The Peter Zeihan Podcast Series - Intel Keeps Playing Catch-Up with TSMC || Peter Zeihan

Episode Date: January 25, 2025

We've discussed what TSMC is up to in a recent video, so let's look at what another big name in the semiconductor space -Intel- is doing to keep up.Join the Patreon here: https://www.patreon.com/PeterZeihanFull Newsletter: https://mailchi.mp/zeihan/intel-keeps-playing-catch-up-with-tsmc

Hey everybody. Peter Zion here. Today we're going to talk about the semiconductor sector, specifically the American company Intel. Now, Intel has had a rough few years. 20 years ago, 20 years ago, 25 years ago, they used to be the industry leader. And they were so far ahead that they would release designs that were nowhere near the most sophisticated because they knew it would take forever to the market to catch up. So they had years of work planned in front of them. Unfortunately, they rested on their laurels, and they failed to invest in a technology called Extreme. ultraviolet that instead a Taiwanese firm TSM picked up on, which allowed for much faster fabrication

and much more accuracy and much less waste. And in a few years, TSM over took Intel become the world leader. Intel did get on the extreme ultraviolet bandwagon eventually, but it took them while to master the technology and they've been behind ever since. Now, that said, the semiconductor industry is really weird and that we really do only have that one world leader TSM that makes almost all of the high-end chips. Intel is trying to catch up from behind, and Samsung out of Korea has picked up some fabrication facilities from a merger and is doing their best to play, but they're a distant, distant, distant, distant, distant, distant, distant, distant second. So I thought it would be worth understanding where this technology is going to evolve and where corporations are going to involve. American politicians like to focus on the fabrication facilities, the places where the semiconductors

are actually grown, etched, doped, and built, but that's really not the hard part of the process. I mean, no offense to the fabricators. They do amazing work and in a difficult condition in a technologically challenging field. But the harder work is in design. A basic design for high-end chip can take upwards of 20-forming months to really get good. going, and that's assuming you're not really incorporated any fundamentally new technology into it. And most of this design work is done in the United States, with a lesser degree in Japan, but that makes it sound like it's just some guy with a protractor basically scribbling on paper,

and that's not what it is. You're bringing together literally tens of thousands of elements into a design to try to do something new, process faster, manage heat better, use less electricity, use different materials, and so on. And all the new technologies have to, to be incorporated into this theoretical construct, which is then taken to Taiwan, where they work with TSM in order to basically make an instruction booklet that the TSM staff will follow. And then you have to worry about all of the inputs coming from around the world, because it's not like you just take some silicon and you're off to the races. No, no, no, no, no. There's copper. There's palladium. There's all kinds of different inputs, things like transistors that have to be

very, very specially designed and produced. TSM doesn't do any of that. The logistics and the design companies in the United States do. So even today, with TSM producing 90% of the world's high-end chips, most of the real work, most of the value-added work, most of the high-paying jobs, are actually done in the United States. And operating a fab facility while it's still highly skilled work, it's not nearly as highly as skilled as what happens on the other side of the Pacific.

The problem, and what brings me back to Intel, is what happens on the other side of the equation. Once you have all of your raw semiconductors, you then break them into their individual components and test them and package them. And then you have to put them into an intermediate product like a motherboard or a memory drive or a chip within a sensor system. And only then can you go into the proper manufacturing process where it is put into a car or a plane or a satellite or whatever else. So this one step fabrication, obviously unavoidable, obviously important, but it's not really where the money is. Now, TSMC is a little obsessed with its security because it is a Taiwanese company and you can understand why. And so the concern a lot of people have in the sector and more broadly is that if something happens to Taiwan, we lose all the IAN semiconductors. And that is true.

But if something happens to South Africa, we lose a lot of the rarer. materials that will go into it. If something happens to North Carolina, we lose the ability to purify the silicon that goes into it. God forbid, something happens to San Jose. We lose the ability to do a lot of the software work on the back end. There are thousands of single point failures throughout the system. What makes Intel unique, from my point of view, is that they have a number of these other steps under their umbrella. There are still literally thousands of single point failures throughout the Intel system, but probably about a third to half less than what TSM has.

So in a world that is on the verge of rapid declobalization, the idea that we're going to be able to make these high-end chips at all is kind of a stretch to me because there's just too many places where a single break means the whole thing falls apart. But Intel has three advantages. Number one, more of the steps are under the umbrella. Number two, its fab facilities are in the United States. And number three, if we're going to have to rebuild this environment anyway, it'd be easier to do it if you only have to repay a 3,000 steps instead of 4,500.

So one way or another, regardless of the corporate success or failure of Intel in the months and years ahead, the fact that more of this stuff is concentrated in Intel and in the United States suggests that some version of Intel is actually going to be a bigger part of the semiconductor future globally than TSM over the long run. Of course, we're all dead in the long room, so this is all about timing.