The Peter Zeihan Podcast Series - The Bleeding Edge of Semiconductors: A Tale of Three Companies || Peter Zeihan

Episode Date: February 12, 2024

The semiconductor industry is one of ever-growing importance and its leaders—Intel and TSMC—are now fighting to be the first to bring the next generation of advanced chips to market. Full Newsletter: https://mailchi.mp/zeihan/the-bleeding-edge-of-semiconductors

Good morning from a frigid Colorado. It's a balmy zero degrees this morning. And today I'm going to tell you a tale of three companies and the state of the semiconductor industry from a technological and production point of view. Now, if you go back to the world before 2017, the technology of the day was something called Deep Ultraviolet, which was basically a way of producing microchips. And Intel, the American technological giant, was the world leader by pretty much any measure. And they had gotten a little cocky and they had gotten a little bit lazy. So they would design chips two, three, four models out, but it would only produce the next one up. Because they were so far ahead of everybody else, they didn't feel the need to jump steps.

So they would use DUV and they would make a chip that was marginally better than the one before. And then at the end of the year, no one was head caught up. So they'd do it again and again and again. And they did this for like 15 years. I mean, they were very good at what they do, but they could have pushed the technological envelope a lot more, had they chosen to. In part, that was because of the nature of DUV technology.

The problem with it is you have to kind of make micro adjustments and physically adjust the equipment for each type of chip, and you have to do that manually and physically. And so with every design, you had to do it all over, and with every machine in a fabrication facility, you would have to do it independently. So no new chips from different machines are going to be quite exactly alike.

and it generated a relatively higher loss rate from the final semiconductors than what we have today, and so generated a little bit more waste. But, again, they were the industry leader. No one was close. Well, they were always had their eye on the future, however. And so they invested in new technologies that would take them beyond DUV, one of which is EUV, extreme ultraviolet. And the company that developed that technology is ASML out of the Netherlands.

And back in 2016, ASML thought the... stuff was ready to go. So they're providing demonstrations for Intel, showing them how this technology is better. You can not only get more nodes on a chip and get to smaller and smaller nanometers, but it's all digital. So you kind of type in what you want to the machine over the course of a few days to a few weeks, and then the machine doesn't actually have to be physically manipulated in the way that DUV did. Now, what that would mean is you'd have a higher success rate and more efficiency, but back in 2016, Intel was like, I don't think this technology is quite right,

and we're the industry leader, we're going to give it a few more years. Well, ASML, not very happy with that, marketed the technology to everybody else, and a company decided to take the plunge. That company is TSMC out of Taiwan. And when we get to 2017, TSMC suddenly hits the ball out of the park and proves that EUV is ready for mass application, and over the next couple of years very rapidly overtakes Intel

because they have a shorter turnaround time for their chips and they can make chips with smaller nanometer sections. It isn't until 2022 or 2023 that Intel finally makes its first extreme ultraviolet chip. So TSMC in Taiwan has been the industry leader now for several years. Now, we've had a kind of a reverse in the roles now. ASML, the Dutch, have another new technology called high numerical operature, whose physics I'm not even to pretend to understand, and they have marketed again, and this time, Intel is the

one that's behind, and they're kind of desperate and kind of hungry, and TSM is the one that's resting on their laurels. So the first delivery of those new machines, the high N-A chips, went to Intel in the second week of January of this year, and Intel expects two things. Number one, they plan to overtake TSM using the EUV technology in 2024, hoping to get down to two nanometers. Right now, the industry lead is it about three nanometers, and that's a TSM product. And then next year, they hope to leak frog even further, provided that these new high N.A. machines work, which we'll find out pretty soon. Anyway, that's where we are right now. In terms of the overall geopolitics, it's pretty straightforward. Right now, 90% of all I-Hen chips are made by one company, TSMC, in one city in Taiwan.

It's a high concentration. But if Intel working with ASML can pull this off, all of a sudden we will have facilities in the United States that are working on the higher-end stuff with some of the first facilities that are going online outside of Phoenix and Columbus, Ohio. So stay tuned because the geography of these chips is about to evolve pretty significantly. if high N.A. works. And if not, we're still stuck with Taiwan, it could be worse.