1. Podcasts
  2. The Vergecast
  3. The impossible process of making a computer chip
The Vergecast

17 Nov 2026 · 46m · 17 chapters

Episode artwork

The impossible process of making a computer chip

In short

How advanced computer chips are made, why progress like Moore’s Law has been “unbelievable” rather than inevitable, and why reshoring chip manufacturing to the U.S. is far harder than building a single factory.

Guest backgrounds

Gideon Lewis-Krauss is a writer (notably for The New Yorker) who reported an “epic worldwide journey” to understand chip production. Host David Pierce interviews him.

Key claims

Chipmaking depends on a global, interlocking supply chain and specialized know-how; “chips aren’t made anywhere” in the way cars are. Moore’s Law is treated like a self-fulfilling industry roadmap. ASML’s extreme ultraviolet lithography is the critical bottleneck; China’s DUV progress is still far from cost-effective throughput. TSMC’s value includes decades of community knowledge and strict IP separation.

Notable examples

ASML’s EUV machines; Zeiss mirrors with atom-by-atom fabrication; TSMC clean rooms (air showers, “thousand times cleaner” than surgical theaters); a “printing an entire city on a chip” metaphor; digital pregnancy tests outperforming Apollo guidance computers.

Written by AI. May contain mistakes.

Chapters

2:55

Understanding Moore's Law and its Critiques

A discussion on Moore's Law and the perspectives of industry experts.

“Gideon Lewis-Cross, welcome to The Verge Cast.”
4:00

The Illusion of Inevitable Progress in Chip Manufacturing

Exploration of the challenges in chip manufacturing and the role of industry players.

“all that like, you know, they really see it as a kind of natural byproduct of their struggles against the laws of physics.”
5:15

ASML and TSMC: Key Players in Chip Production

In-depth look at the roles of ASML and TSMC in the semiconductor industry.

“standard thing that everybody says about Moore's law, which is, you know, it's not a law, like, it's not a physical law of the universe, like gravitation, that it's just, it's merely an observed regularity.”
7:40

The Critical Process of Lithography

Explains lithography's importance in chip making and the challenges involved.

“You can basically go back like 10 years and every single year there's a major news organization that has a story about ASML that has some title like the most important company you've never heard of.”
9:30

Metaphors for Understanding Chip Size and Complexity

Discussion on metaphors that illustrate the scale and intricacy of chip designs.

“But of all of these, you know, special elements of all these special children of the chip making process, the most special is lithography.”
11:55

The Evolution of Transistor Sizes and Their Impact

A look at the historical context of transistor sizes and their implications.

“is the previous generation where ASML also dominates these machines.”
14:00

The Impossibility of Scaling Down Chip Sizes

Exploring the challenges of understanding nanometer-scale chips and their implications.

“in the real world in a city the size of Tokyo would be like the size of a golf ball.”
21:12

Inside TSMC: The Clean Room Experience

A detailed account of the clean room process and the unique environment at TSMC.

“If you've been kicking around a business idea but have yet to make it happen, Black Friday is a great deadline to give yourself.”
24:38

Access and Secrecy in Chip Manufacturing

Discovering the balance of transparency and secrecy in chip manufacturers like TSMC.

“So what did you run into between these companies?”
28:00

TSMC's Role in Chip Manufacturing

Learn about TSMC's value proposition and the complexity of IP protection.

“So you didn't really have anywhere to go.”
Show all 17 chapters
29:10

Challenges of U.S. Chip Manufacturing

Explore the difficulties in transferring chip manufacturing to the U.S.

“So the ecosystem piece of this, I think really comes into play as you get to the question about how is this all going to work in the United States, right?”
32:20

The Generational Nature of Chip Production

Understand the long-term commitment required for U.S. chip production.

“And because like there is something special about this activity and the way it kind of like binds communities together in like pursuit of this impossible goal.”
34:36

Collaboration in the Chip Industry

Discover how companies collaborate in the complex chip supply chain.

“And actually the only responsible thing we can do is try to build more of all of this everywhere as fast as we possibly can.”
36:50

Understanding the Complexity of Chip Manufacturing

Delve into the intricacies of how chips are made and the misconceptions around it.

“I mean, like ASML has their make their light source in San Diego.”
39:25

The Wonder of Chip Production

Reflect on the miraculous nature of global chip production.

“Like, there's nobody who really understands the entirety of how one ASML machine works, let alone like the entire thing.”
42:00

Chips: A Global Cooperative Effort

Explore the complexities and global nature of chip manufacturing.

“And the idea that you can just stand up a factory and start building chips is just dead wrong.”
45:12

Documentary Recommendation: Funny Money

A deep dive into the documentary about Dane Cook's rise in comedy.

“All right, before we go, I have one more thing to tell you about, which is a documentary that's on Netflix.”
A man walking through an airport, asking his phone a question

Ask about this episode

Get the gist, explore an idea or ask about something said.

Claude

On The Vergecast, what are the key points of this episode?

Searching your listening

Transcript

Automatic transcript. May contain errors.

0:00Nilay Patel: Support for the show comes from Satelli. If you've ever thought there has to be a better way to manage my pet's health and safety, then this message is for you. The Satelli Collar Go is an AI-powered GPS dog tracker that helps you keep tabs on your dog, especially in locations where they might run away. And Satelli, since AI, can help you spot challenges in your pet's health. While others just track locations, Satelli decodes lives. Go to satellite.com to get 35 % off with promo code VERGECAST. That's S-A-T-E-L-L-A-I.com to get 35 % off with promo code VERGECAST. satellite.com, promo code VERGECAST.

0:43Nilay Patel: This episode is brought to you by On Investing, an original podcast from Charles Schwab. Each week, hosts Liz Ann Saunders, Schwab's chief investment strategist, and Colin Martin, head of fixed income research and strategy for the Schwab Center for Financial Research, bring you fresh insights on what's happening in the markets and why, and what the implications might be for your portfolio. Join Colin and Liz Ann as they explore questions like, how do you evaluate corporate bonds that look interesting? And what sectors are on the move right now? Download the latest episode and subscribe at schwab.com slash oninvesting or wherever you get your podcasts.

1:48Nilay Patel: customer accounts across 48 states in Washington, D.C. That includes 80 % of Fortune 100 companies covering healthcare, manufacturing, and data. Explore how Constellation is advancing reliable energy for America's future at constellationenergy.com.

2:06Hello and welcome to The Verge Cast, the flagship podcast of the difference between a fab and a foundry. I'm your friend David Pierce, and today on the show, we're talking about chips. Specifically, the incredible economy and pipeline and just unending series of research innovations that have led to the chip industry as we know it now. Gideon Lewis-Krauss, who's a writer for The New Yorker and elsewhere, recently went on this sort of epic worldwide journey into some of the most important companies in chips to figure out how they get made, why they get made the way that they do. and whether if you're a country like the United States, you can just throw a bunch of money at the problem and suddenly become a powerful chip manufacturer in your own right.

2:45Turns out it is much harder than you think for maybe reasons that aren't very obvious. I asked Gideon to come on the show and walk us through some of his journey. So let's get into it. Gideon Lewis-Cross, welcome to The Verge Cast. Thank you so much, David. It's a pleasure to be here. I'm a frequent listener. I have been a fan of your writing for a very long time. I've read you. You're one of those people who has written some cool thing about technology for every great publication on Earth. So I'm very happy to have a reason to finally have you here, which is to talk about chips and semiconductors.

3:14And I want to talk about a lot of things, but one of the things that most stuck out to me in your reporting, and I think is a useful place to start, is with Moore's Law, which is this long-held idea in the technology industry that computing power roughly doubles roughly every two years. This is a thing that has sort of held true for, what, 50-plus years at this point. longer. You talked, yeah, you talked to a lot of people who hate the existence of Moore's law and are extremely annoyed by it. Can you tell me why? Well, there was one very funny comment that a longtime physicist at ISML told me in Veldhoven where he was like, Moore's law, I do not give a shit about it.

3:56Moore's law is for people who never worked on it. It's just marketing. And what he meant was, you know, that it's not something that they think about at all that like, you know, they really see it as a kind of natural byproduct of their struggles against the laws of physics. And, you know, they like they hate the prognostication about it. They hate the kind of like constant, you know, terminal diagnosis of Moore's law on the installment plan. Like they just it seems like people in the industry just really don't think about it. I mean, they like to look back and say like, hey, we keep kind of pulling off his magic trick, but it's not something that they talk about.

4:29I picked out the Moore's law thing because it feels like a sort of perfect microcosm of everything that you discovered, which is we treat semiconductors and chips and all this stuff is this sort of inevitable march of progress that like, of course, everything keeps getting better and everything gets smaller and everything gets more efficient. And like, of course, that's what happens. It's what happens. And if you went on this journey and uncovered that actually it's kind of unbelievable and almost impossible that that keeps happening. And it is this like insane, almost incredible, almost unlikely confluence of things that has actually been what has made all of this progress happen.

5:05And it's a bunch of like weirdos who barely understand why any of this works, who are the ones who have made this progress happen, that there's nothing remotely inevitable about it at all. Well, I mean, even kind of beyond that, you know, I had heard the sort of standard thing that everybody says about Moore's law, which is, you know, it's not a law, like, it's not a physical law of the universe, like gravitation, that it's just, it's merely an observed regularity. And it's an observed regularity that seems to have continued somehow for 60 years. But that's not even accurate. It's like beyond that, like, I think the only way to describe Moore's law is as this like collective delusion in the that has become like a self fulfilling prophecy that a bunch of people get together, like from the industry in a room, and they make their roadmap for the next 15 years.

5:52And each guy says, Well, you know, I can build this thing that doesn't exist yet and might never exist as long as you at your company, my supplier, can build a thing that doesn't exist yet and your supplier can build a thing that doesn't exist yet. So everyone, it's a pre-commitment mechanism where everyone is like, okay, we are going to do this thing that maybe can't be done, but we're all, the reason we're going to do it. I mean, this was one of the things that really struck me. It's not that, oh, we're going to do this because we need better AI accelerators. Like, it doesn't seem like they really think about the end customer that much at all.

6:27I mean, clearly like TSMC does, but a lot of the other people in the chain don't. For them, it's just a commitment they're making to like their friends that it's like, oh, you know, like, you know, somebody in place will think, well, if we can't pull this mirror off, our buddies at ASML are going to be really disappointed in us. So everybody has to do it because everybody has agreed to do. Right. Yeah, it does seem like there is something kind of purely sporting about a lot of this research that is just like, what if I can just do this thing that no one else has ever been able to do? Won't that be neat?

6:56And it is there's something really sort of perfectly scientific about it that I really appreciate. But for this story, you you picked out and got, I would say, essentially unprecedented access to two companies, ASML and TSMC. And I think you kind of tell the whole story of chips through those two companies. Why those two companies? Well, I mean, TSMC is sort of obvious because they make, you know, 90 % of the world's most advanced chips. So and I think people had like kind of a vague idea that like, oh, most of these chips are made in Taiwan and like they're made by this one company. You don't really know that much about that company.

7:29And this company's like, you know, sits at like one of the great kind of choke points of the international system. So TSMC was a little obvious. And then I had always just been really curious about ASML because if you go back and you look at the headlines about ASML, I think I have one headline in my piece, but I originally had a string of 10 headlines. You can basically go back like 10 years and every single year there's a major news organization that has a story about ASML that has some title like the most important company you've never heard of. Yes. But it's been going on for 10 years. And in fact, they're like the second most popular item in the company store is a hoodie that says like an obscure little company you never heard of quoted like by the BBC in 2021 or something.

8:14And so there was there was this aspect of like this is, you know, like that, like as as all these headlines have said, kind of arguably the most important company in the world that most people have never heard. Yeah. So and I think to that end, it's probably still true that most people have never heard of it. And I think you're right that TSMC, we'll come back to TSMC, but I think most people at least watching or listening to this roughly know TSMC is the one that makes all of the chips. It is the best company at making chips in the world. I think that's a fairly safe thing to say at this point.

8:43ASML, I think most people don't know. And I think it is still true that none of this works without ASML in particular. So just real quick, give me a sense of what ASML does that is so irreplaceable in the chip making process. So there are thousands and thousands of steps that go from turning silicon into a finished wafer. And almost all of these steps along the way are controlled by essentially like monopolies or duopolies because they're so specialized that we've just kind of assembled. Like we decided to concentrate all of our expertise in a handful of companies because it doesn't kind of doesn't make sense to have a bunch of companies competing to do these incredibly niche things.

9:28And none of these steps are easy or replaceable or kind of available off the shelf in some kind of like commoditized form. But of all of these, you know, special elements of all these special children of the chip making process, the most special is lithography. And lithography is the process by which chips are basically printed more or less like photographs, that you have the pattern of a chip's design, and then you have light shown through this pattern, which then allows you, I mean, I'm radically simplifying here, and everybody in the industry is going to write you mean emails about all the, like, my coarse-grained description.

10:06but light is shown through the pattern and projected onto the chip and that allows you to draw the pattern of the architecture onto the chip and lithography was the place where it seemed like we were just gonna like that moore's law was just gonna run out of steam so about like 30 35 years ago it became clear that all of these elements which were getting increasingly precise in order to do increasingly miniaturized chip patterns because they used increasingly shorter wavelengths of light, kind of like, you know, wielding light like a sharper and sharper scalpel in order to draw these really intricate chip designs as transistor density increased.

10:43And at a certain point, we were just going to run out of light that there were there was not going to be any more usable light at a shorter wavelength. And essentially, the whole chip industry got together in the early 90s under the leadership of Intel and an American consortium with the help of the the American government to try to figure out any possible way around this light issue. And what eventually emerged from that, I mean, it took 15 years or 25 years to get it to work, really, was what's called extreme ultraviolet lithography, which is the ASML is the only company in the world that makes these machines.

11:17And it is the most irreplaceable part of this entire process because you cannot do the chip without the lithography and you cannot do advanced chips without asmls specific uv machine and still at this point that is the only company that has figured out how to do this at the at the scale and efficiency that is required to do it yeah because it's not even a matter of like can you print a chip it's a matter of can you handle a throughput of 300 wafers an hour or something because you know like otherwise it's just not cost effective for these companies unless you can hit a certain measure of productivity.

11:54So even in China, China recently announced advances in deep ultraviolet technology, which is the previous generation where ASML also dominates these machines. And so it seems like China has made progress in being able to do DUV, but the reports are that China can do these prototypes that print, I don't know, 30 or 50 wafers an hour. And that's still a very far cry from like having an actual cost productive system or cost effective system rather. Yeah. So I don't want to dwell too long on the tininess of the chips. Yes. But I do think you probably in the course of writing and reporting this piece had to come up with the right metaphor to describe exactly how small and how remarkable this process is.

12:40what is what is your favorite metaphor for like you had one line in the story for how they're built that's like it's like building an entire city skyline story by story which i really liked as a way of thinking about the printing but just in in terms of trying to think about the scale do you have a favorite metaphor for how remarkably tiny and remarkably intricate these things actually are that you've come up with i really like the idea of a cityscape because a cityscape really is actually pretty accurate where you have a kind of like base layer of the transistor is a kind of like the footprint of all of the buildings in the city.

13:11And then you have all of these layers of infrastructure on top of that, of the interconnections and the wiring. And so I started to think like, it really is, I'm certainly not the first person who's used this metaphor, but that like, occasionally you would see something like, it's like printing an entire city on a chip. And then I went down, me and my buddy, Claude, went down like a really like deep rabbit hole about like, exactly how this would translate if you were trying to like, you know, work across these 20 orders of magnitude or whatever and i mean it turns out actually like the math wasn't that hard but that basically you like if you want it's like etching an entire perfect replica of some vast metropolis like tokyo onto an area the size of a fingernail with the detail that you would capture would be like the smallest possible feature on a chip in the real world in a city the size of Tokyo would be like the size of a golf ball.

14:10Wow. Good Lord. Okay. That's that. So yeah, that's, that's pretty, that's pretty astonishing actually. Well, because you know, so much of this chip world stuff, like, you know, I certainly don't have to tell you is like working with the, like describing these units that are just like completely meaningless units of abstraction to most people. When you talk about the two nanometer node or whatever. People have no way to even think about how small that is. And at least in the past, 60 years ago, when Gordon Moore was first projecting Moore's Law, you could do it relative to human hair. You could say a transistor was a third the size of human hair.

14:53But now that you're at the point where these transistors are rendered and like literally atomic dimensions, you can't say it's like, oh, one 50 ,000th of human hair. Like that doesn't mean anything to anyone. So it's very hard to try to make this stuff material for people. Totally. And especially at this moment, like I think the metaphors are so interesting because we are at this point now where the gap from impossibly small to the point where you can't even fathom that as a person to slightly smaller than that is like a step change in how technology works. We are so far beyond our ability to actually understand the changes being made here because they are literally beyond our ability to perceive them, which just it just blows my mind at every at every step of this process.

15:37This detail didn't make it into the story, but when I was in Veldhoven at ASML's headquarters, I was talking to this one researcher who told me, he was describing, it was more of kind of a comment about his organizational and managerial style. but and he was talking about kind of pitting elements of his team against each other to see like who could further refine the process and he was like i gave this you know this entire big team or collection of teams under me a budget of four atoms and they had to sort out among themselves how they were going to divide up a budget of four apps wow yeah good lord that is that is wow so So speaking of you, you said you went to Verhoeven, you also went and saw TSMC in Taiwan.

16:24Tell me about these tours. What is it like to actually be inside these places? You know, it's one of these funny things where like on some level, it is just a factory, right? I mean, like some of these sections went on much, much longer in my earlier drafts. And one of the comments my editor made was like, it's, you know, it's a little bit boring. Just it feels like it's just like a tour of all these machines. And I was like, point taken. But that's also what this piece is. Like this piece is a tour of machines. You wear a funny suit and you go into a clean room. Like that's the, yeah. Well, the clean room stuff is really extreme.

16:55You know, I was, I was warned in advance. I couldn't have any metal on my body. I mean, that was actually more of a security issue. But then like they, they just really encase you head to toe in the clean room. They gave me this like really special clean room paper. I think I have it some here. Like for taking notes on, you mean? Yeah, for taking notes. Here I have a visual. this was the clean room notebook they gave me because it doesn't shed shed particles and it like you can't really tell but like it has like this really weird starchy crinkly feel to it it looks like a high school test booklet but it says clean note on the front that's wild yeah and you know then you go through these like they check your body's static electricity and then you go through these air showers and then you go inside it's very very quiet um you know the other factories I've been in are definitely not quiet.

17:42My father-in-law has a factory and it's extremely noisy. And you can feel, it's a little bit like walking on an air hockey table because they have these floor tiles with vents in them because they're replacing the entire atmosphere in this, in the clean room, you know, every couple of hours because they're, so they're constantly recirculating the air. And then the light they have, for some technical reason, I forget, they have um yellow lights so there's this kind of like sulfurous glow and so between like the whisper of the vents and the sulfurous glow it has this feeling of like you know like one didn't in the blitz except it's also immaculately clean it's literally we went in fact i made this comment in the piece about how it's a thousand times cleaner than a hospital surgical theater and the fact checker you know fact checkers don't like absolute numbers they always like to give you a little wiggle room so he was like should we say like generally about a thousand times cleaner and i was like no no this is that this is actually the metric that in terms of like parts per million of impurities in the air it's a thousand times cleaner than a hospital theater like this isn't just like generally like general ballpark order of magnitude and then you know they they've done so much automation at tsmc in order like you know both because any error is so costly that you know They told me when people used to push wafers around in these hand carts, there would be signs that said, like, if you nudge this cart, your pay is going to be docked and, you know, X dollars because it's so easy to crack a wafer.

19:13And so, but then also, one of the reasons they do a lot of automation, actually, it doesn't have anything to do with labor saving. In fact, TSMC has grown a lot. It has to do with the reduction in variance because any variance is going to kill you in this process. And so they have to eliminate as much variance as possible. And so inside the factories, there are, I forget what the numbers were, like thousands and thousands of miles of rails above your head, like 20 feet above your head. And they're branching trackworks. And they have these what look like arcade game claws that are like racing around.

19:50These claws are holding the wafers and these special carriers called hoops below them. And then so the claws will race around. They will descend. They will deliver the wafer into one end of the machine. They'll go to the other end of the machine. They'll pick the wafer back up. And, you know, it's one of these things where there were just so many wild numbers and details that you couldn't possibly include them because at a certain point, it was going to beggar credibility. that it's like the distance traveled by these wafers inside a fab in one day is like new york to taiwan and back or something like that i mean just like crazy distances yeah and then these these fabs go on forever i mean they are um um you know i think that they told me that this one fab i visited has enough clean room space to host 22 simultaneous full-scale soccer games good lord And one of these, I turned to...

20:46I read that number and didn't really process it. And now I'm realizing just how large that actually is. And this is only one of the dozens of fabs that they have. And after we went on this first tour, I turned to one of the TSMC guys. He was wearing an Apple Watch. I obviously didn't have any devices. And I said, like, what did your watch say? We just walked. And he looked down. He was like, we just walked like five miles inside this building. Wow. Wow.

21:11Nilay Patel: Support for the show comes from Shopify. If you've been kicking around a business idea but have yet to make it happen, Black Friday is a great deadline to give yourself. There's lots of extra traffic to take advantage of, and Shopify can help you take advantage of it. Shopify's pre-built website themes and AI's site builder make it easy to launch a beautiful, professional-looking store without ever touching a line of code. Shopify's mobile app lets you run your store from anywhere, use it to track inventory, offer customer support, handle payments, and more. You also get your own AI e-commerce expert, Sidekick, that can help you manage almost any business task with a single prompt, including holiday discounts.

21:51Nilay Patel: Shopify powers millions of businesses worldwide, from household names to small businesses just getting started. Launch before Black Friday, a time of year when more people are already shopping and ready to buy. If you're ready to launch your business before Black Friday, go to shopify.com slash Vergecast to start your free trial. That's shopify.com slash Vergecast. Go to shopify.com slash Vergecast today to hear your first.

22:21Nilay Patel: Support for the show comes from Framer. When a new landing page turns into a pile of tickets and handoffs, Framer can help your team move faster. Teams can collaborate in real time, iterate on the same page, and publish instantly. Agents help you go beyond the vibe-coded first draft to bring you a production-ready site faster than ever. That's because agents and humans work in tandem. Agents bring speed and scale. People bring taste, judgment, and control. Agents and Framer work alongside teams to streamline collaboration on the same canvas, build custom code components, create and manage CMS content, optimize SEO settings, audit for issues, and ship everything all in one place.

23:00Nilay Patel: Bottom line, Framer is the AI website builder that helps creators, teams, and businesses ship production-ready sites faster than ever while getting every detail right. Prompt, inspect, edit, and publish in one place at a whole new pace. Learn how you can get more out of your site from a Framer specialist or get started building for free today at framer.com slash verge for 30 % off our Framer Pro annual plan. That's framer.com slash verge rules and restrictions may apply.

23:34Nilay Patel: Support for the show comes from Superhuman. There's a lot of potential for AI assistants to make your job easier, but not if every ask starts with you debriefing the entirety of the project from scratch. Ideally, it would be following along from the start and ready to jump in as soon as you need help. Superhuman says they've got the solution. Superhuman Go comes from the makers of Grammarly. It can embed in the tools you and your team already use, always ready to help without having you switch apps or break your flow. It can help make sense of a dense message thread, help you get ready for a meeting, or catch a typo before you hit send.

24:09Nilay Patel: And when you get stuck, Go can help you brainstorm, outline, or review relevant contexts so you don't lose your place. It's basically an AI assistant that's always on, always ready to help. Bottom line, Superhuman is built for people who take their work seriously and expect their tools to do the same. You can try Superhuman Go from the makers of Grammarly. Find out more at superhuman.com.

24:38So what did you run into between these companies? Obviously gave you access. They are proud of their work. They're doing really interesting stuff. they're also some of the most secretive enterprises you will ever discover obviously they have customer secrets to protect but they also have trade secrets to protect these companies like again most people do not get in these rooms that you got into what did you encounter in terms of like what they would show you versus what they wouldn't show you i mean it's so funny because i mean i should i should start by saying that the comms groups at both companies were fantastic particularly they at asml um they're the chief of head of us comms and one of the sarah is just an incredible comms person and like it was the kind of person who wanted you there because she just thought what they did was so magical that she wanted people to see but they also were not particularly well calibrated in terms of the kinds of things that i would care about or even the kinds of things that i would notice so the first room we went into at the first fab at tsmc um and you know it's just like machines as far as you can see and everything is is worrying and quiet um and cold and the only sign of any human presence was that they had put up little paper coverings over the brands and the names of the machines and i was like okay why did they do that and one of the tsmc guys was like well you know like technically you could look at the layout of these machines in this production process and like you could kind of determine which customer this is and maybe even which product this is by the layout of the machine and i was like you think that i'm gonna be like oh that's a surf scan 5x xp next to a canon 510 411i and like i'm immediately gonna be like oh these are rear rubens you know like like there was no way i mean they had like radical there probably are you know like dylan patel or something probably could have like walked into that room had done that i certainly couldn't have um so there was you know it there was stuff that they were trying to keep secret that was like there was no reason for them to keep secret because i couldn't have possibly processed it but you know tsmc's whole like their founding shtick from the beginning was we are never going to compete with our customers because like this is why tsmc existed in the first place if you were a young designer but you know if you were like a jensen in the early 90s um if you were a young designer working for intel and like you had some great idea for a new chip design you had two options one was that you could go out and try to raise like the like you know 20 billion dollars to build a fab which was never going to be easy because that you know and the chip industry at that point was so vertically integrated that even if you bought a fab there's all kinds of other stuff you wouldn't have access to you would have to like recreate essentially the entire supply chain within the walls of that fab so it was just like prohibitively expensive to even think about doing that or you could go to your bosses um and say like would you should we make this and like typically you know if like you know if you were designing like a gpu in the early 90s like the bosses would be like what do we need this for like like well you don't need to speed up video games um or your other option was you know there there were some of the big um chip makers would use a little bit of their excess capacity to like you know do contract runs for other people So like you could go to an Intel and say like, would you build this for me?

27:59But then of course, if it was something good, like Intel was just going to steal, like going to help themselves to your idea. So you didn't really have anywhere to go. So TSMC is a big part of their value proposition was this is a way that you can get your designs executed without spending all the money on the fab. And we are never going to compete with you. So we are never going to take your IP and make our own chip and undercut you. And, you know, the only way that they can work both with, you know, like they can have one building where they're serving Apple and NVIDIA and Broadcom is by promising to all their customers that there's like never going to be any, you know, leakiness of IP between those places.

28:36So even their customers, when their customers go to visit, they don't get to see the whole fat. They get to see their own production line doing something because like they're so paranoid that there could be some knowledge transfer there. And I mean, obviously it's a blurry question because obviously on some level, like some knowledge transfer is good because if TSMC figures out some new aspect to like optimize their process engineering on an Apple chip, like that's going to help them across the board. But then there are other aspects of it where like they really, you know, Apple would never want NVIDIA to know anything about like their M7 design or whatever.

29:09Yeah. So the ecosystem piece of this, I think really comes into play as you get to the question about how is this all going to work in the United States, right? There's been this big push toward We need to do more chip manufacturing in the United States. And I feel like your reporting seems to have made you skeptical of the idea that this is possible or plausible to do in the U.S. And in part not because we can't like TSMC couldn't open a plant in Phoenix and start making chips in Phoenix under some amount of time and amount of money, but because that's not nearly the whole story. Can you explain what you mean?

29:46Like, why is this such a hard thing to transfer to a new place or open up in a new place like Phoenix, Arizona? I mean, I'll say two things about that. Like the first part is like, you know, to your point about like much more skeptical that this is possible. But then then I have a broader point about like what I see of the effort. The first point is that chips are not made anywhere. Like you cannot like you, you know, you can like point to a Ford plan and say like, oh, you know, these cars are made in Detroit or whatever. But you cannot really point to a place and say this is where chips are made because, you know, they go, they cross borders.

30:2370 times. It's the thousands of companies along the way. And so many of these companies are by necessity, you know, like the whole thing starts out with mining quartz and to make these crucibles that are used to make the silicon ingots. And it is, there is only one quartz mine in the world where the quartz deposit is pure enough that they can do this. And it's in like the mountains of Western North Carolina. And then, you know, the next step of the process, you know, the like the companies that have to like refine the polysilicate to be, you know, to make into the ingots. It's like three companies in the world.

31:03It's like one in Michigan and one in Germany and one in Japan. And like every step of the way, it's some hyper specialized place, some in some far flung locale doing this and so it doesn't really like yes you can move the one final factory that is kind of like putting it all together but the rest of it is just this global super organism that like exists kind of everywhere at once and like you can't possibly just like localize it and like we're going to bring this home like what do you mean you would be bringing home like an entire industry and like millions of people and like an incredible like that that are very rooted in places like you know part of tsmc's value is like it's this community of engineering practitioners that has been doing this for 40 years and they have feeder schools and everyone knows each other and they all went to the same schools and like every like so there's a lot of tacit knowledge that exists in these communities but for me one of the kind of like updates about so i i had gone along sort of thinking like this just like makes the chips act look like kind of silly especially it's relatively small amounts of money, all things considered.

32:10And it seems sort of futile to be like, well, we're going to build a Samsung fab here and a micro-on fab there and a TSMC fab there. But when I talked to Mike Schmidt, who ran the implementation team for the Biden administration and sort of put this to him baldly, like, you know, TSMC moved to Phoenix and they couldn't even get there or like set up in Phoenix, they couldn't even get their bunny suits wandered nearby they had to like drive them across the desert to find like a suitable clean room laundry and he was like well that's great like that means that like somebody's going to set up a clean room and he was like this wasn't supposed to be an overnight thing that was going to happen on like a standard short-term like electoral cycle like this was like what maris chang did when he founded tsmc 1987 he knew that it was going to be a generational project and he was like we knew that this was going to be a general generational project too like this was about what's going to happen over the next like 30 or 40 years, like in order to build all of this up, build up this expertise, build up the vocational schooling, build up these communities, build up all these suppliers such that we can have a more resilient supply chain.

33:15And because like there is something special about this activity and the way it kind of like binds communities together in like pursuit of this impossible goal. And I was like, wow, first of all, it's so great to hear, you know, a former member of an administration talking and like at that kind of timescale and those kinds of commitments. But it also made it clear that people who were doing this knew what they were doing. It wasn't just like, throw up some walls and fill it with machines. It really, they knew exactly what they were doing. Yeah, and I think I was so torn reading that part of it because on the one hand, you go through this whole process, you learn about this broad economy, and there's something about it, again, just to go back to what we were saying at the beginning, that feels almost impossible and like a sort of miracle of innovation in so many different ways that all of this stuff happened to come together to support these things.

34:04Like these people saying, what am I going to be able to do in 20 years and then just working on it? Which never happens. Like so much of this is sort of profoundly anti-capitalistic that it's almost amazing it happened at all. And looking at that and you say, okay, maybe this is actually something rather than try to build a really efficient market out of, we need to sort of protect because it is this beautiful, like natural organism we've created that is a triumph of globalism and is everything that is going right in the world. But at the very exact same time, you can look at it and say, oh, it is an unbelievable threat to everything that TSMC exists in Taiwan.

34:37And actually the only responsible thing we can do is try to build more of all of this everywhere as fast as we possibly can. Is this just cognitive dissonance we have to live with? Like, is there a way through those two things? I think there is a way through those two things. I mean, it's a great point and you're exactly right. Like you've gone to kind of the crux of the issue. I think that there are ways to pursue those two things in parallel, as long as you have some respect for the equilibrium that exists now. So like one of the things that, you know, sometimes you'll listen to observers of this stuff.

35:14Typically, people like don't know a ton about it, who will say things like, well, you know, especially economists who will say like, look, there are some like real economic inefficiencies here in the way that this market functions. And like clearly what like either ASML has to like double their margins, like to bring it into like a better competitive market equilibrium, or like what you actually need is like some outside vulture who's going to like get in and buy the machines and then resell them at some value because like there's all this value that's not being captured. and like sure from a kind of like econ 101 like i pencil perspective like maybe that's true but it would also just break everything like because like these margins you know and the margins in the chip industry are not totally fair like from one end to the other like apple's taking like much bigger margins than tsmc and tsmc has been taking bigger margins than asml and like all down the line um but that if one company kind of unilaterally did that it would just break the whole thing.

36:15And that there is some aspect of it where it's supposed to be. And like, it's not like there haven't been conflicts over the years, there have been plenty of conflicts between these companies, but that like, they have settled into this idea that like, we are going to kind of like spread the margins out and do this cooperatively, because we understand that like, we're not actually even really talking about like, kind of individual firms in like the Coase model, like interacting at arm's length. And we were talking about this, like very blurry spectrum of companies that are always like running into each other like i talked about this company zeiss this this legendary german german um lens maker and they make these mirrors for the asml machine and these mirrors are like the process is unbelievable it's like 40 alternating layers of um of carbon and molybdenum and each one is put down like like literally like practically atom by atom and they take two years to make and they cost millions of dollars and like the statistics on them are crazy like if you blew them up to the size of the world the highest elevation these are so flat the highest elevation would be a face down playing card same and that some of the other mirrors are like can can reflect light so accurately they could hit one or the other side of a dime on the moon i mean just like it doesn't seem like this stuff should be real and so so asml couldn't make their machine without zeiss making these mirrors and when i and you know they spend it's a very collaborative relationship between them and like when the asml employees like one of them would tell me like when we're down at zeiss which is in southwestern germany like i couldn't tell you at any given time like who worked for asml and who worked for zeiss like we were all just like there in it together i mean it does seem like if you would be committed to some like if you looked at the resiliency issue in the light of all of those existing relationships like you probably could do something about it And, you know, another part of it that it sort of like belies the common understanding is that plenty of this stuff already happens in the U.S.

38:16I mean, like ASML has their make their light source in San Diego. They make their top module in Connecticut. Like it's it's already like there's already a lot of stuff that happens here. But it also just seems like actually maybe we should just try to like solve for a world peace equilibrium. And then like we don't have to worry about stuff as much. I mean, then we'd still have to worry about like earthquakes and typhoons in Taiwan, you know, wouldn't solve those issues. I do sort of like the idea. And I think it's not entirely wrong that solving world peace might be easier than reverse engineering the chip making pipeline.

38:51Exactly. Maybe the prudent problem to solve is just fix the world. OK, last question that I'm going to go. I was telling you before we started recording that as I was reading the story, I kept thinking about the schoolhouse rock, how a bill becomes a law thing. And I feel like you were sort of in a backwards way trying to figure out, like, can I actually understand how a chip gets made? Is it even possible to fully understand this process? Do you feel like you got there at the end? Like, do you think you know how chips are made now? Do you think anyone knows exactly how chips are made now? No one knows exactly how chips are made.

39:22I mean, like, these people don't even really know how, like, the rest of their machine is made. Like, you know, if you went to the people in the physicist in San Diego who are working on the light source for the UV machine and you ask them to explain like the reticle stage, like, you know, they could give you a course description, but they couldn't really tell you how it worked. Like, there's nobody who really understands the entirety of how one ASML machine works, let alone like the entire thing. I mean, I certainly understood it. I mean, I'm not, I'm just a professional dilettante. Like I could never claim to like have any real depth of expertise here, but I felt like I understood it well enough to answer my basic questions, which are like, you know, whenever you hear people talk about these like reshoring issues, like, and they're like, why?

40:11And it's like, well, you can just build a factory and make it here. Like, I don't get what's so hard. And like, by the end of it, I was like, oh no, now I get why it's so hard. Now I get why you can't just do that. And like, that was satisfying to me. And I felt like I had a much better understanding of the sweep of this that's going back, you know, 65 years or a little more and how like how continuous the project has been. And and also just as just something we really take for granted. Like, you know, I one of the things that was shocking about this. So Gordon Moore in the late 70s is reflecting on like a little over a decade of Moore's law.

40:49And he said, if any other industry had moved this fast, we would be driving like 100 ,000 miles an hour and getting 50 ,000 miles a gallon. And in fact, it would be so cheap that we would just like drive our Rolls Royce downtown and throw it away for the evening and buy another one to go back. And I mean, it's so vivid and it also seems so ridiculous. And then it turns out that it's true. I mean, like, I think everyone, like, sort of vaguely knows that, like, your iPhone is just orders of magnitude more powerful than, like, the Apollo guidance computers. But I think people don't know that, like, just looking around, like, a car key fob could easily, like, plot, like, you know, an Apollo mission to the moon.

41:32And, like, but to me, I was looking for an example. I was like, is there something, like, you know, Gordon Moore says, like, it's like throwing away a Rolls Royce. I was like, is there an example of something that like we actually throw away? And I was talking to people and somebody was like a digital pregnancy test. Like people throw these away every single day and they are considerably more powerful than the like Apollo guidance. And all it is, it displays a plus or a minus sign. That's what it does. That's wild. That's a really good one. Yeah, there was a line in your story that was, I think, that will stick with me for a long time, which is basically like chips aren't made anywhere is the thing is such an important takeaway of all of this to me that it's like you have no idea how many of these there are and also how frankly impossible it is that any of them get made.

42:19And the idea that you can just stand up a factory and start building chips is just dead wrong. And that, in fact, it is it is a global like you you start the story by comparing it to like the wonders of the world, which I thought was frankly ridiculous at the beginning. And at the end, I was like, oh, I actually get it. Like, I think there's there is something there is something equally global and equally like how on earth do we do this about chips? And cooperative, like people were kind of quibbling with that line online and they were just coming up with like individual inventions. And I was like, no, the point is not an individual invention.

42:50The point is a cooperative enterprise that this is like thousands of companies and more than two million people who are all doing this together. Yeah, it's pretty good. All right, Gideon, it's great to have you here. Thank you for coming on and explaining all of this to us. It's great to see you. Appreciate it. David, it was such a pleasure. Thank you so much for having me.

43:09Nilay Patel: Support for the show comes from AWS AI. AWS AI is powering transformation across the world's leading organizations, moving AI from possibility to production at scale. Like how AWS AI helps Formula One turn race day data into real-time insights that bring fans closer to the action. Or how it helps Booking.com personalize every trip for travelers around the world. AWS AI is even transforming how things get built, helping Siemens power the production of smarter cars. Across industries, AWS AI is helping businesses innovate faster, lead with confidence, and transform what's possible today. AWS AI is how.

43:52Nilay Patel: Support for the show comes from Jira by Atlassian. AI is only as useful as the context it has to work from. But when that context is scattered across tools, threads, and DMs, that means your team and your AI agents end up flying blind. That's the problem Jira solves. What's the goal tied to your project? What got decided last week in Slack DMs? Atlassian's teamwork graph pulls all the pieces together, from Jira, Confluence, GitHub, Slack, and more, so nothing falls through the cracks. With Jira, you can actually share the context with agents you already use, like Claude, Cursor, and GitHub Copilot.

44:32Nilay Patel: Assign them work directly or connect your agents through MCP. Spend less time digging through endless links and DMs, less time chasing down what got decided and by who, and more time actually shipping with Jira. Learn more at jira.com. That's J-I-R-A dot com. Company spend can be complicated. Managing it shouldn't be. Amex Corporate connects cards, expenses, and payments all in one place. That's less time managing work and more time doing work that matters. Built to simplify, designed to scale. This is Amex Corporate. Get started at go.amex slash Amex Corporate. Terms apply.

45:23All right, before we go, I have one more thing to tell you about, which is a documentary that's on Netflix. You've probably seen it. I think it's like at or near the top of the charts on Netflix, but it's called Funny Money. And it's about Dane Cook, who 20 years ago was like the biggest thing in comedy. If you have deep Dane Cook memories and you can like quote things and you used to say things about the BK Lounge to your friends, like know that you and I are the same age and know that you and I have a lot of references in common. This documentary is really fascinating. It's fundamentally about Dane Cook's rise in comedy and his relationship with his brother who stole a lot of money from him.

46:00It's a really intense, surprisingly emotional story about a stand-up comedian and his relationship with his family and his relationship with comedy. It's very good on that front. But I think you should also watch it because there's a lot of technology in it. So one of the ways that Dane Cook came up in the early 2000s was on MySpace. He was really early to identify that MySpace was a place full of people he could reach directly. Like, remember, 25 years ago, the idea that you could just have a direct relationship with your fans didn't really exist. And so he spent a lot of time trying to find ways to get on Comedy Central and go on late night shows.

46:35And like, that was how you reached an audience. And he was one of the first people in that space to really discover that the internet connected you to people directly. And that if you could build a big fandom online, suddenly you didn't have to get through the bookers on Comedy Central just in order to make a living as a standup comedian. You could fill rooms around the world just by having fans on the internet. Again, this is how the economy works now. This is how the creator business has been built for countless people all over the world. but you can see it in so many of its early days just looking at Dane Cook's story.

47:08He had really interesting ideas about what touring would look like, really interesting ideas about how to book arenas. He, in interviews around this documentary, has kind of taken credit for the way that Live Nation and Ticketmaster tried to build a monopoly. I don't know if I would give him quite that much credit for changing the way that the stadium and ticket booking economy works, but it's a really interesting idea. and his focus on how do I connect directly to my fans is really fascinating and really important. And he was really early to it and it made him hugely, hugely famous. Like, I don't think there is a comedian now who is big in the way that Dane Cook was big 20 years ago.

47:47And I think there is something he tapped into that the rest of the comedy ecosystem now really operates around. Anyway, the documentary is really good. It's really interesting on a bunch of different levels, but even from a tech story about what the internet means and especially what it meant in those early days for the first few people who really figured out how to use it, really great. All kinds of good stuff. He's also done some really interesting interviews around it, talking more about the tech side of it and how he found fans and how he built these direct relationships with his fans and how that changed the whole trajectory of his career.

48:21Really good, highly recommend it. It's called Funny Money. I think it's the Funny Money, the Dane Cook story or something like that on Netflix. Go check it out. I think you'll enjoy it. All right, that's it. Let's get out of here. That is it for the show. Thank you to Gideon for being here. And thank you as always for watching and listening. As a reminder, if you have thoughts, questions, if you can explain how chips are made in the schoolhouse rock, how a bill becomes a law kind of way, I would love to hear from you. Vergecast at theverge.com is the email address. 866-VERGE-11 is the hotline.

48:48We love hearing from you about anything and everything all the time. So keep reaching out. And the best thing you can do to support all of this that we're up to here is to subscribe to The Verge. You get all of our podcasts ad free. You get all of our exclusive of email newsletters. You get all of our coverage of Chips and Jensen Wong and everything else going on. Theverge.com slash subscribe. Thank you in advance. The Vergecast is a Verge production and part of the Box Media Podcast Network. This show is produced by Josh Cajas, Eric Gomez, Brandon Kiefer, and Travis Larchuk. See you tomorrow.

49:14Rock and roll.

From the publisher

Computer chips are so utterly everywhere that we often don’t think about them at all. But we should. Gideon Lewis-Kraus, a writer and reporter, recently traveled the world attempting to figure out how chips get made — he joins the show to explain the wildly complicated, unbelievably expensive, non-stop miracle that is chip manufacturing. He also explains why no one, no matter their budget, can simply decide to become a chip giant. It’s just not that easy.

Further reading:

From The New Yorker: Computer Chips Are the Most Magnificent Things Humanity Has Ever Made

Subscribe to The Verge for unlimited access to theverge.com, subscriber-exclusive newsletters, and our ad-free podcast feed.

We love hearing from you! Email your questions and thoughts to [email protected] or call us at 866-VERGE11.
Learn more about your ad choices. Visit podcastchoices.com/adchoices

More from The Vergecast

All episodes