# The Story of How TSMC Came to Dominate the World

- **Show:** Odd Lots
- **Published:** 25 Jan 2021
- **Source:** https://omny.fm/shows/odd-lots/the-story-of-how-tsmc-came-to-dominate-the-world
- **Transcript:** https://podcasts.opennash.com/shows/odd-lots/the-story-of-how-tsmc-came-to-dominate-the-world-ba058e/
- **Provider:** OpenNash Podcasts (machine-generated)

## In brief

TSMC dominates chipmaking because it pioneered the foundry model, in which a company only manufactures chips and never designs them. Tim Culpan argues the company grew out of Taiwan's state-backed industrial policy and an American-style management culture. Years of heavy spending on equipment and engineers created an advantage that rivals, and even rich governments, struggle to copy with money alone.

The central tension is that TSMC wants to stay neutral between the US and China but is likely to be forced to pick a side. Culpan expects it to choose America. Its other vulnerabilities are being too advanced for customers to need, a deep tech downturn, and Taiwan's power shortage. Meanwhile China and the US try to build rival chip industries from scratch.

## Key insights

1. **Pure contract manufacturing was once a radical idea** Culpan says that when TSMC started, chip firms both designed and manufactured their own chips, summed up by the line 'real men have fabs'. TSMC's offer to manufacture for others reshaped Silicon Valley.
2. **Philips and Morris Chang's credibility won early clients** Chang sought an industry investor and found Philips, which took roughly a 30% stake for almost 20 years. His long career at Texas Instruments also meant American chip firms trusted him.
3. **Fixed costs force TSMC to spend relentlessly** Its costs are mostly equipment depreciation, which accrues whether or not factories are busy, plus engineers' salaries. That pushed it to price competitively and fill capacity, and to keep investing $10-20 billion a year.
4. **The foundry model won as smartphones replaced PCs** Chip designers like Qualcomm and MediaTek succeeded without factories, while companies that built their own fabs carried the same depreciation in good years and bad. Some, like AMD, gave up manufacturing and others disappeared.
5. **Money alone cannot replicate TSMC's manufacturing know-how** Culpan says EUV equipment costs around $200 million per machine and few firms can afford it or know how to run it. Even if China seized TSMC's factories, he says, it would not know what to do with them.

## Questions this episode answers

### Why is TSMC so important to the world economy?

Tim Culpan says almost every electronic device, from iPhones and cars to temperature sensors, contains a TSMC chip somewhere. He believes the global technology industry would grind to a halt very quickly if TSMC stopped production.

### What is the foundry model and why did it matter?

A foundry only manufactures chips for other companies that design them, instead of doing both jobs itself. It let anyone start a chip design company without building factories, which is how firms like Qualcomm grew.

### Is TSMC a Taiwanese or an American company at heart?

Culpan argues its roots and culture are far more American than Chinese. Its two chairmen have been American citizens, its governance style is American, and its biggest paying customers are American.

### Can TSMC stay neutral between the US and China?

Culpan thinks not forever. He points to the US effectively telling TSMC to stop making chips for Huawei, and says that when forced to choose it will pick the American side.

### What could knock TSMC off the top?

Culpan lists three risks: its technology getting so good that customers stop paying for the newest generation, a huge tech slowdown, and Taiwan's electricity shortage. He also says Samsung is the only other company that might overtake it.

## Facts

- **Tim Culpan** [0:06:19]: "It was founded by Maurice Jiang himself back in 1987, but the seeds of Taiwan's Semiconductor industry predated that by about 14 or 15 years"
- **Tim Culpan** [0:06:19]: "when an organization was set up by the Taiwan government back then in 1973 called ITRI, the Industrial Technology Research Institute."
- **Tim Culpan** [0:08:11]: "at the time, Jerry Sanders of AMD later said in that famous line that real men have fabs."
- **Tim Culpan** [0:09:52]: "He was born in China and left China at the height of the Civil War in 1948."
- **Tim Culpan** [0:09:52]: "who went to Taiwan at the ripe old age of 54 to take up a retirement job."
- **Tim Culpan** [0:12:17]: "Philips was actually a founding investor and had, I think it was about 30 % stake."
- **Tim Culpan** [0:16:59]: "but born out of ITRI in around 1980 was another company called UMC, United Microelectronics."
- **Tim Culpan** [0:16:59]: "it then spun out of UMC in the late 90s was a company called MediaTek"
- **Tim Culpan** [0:17:37]: "There was a company called Chartered that was also looking quite strong for a while until they just couldn't keep up with TSMC."
- **Tim Culpan** [0:18:41]: "the cost base for the TSMC model is it's mostly depreciation. You spend buckets of money, $10 or $20 billion a year at the current rate"
- **Tim Culpan** [0:08:54]: "And Foxconn, which was getting going at around about the same time, started off making basic plastic TV knobs."
- **Tim Culpan** [0:24:39]: "The two people who've held the chairmanship of TSMC are both American citizens."
- **Tim Culpan** [0:26:12]: "As of right now, you know, January 2021, like 95 % or probably 98 % of their production capacity is in Taiwan."
- **Tim Culpan** [0:28:21]: "the US basically said almost directly to TSMC without naming, like, you shall not make chips for Huawei anymore."
- **Tim Culpan** [0:33:16]: "there's very very few companies that can a afford to buy this euv equipment and b have the know-how including the staff and the processes to run it"
- **Tim Culpan** [0:35:05]: "And Samsung is a very, very strong competitor. And they were neck and neck for a while."
- **Tim Culpan** [0:38:12]: "And Taiwan actually has a power problem. Taiwan, for political reasons, is starting to shut down its nuclear power stations."
- **Tim Culpan** [0:45:24]: "In fact, there's more silicon made in Arizona where Intel has a factory than is made in California."
- **Tim Culpan** [0:44:51]: "SMIC, Semiconductor Manufacturing International, which is kind of, I guess, the most famous of the flag bearers in China, is really still struggling to get close to a company like TSMC."
- **Joe Weisenthal** [0:50:22]: "And I think it's now a 600 or 700 billion dollar company almost, which puts it in league with some of the real, you know, huge tech winners."
- **Joe Weisenthal** [0:03:20]: "car manufacturers around the world are actually running into supply constraints because they can't get chips because the chip manufacturers decided to reduce their production of automotive chips during the crisis in the spring"

## Transcript

**[0:00:00] Tim Culpan:** Finance is increasingly being built on digital rails, but as the adoption of virtual assets, tokenization and stable coins grows, can regulation not only keep pace, but help accelerate it? Find out more later in the podcast.

**[0:00:14] Tracy Alloway:** When you're a business owner interested in expanding to new markets, there's a lot to consider and countless questions to answer before making the call. The most important one being, are you ready? That's where ChachiPT comes in. Give it a goal, add your files, and ask chat to help you map out different scenarios for the future of your business. So when growth is on the table, you've got an expansion plan ready to implement. Get started at ChachiPT.com.

**[0:00:51] Joe Weisenthal:** hello and welcome to another episode of the odd lots podcast i'm joe weisenthal and i'm tracy alloway so obviously uh tracy we've been uh talking a lot about chips lately but for all the episodes we've done we haven't hit like what is sort of a um i guess i would say the elephant in the room or the gorilla in the room that keeps coming up over and over again?

**[0:01:17] Tracy Alloway:** Yeah, we've been going at it, I would say, from a U.S. perspective, very focused on the troubles at Intel, but we haven't really talked about the success story that is Taiwan Semiconductor Manufacturing Company, better known as TSMC. Yeah, exactly right. So every episode that we

**[0:01:37] Joe Weisenthal:** do, it sort of comes back to them. What a juggernaut they've become. And of course, you know, the sort of basic story is that manufacturing chips is extremely hard. It's extremely expensive and it's very difficult to do at scale. Intel is one of the rare companies that designs and manufactures its own chips, but that for a lot of these companies that are sort of exploding, right gaining market share they're um having taiwan semi manufacture them and taiwan semi is getting extremely good or is extremely good at uh manufacturing and maybe uh pulling away from intel to some extent in terms of its manufacturing capabilities right and i think

**[0:02:23] Tracy Alloway:** they actually invented the foundry model which you know this idea of just manufacturing chips which keeps coming up over and over and over again in all of our conversations as you know one reason the entire semiconductor industry has changed and one reason why Intel is struggling. But the thing I find kind of amazing, like in 2021, we all take it for granted that TSMC is this massive player in the semiconductor industry in the world, really. But I find it really, really noteworthy and, you know, somewhat surprising in retrospect that what is a single company on an island has emerged to really dominate chips that are now vital to all sorts of things.

**[0:03:07] Tracy Alloway:** So computers, smartphones, cars, everything. I think TSMC manufactures a little over half of the world's chips. So it's not quite a monopoly, but again, like they are the juggernaut

**[0:03:20] Joe Weisenthal:** in the room as you put it. Yeah, absolutely. And it's such a, I mean, it's such a pivotal company for a lot of reasons. I mean, you mentioned cars. There was just a story this week, we're recording this January 12th, by the way, there was just a story this week about how car manufacturers around the world are actually running into supply constraints because they can't get chips because the chip manufacturers decided to reduce their production of automotive chips during the crisis in the spring on the expectation that demand wasn't going to be there, and now they haven't ramped up.

**[0:03:52] Joe Weisenthal:** Taiwan Semi, they're actually reporting earnings this week expected to be very strong. Earnings are soaring. The stock is soaring. And then, of course, because it's in this strategic position and because Taiwan itself is in this strategic position, obviously, between the U.S. and China and the rollover, its status, it's just an incredibly central player and sort of necessary to understand the story.

**[0:04:18] Tracy Alloway:** Right. A big player in tech, a big player across multiple supply chains in a bunch of different industries and certainly a player in geopolitics as well. So we really got to talk more Taiwan Semi.

**[0:04:30] Joe Weisenthal:** And we have the perfect guest to do that. Someone who knows all about the company, who's been covering them for a long time, since long before they were the dominant player they are today. We're going to be speaking to Tim Culpan. He is a tech columnist at Bloomberg Opinion. He's been with us at Bloomberg for 15 years. And he's actually been covering the tech industry from Taipei specifically for the last 21 years to news everything about Taiwan Semiconductor, where it came from, how it got to be this important player.

**[0:05:04] Joe Weisenthal:** So, Tim, thank you very much for joining us. Hey, thanks for your time, guys. Good to talk to you. Tim, how would you describe the importance of Taiwan Semiconductor in the world right now?

**[0:05:17] Tim Culpan:** well as you assessed just saying they are so huge tracy pointed out they they create uh you know such a large share of the world's uh chips there is probably not a single device in the world that doesn't have tsmc in there somewhere well whether it's a car the iphone of course famously uh but even some really unimportant boring things like a temperature sensor here or some LED light array there. They do so much of the world's products.

**[0:05:47] Tim Culpan:** They're basically everywhere. And so if they were to stop production tomorrow, I think the global technology industry

**[0:05:54] Tracy Alloway:** would grind to a halt very, very quickly. I mentioned this in the intro, but we take it as a given that this one country has such a big position in global manufacturing, in the global supply chain. How did this actually come about? Or let me rephrase that. Let's start at the beginning. How did TSMC actually start?

**[0:06:19] Tim Culpan:** Well, the story of TSMC is in many ways the story of Taiwan's electronics industry. It was founded by Maurice Jiang himself back in 1987, but the seeds of Taiwan's Semiconductor industry predated that by about 14 or 15 years when an organization was set up by the Taiwan government back then in 1973 called ITRI, the Industrial Technology Research Institute. And at the time, the government realized that technology, and not specifically semiconductors, because semiconductors were relatively new, but integrated circuits were going to be big.

**[0:06:59] Tim Culpan:** The other big area for Taiwan at the time was plastics. And Formosa Plastics is one of the world's behemoths in the plastics industry. And of course, you may remember back when you were a kid, you'd have these cheap plastic toys that were made in Taiwan. Barbie dolls, for example, were made in Taiwan for many, many years. So they were the two kind of twin engines of Taiwan's future. And the government knew that. And so they started getting into it through a government institute, spending money on it, and then working with the Americans and to a lesser extent, Japanese and Europeans to transfer that technology to Taiwan to be made in Taiwan.

**[0:07:37] Tim Culpan:** So the first ships made out of Taiwan really were rolling off the presses or the fabs in the mid-70s. And TSMC came along in the mid to late 80s. So it was already underway. But as Joe pointed out, it was the foundry of foundries. It was the first one to decide that offering a pure manufacturing solution so that clients don't need to have factories and TSMC would do it for them without designing their own chips, only purely manufacture for somebody else.

**[0:08:11] Tim Culpan:** That was very, very new. It was a ridiculous concept at the time because, you know, at the time, Jerry Sanders of AMD later said in that famous line that real men have fabs. At the time, you designed and you manufactured. You did them both. Cypress and Faraday and, sorry, FreeScale and all those other companies, of course, Intel being one of them, they did both. AMD did both. And it was only after TSMC came along and said, hey, actually, you don't need to.

**[0:08:40] Tim Culpan:** We can do it for you. That changed a lot of people's minds and it really turned Silicon Valley around. So where did that idea come from? I mean,

**[0:08:49] Joe Weisenthal:** what was the sort of realization that this model could be powerful?

**[0:08:54] Tim Culpan:** Making to order has always been part of Taiwan's industrial DNA going back many, many years. You probably also know that Foxconn, which makes your iPhone, is also essentially an OEM. They manufacture for others. And Foxconn, which was getting going at around about the same time, started off making basic plastic TV knobs. Terry Guo, who founded that, was really a plastics guy. He wasn't really an electronics guy. He was a plastics guy. And so that was pretty much Taiwan's DNA.

**[0:09:24] Tim Culpan:** don't do the design work don't do the branding don't do any of that just hey america hey japan hey europe bring us your stuff your designs we'll make it for you cheaper quicker and all of that and and so semiconductors at the time were the next thing in that area and you know it was it was also in a similar period of time that uh what is now known as asa started doing ibm pc clones and, well, actually manufacturing for IBM.

**[0:09:52] Tim Culpan:** So it's the same kind of period. Acer was a little bit later, but it's the same general generation of Taiwan. But really the founding idea came from Morris Jung himself, who went to Taiwan at the ripe old age of 54 to take up a retirement job. It's not well known, but in fact, Morris Jung is not Taiwanese. He was born in China and left China at the height of the Civil War in 1948. He was in Hong Kong for a brief stint and then found himself in the US and went on to study at MIT.

**[0:10:23] Tim Culpan:** And he spent a quarter of a century at TI, Texas Instruments. And it was only after leaving Texas Instruments that the Taiwan government invited him to Taiwan to lead ITRI. He wasn't the founding chairman of ITRI, but he was, I think, the third president of ITRI. And he came in and was brought to Taiwan, not as a Taiwanese, but as actually an American citizen. in it. And Morris Jung got the idea of doing a foundry model.

**[0:10:52] Tim Culpan:** And that was really the start of

**[0:10:53] Tracy Alloway:** the whole process. So how difficult was it to get the foundry model off the ground at the very beginning? Because as you point out, that wasn't how people had been making semiconductors. Historically, they'd been doing the designing and manufacturing in-house. How much of an adaptation or change was it for the industry? And how difficult was it to get clients on board, to actually go to them and say, hey, we're going to make your chips for you?

**[0:11:22] Tim Culpan:** It was very difficult. And Morris Jung recognized that. He'd worked in the American semiconductor industry for 20 or 30 years. He knew that they were his people. And so when he started TSMC, one thing that he knew early, early on is that although he had Taiwan government backing and there was very deep pockets. He knew that he needed to get industry on board. So he went hunting for an industry player that would be a cornerstone investor.

**[0:11:50] Tim Culpan:** He wanted to get someone on board that would take a 20, 30, 40 % early stake in the company so that he could really make sure that they were vested in the company. It was a vested interest to have TSMC succeed. So he really wouldn't buy it. And eventually he stumbled across, well, didn't stumble across, but he came across an investor that was willing, that really did see the vision and was willing to take a punt.

**[0:12:17] Tim Culpan:** And that was Philips, the Dutch company, which was for a while in the semiconductor industry. And so Philips was actually a founding investor and had, I think it was about 30 % stake. And they had that 30 % stake for, I think, almost 20 years until they started selling it down when they did their own kind of revamp in the early 2000s. So that really, really helped. It showed that you had someone in the industry on board and you had a certain amount of gravitas.

**[0:12:46] Tim Culpan:** And I think that was the key thing for Morris Jung. And the other thing was that he was American. He grew up through the ranks of Texas instruments. He was a general instruments for a brief time before he went to Taiwan. So the industry in America knew who he was. They knew he was the real deal. There wasn't some unknown coming out from a far off land pitching a crazy idea. And so I think those two things together helped him get traction with the early clients who were willing to take a punt.

**[0:13:13] Tim Culpan:** And they did so and they were rewarded with it. What's next for virtual assets? I'm Jacob Greaves, host of Economy Scale, a series exploring growth in key global sectors. And in this episode, we're diving into the digital forces shaping finance. Sean McHugh from Dubai-based Virtual Assets Regulatory Authority has this to say. We're seeing a lot of institutional players come in, digital asset dedicated hedge funds, high frequency traders, and that sort of thing, and folks who service institutional clients.

**[0:13:58] Tim Culpan:** As more capital enters the market, adaptive regulation is likely to become increasingly key. Rhys Merrick of Ripple says, A lot of regulators globally would like to keep the industry at arm's length. That's the opposite here. I think also what you'll find is that regulation and frameworks can come out, but some of them are maybe a slight iteration on some 80-year-old policy and not actually in line with what the technology can do and offer. And again, that's very different in Dubai.

**[0:14:25] Tim Culpan:** Listen to the full episode wherever you get your podcasts. This series is brought to you by Bloomberg Media Studios and the Dubai Department of Economy and Tourism.

**[0:14:34] Joe Weisenthal:** So I want to just, you know, go back to that founding a little bit more. I mean, you mentioned that prior to TSMC, that the Taiwan government for already several years had been sort of like researching and building up this capability. Were there other players who could have been TSMC? Were there other entities trying to vie to be sort of like a domestic national champion? And also, you know, like I've read a little bit about Asian development models, like particularly in Korea, and this idea of export subsidies for companies that were finding traction with sales outside.

**[0:15:13] Joe Weisenthal:** And I'm curious, like how much that effort to attract that foreign investor, Phillips, in that case, was also about proving themselves domestically that TSMC could be the international player that the country should get behind.

**[0:15:29] Tim Culpan:** Yeah, I think that's a valid point. I think that they needed to sell it to the masses, but Taiwan was not democratic at the time. It was under martial law, so it's not like the government of the Kuomintang at the time really needed to worry about elections or anything like that. But they did very much have concerns about the Communist Party on the other side of the strait. That tension continues to this day. And so there was definitely a feeling that Taiwan had to be a big, strong place.

**[0:16:00] Tim Culpan:** It was called Republic of China, and it needed to be strong and powerful and make money so that it could fend off the communist insurgency. And so these industrial policies were in a kind of war mentality, not unlike what Korea went through. You know, they went through their own civil war and had a kind of war mentality after the Korean War ended and had the tensions between North and South. And of course, Japan was rebuilding after its own war with America and much of Asia.

**[0:16:32] Tim Culpan:** So there was a war mentality for a lot of these countries. And that really, I think, directed industrial policy. And without having to worry about democracy, to put it bluntly, they could go in and basically make decisions, make decisions for the long term, and really lay down what they were going to do for the next 10 or 20 years. And so I think that, yeah, getting the early investor on board in Philips was very, very important.

**[0:16:59] Tim Culpan:** But to your earlier point, TSMC wasn't the earliest one, in fact. Born out of ITRI, ITRI was started in the 70s, but born out of ITRI in around 1980 was another company called UMC, United Microelectronics. And that company was spun out of ITRI. And that was in the semiconductor industry, but it was also designing its own chips. And in fact, it then spun out of UMC in the late 90s was a company called MediaTek, which some people might have heard of because they're actually quite a competitor to Intel and Qualcomm and others today.

**[0:17:37] Tim Culpan:** They did have other competitors. It wasn't a given that TSMC would win. They had a very strong competitor in UMC. And at the time, there was also competitors overseas in Singapore. There was a company called Chartered that was also looking quite strong for a while until they just couldn't keep up with TSMC. So TSMC had competition in the early days in foundries, but they were the ones who really stuck to their guns and managed to win out.

**[0:18:05] Tracy Alloway:** So just on that winning point, there's one other thing from what I understand that Chung actually did that helped TSMC get itself into first place in the long run. And that was that they sort of sacrificed profits early on in order to grow their market share very quickly, which is something that we've seen in other big tech companies. I mean, notably Amazon and others like that. How was that decision made and how important was it to TSMC's long term success?

**[0:18:41] Tim Culpan:** Well, the key really for TSMC cost base, it really comes down to what is the cost. And so, of course, if you're trying to win over customers, but you're offering a price that is not acceptable or not competitive to them doing it themselves, then you're not going to win. And the cost base for the TSMC model is it's mostly depreciation. You spend buckets of money, $10 or $20 billion a year at the current rate, and that depreciates pretty quickly, to be honest, in only a few years.

**[0:19:12] Tim Culpan:** So your depreciation is very, very high. And once you've built that and bought all the equipment, if it sits idle, then you're not making money, right? So it's immediate cost off the bat. The other one is R &D. And that's where they really haven't skimped. They've been willing to spend buckets of money on R &D, which is engineers. At the end of the day, it's the best engineers money can buy. And so those two things, they basically had to take it.

**[0:19:35] Tim Culpan:** They really didn't have a choice. They had to spend money to build the capacity and get the technology going. And they needed to spend money on the staff, the engineers, to make the best they could of this equipment. And then on the flip side, when they're going to their customers and saying, well, we want to charge you this much, well, you have to be price competitive. So in a way, it was a deliberate tactic, but it was also an issue of the circumstances of the time.

**[0:20:04] Joe Weisenthal:** So when we think about contract manufacturing in any industry, and of course, the basic model, as you identified, was not new. It's not new to chips. There have always been setups where one company designed something, another country does the building. I think for a long time, the business of, and you can correct me if I'm wrong, but the business of sort of being a fab, I mean, I don't think it was seen as particularly sexy. It may have been a profitable business model, an important one.

**[0:20:36] Joe Weisenthal:** But it seems like in the discussions we're having lately, like something flipped, like it went from being like, OK, here's the here's the company that takes in your designs and then spits them back out to here's a company that's like a true tech leader, like they really have something in their manufacturing know how that is increasingly difficult to be replicated. was there some inflection point, some flip where like, when did that happen? When did that sort of

**[0:21:07] Tim Culpan:** change of perception start to occur? I would argue that it happened during the period of the crossover from PCs to smartphones, because as we know, PCs were Intel's domain and AMD's domain. AMD turned up to the party and they made their own stuff. AMD doesn't anymore, but for the longest time, they did make their own stuff. They designed it and made it themselves. They call them IDM, right? And then we went into the smartphone business. There was a short period of time of, you know, the 2G phones, the, you know, the Nokias and so forth.

**[0:21:39] Tim Culpan:** But then we moved into smartphones. And those were powered basically by companies like Qualcomm and MediaTek and actually TI, if you might remember, were in that business for a little while. And these were companies that didn't have their own factories. Well, TI did, but Qualcomm certainly didn't. And MediaTek didn't. And a lot of companies didn't have their own factories. Because they didn't need to have their own factories, a company like Qualcomm could start up. Anybody could get together a few friends fresh out of grad school and try and start a chip design company.

**[0:22:13] Tim Culpan:** And Qualcomm was able to get away with doing that and was very successful. So you have a company like Qualcomm becoming very, very successful by saying, you know, we design the best chips for mobile phones and we don't need to manufacture them. And I think the rest of the industry cottoned on. And as we're moving away from the PC era into the smartphone era and then electronic devices and we've got PlayStations and handhelds like the Switch and all sorts of other devices, a lot of these chips were no longer the big hunky Intel processors that were smaller and lighter.

**[0:22:51] Tim Culpan:** And then, of course, Apple came along and decided to do their own chips as well. And so I think that's really where we saw the flip from having to have your own factory to being able to do it without. And at the same period of time, the economics just wasn't working out for companies like AMD anymore. And a lot of companies who are making their own were being burdened and saddled by these massive factories and the depreciation costs they had to bear.

**[0:23:15] Tim Culpan:** No matter whether they sold one chip or a million chips, the depreciation was going to be the same. And so if you have a bad year or two or there is a shift away from your business, such as PCs to something else, and you don't have the chips to keep up, then you're going to be in debt very, very quickly. And so we saw a lot of the companies that were designing and manufacturing for themselves just couldn't keep up economically or financially.

**[0:23:38] Tim Culpan:** And some went fabulous like AMD and some just disappeared.

**[0:23:43] Tracy Alloway:** Can we talk a little bit about the role of the U.S. and China in all of this? So we mentioned that Taiwan itself is sort of perennially caught between these two forces, the US and China, which doesn't actually recognize it as an independent country. What influence has that relationship had on TSMC and how does TSMC feed back into that geopolitical relationship?

**[0:24:12] Tim Culpan:** TSMC's view of the geopolitical relationship is quite smart and diplomatic in almost the same way that they view their customers, where they want to be the neutral supplier to everybody. They want to be everybody's foundry. They want to be everybody's partner. And they take that view to the broader, the global geopolitical situation by saying, well, you know, we're not in the American camp. We're not in the China camp. We're just in the camp of whichever clients need us.

**[0:24:39] Tim Culpan:** And that's the line that they've been taking for the longest time. Now, what's interesting is I think people misunderstand TSMC because Taiwan and TSMC is so close to China. And of course, many devices are actually assembled in China. Most electronic devices are assembled in China. People see it as being, oh, it must be kind of close to China. But in fact, the history and the genesis of TSMC is far more American than it is Chinese. The two people who've held the chairmanship of TSMC are both American citizens.

**[0:25:14] Tim Culpan:** Morris Zhang, the first one, was not even Taiwanese. And the current one, Mark Liu, is if you look at the annual report of TSMC, it says that he's American citizen. I suspect he's a dual national. I suspect he's got Taiwan citizenship as well. But in actual fact, they're both Americans. The style and the culture and the processes of the company are very, very American. The corporate governance approach is very, very American. It's very, very different to the way most Taiwanese or Asian companies are run.

**[0:25:42] Tim Culpan:** And so that culture, of course, is very, very important. And the clients for the longest time have been American. Yes, the devices get assembled in China, but all those big name semiconductor clients, the ones that pay the bills, they're all American. And so the relationship is much, much closer to America than it is to China, even though China, of course, is a large buyer of chips now and, of course, is the place where so many semiconductors are then installed inside devices.

**[0:26:12] Tim Culpan:** The relationship is much, much closer to America than China. Of course, it may change in the future. And both sides want to kind of drag TSMC to their side. But you know what? As of right now, you know, January 2021, like 95 % or probably 98 % of their production capacity is in Taiwan. They've got pretty small production facilities or capacity in China. And of course, they're going to do something in Arizona, which would be at a similarly small scale.

**[0:26:42] Tim Culpan:** But Taiwan still is

**[0:26:44] Tracy Alloway:** the heart and soul of TSMC. So just on that note, on the idea of TSMC being quite smart in maintaining a neutral stance or trying to maintain a neutral stance between the US and China, How long do you think they can keep doing that, given that semiconductors seem to be becoming rapidly politicized and tech in general is becoming politicized? So we just saw, well, over the past couple of years, we've seen the U.S. announce all these restrictions on semiconductor technology in the past week or so.

**[0:27:20] Tracy Alloway:** Again, we're recording this on January 12th. We've seen the U.S. announce fresh sanctions on Chinese military firms. And meanwhile, China is saying that it is going to allow people to sue international companies if they comply with U.S. sanctions. it feels like remaining neutral is becoming more and more difficult in the current environment. Of course, everything could change when the new U.S. administration comes in. But in the meantime, it does feel like companies are going to be forced to choose sides.

**[0:27:53] Tim Culpan:** Yeah, Tracy, I think that's the key issue. And I think that's one of the most important things that we need to consider in this whole situation. In fact, I've been arguing for more than five years that there's this Berlin Wall of tech brewing between America and China or American and Chinese fears of influence. Even before Donald Trump came to power, that was already going to happen. Now, Trump clearly accelerated a lot of those moves, but it was already underway.

**[0:28:21] Tim Culpan:** TSMC wants to sit on that fence. They've been very good at sitting on that fence. But I've argued in columns over the last couple of years that they can't do it forever. They are going to have to choose sides. Now, I've got a lot of pushback from people at TSMC and others in the industry saying, no, no, no, we're neutral, always be neutral. We can manage to be neutral forever. And I just don't agree. I think that we saw that with the Huawei situation over the last couple of years, when at one point the US basically said almost directly to TSMC without naming, like, you shall not make chips for Huawei anymore.

**[0:28:52] Tim Culpan:** And they had to stop, right? And ground to a halt. Now, that may open up again under a new administration in D.C., but that was really should have been a wake up call to management at TSMC that you can't sit on the fence any longer. And I think that they are they're realizing now they've got they've hired some new lobbyists, a former Intel lobbyist amongst others in D.C. to help their cause. I think they now realize they can't sit on that fence forever, but they're going to try to hold on as long as they can before they have to choose sides.

**[0:29:25] Tim Culpan:** And when they do, they're going to choose the American side. That's where their bread is buttered.

**[0:29:30] Joe Weisenthal:** It really seems kind of stressful, though, for the world. Like if literally you can't come up with a piece of electronics that in some way hasn't been touched by TSMC and the government can just say don't work for that company anymore and there's no alternative to TSMC anywhere near its scale. So it seems like a pretty big sort of choke point for everything.

**[0:29:54] Tim Culpan:** Yeah. I mean, if you were to go back to the year 2000 when everyone used to using Windows PCs, can you imagine a government saying you can't use Windows anymore? And Android, sorry, not Android, but Linux wasn't really coming along yet. Can you imagine the problems and the trouble that would happen? That'd be the kind of equivalent today. If you were to be told you can't use TSMC and Huawei struggled. Oh, my God, the problems they faced in not being able to use TSMC, they're putting in rush orders, rush orders to the last minute so that TSMC would churn stuff out and get it out the door before the barriers came down.

**[0:30:27] Tim Culpan:** And that's really caused a lot of problems for Huawei and they don't really have a backup source. As your wealth grows and your needs change, finding the right financial advisor can make all the difference. Edward Jones offers comprehensive wealth management to help give you a greater sense of confidence and control while you focus on living a life you love.

**[0:31:01] Tracy Alloway:** Your Edward Jones advisor takes the time to truly get to know you and provide personalized strategies to help keep your goals on track. Whether you're passionate about charitable giving, considering business succession, or legacy planning with the next generation in mind. With more certified financial planner professionals than nearly any other firm, Edward Jones Financial Plans are custom built around what matters most to you because the key to being rich is knowing what counts. Edward Jones, member SIPC.

**[0:31:31] Tracy Alloway:** It's 9 a.m. Monday morning. Time to report out on last week. You're clicking around the weekly dashboard, looking at the same charts you've looked at hundreds of times, but still can't figure out what happened. Was it one region? A particular customer segment? Did conversion drop? Was there a change in the marketing funnel? Instead of waiting on the data team or spending the next two hours digging through spreadsheets, you just ask. With Hex, teams across sales, product, marketing, and operations can ask questions about their data in plain English, explore what's driving the numbers, and quickly get to the answer.

**[0:32:07] Tracy Alloway:** And because Hex works with your data team, not around them, the analysis is built on trusted data and governed metrics your company can actually rely on. So the next time someone asks, what happened last week? You don't have to say, let me get back to you. Just ask Hex, the AI analytics platform for exploring your data together. Join over 2000 leading organizations at hex.ai. So, you know, one of the themes,

**[0:32:35] Joe Weisenthal:** One of the big things we're talking about is this sort of like tech advantage that they're building up such scale, such manufacturing know-how, ability to do it at excellent quality, at improving price. Intel, as we know, stumbling with their own manufacturers. In fact, there are reports that Bloomberg ourselves have reported about perhaps Intel perhaps outsourcing some of its manufacturing to TSMC. Talk to us about the tech edge that the company is building and how difficult it is for even one of the largest competitors, let alone some of the smaller competitors, to match them in terms of price and quality.

**[0:33:16] Tim Culpan:** Semiconductors are really tough. They've always been tough, but they're getting tougher and tougher because we're now at this, you've heard about Moore's Law and the idea that we're getting to the end of Moore's Law or beyond Moore's Law. but we're getting some really really tough ends of semiconductors are essentially a mix of chemical engineering with physics and electronic electrical engineering and they're getting to the point now with this new euv technology that there's very very few companies that can a afford to buy this euv equipment and b have the know-how including the staff and the processes to run it so even if someone else was managing to get their hands on one or $200 million piece of equipment that

**[0:34:00] Tim Culpan:** can do the latest stuff, there's not many people who could run it. And so it becomes this self-fulfilling prophecy that TSMC is ahead of everybody else. And because they're ahead of everyone else, they get all the best orders and all the money, and they've got really nice fat margins. They've got one of the largest operating margins in the tech sector. And nobody else can compete because they don't have the money to keep up, which means they don't have the money to buy all the equipment, which means they don't have the money to hire the best engineers and so on and so forth.

**[0:34:31] Tim Culpan:** And we really are at that point where TSMC, a lot of people see that they're unassailable. I mean, of course, they can mess up and they probably will mess up. But we're now at this point where money talks. And it's not just money, it's the know-how, it's the store of knowledge that they've built up over 30 years that no one else can get near, not even Intel. And when Intel stumbles, you know that things are getting really difficult in the chip industry.

**[0:34:56] Tracy Alloway:** So what would a mess up actually look like for TSMC? Or what would it take for them to lose this top spot that they currently have in the industry?

**[0:35:05] Tim Culpan:** Well, I think there's one thing that could be losing the top spot, I guess, to Samsung. He's probably the only other company that might take over from them. And Samsung is a very, very strong competitor. And they were neck and neck for a while. And I think Samsung will come back again. They've got a lot of money. they've got a lot of their own orders in the NAND and DRAM business. So that really helps them push their own semiconductor manufacturing.

**[0:35:29] Tim Culpan:** But in terms of TSMC, you know, having a bit of a nosedive or any fall, I think there's actually three areas that they could face problems. The first one, I think, is that they could be a victim of their own success in that they are always at the cutting edge with the most incredibly good foundry technology to the point where their technology is so good that nobody really needs it. Like if you think of the CPU wars between Intel and AMD of 20 years ago, it was the coolest thing to be the one with the fastest clock speed.

**[0:36:08] Tim Culpan:** But today, nobody cares about clock speed, right? Nobody really pays attention. Maybe they care about how many cores are in a CPU. But all that kind of chest thumping is gone in the CPU industry. And I think we'll see a point in the future where maybe it won't be the leading companies like Apple or Qualcomm or Huawei who say, we don't need the latest technology, but maybe one of their top 10 customers, maybe the 10th largest customer says, you know what, we're going to sit out this round of new technology.

**[0:36:39] Tim Culpan:** The previous round is good enough for us. The N minus two, which is two generations old, that's good enough for now. We'll stick with that. But TSMC, of course, has spent all this money to upgrade. And then, of course, another client might drop off and say, you know what, it's getting a bit expensive. What we've got now is good enough. We'll stick with that. And so I think we could see TSMC being too good and being a victim of their own success in that regard.

**[0:37:07] Tim Culpan:** The other area where I think that they could really kind of come down is if there is a slowdown in the tech industry that is so vast that it could really cut into their margins and their costs and so forth. Because even though they're the best in the world, people just don't have the money to buy things. Now, of course, we all thought this pandemic might be yet a year ago when it was starting to unfold, we thought that the tech industry was going to come to its knees.

**[0:37:36] Tim Culpan:** It didn't pan out that way. And then the other thing actually that I think TSMC has to worry about is power, like energy, electricity. The new technologies, the new EUV machines guzzle a lot of energy. In fact, there's three science parks in Taiwan, the north, center and south, South Science Park, which is where the Apple chips are made, actually had to rush in. The power company had to rush in new transformers to feed enough electricity into the science park, into TSMC's factories, because they are such power hogs.

**[0:38:12] Tim Culpan:** And Taiwan actually has a power problem. Taiwan, for political reasons, is starting to shut down its nuclear power stations. Nobody wants coal. And the renewable energy ideas that are coming on are not coming on quickly, enough. And so Taiwan, in fact, just a week ago, UMC arrival faced a power outage and Taiwan has faced power outages. And this is a pending crisis. This is a looming crisis for the entirety of Taiwan. And if they don't get their act together, this could really hurt TSMC big time.

**[0:38:45] Tim Culpan:** So there are a few landmines out there that TSMC could step on and not all of them will be up to itself to

**[0:38:53] Joe Weisenthal:** avoid. It'll need help from the government. Another question that I have is, so we've been talking about how some of these incumbent tech giants are, they're going further up the tech stack themselves, some of these huge cloud players designing their own servers, designing their own silicon. We had an episode where we talked about Apple. Could some of these companies, these tech giants that have tremendous amounts of money ever have a competitive reason to get into

**[0:39:25] Tim Culpan:** manufacturing chips themselves? I don't think so. I don't see why a Google or a Facebook or a Microsoft or anyone like that would want to get into the business. It would take probably in the order of a willingness on the part of a CFO or a CEO to burn maybe 20 or $30 billion. like i'm not joking because tsmc is going to be spending 10 to 20 next year it spent 10 to 20 last year that's just on the equipment and then you've got to get the engineers and sad to say the best semiconductor manufacturing engineers not the design engineers the best semiconductor manufacturing engineers are really all in asia now distributed amongst the koreans the japanese and the

**[0:40:11] Tim Culpan:** taiwanese and more and more in china now there you do still have intel and and we shouldn't count out Intel now. Intel is still a very, very solid company. They're definitely struggling right now, but they have very, very good engineers. So if you're an American company thinking, I want to get into the chip game, I don't think any board of directors would sign off on that, to be honest. I think you'd be burning a lot of money for not a lot of upside.

**[0:40:33] Tim Culpan:** Why do they need

**[0:40:34] Joe Weisenthal:** it? They've got TSMC to do that for them. So here's something that I'm still struggling with a little bit, and we've talked about on other episodes, is this idea of learning by doing the link between semiconductor design and semiconductor manufacturing, some of these red hot chip companies like, say, an NVIDIA, are there ever going to be risks to their business model that they aren't more intimately involved with the manufacturing aspect of the job and are sort of strictly design?

**[0:41:07] Joe Weisenthal:** Or is that sustainable? No, I think that is a risk. I don't think it's a

**[0:41:13] Tim Culpan:** a risk for the near term. I think it's a risk for maybe 10 or 20 years out. I mean, quantum computing is kind of around the corner. I don't think we're going to see quantum computers next year, but that's the thinking is in the next generation, so say 20 years, quantum computers will be out there. And quantum computing is just, it turns on its head the semiconductor industry, because instead of having ones and zeros, you've got qubits and some kind of mid-state between off and on that is different to the base of how semiconductors work now, which is very binary.

**[0:41:51] Tim Culpan:** It's literally a transistor is on or a transistor is off. And you write all of your code and you design semiconductor circuits around these basic logic gates, this idea of on and off. And quantum computing is very, very different. And so if that comes through, it may be in the best interest of someone like an NVIDIA to say, you know what, we want to dabble in making it ourselves because we think we know the best way to do it.

**[0:42:18] Tim Culpan:** And maybe someone like NVIDIA would say, you know, we're going to start a little test fab just, you know, here in Silicon Valley and play around with it and see how we do. And maybe they'll discover that it works well for them and they get a lot orders and they decide that this is something they can scale up. But they would have to be designing and kind of discovering the technology themselves to do so. I don't think you can set up shop tomorrow and get that technology off the shelf.

**[0:42:44] Tim Culpan:** So much of what any semiconductor maker is doing now is very, very proprietary to the point where nobody else could copy it, even if they were to take over the factory. Like if, for example, China invaded and got access to TSMC factories and TSMC engineers disappeared, I don't think they would know what to do with it. So you'd have to have that fundamental knowledge and really be building that knowledge from the ground up. And I think there's a possibility that someone like an NVIDIA, maybe AMD, or maybe a chip company we haven't come across yet decides that actually we want to have a closer relationship with manufacturing.

**[0:43:25] Tim Culpan:** And in fact, I think the company that is very, very closely connected to manufacturing is probably Apple. As Apple goes down the Apple Silicon path, I think we'll see them get a bit more involved in the manufacturing process.

**[0:43:39] Tracy Alloway:** What scope is there for government policy to play a role when it comes to encouraging domestic industries that are able to compete with TSMC? So for instance, in China, we know that the government there has already announced its intentions to build up its own chip manufacturing capacity. We've talked a lot in this podcast about the idea that, you know, more than half of the world's semiconductors are coming from this one company that's on a rather vulnerable island caught between two very, very big geopolitical forces.

**[0:44:13] Tracy Alloway:** If countries recognize chips as strategically important, will they be trying to encourage their own domestic production

**[0:44:22] Tim Culpan:** and how successful do you think they might be in that? Yeah, I think that's very, very true. I think both Beijing and Washington have recognized semiconductors as being of national security and geopolitical interest to them. China has recognized that for 20 years, but you know what, they're still, to be blunt with you, they're not really doing that well. People say, oh, China's catching up. China's catching up. And that's true. But it's not like TSMC and everyone else is standing still.

**[0:44:51] Tim Culpan:** They're still moving forward in leaps and bounds. And SMIC, Semiconductor Manufacturing International, which is kind of, I guess, the most famous of the flag bearers in China, is really still struggling to get close to a company like TSMC. And they really can't replace TSMC, for example, if Huawei needs to outsource to somebody else. So I think China is a really good example of how difficult it is that even with endless amounts of money and very, very smart, China has really good engineers, right?

**[0:45:24] Tim Culpan:** But they're just not as good as the Taiwanese or even the Americans or the Japanese or Koreans, right? And then the other side is, of course, the United States. They would like to get their chip glory back. But one of the big problems is if you look at Silicon Valley, like literally that stretch of land we call Silicon Valley, almost no silicon is made there anymore, right? In fact, there's more silicon made in Arizona where Intel has a factory than is made in California.

**[0:45:51] Tim Culpan:** That's data I checked out. Yeah, it's amazing. Arizona is the new Silicon Valley. So that's part of the problem that any government will face. Obviously, Donald Trump wanted to get things going using state and commerce and various other departments to push that forward. It remains to be seen whether Biden will take that forward. But America does have a chance to bring back and revitalize its chip industry. And I think the way they will do it will be semi-national interest protectionist policies, such as any new equipment that goes into Department of Defense, serve a farm, must have chips made in America, that kind of thing.

**[0:46:31] Tim Culpan:** You put up national security policies and that basically means if you want to sell to these companies, you have to make it on our terms, which probably means on American soil. So that's what they're doing with TSMC. They're trying to get TSMC to set up and TSMC will set up in Arizona. And I think if you were to take that to its logical conclusion, I can imagine where there would be some kind of technology transfer where companies outside of the United States, Taiwanese or Japanese or Korean companies will be incentivized and pushed to transfer some of the technology or do joint ventures with American companies to help transfer that technology back, which would be a beautiful irony because the Asian

**[0:47:15] Tim Culpan:** countries got their technology from America 50 years ago. So So it would really be a nice kind of closing of the loop if America then started to have the technology transfer come back the other way.

**[0:47:28] Joe Weisenthal:** Tim, that was that was fantastic. It's it answered a bunch of questions that we have is sort of the perfect way to move the story forward for us. So really appreciate you joining us. No problem. It's fun to talk to you guys.

**[0:47:41] Tracy Alloway:** Thanks, Tim. Crash course in TSMC.

**[0:47:43] Joe Weisenthal:** Yeah, that was fantastic. I really enjoyed that. You know, I think that last question that you asked and his answer really helped put some pieces together for me about, say, like the challenge that China has in sort of replicating a world class chip system. And it really sort of the whole story of TSMC and really all of this and how much of this knowledge is just embedded in the engineers and the processes that have been going on forever.

**[0:48:29] Joe Weisenthal:** This idea like you cannot replicate it just with money. You could throw an infinite, literally infinite amount of money at something or virtually infinite, but there's no way to sort of recreate that know-how overnight. I think it's just a really, really important thing.

**[0:48:45] Tracy Alloway:** Yeah, I think one of the things that's standing out in all of these episodes is just how deep the moat is around semiconductor expertise. The more I learn about it, the more difficult it seems to me to catch up to these players, and in particular TSMC, which, as we just discussed, has this big leading position in the industry. The other thing I loved was that last bit of irony from Tim, this idea that the U.S.

**[0:49:15] Tracy Alloway:** might end up doing tech transfers to import semiconductor expertise after having exported it many, many years ago. I love how we come full circle in that way. But I guess just going back to the other point about the idea that there's this deep moat around foundries and semiconductor manufacturing. I mean, 20 or 30 years ago, a lot of people wouldn't have expected Taiwan to have built up an industry like this. So I guess things can change over the long term.

**[0:49:50] Joe Weisenthal:** Yeah, things can change over the long term. And Tim sort of identified some things that over time could perhaps erode TSMC's moat or maybe some different models whereby, as he mentioned, maybe companies don't feel they need to pay for always the state of the art as tech changes. But I really do think that drove home this idea. It's like the sort of the whole thing, just the compounding advantage that TSMC is building up. And all you have to do is look at the company's stock price.

**[0:50:22] Joe Weisenthal:** And I think it's now a 600 or 700 billion dollar company almost, which puts it in league with some of the real, you know, huge tech winners. You can see this is increasingly this moat, this advantage is increasingly reflected in the investor perceptions of the company.

**[0:50:41] Tracy Alloway:** Absolutely. All right. So now my question is, how many more semiconductor episodes are we going to be doing?

**[0:50:48] Joe Weisenthal:** I think we have one more. I think we maybe that we could maybe pause. I think we're going to be talking to one of our old favorites, Dan Wang, pretty soon to really sort of figure out, OK, basically a year post-COVID, how is the state of China's tech industry and where they really stand with all of this? So but I think that'll be the last one for now.

**[0:51:08] Tracy Alloway:** For now is the key here, because I mean, I'm joking a little bit because you wanted to turn all thoughts into a semiconductor podcast. But I do think another one of the standouts from all these conversations is just how important this technology is going to be in the future. And this idea that it really is becoming a strategic asset. I've seen some people compare it to oil in the sense that it underpins all these different things in the global economy.

**[0:51:38] Tracy Alloway:** So I do think this is going to come up again and again and again.

**[0:51:43] Joe Weisenthal:** Yeah, no, totally. And I hadn't really even thought about that point about the U.S. putting pressure on TSMC not to sell chips to Huawei. That's super interesting in its own right. Like here is this company that's in Taiwan that China itself would, you know, Taiwan from China's view is part of China. And yet still from the business perspective and sort of the way things are, that it's the U.S. that can pressure TSMC not to build for Huawei than the other way around.

**[0:52:16] Tracy Alloway:** Right. So we've spent all this time talking about resiliency of supply chains and worrying about what happens if there's a pandemic in the country where you manufacture something that's vital for your final product, or if sanctions come in that suddenly cut off a supplier or something like that. And here we have an area where the impact could be really, really strong, but also it seems likely that because of the strategic importance of semiconductors, they will actually be targeted by countries that are trying to gain a tech advantage.

**[0:52:54] Joe Weisenthal:** Absolutely. It's super, super interesting dynamics that I hadn't really thought of at all before prior to this conversation.

**[0:53:02] Tracy Alloway:** All right. So we're all semiconductor enthusiasts and nerds now. Shall we leave it there?

**[0:53:08] Joe Weisenthal:** Yeah, let's leave it there.

**[0:53:08] Tracy Alloway:** All right. This has been another episode of the Odd Lots podcast. I'm Tracy Alloway. You can follow me on Twitter at Tracy Alloway.

**[0:53:16] Joe Weisenthal:** And I'm Joe Weisenthal. You can follow me on Twitter at TheStalward. Follow our guest, Tim Culpan. He's on Twitter at T. Colpin. And check out all of his columns at Bloomberg Opinion. Follow our producer, Laura Carlson. She's at Laura M. Carlson. Follow the Bloomberg head of podcast, Francesca Levy, at Francesca Today. And check out all of our podcasts under the handle at podcasts. Thanks for listening. The world of business is constantly evolving and Comcast Business keeps you totally in step with secure AI-backed networking in more than 100 countries.

**[0:54:24] Joe Weisenthal:** They're powering over 90 % of the Fortune 500 and millions of small businesses. That's a lot of muscle. And behind it all, thousands of experts answering your call at 2 a.m. like it's 2 p.m. One partner powering how business gets done for companies around the globe. When you add it all up, no one does business like

**[0:54:42] Tracy Alloway:** Comcast Business. Most meeting note takers get the easy words right and the important ones wrong. The client's name, the acronym. Whisperflow NoteTaker is built to get the hard words right. It takes notes on any meeting platform, Zoom, Google Meet, Teams, even WhatsApp calls, keeps every meeting searchable, and connects to Claude or Chat GPT so you can ask, what did we agree to last Tuesday? Free to try on Mac and Windows at whisperflow.ai. That's W-I -S-P -R flow dot A-I.
