WEBVTT - How the U.S. Lost Chip Dominance and How It Can Be Regained 0:00:11.039 --> 0:00:15.120 Hello, and welcome to another episode of the Odd Lots podcast. 0:00:15.200 --> 0:00:19.520 I'm Joe Wisn'tal And I'm Tracy all Away. Tracy, remember 0:00:19.520 --> 0:00:21.640 when I said a few weeks ago that I thought 0:00:21.680 --> 0:00:26.599 we should do a semiconductor Odd Lots spinoff series. No whole, 0:00:26.680 --> 0:00:28.560 not a whole, not a series, like a whole new show. 0:00:29.120 --> 0:00:31.280 The idea, a whole new show on the subject. No, 0:00:31.480 --> 0:00:33.040 that's not what you said. You said you wanted to 0:00:33.080 --> 0:00:37.640 transform All Thoughts into a semiconductor podcast. So this is 0:00:37.680 --> 0:00:42.280 the latest step in your plan to do that, I assume. Yeah, 0:00:42.440 --> 0:00:45.000 so we're not going to do that, And realistically, I 0:00:45.040 --> 0:00:47.760 don't think we're going to have a new spinoff Odd 0:00:47.800 --> 0:00:52.640 Lots branded uh semi conductor show, but I do think 0:00:53.000 --> 0:00:54.880 but we are going to do some more episodes on 0:00:54.920 --> 0:00:58.480 the topic because it really does just intersect with a 0:00:58.560 --> 0:01:02.280 lot of different themes. There's sort of the pure tech 0:01:02.400 --> 0:01:05.920 question of their role in the economy, why different companies 0:01:05.959 --> 0:01:09.360 are rising and falling, but there's it also intersects with 0:01:09.720 --> 0:01:14.399 um things like national security and sort of economic strength 0:01:14.560 --> 0:01:17.560 and some of the questions we've seen during the COVID 0:01:17.600 --> 0:01:23.120 crisis about say domestic manufacturing and reliance on manufacturing in 0:01:23.160 --> 0:01:27.680 other countries, So it's a it's a rich load to mine. Yeah, well, 0:01:27.680 --> 0:01:29.800 I mean you could go even further. So it kind 0:01:29.840 --> 0:01:32.760 of gets to, I guess, the heart of a large 0:01:32.760 --> 0:01:36.160 body of economic thought, right, how do you balance domestic 0:01:36.280 --> 0:01:40.000 interests with market forces? So when the entire market is 0:01:40.000 --> 0:01:43.920 telling you to outsource manufacturing to a cheaper place and 0:01:43.959 --> 0:01:46.960 they can do it more efficiently, is that the way 0:01:46.959 --> 0:01:50.320 to go for something as important as a sensitive technology 0:01:50.880 --> 0:01:55.200 like a semiconductor. And I think we're seeing those tensions 0:01:55.280 --> 0:01:57.280 over and over. I think we see them reflected in 0:01:57.480 --> 0:02:00.320 a lot of US politics and the way people feel 0:02:00.360 --> 0:02:02.560 at the moment. But of course we also see that 0:02:02.640 --> 0:02:07.800 reflected in the U. S. China trade war, and technology 0:02:07.880 --> 0:02:11.880 is really ground zero of that. And within technology, semiconductors 0:02:11.919 --> 0:02:16.280 are the epicenter of the ground zero. I'm mixing my metaphors, 0:02:16.320 --> 0:02:18.519 but I think you get it. The epicenter of the 0:02:18.560 --> 0:02:20.840 ground zero. No, I get it, And I think you're 0:02:20.840 --> 0:02:23.640 absolutely right. Like it's kind of a it's a business topic, 0:02:24.160 --> 0:02:27.320 but it intersects with a lot of our favorite macro 0:02:27.720 --> 0:02:30.160 themes that we discuss all the time. And I think 0:02:30.160 --> 0:02:33.680 another sort of macro theme is you know, we talked 0:02:33.680 --> 0:02:37.840 about it a lot in but the sort of rethinking 0:02:37.880 --> 0:02:40.520 the role of the public sector and rethinking the role 0:02:40.520 --> 0:02:44.720 of government spending, rethinking the role of government investment, and 0:02:44.760 --> 0:02:46.519 I think that's also I mean, that's a that's a 0:02:46.560 --> 0:02:49.480 big topic for us, it's a big topic nationally. I mean, 0:02:49.520 --> 0:02:51.640 I think you know, if you look at one of 0:02:51.680 --> 0:02:55.320 the successes that we've had in this crisis, most people 0:02:55.320 --> 0:02:58.519 would say the speed with which we were able to 0:02:59.120 --> 0:03:02.560 develop a X seeing for covid um was in part 0:03:02.680 --> 0:03:06.320 due to the government investment and operation warp speed, and 0:03:06.360 --> 0:03:10.079 of course the longstanding investment in pharma. So there's all 0:03:10.160 --> 0:03:13.840 kinds of reasons to sort of use this moment as 0:03:14.000 --> 0:03:18.160 a chance to reconceptualize what is the role of the 0:03:18.160 --> 0:03:22.079 public sector. How can the public sector spur more innovation, 0:03:22.720 --> 0:03:26.280 manufacturing production. So it's a it's a great topic because 0:03:26.320 --> 0:03:29.640 it hits a lot of odd themes. I would say, yeah, 0:03:29.639 --> 0:03:32.600 that's right, And I think we have seen a proposal 0:03:32.680 --> 0:03:38.720 in Washington about how to boost domestic semiconductor capabilities, So yes, 0:03:38.840 --> 0:03:41.560 it hits all the sweet spots. It's something that's happening 0:03:41.760 --> 0:03:43.880 right now as well. And I think it's going to 0:03:43.920 --> 0:03:47.160 be an interesting conversation. Yeah, I'm really excited about this. 0:03:47.400 --> 0:03:51.000 Um our guest today is um Professor of Management Practice 0:03:51.080 --> 0:03:54.720 at Harvard Business School, Willie She. He writes a lot 0:03:54.760 --> 0:03:59.840 about the topic about semiconductors, about manufacturing, US manufacturing, and 0:04:00.000 --> 0:04:02.960 particular what it takes to make it work. So we're 0:04:02.960 --> 0:04:05.400 going to dive into all of these questions with him. 0:04:05.920 --> 0:04:09.400 Professor She, thank you so much for joining us. Well, 0:04:09.440 --> 0:04:12.040 thank you for having me, Joe. So I'm trying to 0:04:12.080 --> 0:04:14.640 think like where the best place to start is. But 0:04:15.080 --> 0:04:18.440 you know, obviously, uh, we're talking about Intel, and for 0:04:18.480 --> 0:04:21.840 a long time, Uh, Intel is not just the leader 0:04:22.560 --> 0:04:26.520 in semiconductors, but the sort of like the vanguard of 0:04:26.839 --> 0:04:32.080 US tech and tech manufacturing capability. And of course the 0:04:32.160 --> 0:04:34.920 story over the last couple of years has been the 0:04:35.000 --> 0:04:39.960 problems they've run into in manufacturing. Also business model questions. 0:04:40.000 --> 0:04:44.520 But to start, why is it important for a country 0:04:44.600 --> 0:04:47.279 like the US or a country like China for that matter, 0:04:47.320 --> 0:04:49.960 But why is it an important for country like the 0:04:50.080 --> 0:04:55.400 US to have this capability at scale to produce high 0:04:55.400 --> 0:04:58.599 tech goods such as semi conductors. Well, I think the 0:04:58.680 --> 0:05:03.200 reason semiconductors have attracted so much focus in this regard 0:05:03.800 --> 0:05:07.440 is because they are a building block for so much 0:05:07.560 --> 0:05:11.159 modern technology. In fact, if you look at and it's 0:05:11.160 --> 0:05:14.280 been really highlighted by the pandemic kind of the role 0:05:14.520 --> 0:05:20.279 of electronics and technology communications technology, whether it is conducting 0:05:20.279 --> 0:05:25.120 classes and meetings and business on Zoom or it's ordering online, 0:05:25.279 --> 0:05:28.920 underpinning all this sort of the the core building block 0:05:29.240 --> 0:05:34.000 are semiconductors. When you start to look at the numbers 0:05:34.880 --> 0:05:38.880 on what percentage of the manufacturing happens in the US 0:05:38.960 --> 0:05:42.440 and how that has declined over the years, it's only 0:05:42.440 --> 0:05:47.320 about twelve for the microelectronics that we're talking about, for 0:05:47.440 --> 0:05:51.360 things like Intel microprocessors or the chips that go into smartphones. 0:05:52.080 --> 0:05:54.000 And you look at how much of the dependence is 0:05:54.040 --> 0:05:59.039 in Asia, and combine that with the rhetoric around you know, 0:05:59.160 --> 0:06:04.040 dependency on China or areas that China claims like Taiwan. 0:06:04.440 --> 0:06:08.480 I think that has really brought attention to a concern 0:06:08.600 --> 0:06:11.560 that I would say has been around in some circles 0:06:11.600 --> 0:06:16.360 for the past twenty years. Mhm. So, I think there 0:06:16.440 --> 0:06:21.720 is a recognition that semiconductors are strategically important for the 0:06:21.760 --> 0:06:25.359 reasons that you just laid out. But we've we've discussed 0:06:25.360 --> 0:06:29.440 this in some previous episodes. It's an incredibly competitive industry 0:06:29.839 --> 0:06:34.560 and it has huge startup costs. How difficult is it 0:06:34.600 --> 0:06:42.560 to actually um set up and create a semiconductor manufacturing capability? 0:06:42.880 --> 0:06:45.880 And what are the business decisions that kind of go 0:06:46.000 --> 0:06:50.720 into that. Well, there are a couple of things that 0:06:51.320 --> 0:06:54.440 I think have evolved as you get to the more 0:06:54.480 --> 0:06:58.640 advanced technologies. The cost of setting up kind of a 0:06:58.720 --> 0:07:04.080 minimum efficient ale manufacturing facility. It used to be, you know, 0:07:05.160 --> 0:07:07.279 back thirty years ago, you could do this for a 0:07:07.279 --> 0:07:10.480 few hundred million dollars, okay, or at the beginning it 0:07:10.520 --> 0:07:14.280 was a few million dollars. Now to set up the 0:07:14.360 --> 0:07:18.240 most advanced, leading edge fab will probably cost about twenty 0:07:18.280 --> 0:07:23.440 billion dollars, right, And that's a substantial capital cost That 0:07:24.320 --> 0:07:29.360 really strikes fear in the hearts of many uh CFOs 0:07:30.040 --> 0:07:33.240 as they look at the level investment. And and when 0:07:33.280 --> 0:07:36.080 you make that level investment, you have to make it 0:07:36.120 --> 0:07:39.320 successful in order to earn a return. That means you 0:07:39.520 --> 0:07:44.040 have to be able to learn all those processed technologies. 0:07:44.560 --> 0:07:49.400 You have to be able to manufacturer to high yield, okay, 0:07:49.440 --> 0:07:53.840 because if you don't have high yield, then you're throwing 0:07:54.120 --> 0:07:56.240 out half of your output and then your cost just 0:07:56.920 --> 0:07:58.880 you know, you don't have anything to absorb your costs. 0:07:58.960 --> 0:08:03.240 So it's a technological challenge in this the financial challenge 0:08:03.240 --> 0:08:08.600 as well. So, uh, you know we're talking about Intel. 0:08:08.640 --> 0:08:11.880 You're the co author of a book Producing Prosperity, why 0:08:11.920 --> 0:08:17.119 American needs a manufacturing renaissance. So when when thinking about 0:08:17.200 --> 0:08:21.000 this question, it seems like there's kind of two parts 0:08:21.000 --> 0:08:24.680 to it. One is, Okay, something is going on at Intel, 0:08:25.280 --> 0:08:29.360 they're having tech problems, they're having business model problems as well, 0:08:29.760 --> 0:08:35.520 But there's also a sort of broader US manufacturing malaise 0:08:35.760 --> 0:08:38.120 or you know, a reason that there's less and less 0:08:38.160 --> 0:08:42.000 of it happening here. How do you sort of disentangle 0:08:42.120 --> 0:08:43.960 the two things? How much of it is sort of 0:08:44.000 --> 0:08:47.360 macro factors about something that's going on in the US 0:08:47.440 --> 0:08:51.920 that's causing this decline in high tech manufacturing capability and 0:08:51.960 --> 0:08:56.920 know how versus specific micro things going on at a 0:08:57.000 --> 0:09:00.600 company that happens to be stumbling. Well, so let me 0:09:00.840 --> 0:09:04.959 start with the macro thing, because I think in essence, 0:09:05.000 --> 0:09:10.199 the financial risk and the technological challenge a are linked. 0:09:10.480 --> 0:09:15.760 Because what we saw, we've seen this type of crisis before. 0:09:16.120 --> 0:09:18.280 It was actually in the nineteen eighties, and at the 0:09:18.360 --> 0:09:22.960 time it wasn't China. It was a Japanese who were 0:09:23.080 --> 0:09:28.720 manufacturing memory chips so called d rans dynamic random access memories. 0:09:29.280 --> 0:09:32.680 And in the nineteen eighties, the Japanese had invested in 0:09:32.760 --> 0:09:37.840 manufacturing and they were actually much better at it than 0:09:38.000 --> 0:09:42.800 American companies, even like Intel. Right, Intel was founded as 0:09:42.800 --> 0:09:47.120 a memory company. Many people forget that, right, but Intel 0:09:47.760 --> 0:09:51.080 Texas Instruments was a big player in memories, moss tech 0:09:51.800 --> 0:09:56.920 companies like that, and the Japanese out executed American manufacturers 0:09:56.960 --> 0:09:59.520 because they were delivering higher yields. Right, So if you 0:09:59.559 --> 0:10:02.040 think out, I'm going to sink all this money into 0:10:02.080 --> 0:10:05.280 a factory. If I have higher yields, that means my 0:10:05.400 --> 0:10:11.320 cost is lower. And uh, Intel really was pushed the 0:10:11.360 --> 0:10:14.840 wall on memories and they decided to get out of 0:10:14.920 --> 0:10:19.320 memory manufacturing and focused on microprocessor manufacturing. At the time, 0:10:19.559 --> 0:10:23.240 you know, IBM had picked UH their microprocessor for the 0:10:23.240 --> 0:10:27.840 IBM PC. There were early comments in Intel where they 0:10:27.840 --> 0:10:30.760 weren't sure if you know, this PC thing was going 0:10:30.800 --> 0:10:34.360 to be big, Okay, But in some sense they were 0:10:34.400 --> 0:10:39.760 smart and they were lucky in moving two microprocessors, okay, 0:10:39.800 --> 0:10:43.120 and then they were able to escape this problem of 0:10:43.200 --> 0:10:47.000 being the best manufacturer. Okay. Now, at the same time, 0:10:47.520 --> 0:10:51.679 what we saw in American industry was this separation of 0:10:51.720 --> 0:10:56.439 design from manufacturing, which was really enabled by a lot 0:10:56.480 --> 0:10:59.800 of these computer aided design tools, right, which you know, 0:11:00.120 --> 0:11:04.120 separated out the manufacturing process. And then we saw the 0:11:04.240 --> 0:11:07.840 rise of some Asian companies, in particular in Taiwan, but 0:11:07.880 --> 0:11:10.880 also Korea and in Japan at the time, who said, well, 0:11:10.920 --> 0:11:14.160 we're very happy to do the manufacturing because we're not 0:11:14.240 --> 0:11:16.640 the experts of the design, right. So we saw that 0:11:16.760 --> 0:11:21.640 bifurcation okay. And now as that has developed, what we've 0:11:21.679 --> 0:11:26.640 seen is as you go to the more advanced manufacturing processes, 0:11:27.240 --> 0:11:32.040 there were very few companies who could kind of keep 0:11:32.080 --> 0:11:35.560 in that game. Intel has historically been one of them, right, 0:11:35.559 --> 0:11:39.080 because after they left the memory business, they invested a 0:11:39.120 --> 0:11:44.959 lot in manufacturing to give their microprocessor business a lead, right, 0:11:45.000 --> 0:11:50.120 And historically Intel lead with manufacturing in that regard, okay, 0:11:50.200 --> 0:11:53.800 and they were able to beat competitors like a m 0:11:53.880 --> 0:12:00.200 D because their manufacturing processes were a generation or two ahead, okay, 0:12:00.200 --> 0:12:02.720 So it didn't matter if you had a design that 0:12:03.040 --> 0:12:06.319 might have not been quite as efficient. Okay. Now a 0:12:06.559 --> 0:12:10.200 m D also had their own manufacturing, but then they 0:12:10.280 --> 0:12:14.320 spun it off a decade ago into Global Foundaries. And 0:12:14.440 --> 0:12:18.880 once a m D separated their manufacturing from their design 0:12:19.720 --> 0:12:22.640 and uh, you know, they left Global Foundaries and they 0:12:22.640 --> 0:12:27.720 shifted over to Taiwan Semiconductor. Okay. Once am D now 0:12:27.880 --> 0:12:33.280 had access to a more advanced process than Intel did, 0:12:33.320 --> 0:12:35.439 they have been able to take a lot of market 0:12:35.440 --> 0:12:39.640 share from Intel. Right. So you know this this problem 0:12:39.679 --> 0:12:42.400 has been a long time in developing. But the roots 0:12:42.440 --> 0:12:46.679 of it are the manufacturing part is capital intensive and 0:12:46.760 --> 0:12:49.520 it's hard, it takes a lot of discipline, Okay, and 0:12:49.559 --> 0:12:52.439 you can make more money on the design side, right. 0:12:52.480 --> 0:12:55.320 So hey, if somebody else wants to do the manufacturing, 0:12:55.520 --> 0:12:57.480 let them do it. I'll be more profitable. And a 0:12:57.480 --> 0:12:59.800 lot of people have said, well, the U S. Semi 0:12:59.800 --> 0:13:04.040 can actor industry had a renaissance in the nineteen nineties 0:13:04.080 --> 0:13:07.640 because a lot of those companies let go of the 0:13:07.679 --> 0:13:10.959 manufacturing part. Not everybody did, but a lot of them did, 0:13:11.080 --> 0:13:14.040 and they've been able to be very successful. Companies like 0:13:14.080 --> 0:13:19.400 Qualcom and in Video in particular. Uh, and now am 0:13:19.480 --> 0:13:25.000 d by letting go of manufacturing. So this is the 0:13:25.840 --> 0:13:29.560 I think it's called idem model, where you design and 0:13:29.600 --> 0:13:33.319 manufacture chips like what Intel has been doing, versus the 0:13:33.400 --> 0:13:37.480 fabulous foundry model like in Video and Qualcom. We got 0:13:37.520 --> 0:13:40.040 into this a little bit with um Well in the 0:13:40.080 --> 0:13:44.800 episode with Stacy raskin the semiconductor analysts over at Bernstein. 0:13:45.840 --> 0:13:48.880 I'm curious, like the consensus in the industry right now 0:13:48.960 --> 0:13:51.880 seems to be if you move to the fabulous foundry model, 0:13:51.880 --> 0:13:56.000 where you sort of outsource your manufacturing to a foundry 0:13:56.080 --> 0:13:59.000 like t SMC, that that's the way to go. But 0:13:59.200 --> 0:14:03.280 your argue that there are some hidden costs or some 0:14:03.400 --> 0:14:07.720 synergies that a company might lose in that model, Is 0:14:07.720 --> 0:14:11.800 that right? Well, yes, I think so. And here I 0:14:11.800 --> 0:14:15.960 would argue that more broadly speaking, there's a set of 0:14:16.040 --> 0:14:23.000 capabilities that comes with your manufacturing which can feedback into 0:14:23.040 --> 0:14:28.119 design as well. Now, I think in the chip industry 0:14:28.480 --> 0:14:33.200 what we have seen is manufacturers have continued to work 0:14:33.520 --> 0:14:37.320 very closely with the foundaries. Uh. And you know a 0:14:37.400 --> 0:14:42.200 lot of that capability is transmitted through like software, design 0:14:42.280 --> 0:14:45.880 tools and so on. Right, So that chip industry has 0:14:45.920 --> 0:14:48.760 been able to the fabulous companies have been able to 0:14:50.120 --> 0:14:56.120 leverage those manufacturing processes by working closely with their foundry partners. Okay. 0:14:55.920 --> 0:15:01.920 I think the question in Semiconnectors is really what would 0:15:01.920 --> 0:15:05.320 it be like for the US if it was unable 0:15:05.360 --> 0:15:11.040 to manufacture anything domestically okay. And I think here the 0:15:11.120 --> 0:15:16.160 COVID pandemic has really shined a light on that because 0:15:17.080 --> 0:15:22.520 on a lot simpler commodities, things like personal protective equipment okay, 0:15:22.680 --> 0:15:26.520 or ninety masks and things like that, you know, we 0:15:26.640 --> 0:15:30.880 didn't have the manufacturing capability. Uh. And the question would 0:15:30.920 --> 0:15:38.040 be what if Asian manufacturing capability was suddenly turned off? 0:15:38.480 --> 0:15:40.120 You know. I'll point out that an awful lot of 0:15:40.120 --> 0:15:45.000 the semiconductor manufacturing capability is in three science parks in Taiwan, 0:15:45.720 --> 0:15:49.880 which is a seismically active region okay, And there have 0:15:50.040 --> 0:15:54.160 been minor disruptions over the last few decades in terms 0:15:54.200 --> 0:15:57.920 of ta typhoons and earthquakes and so on. But you know, 0:15:58.600 --> 0:16:03.040 there's also the risk of geopolitical disruptions. Okay. So I 0:16:03.080 --> 0:16:09.000 think that has heightened the sensitivity about having manufacturing concentrated 0:16:09.000 --> 0:16:13.280 in such a small geographic area. It's a little different, 0:16:13.360 --> 0:16:17.560 I would say than the the know how from manufacturing 0:16:17.600 --> 0:16:23.360 in a lot of other processes, but you do see 0:16:23.360 --> 0:16:27.680 it in semiconductors. In terms of the tool manufacturers okay, 0:16:27.680 --> 0:16:31.960 and here American companies have a large presence and tool 0:16:32.000 --> 0:16:36.880 manufacturing okay, and being close to the actual manufacturing facility 0:16:37.000 --> 0:16:39.360 is very important, and you see a lot of those 0:16:39.360 --> 0:16:45.480 toolmakers moving their development activities and manufacturing activities closer to 0:16:45.520 --> 0:16:49.400 the market, be at Singapore or Taiwan, or Korea or 0:16:49.720 --> 0:17:08.560 or even China. Can you explain in a little more 0:17:08.840 --> 0:17:15.320 detail the mechanism by which learning by doing actual know 0:17:15.440 --> 0:17:20.040 how in the manufacturing feeds through back to innovation, because 0:17:20.280 --> 0:17:22.600 you know, it sounds good this idea that like, okay, 0:17:22.640 --> 0:17:25.560 if you're actually in the process of sort of building chips, 0:17:25.600 --> 0:17:29.960 if you're not outsourcing it, that also helps the technological innovation. 0:17:30.119 --> 0:17:32.800 And that this idea, this dream of okay, you can 0:17:32.880 --> 0:17:36.040 just separate the design and the manufacturing and still be 0:17:36.240 --> 0:17:39.840 a world leader. That it's problematic. How does that work 0:17:39.960 --> 0:17:42.439 in practice? Walk us through a little bit why that 0:17:42.520 --> 0:17:47.120 link is important? So I think the key idea here 0:17:47.320 --> 0:17:53.879 is tacit knowledge okay, And tacit knowledge is that which 0:17:54.359 --> 0:17:58.320 you know but you maybe haven't written down or haven't 0:17:58.359 --> 0:18:02.280 really explained to people. Right, So there is a lot 0:18:02.320 --> 0:18:06.760 of know how on the shop floor that comes from 0:18:06.840 --> 0:18:10.679 kind of learning as I produce more product, then the 0:18:10.760 --> 0:18:13.679 question is how is that knowledge captured and how is 0:18:13.760 --> 0:18:19.639 it uh communicated? Okay? Oftentimes you will see there's this 0:18:19.720 --> 0:18:24.520 concept called design for manufacture. In other words, I was 0:18:24.560 --> 0:18:28.960 in one factory once where the people assembling this piece 0:18:29.000 --> 0:18:31.800 of equipment and said, you know, the guys who design this, 0:18:32.240 --> 0:18:35.920 they haven't thought about you know, assembly sequence for example. 0:18:36.040 --> 0:18:38.480 It's like, you know, you can't put that part there 0:18:38.520 --> 0:18:42.479 because I can't I can't install it. But there's a 0:18:42.520 --> 0:18:46.000 lot of those kind of tricks of the trade that 0:18:46.119 --> 0:18:50.080 come from practice and manufacturing, which then feeds back into 0:18:50.160 --> 0:18:55.320 design that says, okay, when you design it, recognize that 0:18:55.359 --> 0:18:59.800 we have to do these things in manufacturing, and how 0:19:00.240 --> 0:19:04.600 that change your design? Right? So tacit knowledge is really 0:19:05.240 --> 0:19:08.679 the main concept you know that which I know, which 0:19:08.760 --> 0:19:16.480 I cannot have not or cannot explained necessarily. So we've 0:19:16.520 --> 0:19:20.400 sort of laid out the scene here and laid out 0:19:20.600 --> 0:19:24.480 the differences between the two business models and why semiconductors 0:19:24.480 --> 0:19:28.199 are important, and also why the industry is different to 0:19:28.680 --> 0:19:32.800 a lot of other types of manufacturing and design. What 0:19:32.920 --> 0:19:37.240 can actually be done about this? If if we agree 0:19:37.280 --> 0:19:40.639 that Intel is falling behind on semiconductors, that the U 0:19:40.720 --> 0:19:45.159 S should do more to boost its domestic capacity in 0:19:45.320 --> 0:19:48.639 order to have a more resilient supply chain and you know, 0:19:49.400 --> 0:19:53.920 be able to withstand unexpected disruptions like the pandemic, what's 0:19:53.960 --> 0:19:59.159 the first step towards actually achieving that? Well, Tracy, I 0:19:59.160 --> 0:20:02.080 think this is a tough question, uh, And a lot 0:20:02.119 --> 0:20:06.680 of people I've talked to agree on the problem, but 0:20:06.760 --> 0:20:09.360 you know, I think there's really a searching for what 0:20:09.440 --> 0:20:13.800 kinds of steps would help. Now, what some people say 0:20:14.240 --> 0:20:19.080 is Intel needs to establish a foundry operation. Okay, I 0:20:19.440 --> 0:20:25.000 think I agree with that because in manufacturing, and especially 0:20:25.080 --> 0:20:29.119 in manufacturing semiconductors, you need to have value, right because 0:20:29.960 --> 0:20:33.760 through manufacturing lots of volumes, that's where you get your 0:20:33.840 --> 0:20:37.320 process learning, that's where you get your process improvement, that's 0:20:37.320 --> 0:20:40.520 where you improve your yields, okay, and find out what 0:20:40.640 --> 0:20:46.000 works and what doesn't work. Okay. As Intel's microprocessor volumes 0:20:46.320 --> 0:20:51.800 come under competitive assault from other designs, where people go 0:20:51.960 --> 0:20:55.240 to say t SMC or a foundry to manufacture it, 0:20:55.480 --> 0:20:57.760 Intel needs to keep its volumes up. One way it 0:20:57.840 --> 0:21:00.520 could do that is it could establish a foundry operation 0:21:01.280 --> 0:21:04.720 that would make both its own chips and chips for 0:21:04.800 --> 0:21:10.920 other customers. That has its challenges, though, because Intel has 0:21:10.960 --> 0:21:13.840 its own way of doing things which have served its 0:21:13.920 --> 0:21:16.800 own needs historically, and so what it needs to do 0:21:16.880 --> 0:21:21.240 then it needs to learn how to be a service 0:21:21.280 --> 0:21:27.080 business that will service the needs of other customers, including 0:21:27.119 --> 0:21:31.520 its competitors. Okay. And and that's that's kind of a 0:21:31.680 --> 0:21:35.879 tough transition because that means it has to give priority 0:21:36.000 --> 0:21:42.040 to its uh customer slash competitor needs okay, and it's 0:21:42.080 --> 0:21:48.920 competitors who might use it as a foundry then will 0:21:49.240 --> 0:21:52.480 ask the question of why do I want to help 0:21:52.520 --> 0:21:56.760 Intel because they've been a tough competitor over time, and 0:21:56.880 --> 0:22:01.320 if I give business to this foundry and my helping Intel, okay. 0:22:01.320 --> 0:22:05.600 And the other question as a potential US foundary customer 0:22:05.600 --> 0:22:07.960 I'm gonna ask is like, are you going to give 0:22:08.000 --> 0:22:13.199 me the same level of technology and process capability that 0:22:13.320 --> 0:22:19.400 I can reliably get in Taiwan today? So there there 0:22:19.400 --> 0:22:23.280 are a lot of barriers to making a foundry business 0:22:23.320 --> 0:22:27.960 like that work. But I would argue that Intel needs 0:22:28.000 --> 0:22:30.760 to do something like that in order to keep its 0:22:30.920 --> 0:22:36.200 manufacturing volumes up. You know why Intel wow? Because Intel 0:22:36.480 --> 0:22:40.240 is probably the last best chance for the US to 0:22:40.320 --> 0:22:44.760 have a leader into this technology. Let me just actually 0:22:44.920 --> 0:22:47.600 back up here because I'm still sort of struggling with 0:22:48.040 --> 0:22:52.600 a key question, which is when we talk about the 0:22:52.640 --> 0:22:55.439 stumbles at Intel or the decline. I mean, here is 0:22:55.480 --> 0:23:00.200 a company that has been integrated, it's uh connected, you 0:23:00.280 --> 0:23:03.600 know sort of, it has this lots of tacit knowledge 0:23:03.640 --> 0:23:06.719 and having manufactured its own chips for so long, unlike 0:23:06.720 --> 0:23:10.119 some of its competitors. What is your core diagnosis for 0:23:10.200 --> 0:23:13.879 what's wrong with the company? And also, um, you know, recently, 0:23:13.920 --> 0:23:16.320 I think it was the hedge fund Third Point came 0:23:16.359 --> 0:23:20.159 out with some proposals the activist firm to try to 0:23:20.200 --> 0:23:24.040 revive the company. But a, um, how would you sort 0:23:24.040 --> 0:23:27.359 of summarize your diagnosis of what went wrong there? And 0:23:27.400 --> 0:23:31.280 then what do you make of some of the suggestions 0:23:31.320 --> 0:23:36.600 by them? Well, I think uh, Intel needs to invest 0:23:36.760 --> 0:23:42.720 heavily on process leadership. Ten years ago, they were two 0:23:42.720 --> 0:23:47.280 generations ahead of the SMC. Now ten years later there 0:23:47.320 --> 0:23:50.720 behind t SMC. Right, so they they need to invest 0:23:50.960 --> 0:23:56.840 heavily on that. I think Intel historically has tried to 0:23:56.880 --> 0:24:00.600 be very self sufficient in terms of using a lot 0:24:00.640 --> 0:24:04.600 of their own tools and a lot of their own methods. Okay, 0:24:04.640 --> 0:24:07.560 I mean I think one thing that would be helpful 0:24:08.440 --> 0:24:14.639 would be to work more with others in the industry 0:24:14.680 --> 0:24:20.679 in broadening the funnel of ideas that come to them. Okay, 0:24:20.720 --> 0:24:26.520 because in some sense, trying to do it all yourself 0:24:27.160 --> 0:24:34.280 in this very challenging technological era may not be the 0:24:34.320 --> 0:24:37.359 most competitive path. If you look at companies like t SMC, 0:24:38.640 --> 0:24:44.600 they have heavily leveraged partnerships with a SML who's the 0:24:44.680 --> 0:24:52.320 lithography leader for DPUV. They have heavily leveraged organizations like 0:24:53.320 --> 0:24:57.879 im EC in Europe, which is kind of a shared 0:24:57.960 --> 0:25:03.160 research collaborative. Okay. And t SMC is very good about 0:25:04.000 --> 0:25:07.720 working with a broad range of parties, and I think 0:25:07.880 --> 0:25:12.440 that would help Intel, right, And that that means they 0:25:12.480 --> 0:25:15.920 have to kind of set aside some of their established 0:25:15.920 --> 0:25:20.439 ways of doing things and open themselves up to maybe 0:25:20.720 --> 0:25:26.080 broader approaches to the competitiveness. Okay. Now on the third 0:25:26.119 --> 0:25:31.280 point question, Okay, I think I agree with the need 0:25:31.920 --> 0:25:38.520 for Intel to really establish a foundry business because, as 0:25:38.600 --> 0:25:42.960 I said earlier, I think volume is very important for them. 0:25:43.119 --> 0:25:50.080 But just the simplistic idea of separating their design business, 0:25:50.160 --> 0:25:56.160 which is designing microprocessors, and manufacturing, which is manufacturing semiconductors, 0:25:57.359 --> 0:26:00.920 taking them down the path that am Be went down 0:26:01.040 --> 0:26:05.760 before and many other companies went down before, that will 0:26:05.840 --> 0:26:10.600 create value for shareholders. Okay, but then you have to 0:26:10.640 --> 0:26:15.520 ask how likely is that foundry going to be absent 0:26:16.240 --> 0:26:19.639 some other moves to ensure that it has customers and 0:26:19.720 --> 0:26:23.399 ensure that it is competitive. Right, So at a simple level, 0:26:24.720 --> 0:26:26.480 I could see how it would create a lot of 0:26:26.520 --> 0:26:30.760 shareholder value, but is that the right answer for the 0:26:30.840 --> 0:26:36.359 United States? So one thing I'm wondering is, you know, 0:26:36.440 --> 0:26:40.200 we're having this conversation now, and uh, there is a 0:26:40.200 --> 0:26:44.040 proposal in Washington, as I mentioned, to do some of this, 0:26:44.320 --> 0:26:46.560 and we can talk about that in some more detail 0:26:46.640 --> 0:26:51.760 in just a bit. But why hasn't this been done before, 0:26:52.119 --> 0:26:55.800 or why hasn't there been more of a concerted effort 0:26:56.200 --> 0:27:00.240 to boost America's you know, strategic capability when it comes 0:27:00.280 --> 0:27:07.119 to a technology that people agree is increasingly important. I 0:27:07.359 --> 0:27:14.320 think we have h in some sense been looking at 0:27:14.359 --> 0:27:18.919 a rosy picture of the semiconductor industry, reflected by the 0:27:19.040 --> 0:27:25.879 success of fabulous companies like Nvidia, like Broadcom, like Qualcom, 0:27:25.920 --> 0:27:30.359 like Apple by the way, uh, and we're seeing it 0:27:30.440 --> 0:27:34.280 as well in companies like Amazon and Google who say 0:27:34.359 --> 0:27:37.760 I don't need the manufacturing, but you see the valuations 0:27:37.800 --> 0:27:41.560 of these companies and you see the financial success. As 0:27:41.640 --> 0:27:46.680 long as you know, Asia represents a secure manufacturing supply 0:27:46.800 --> 0:27:52.040 based one could argue if they want to subsidize that, 0:27:52.280 --> 0:27:57.439 or they want to do lower value work, then we 0:27:57.480 --> 0:28:00.000 should let them. And there are people in Silicon Valley 0:28:00.040 --> 0:28:02.119 I've talked to who said, you know, there's a strong 0:28:02.320 --> 0:28:07.000 argument for Hey, if Asian countries want to subsidize manufacturing, 0:28:07.000 --> 0:28:09.160 we should let them. We should take advantage of that. 0:28:09.800 --> 0:28:12.960 And then the only question is what happens if you 0:28:13.000 --> 0:28:16.960 get cut off from that supply. Now, I think we 0:28:17.119 --> 0:28:20.080 tend to believe in the US that people wouldn't cut 0:28:20.160 --> 0:28:24.520