the Angstrom era uh defining them in 0:26:53.739 --> 0:26:56.130 terms of angst drums. We'll have to invent I think 0:26:56.140 --> 0:26:59.859 new terminology to measure the uh minute distances by the 0:26:59.869 --> 0:27:01.819 end of the decade. Uh But at some point, we'll 0:27:01.829 --> 0:27:04.209 hit real physical limits. But even at that point, I 0:27:04.219 --> 0:27:06.890 think the the past decade has actually given us a 0:27:06.900 --> 0:27:09.310 lot of new technique 0:27:09.400 --> 0:27:14.359 to provide better semiconductors um with the same size transistors. So, 0:27:14.369 --> 0:27:17.189 for example, there's new ways to design semiconductors and GP 0:27:17.199 --> 0:27:19.290 U is the best example. What makes GP U a 0:27:19.300 --> 0:27:21.520 great chip for A I is not that the transistors 0:27:21.530 --> 0:27:23.750 are any smaller than the chip inside of your iphone. 0:27:23.760 --> 0:27:25.569 It's that they, they're designed in a different way. And 0:27:25.579 --> 0:27:27.910 so I think we're still in the very early stages 0:27:27.920 --> 0:27:32.280 of exploring ways chip design can produce uh performance enhancements. 0:27:32.459 --> 0:27:34.449 And then second, there's a whole new trend in the 0:27:34.459 --> 0:27:36.349 industry called advanced packaging, 0:27:36.790 --> 0:27:39.219 which sounds very boring, but actually, it is going to 0:27:39.229 --> 0:27:41.969 be revolutionary in terms of letting us put different types 0:27:41.979 --> 0:27:45.000 of chips together in different ways, which will provide huge 0:27:45.010 --> 0:27:48.550 performance improvements in terms of speed of data transfer, in 0:27:48.560 --> 0:27:52.489 terms of managing heat dissipation, huge issue when you've got 0:27:52.500 --> 0:27:53.979 lots of electrons flying around. 0:27:54.329 --> 0:27:56.900 Um and also in terms of cost, letting you combine, 0:27:56.910 --> 0:27:58.780 for example, a low cost ship and a high cost 0:27:58.790 --> 0:28:02.280 ship in a way that brings down your overall device cost. 0:28:02.290 --> 0:28:04.459 And so we're in the early stages of of this 0:28:04.469 --> 0:28:06.329 trend as well. And so when I look at the 0:28:06.339 --> 0:28:10.099 trajectory we have of at least another decade of shrinking 0:28:10.109 --> 0:28:13.859 transistors further lost the design renaissance is underway. Plus the 0:28:13.869 --> 0:28:17.140 advanced packaging shifts. I'm pretty optimistic that we've got a 0:28:17.150 --> 0:28:19.599 long pathway for further advances. 0:28:20.569 --> 0:28:23.530 So even if the exponential increase in the number of 0:28:23.540 --> 0:28:27.399 transistors per chip stops, we can probably have computing power 0:28:27.410 --> 0:28:29.380 per chip continue to increase. 0:28:29.739 --> 0:28:31.219 And that's right. That's exactly right. 0:28:31.709 --> 0:28:33.880 Right. You know, you thanks to your book. I mean, 0:28:33.890 --> 0:28:35.650 I'm doing stuff that I've never done before. I now 0:28:35.660 --> 0:28:38.219 go to the IBM website and they have this newsletter 0:28:38.229 --> 0:28:42.020 from their scientists about to your point about chip design 0:28:42.180 --> 0:28:45.229 and issues related heat dissipation. So, yeah. Um I have 0:28:45.239 --> 0:28:48.650 no idea exactly what they're talking about, but it's nonetheless 0:28:48.689 --> 0:28:50.910 sort of energizing to see that there's all sorts of 0:28:50.920 --> 0:28:53.920 thinking going on for the next and the next next 0:28:53.930 --> 0:28:55.359 uh generation of frog 0:28:56.170 --> 0:28:59.810 um Chris, we talked about military uh issues earlier. I 0:28:59.819 --> 0:29:02.530 want to bring that back. Uh Even in your book, 0:29:02.540 --> 0:29:06.109 uh from the very beginning when rockets were being developed, 0:29:06.119 --> 0:29:09.949 uh military uh experts saw the application of chips in 0:29:09.959 --> 0:29:12.920 the guidance system and you have this uh really cool 0:29:12.930 --> 0:29:14.420 chapter in your book where you talk about the early 0:29:14.430 --> 0:29:17.050 nineties during the Gulf War. First time we saw the 0:29:17.060 --> 0:29:19.189 application of, you know, advanced guidance systems. 0:29:19.459 --> 0:29:23.599 Um So military applications are always, always at the forefront 0:29:23.609 --> 0:29:25.510 of ship demand. Is that a good thing? 0:29:27.380 --> 0:29:28.949 Well, I, I don't know if it's a good thing. 0:29:28.959 --> 0:29:31.650 I think it's an inevitable thing. I think uh it 0:29:31.660 --> 0:29:35.890 is inevitable today that militaries are going to try to 0:29:36.239 --> 0:29:40.050 acquire whatever edge they can. Um And today that means 0:29:40.060 --> 0:29:42.069 developing systems that are more autonomous 0:29:42.589 --> 0:29:45.209 uh autonomous systems are a key focus of all of 0:29:45.219 --> 0:29:49.540 the world's major militaries and deploying A I to defense 0:29:49.550 --> 0:29:53.300 and intelligence systems we already see in the Russia Ukraine War. 0:29:53.310 --> 0:29:56.130 Plenty of uh of examples of ways that A I 0:29:56.140 --> 0:29:59.280 is is being used. For example, militaries have computer vision 0:29:59.290 --> 0:30:01.520 algorithms in their intelligence systems. So 0:30:01.910 --> 0:30:04.280 take a satellite photo and differentiate what's a tank and 0:30:04.290 --> 0:30:06.609 what's a truck, not like be done automatically. That's, that's 0:30:06.619 --> 0:30:11.439 pretty simple. Um But uh more decisions about um communications. 0:30:11.449 --> 0:30:15.280 For example, if your adversary is jamming your frequencies of 0:30:15.290 --> 0:30:17.939 your communication system, can you use an A I to 0:30:17.949 --> 0:30:21.439 rapidly jump between frequency in a way that's faster than 0:30:21.449 --> 0:30:22.579 your adversary system? 0:30:23.140 --> 0:30:25.709 Uh These are some of the use cases of, of a, of, 0:30:25.719 --> 0:30:28.130 of A I and military contacts. And then I think 0:30:28.140 --> 0:30:30.770 if you look at the role that drones have played 0:30:30.780 --> 0:30:33.829 in the Russia Ukraine war again, that provides a very clear, 0:30:33.839 --> 0:30:36.930 I think worrisome but clear pathway for how militaries will 0:30:36.939 --> 0:30:42.430 begin to deploy increasing degrees of autonomy in drones as well. 0:30:42.439 --> 0:30:45.290 And so we're gonna, I think have a very complex 0:30:45.300 --> 0:30:46.449 ethical debates over the next 0:30:46.525 --> 0:30:50.854 several decades about in what ways are we comfortable deploying 0:30:50.864 --> 0:30:53.734 autonomy in military systems? But just as we're running at 0:30:53.744 --> 0:30:56.954 autonomous cars and just as autonomy is coming to all other, 0:30:56.964 --> 0:30:59.515 the range P applications. So too, it's already coming to 0:30:59.525 --> 0:31:02.635 military systems. Uh And there is a, I think very 0:31:02.645 --> 0:31:05.925 real race between the world's leading militaries and above all 0:31:05.935 --> 0:31:08.915 between China and the US to harness this technology for 0:31:08.925 --> 0:31:09.834 defense purposes. 0:31:10.780 --> 0:31:13.599 So robot soldiers here we come. Um Yeah, when I 0:31:13.609 --> 0:31:16.689 see those Boston dynamics videos of those, you know, somersaulting 0:31:16.699 --> 0:31:19.579 uh robots. The first thing I don't think initially everybody 0:31:19.589 --> 0:31:21.949 said that that's like the next, you know, guy who's 0:31:21.959 --> 0:31:24.689 shelving stuff at the Amazon warehouse. But to me, it 0:31:24.699 --> 0:31:27.790 was more like somebody holding an AK-47 and, you know, 0:31:27.800 --> 0:31:30.349 joining the front. Um Chris, you know, I mean, this 0:31:30.359 --> 0:31:33.810 is just a stream of consciousness observation while the Ukraine 0:31:33.819 --> 0:31:36.589 conflict shows all sorts of cutting edge application, 0:31:37.050 --> 0:31:39.689 it still looks like a very first world war type 0:31:39.699 --> 0:31:42.969 of uh terrain. When we look at the videos trenches 0:31:42.979 --> 0:31:45.660 and something called dragon's teeth, which is basically, you know, 0:31:45.670 --> 0:31:48.670 concrete boulders to stop tanks from going over. Um So 0:31:48.680 --> 0:31:50.719 it seems like, you know, we have a lot of 0:31:50.729 --> 0:31:54.750 shiny new toys. Uh but the adaption of them, the 0:31:54.760 --> 0:31:57.119 full usage of them, it takes a while, right? I mean, 0:31:57.130 --> 0:31:59.250 would you agree if I say that the sort of 0:31:59.260 --> 0:32:02.219 compute power that we have right now, even the existing 0:32:02.229 --> 0:32:04.750 computer power, forget about the stuff that's coming. We're barely 0:32:04.760 --> 0:32:05.880 using a fraction of that. 0:32:07.619 --> 0:32:09.750 I, I think that's right. II I think when you 0:32:09.760 --> 0:32:13.510 look at the Russian Ukraine war, it's easy to miss 0:32:13.520 --> 0:32:16.280 a lot of the ways in which computing has actually 0:32:16.290 --> 0:32:20.310 been pretty decisive. Um So for example, there's, there's the 0:32:20.319 --> 0:32:22.329 whole cyber war that didn't happen. 0:32:22.939 --> 0:32:28.849 Um We collectively expected that Russia would launch devastating cyberattacks 0:32:28.859 --> 0:32:31.579 and knock out Ukraine's cyber infrastructure ali accounts, but that 0:32:31.589 --> 0:32:32.349 didn't happen 0:32:32.699 --> 0:32:36.290 in no small part because Ukraine was defended by Google 0:32:36.300 --> 0:32:39.849 and Microsoft and some of the key Western tech giants and, 0:32:39.859 --> 0:32:42.849 and their computing capabilities are key to cyber defenses. And it, 0:32:42.969 --> 0:32:45.630 you know, we've talked about autonomy and uh in, in 0:32:45.640 --> 0:32:48.209 many parts of the economy, well, autonomy in both cyber 0:32:48.219 --> 0:32:51.689 offense and cyber defense is coming soon, it's actually already 0:32:51.699 --> 0:32:54.180 already arrived. And so uh that will be a AAA 0:32:54.189 --> 0:32:57.839 new frontier of the application of, of A I. When 0:32:57.849 --> 0:33:00.729 it comes to the intelligence that Ukraine has received from 0:33:00.739 --> 0:33:01.310 the West, 0:33:01.869 --> 0:33:05.849 we've already uh we've got many examples of um Ukraine 0:33:05.859 --> 0:33:09.260 receiving very accurate targeting data from the West and Signals 0:33:09.270 --> 0:33:12.530 Intelligence is one of the early use cases uh of 0:33:12.540 --> 0:33:14.920 the application of A I to military systems. And so 0:33:14.930 --> 0:33:18.000 we don't have uh great visibility into all of the 0:33:18.010 --> 0:33:22.089 specific platforms that the US is using to uh give 0:33:22.099 --> 0:33:25.569 Ukraine uh targeting data, for example, for its high on rockets. 0:33:25.579 --> 0:33:26.738 But I don't think you can 0:33:27.130 --> 0:33:30.839 um safely. Uh I think you have to assume that 0:33:30.969 --> 0:33:33.810 uh there's a whole lot of computing uh involved in 0:33:33.819 --> 0:33:38.050 the very rapid identification of targets uh transmission of those targets, 0:33:38.060 --> 0:33:41.199 Ukrainian forces and the Ukrainians uh our system, if you 0:33:41.209 --> 0:33:44.130 just press a button and the rocket flies automatically it 0:33:44.140 --> 0:33:45.760 hit its targets. So those are just two ways in 0:33:45.770 --> 0:33:46.630 which computing is made. 0:33:47.030 --> 0:33:49.420 I think a real difference on the battlefield. But you know, 0:33:49.430 --> 0:33:51.689 I think you're right that, uh, in some ways this 0:33:51.699 --> 0:33:53.750 is a 21st century war, in other ways, it's 1/19 0:33:53.760 --> 0:33:58.650 century war. Uh, and I, I think that speaks to the, uh, the, 0:33:59.050 --> 0:34:01.709 the fact that regardless of the technologies or in some 0:34:01.719 --> 0:34:04.819 ways because of the technologies, uh, warfare remains a very 0:34:04.829 --> 0:34:05.699 brutal business. 0:34:07.030 --> 0:34:10.629 Yeah, absolutely. So, we've sort of almost naturally progressed to 0:34:10.639 --> 0:34:13.290 talking about Ukraine. That was supposed to be my final question. 0:34:13.419 --> 0:34:15.719 And I think many fans of Chip War may not 0:34:15.729 --> 0:34:19.590 be aware that your core expertise include Russia and Cold 0:34:19.600 --> 0:34:22.719 War history. And you've written several books on Putin and 0:34:22.729 --> 0:34:25.719 Imperial Russia. I'd love to do a separate podcast one day, 0:34:25.729 --> 0:34:28.229 Chris with just talking about Russia because there's like so 0:34:28.239 --> 0:34:31.050 much to talk about, particularly the historical context that you 0:34:31.060 --> 0:34:34.399 have written so articulately about. But, uh, just give me 0:34:34.409 --> 0:34:36.360 a sense, uh, we've talked just a little bit about 0:34:36.370 --> 0:34:36.719 the 0:34:36.979 --> 0:34:40.379 technology around the Ukraine conflict. But as a historian, you know, 0:34:40.389 --> 0:34:43.010 what's your take on the course the conflict has taken 0:34:43.229 --> 0:34:46.429 and what would be a path to resolution? 0:34:48.790 --> 0:34:51.419 Well, I think the the key driver of, of the 0:34:51.429 --> 0:34:54.459 conflict is, is a question about where does Russia's sphere 0:34:54.469 --> 0:34:57.569 of control end? I think the, the Russian view which 0:34:57.739 --> 0:35:01.050 President Putin has articulated is that Ukraine is an, ought 0:35:01.060 --> 0:35:03.959 to be in Russia's sphere of control. And the Ukrainian 0:35:03.969 --> 0:35:06.399 view and the Western view is that Ukraine should be 0:35:06.409 --> 0:35:08.620 able to make its own choices about whether it wants 0:35:08.629 --> 0:35:10.870 to be part of the Russian sphere. I mean, that's 0:35:10.879 --> 0:35:15.419 the core of disagreement that has led to uh the conflict. 0:35:15.429 --> 0:35:15.860 And I think 0:35:16.239 --> 0:35:20.209 there's no pathway in sight for that issue to be resolved. 0:35:20.219 --> 0:35:24.469 And people speculate, welcome, you know, this territorial settlement, that 0:35:24.479 --> 0:35:25.409 territorial settlement 0:35:25.770 --> 0:35:28.709 resolve the conflict. But I think the core question is 0:35:28.790 --> 0:35:35.020 um is Russia going to continue uh fighting to keep 0:35:35.030 --> 0:35:37.830 Ukraine or try to keep Ukraine in its sphere or 0:35:37.840 --> 0:35:39.550 is it willing to let Ukraine not be uh in 0:35:39.560 --> 0:35:42.989 its sphere? And that's, I think the key uncertainty that 0:35:43.120 --> 0:35:46.158 uh we just don't know uh Russia's willingness to fight 0:35:46.169 --> 0:35:49.189 on to pursue that aim. And the Ukrainians will say 0:35:49.199 --> 0:35:50.840 we're not going to do a cease fire now because 0:35:50.850 --> 0:35:53.459 if we do, the Russians will start fighting again after 0:35:53.469 --> 0:35:55.040 rebuilding for a couple of years. And 0:35:55.600 --> 0:35:57.330 you know, I understand why they think that because that's 0:35:57.340 --> 0:36:00.020 been the track record of, of recent history, there was 0:36:00.030 --> 0:36:02.639 a cease fire in 2014, the Russians rebuilt and uh 0:36:02.649 --> 0:36:06.429 attacked again. And so I, I think that that that 0:36:06.439 --> 0:36:09.770 key question of, of ultimately, is Ukraine an independent country 0:36:09.780 --> 0:36:12.929 or not? And does Russia recognize it is independent or not? 0:36:13.169 --> 0:36:17.090 That's what's at stake? Uh And I don't see right 0:36:17.100 --> 0:36:22.050 now um any clear pathway to 0:36:22.399 --> 0:36:25.689 getting a resolution of, of that question that both Ukraine 0:36:25.699 --> 0:36:26.610 and Russia could agree on, 0:36:28.129 --> 0:36:33.149 right. Um gosh, uh not the most hopeful uh you know, prognosis. 0:36:33.159 --> 0:36:35.330 And Chris, we, uh I, I was talking to a 0:36:35.340 --> 0:36:36.939 friend of mine, she's Russian and she just came back 0:36:36.949 --> 0:36:39.820 from Moscow and she said that you don't really feel 0:36:39.830 --> 0:36:41.860 that there's a war going on. Uh when you walk 0:36:41.870 --> 0:36:44.280 the streets of Moscow and when you go to electronic 0:36:44.290 --> 0:36:46.620 stores that used to be, say, 0:36:46.729 --> 0:36:49.359 you know, filled with Apple and Microsoft products, now they're 0:36:49.370 --> 0:36:51.979 filled with uh made in China electronics products and they 0:36:51.989 --> 0:36:54.639 look really good and they're shiny and people are buying 0:36:54.649 --> 0:36:58.040 those things. So this sort of, you know, pivot toward 0:36:58.050 --> 0:37:01.280 Chinese tech. I mean, can Russia sort of, you know, 0:37:01.290 --> 0:37:05.159 withstand Western sanctions substantially better because of that. 0:37:07.360 --> 0:37:09.929 Well, I don't really think the, the pivot to Chinese 0:37:09.939 --> 0:37:12.429 tech is the key to Russia's withstanding of sanctions. I 0:37:12.439 --> 0:37:15.840 I think the key is Russia's oil exports. Um That's 0:37:15.850 --> 0:37:17.770 been the key and I think we've learned over the 0:37:17.780 --> 0:37:21.370 course of the last 18 months of war. Um several 0:37:21.379 --> 0:37:27.899 key things. Um One is that the Russian war effort 0:37:27.909 --> 0:37:30.129 has been very stimulative for the economy. 0:37:30.689 --> 0:37:32.750 Um So the Russian government is, is running a pretty 0:37:32.760 --> 0:37:36.360 large budget deficit now. Um but that is produced a 0:37:36.370 --> 0:37:39.560 sort of Keynesian impact on the Russian economy, 0:37:39.739 --> 0:37:42.280 uh which I think ought to have been expected. But 0:37:42.290 --> 0:37:45.509 most people myself included didn't fully uh expect in the 0:37:45.520 --> 0:37:48.729 early days of the war. And so rather than unemployment 0:37:48.739 --> 0:37:52.209 or declining incomes. We actually have very rapid income growth 0:37:52.219 --> 0:37:55.629 driven by very low unemployment in Russia because of labor 0:37:55.639 --> 0:37:58.229 shortages now. So that's, that's one surprise to the war. 0:37:58.239 --> 0:38:02.209 I think the second surprise that is about the West's 0:38:02.219 --> 0:38:05.469 willingness to impose sanctions because one of the really striking 0:38:05.479 --> 0:38:06.689 acts of the war is that 0:38:06.939 --> 0:38:10.510 uh the West at first started to impose very tough 0:38:10.520 --> 0:38:12.790 oil sanctions on Russia and then immediately walked them all 0:38:12.800 --> 0:38:15.839 back through the price cap scheme. Um And so now 0:38:15.850 --> 0:38:19.029 Russian oil exports are basically unchanged since before the war. 0:38:19.040 --> 0:38:22.929 Um precisely because the West has collectively decided that it 0:38:22.939 --> 0:38:25.810 doesn't actually want to sanction Russian oil exports. It wants 0:38:25.820 --> 0:38:28.989 more Russian oil exports to keep prices as low. 0:38:29.260 --> 0:38:31.100 That is possible. And so if you look at the 0:38:31.110 --> 0:38:34.510 Russian economic situation, that's the key dynamic is that a, 0:38:34.520 --> 0:38:39.060 the domestic economy is being supported by deficit spending domestically 0:38:39.070 --> 0:38:42.600 and b the external accounts are um they look relatively 0:38:42.610 --> 0:38:46.070 stable because oil exports and prices fall on oil. But 0:38:46.080 --> 0:38:48.649 uh volumes are uh almost as high as ever, 0:38:49.949 --> 0:38:52.439 right? But uh I wanna stick with that tech issue. 0:38:52.449 --> 0:38:54.520 I agree with you. I agree that the oil export 0:38:54.530 --> 0:38:58.659 revenues are basically what's driving, you know, Russia's resilience. And 0:38:58.669 --> 0:38:59.739 of course, you know, if they want to buy stuff 0:38:59.750 --> 0:39:02.000 from China, they still have to pay and uh and 0:39:02.010 --> 0:39:04.239 then that's where the revenue is coming from. But going 0:39:04.250 --> 0:39:06.259 back to the tech issue, um 0:39:06.899 --> 0:39:09.000 the, the view is that, you know, if you don't 0:39:09.010 --> 0:39:12.100 have access to Western technology, if the Apple I OS 0:39:12.110 --> 0:39:16.080 is not, you know, upgrading and you're not getting the 0:39:16.090 --> 0:39:20.790 Microsoft 3 65 support that day to day commerce and 0:39:20.800 --> 0:39:23.570 inventory management. Everything sort of comes to a standstill, 0:39:23.840 --> 0:39:25.810 but clearly, that's not the case for Russia. So I'm 0:39:25.820 --> 0:39:30.589 assuming that they're not getting those regular updates from Western tech. 0:39:30.659 --> 0:39:33.449 And still, I don't hear any logistics failure or any 0:39:33.489 --> 0:39:36.409 difficulty in running power plants or for that matter, running 0:39:36.419 --> 0:39:40.459 the military in the case of Russia. So it's basically 0:39:40.469 --> 0:39:42.639 what smuggling is that is that the way they're sort 0:39:42.649 --> 0:39:43.669 of keeping themselves going. 0:39:44.820 --> 0:39:45.929 Well, I think for, for 0:39:45.939 --> 0:39:50.259 software tools, uh there, it's always been relatively easy to 0:39:50.270 --> 0:39:52.830 pirate software tools and, and rush actually has been a 0:39:52.840 --> 0:39:56.350 big market for Pirated software. And so if you can't 0:39:56.360 --> 0:39:58.709 get the latest IO I OS update, you just use 0:39:58.719 --> 0:40:02.610 the old version, it's slightly more vulnerable to being hacked 0:40:02.620 --> 0:40:05.049 your performance declines over time. 0:40:05.310 --> 0:40:09.699 Um But that's, that's pretty manageable. Um I think for, 0:40:09.709 --> 0:40:13.320 for enterprise software, one of the surprising things has been 0:40:13.360 --> 0:40:18.479 how long many enterprise software firms kept servicing Russian customers. 0:40:18.570 --> 0:40:19.320 Um 0:40:19.870 --> 0:40:22.699 And uh there's, it's been a, a matter of some debate, 0:40:22.709 --> 0:40:24.080 I think in the West as well. That was the 0:40:24.090 --> 0:40:28.770 right approach, but many enterprise software companies, um, essentially kept 0:40:28.780 --> 0:40:32.620 Russian clients um, up to date with functional software for 0:40:32.629 --> 0:40:34.729 a very long time, some even up to the present 0:40:34.899 --> 0:40:37.129 uh day. So I, I think on the software side, 0:40:37.139 --> 0:40:41.199 Russia has actually had very few difficulties um in accessing 0:40:41.209 --> 0:40:44.639 maybe for certain niche types of industrial software. But generally speaking, 0:40:44.649 --> 0:40:47.889 it hasn't, doesn't appear to have been a problem um for, 0:40:47.899 --> 0:40:49.459 for tech components, 0:40:50.169 --> 0:40:53.799 you know, there have been open discussions in the Russian 0:40:53.810 --> 0:40:57.439 press about certain types of component shortages. Um but the 0:40:57.449 --> 0:41:02.469 Russian manufacturing base is actually relatively small um outside of 0:41:02.479 --> 0:41:06.669 the auto sector which has gotten uh clobbered by um 0:41:06.679 --> 0:41:09.629 by sanctions of export controls. The other big sector for 0:41:09.639 --> 0:41:11.000 manufacturing is the military. 0:41:11.399 --> 0:41:14.259 Um And the military has got the budget, it needs 0:41:14.270 --> 0:41:18.600 to uh import components um from abroad. It probably does 0:41:18.610 --> 0:41:20.638 sort of an inefficient manner. It probably faced supply chain 0:41:20.649 --> 0:41:23.229 difficulties due to sanctions, but it's been able to import 0:41:23.239 --> 0:41:26.800 at least some volume because we know that multi production continues. 0:41:26.810 --> 0:41:29.409 Um And so I think those are kind of the 0:41:29.419 --> 0:41:31.859 autos in the in the the defense industry, the two 0:41:32.000 --> 0:41:35.070 sectors of Russian manufacturing that are most reliant on uh 0:41:35.080 --> 0:41:36.259 foreign components. 0:41:37.649 --> 0:41:40.639 But there's no area of sort of symbiosis that there 0:41:40.649 --> 0:41:44.110 are certain areas of tech excellence in Russia that China 0:41:44.120 --> 0:41:45.658 can exploit or vice versa. 0:41:47.659 --> 0:41:51.429 I there are a, a small number of discrete examples. 0:41:51.439 --> 0:41:55.739 One could point to um specific computer vision firms that 0:41:55.750 --> 0:41:59.850 have been invested in by Chinese firms, but they're extraordinarily 0:41:59.860 --> 0:42:03.179 small in number. I think that the tech dependence is 0:42:03.189 --> 0:42:07.029 almost exclusively unidirectional Russia is becoming dependent on China. 0:42:07.550 --> 0:42:11.479 OK. Totally different question. So I'm talking to you from Singapore. Actually, 0:42:11.489 --> 0:42:13.120 I don't know where in the world you are right now, Chris, 0:42:13.129 --> 0:42:13.709 where are you? 0:42:14.540 --> 0:42:16.590 I'm, I'm today in Wyoming. 0:42:16.770 --> 0:42:19.419 Oh My. OK. Uh I thought you were like in 0:42:19.429 --> 0:42:22.840 China or something. Uh OK. So um I'm talking to 0:42:22.850 --> 0:42:26.290 you from Singapore, Singapore over the last 50 years have 0:42:26.300 --> 0:42:30.439 had a manufacturing footprint with respect to chips and semiconductors 0:42:30.449 --> 0:42:34.669 and so on. Uh any sense of the excellence of 0:42:34.679 --> 0:42:37.969 production in Singapore vis a vis a Korea, Japan and 0:42:37.979 --> 0:42:38.409 so on. 0:42:40.820 --> 0:42:43.949 Well, I think in terms of cutting edge production, the 0:42:43.959 --> 0:42:46.370 the cutting edge has been in recent years in Taiwan 0:42:46.379 --> 0:42:49.929 and Korea uh in the US, not in Singapore, but 0:42:49.939 --> 0:42:54.149 there's a very uh substantial manufacturing base in Singapore. And 0:42:54.159 --> 0:42:55.860 just because something is not cutting edge doesn't mean it's 0:42:55.870 --> 0:42:59.629 not excellent. Um indeed, there are lots of very profitable 0:42:59.639 --> 0:43:03.090 uh companies that do non cutting edge chip making often 0:43:03.100 --> 0:43:06.169 with unique mix signal or analog or radio frequency 0:43:06.245 --> 0:43:09.915 requirements as well. Uh And so III I would say 0:43:09.925 --> 0:43:12.495 that Singapore's chip making has not been cutting edge, but 0:43:12.504 --> 0:43:14.674 I wouldn't say that that means it has been excellent. 0:43:14.774 --> 0:43:16.995 Um because there are many different ways you can actually 0:43:17.004 --> 0:43:21.074 define what's a profitable business. What's a um what's a, 0:43:21.084 --> 0:43:24.404 a technologically viable uh uh product? And I think Singapore 0:43:24.415 --> 0:43:26.915 is likely to remain an important hub of chip making. 0:43:27.175 --> 0:43:30.014 Uh and also uh a hub of assembling some of 0:43:30.024 --> 0:43:31.594 the tools that are used to make chips, 0:43:32.179 --> 0:43:35.929 right? OK. So final question since you mentioned Korea, uh 0:43:35.939 --> 0:43:38.669 when people sort of, you know, dissect it through the 0:43:38.679 --> 0:43:40.589 uh Huawei phone. So one of the things that both 0:43:40.600 --> 0:43:43.179 mentioned that there are some Korean parts in it, I 0:43:43.189 --> 0:43:46.109 think he, right, that's the company that produces those things. 0:43:46.120 --> 0:43:50.449 Um So yes, Taiwan is at the absolute, you know, 0:43:50.459 --> 0:43:53.750 singular you know, security with respect to cutting it chips 0:43:53.760 --> 0:43:56.550 just gives a sense of how things are in Korea. 0:43:56.560 --> 0:43:57.489 Perhaps also in Japan. 0:43:58.879 --> 0:44:02.830 Korea is really critical as a producer of memory chips. Now, 0:44:02.840 --> 0:44:04.510 there's two types of memory chips, both of which are 0:44:04.520 --> 0:44:07.189 produced at the cutting edge in Korea. There's a number 0:44:07.199 --> 0:44:10.379 of firms that produce cutting edge memory chips. Um uh 0:44:10.409 --> 0:44:13.750 one in the US, uh one in Japan and, and 0:44:13.760 --> 0:44:18.330 two in Korea, primarily um memory Chis are basically a commodity, 0:44:18.340 --> 0:44:21.010 you can more or less swap out one company's memory 0:44:21.020 --> 0:44:22.070 Chis for another. 0:44:22.360 --> 0:44:25.050 And as a result, there's not nearly as much differentiation 0:44:25.060 --> 0:44:28.439 between uh the firms and the pricing dynamics are also 0:44:28.449 --> 0:44:32.689 quite different profitability is uh generally lower. Um But Korea 0:44:32.699 --> 0:44:35.570 is by far the world's largest producer of uh memory 0:44:35.580 --> 0:44:38.830 chips and Korean firms, Samsung and Hex are are absolutely 0:44:38.915 --> 0:44:42.533 critical suppliers of of of memory chips when it comes 0:44:42.544 --> 0:44:45.435 to Japan, Japan is actually not primarily important as a 0:44:45.445 --> 0:44:48.014 chip maker. There, there are chips made in Japan, especially 0:44:48.024 --> 0:44:51.404 for auto and industrial uses. Um But Japan is is 0:44:51.415 --> 0:44:54.665 really unique as a producer of highly specialized chemicals used 0:44:54.675 --> 0:44:57.495 in ship making and also some of the machine tools 0:44:57.504 --> 0:45:01.125 needed to manufacture advanced ships. And that's really where Japanese 0:45:01.135 --> 0:45:04.104 firms have uh pretty unique uh market positions. 0:45:04.820 --> 0:45:10.449 I remember during the Fukushima earthquake and tsunami, uh this 0:45:10.459 --> 0:45:13.169 is a power plant issued that there were that one 0:45:13.179 --> 0:45:15.350 note of vulnerability that there were some chips that were 0:45:15.360 --> 0:45:17.949 made for navigation systems which were not coming in. And 0:45:17.959 --> 0:45:21.000 as a result, Thailand's auto production came to a standstill. 0:45:21.080 --> 0:45:22.429 Uh And that was the first time I sort of 0:45:22.439 --> 0:45:25.049 started thinking about these, you know, nodes of vulnerability. 0:45:25.260 --> 0:45:28.350 Um Chris uh I believe in your book, you had 0:45:28.360 --> 0:45:30.139 mentioned as far as the memory chips, you know, the, 0:45:30.149 --> 0:45:32.409 the DRS and the NANS are concerned that is one 0:45:32.419 --> 0:45:35.149 area where the Chinese can probably eat Korea's lunch. 0:45:36.860 --> 0:45:37.780 Well, it's, it's, 0:45:37.790 --> 0:45:41.229 it's probably true if China can get access to the 0:45:41.239 --> 0:45:44.610 tools needed to make those ships. And until recently, China 0:45:44.620 --> 0:45:49.229 could uh now, however, uh those tools are restricted, uh 0:45:49.239 --> 0:45:52.330 Chinese firms can't purchase them. And actually the leading Chinese 0:45:52.340 --> 0:45:55.429 memory producer, a company called YMTC is uh is uh 0:45:55.439 --> 0:45:59.250 under additional us sanctions in addition to the the broad ones. 0:45:59.639 --> 0:46:00.989 And so actually, 0:46:01.239 --> 0:46:04.270 um I think one could argue that Korean firms have 0:46:04.280 --> 0:46:08.089 seen a potential major competitor really uh impacted by these 0:46:08.100 --> 0:46:11.850 controls in a way that might protect their market share 0:46:11.860 --> 0:46:13.388 uh for some time to come. So there are lots 0:46:13.399 --> 0:46:17.489 of unexpected um knock on effects of the controls that 0:46:17.500 --> 0:46:21.250 have interesting implications uh for for a number of different countries. 0:46:21.260 --> 0:46:22.469 But Korea being one of them, 0:46:23.379 --> 0:46:26.169 lots of unexpected knock on effects. I think that's a 0:46:26.179 --> 0:46:29.169 very good way to sort of summarize the the world, 0:46:29.179 --> 0:46:32.979 the chip war dynamic. Uh Chris, I, I really appreciate 0:46:32.989 --> 0:46:35.409 your time and thank you so much for your insights. 0:46:36.330 --> 0:46:37.209 Well, thank you for having 0:46:37.219 --> 0:46:40.799 me. Fantastic time with the uh talking about this and 0:46:40.810 --> 0:46:43.509 great job with the book. Uh Kobe Time is for 0:46:43.520 --> 0:46:47.629 information only and does not represent any trade recommendations. All 0:46:47.639 --> 0:46:50.939 108 episodes of the podcast are available on youtube and 0:46:50.949 --> 0:46:54.959 all major podcast platforms including Apple Google and Spotify. As 0:46:54.969 --> 0:46:57.629 for our research publications, webinars and live streams, you can 0:46:57.639 --> 0:47:01.219 find them all by Googling devious research library. Have a 0:47:01.229 --> 0:47:02.060 great day.