Planet Money - The strange way the world's fastest microchips are made

Episode Date: November 13, 2024

This is the story behind one of the most valuable — and perhaps, most improbable — technologies humanity has ever created. It's a breakthrough called extreme ultraviolet lithography, and it's how the most advanced microchips in the world are made. The kind of chips powering the latest AI models. The kind of chips that the U.S. is desperately trying to keep out of the hands of China.For years, few thought this technology was even possible. It still sounds like science fiction: A laser strong enough to blast holes in a bank vault hits a droplet of molten tin. The droplet explodes into a burst of extreme ultraviolet light. That precious light is funneled onto a wafer of silicon, where it etches circuits as fine as a strand of DNA. Only one company in the world that can make these advanced microchip etching machines: a Dutch firm called ASML.Today on the show, how this breakthrough in advanced chipmaking happened — and how it almost didn't. How the long-shot idea was incubated in U.S. nuclear weapons laboratories and nurtured by U.S. tech giants. And, why a Dutch company now controls it.This episode was hosted by Jeff Guo and Sally Helm. It was produced by Willa Rubin and edited by Jess Jiang. It was fact-checked by Dania Suleman, and engineered by Patrick Murray. Alex Goldmark is Planet Money's executive producer.Help support Planet Money and hear our bonus episodes by subscribing to Planet Money+ in Apple Podcasts or at plus.npr.org/planetmoney.Learn more about sponsor message choices: podcastchoices.com/adchoicesNPR Privacy Policy

This message comes from TELUS Business. This Black Friday, TELUS Business is helping small businesses get more done and save. Get the latest technology for your business at exclusive Black Friday prices. Shop these limited time offers today at telus.com slash business Black Friday. Conditions apply. TELUS, proud supporter of Canadian business. Hey, a quick word before the show. The 2024 election is over and as a new administration prepares to assume power, it is our job across

the entire NPR network to report on what they do with that power. That's why we're here. And your support makes it possible for us to break down big stories, to fact check, and to make sure you understand what's going on. When you donate, you make a difference in our ability to do this work. If you're already a supporter, we're going to take this moment to just say thank you truly. And if you're not, you can go to donate.npr.org to give.

That's donate.npr.org. Okay, here's the show. This is Planet Money from NPR. So there is this technology that when I first heard about it, I thought, you gotta be kidding me. This has gotta be science fiction. It's this new way of making microchips. There is only one company that has figured out this technology, a Dutch company called ASML.

Recently, Jeff, you went to their labs in San Diego. How many people have seen what we're about to see today? Well, a lot of ASML engineers, and that's most of the list. It's a working clean room, so it's very rare for people to go inside. That's Sarah deCrisenzo. She's a comms person for ASML, and she's about to show me the heart of one of their new microchip etching machines.

A microchip is basically a bunch of circuits that are etched onto a piece of silicon. The more circuits on the chip, the more powerful it is. And for almost the entire history of microchips, the circuits have been getting smaller and smaller, and the chips have been getting more powerful. But about 10 years ago, that progress slowed pretty dramatically. The industry was staring down a dead end. Until ASML made a breakthrough with this new technology. These new machines, they were able to etch billions

and billions of circuits onto a single chip. But they're also incredibly delicate. Sarah says it's been more than a year since they've let a journalist inside this lab. They don't love to do it, because even the tiniest speck of dust could ruin the machine. And journalists are not sterile.

Absolutely not. So we have to go through this whole decontamination procedure. There is a machine where you stick your foot into to clean your shoes. This is great. Oh, it untied my shoes. That's how powerful it was. Next, a technician named Blaine Howarth comes over to wipe down my microphone and recording equipment.

This is all we got to clean? Yeah, all this equipment. What about the foam? Blaine is pointing to this black foam windscreen that's on my microphone, which I will admit was a little dirty. It had some cat hair stuck to it. What don't you like about it?

The particles it gives off. What particle? Oh, you mean this cat hair? Yeah, yeah. Yeah, cat hair is a problem for us. It really is. Okay. I put on what looks like a thin plastic spacesuit, and now I am ready to walk into the clean room. I'm about to see one of the most complicated technologies ever invented.

And Jeff, at that point, you had been talking about this amazing microchip etching technology for like months. amazing microchip etching technology for like months. Yes, because this is the technology behind all the new chips powering the most advanced AI models in the world. It is so important that the US has been lobbying the Dutch government not to let ASML sell any of these machines to China. They see it as a matter of national security. But most of all, I've just been fascinated by these machines themselves.

Like, let me briefly explain how they work. Alright, take it away. You start with a laser. A laser that is so powerful it could cut through a bank vault like butter. And then you focus that massive laser on a tiny droplet of molten tin like the metal tin. And then, blammo! The tin vaporizes into a plasma that gives off a beautiful, intense light. They call it extreme ultraviolet. And this light bounces off a series of mirrors, which have to be like the smoothest mirrors on the planet,

and it etches billions and billions of tiny little microscopic circuits onto a wafer of silicon, which, by the way, is magnetically levitating, culminating in the most powerful microchips that have ever existed. I mean, I can see why they're a little bit touchy about the cat hair. Yes. And Sally, this extreme ultraviolet technology for making chips, it is such an accomplishment.

It is like our generation's moon landing. For years, a lot of people thought it was impossible. And one of the things I've been obsessed with is how. How did we even do it? How did humanity pull this off? Hello, and welcome to Planet Money. I'm Jeff Guo. And I'm Sally Helm.

It is easy to take for granted that microchips are just going to get more and more powerful. But they don't get powerful all by themselves. It takes a lot of people over a lot of time. Often they make bets that, in the moment, seem unlikely, even foolish. Today on the show, how one of the most complicated, most improbable technologies in the world, this breakthrough in how we make microchips came to be. And how it almost didn't. This message comes from WISE, the app for doing things in other currencies.

Send, spend or receive money internationally and always get the real-time mid-market exchange rate with no hidden fees. Download the Wyse app today or visit Wyse.com, T's and C's apply. Alexi Horowitz-Ghazi here. Sure, subscriptions offer convenience, but are they bad for competition? When you're not canceling because you forget about it or it's difficult to cancel, those forces of consumers taking their business to another product are blunt. That's from our recent Planet Money bonus episode, my extended interview with Stanford economist Neil Mahoney. Listen with NPR Plus at plus.npr.org.

Joe Biden's on his way out and Donald Trump's on his way back. Want to know what's happening as the presidential transition is underway? The NPR Politics Podcast has you covered with the latest news and analysis. Listen to the NPR Politics Podcast. Every technology starts with an idea. And in the beginning, that idea is almost like a dream. One of the first people who believed in this revolutionary way of making microchips was Andy Havrluk. You're kind of a big deal in the world of microchips and stuff, right?

That's for somebody else to decide. I don't consider myself a big deal. I just consider myself one of many. Andy's big idea was that he thought it was possible to etch microchips using extreme ultraviolet light. And this idea actually came to him when he was working on something completely different. It was the 1980s and he was a young scientist at the legendary Lawrence Livermore National Labs in California. At the dawn of the Cold War, the United States had started putting billions of dollars into

science and research. And it created these big national labs. They're all over the place, and they all have different functions. The primary function for Lawrence Livermore is nuclear weapons research. Yeah, the government scientists at Lawrence Livermore have designed more than a dozen different types of nuclear warheads. They study nuclear reactions, like nuclear fusion. A fusion reaction generates a lot of powerful light, including extreme ultraviolet.

Andy and his team were working on new kinds of mirrors. Mirrors that, for the first time, could reflect and control the light coming off of a fusion reaction. And Andy realizes that these new special mirrors might have a use outside the lab. Because extreme ultraviolet light is super precise, and if we can now control it with these mirrors, we could etch like the tiniest circuits ever and make microchips way more powerful. This is chapter one in the life of many technologies.

Someone looks at a scientific breakthrough or a discovery in a lab and imagines how it might apply to something totally different, solve some problem in the real world. Andy goes to a conference to present this idea. He remembers feeling nervous. All the top microchip researchers are there. Do you remember the moment where you went up and presented it? What was that like? I remember the moment after I presented it. It was not well received.

So many naysayers got up and basically said, stupid idea, crazy idea, it'll never happen. Andy says it was brutal. Just everyone piling on him, telling him all the different reasons why his idea wouldn't work. Like these mirrors would have to be the smoothest mirrors on the planet. And was it even possible to generate extreme ultraviolet consistently? It's all just too complicated. Andy goes back to his lab. His boss asks how his presentation went.

I said, I don't want to talk about it. I will never speak of it again. Oh my God. I was just, I felt humiliated and embarrassed. But a few days later, Andy gets a phone call. It's from this guy, Bill Brinkman. He'd heard about Andy's presentation and wanted to know more. And he's like, who is this guy?

And so I went to my boss and I said, who's Bill Brinkman? And he looked at me and said, he's the executive vice president of AT&T Bell Labs. And I went, oh. Bell Labs, it's the place where the laser was invented, where fiber optics was invented. It's one of the most famous private laboratories in the world. And Bill Brinkman was one of their top scientists. And I said, well, he just called me. My boss looked at me and said, he called you?

And I went, yeah. And he said, tell him you'll be on the next plane out. The folks at Bell Labs were also looking into ways to etch microchips using extreme ultraviolet light. They flew Andy out to their headquarters in New Jersey. They had an auditorium with, there were 50, 75 people there to listen to me. It was a huge reception for us, you know, for that.

And I went, holy crap. Andy had found a group of fellow believers, people who thought maybe this extreme ultraviolet technology really could work. There was also a researcher in Japan, Hiro Kinoshita, who had started tinkering with the idea even before Andy, but most of the microchip industry doubted the idea. What Andy and the other believers needed to do was to prove that this technology was possible. And to do that, they needed someone to take a gamble. They needed seed money.

This is the start of Chapter Two in the life of many technologies. And in this case, a lot of that early R&D money would end up coming from the U.S. government. Yeah, you see, this was the 1980s. And the government was beginning to think differently about its role in R&D and science. They had been spending billions of dollars on scientific research, which gave them an edge in the Cold War. A lot of nuclear weapons stuff.

Right. But the Cold War was winding down. The Berlin Wall was about to crumble. And Congress realized that the national labs were sitting on all this research that could have a lot of practical applications, could stimulate the economy, help make U.S. companies a lot more competitive.

And so Congress told the National Labs, we want you to partner up with U.S. companies. We want you to work together to explore commercial uses for your research. And we will give you some seed money to do it. It was perfect timing for our band of believers. Andy and his team at Lawrence Livermore eventually signed a deal with several US companies to research extreme ultraviolet chip making. Bill Brinkman and AT&T Bell Labs signed a deal too.

And it wasn't just them. By the early 1990s, a bunch of companies and National Nuclear Weapons Labs were working together to see if this technology was viable. One of the people involved was Rick Stuhlin. Rick was in charge of the research at Sandia National Labs. Was there like a quarterback for all of this or were you all just working on individual projects trying to chip away at this huge problem? Yeah, that's a great question. We did not have a quarterback, yeah. Rick said the government seed money had unleashed all these different teams, each working on their own piece of the extreme ultraviolet puzzle.

And bit by bit, they start to show that they can overcome the technological obstacles here. Using extreme ultraviolet to etch microchips? It looks like that's really possible. But then in early 1996, Rick gets a call from Washington. Rick, we're terminating the program. Congress is no longer interested and has some concerns about this looking like corporate welfare. So you basically have about six months to wrap things up and move on. President Clinton is trying to balance the budget

and that seed money for projects like extreme ultraviolet research dries up. Meaning that government scientists like Rick and Andy, they might have to go back to working on nuclear fusion or national security projects. And so we were stunned and at the same time we knew we were onto something. I mean we knew this was something that was going to make a difference. So I was, you know, obviously scrambling.

Like the future of microchips was kind of hanging in the balance. The world had been transformed by faster and faster chips, faster and faster computers. But with the current technology, there was a limit to how fast chips could get. The industry would soon be facing a dead end. Rick realizes that to save all the extreme ultraviolet research they'd been working on, they needed to raise a lot of money fast. So he goes to the big US microchip companies and says, the government is out. Will you make up the difference here? Take on all the financial risk yourselves?

And it was incredible to watch them sort of sit up and say, this cannot happen. We're going to figure out a way to continue to fund this because we think you're going to make it. Wow. So they like made basically a giant industrial GoFundMe. Well, they did, they did. That's exactly right.

The government seed money had worked. The labs had made so much progress that the biggest companies in the US microchip industry, companies like Intel, wanted them to keep going and were willing to make a much bigger investment in this extreme ultraviolet technology. Intel and other companies super-sized the R&D budget from a couple of million dollars a year

to more than $40 million a year. And they define a clear goal. They want the national labs to work together to build a prototype, an actual machine that can etch a microchip from start to finish using extreme ultraviolet light. It's called a lithography machine. Rick is appointed the chief operating officer for this huge, kind of unprecedented partnership between the national labs and private industry. Basically, he is now the quarterback. Was that a daunting challenge?

Yes. Yes, it was. We'd never built a lithography tool in our lives. None of us, right? And Extreme Ultraviolet Light is so tricky to work with that it takes years. But by 2001, they pull it off. They have a prototype. Yeah, it looked very, very much like a research tool. Wasn't pretty. Didn't have a pretty exterior. There were a lot of cables all over the place. We thought it was beautiful in every way.

Rick remembers feeling just over the moon. Like they had finally proven that this technology was really possible. The mood was giddy. It was exuberance and it was a sense of pride. It was like a moonshot and we landed, right? Now that the prototype was built, Rick's job was mostly over. But now comes the final chapter in the life of a technology. That kind of ugly, cable-y prototype has to become a slick, real-world machine.

One that can sit on the factory floor in companies all around the world, etching tiny circuits onto reams and reams of microchips. And our original slogan for the program was, on the floor in 04. We wanted these tools to be on the semiconductor manufacturer's floor in 04. Well, that was ridiculous. We were way, way, way, way too optimistic.

The toughest challenges would still be to come. After the break, it is one thing to prove that a technology is scientifically possible, but to prove that a technology is commercially viable, that you can make money off of it, that is something else entirely. This is Ira Glass with This American Life, each week on our show. We choose a theme, tell different stories on that theme. All right, I'm just going to stop right there. You're listening to an NPR podcast, chances are you know our show.

So instead, I'm going to tell you, we've just been on a run of really good shows lately. Some big epic emotional stories, some weird funny stuff too. Download us, This American Life. Okay, so does this sound like you? You love NPR's podcasts, you wish you could get more of all your favorite shows, and you want to support NPR's mission to create a more informed public. If all that sounds appealing, then it is time to sign up for the NPR Plus bundle. Learn more at plus.npr.org.

I'm Rachel Martin, host of NPR's Wild Card Podcast. I'm the kind of person who wants to skip the small talk and get right to the things that matter. That's why I invite famous guests like Ted Danson, Jeff Goldblum, and Issa Rae to skip the surface stuff. We talk about what gives their lives meaning, the beliefs that shape their worldview, the moments of joy that keep them going. Follow Wildcard wherever you get your podcasts, only from NPR.

This message comes from the Kresge Foundation. Established 100 years ago, the Kresge Foundation works to expand equity and opportunity in cities across America. A century of impact, a future of opportunity. More at Kresge.org. You hear it all the time. Some lab announces a revolutionary breakthrough. Scientists have discovered a way to encode data using holograms. Or they're able to levitate frogs using magnets. Or they've created nanowires

stronger than maybe anything in the universe. But then years later, like nothing seems to come of it. Where are the frog levitating machines, Jeff Woe? Where are they? Where are they? And the main reason that most technologies never make it into the real world boils down to economics. Like, a lot of really cool technology just ends up being too expensive or

too impractical or both. Yeah, and for extreme ultraviolet chip making, that reckoning came in the early 2000s. By that point, the US microchip industry and the US government had spent hundreds of millions of dollars proving that this next generation technology could work in the lab. And now they turned to that final chapter in the life of a technology, getting it to market.

Like some company needed to take all the research and the prototype and figure out how to make the technology profitable, basically make a commercial version of the machine. And for this challenge, the US effort turned to a Dutch company, ASML. I think people now, in retrospect, think, oh, like this could have been an American technology, but we passed on the torch to a Dutch company. Well, that's an expression, but you could also say

that the US dropped the ball on lithography. Sorry for being blunt in a Dutch way. Very Dutch, yes. Mark Hajink is a Dutch business journalist. He's kind of the world's expert on ASML, wrote a book about it recently. Mark says, sure, some politicians wanted a US company to bring these machines to market.

The problem was, there weren't any great options in the US. The major players at the time were ASML and two Japanese companies, Canon and Nikon. But Canon and Nikon were out because the United States saw Japan as its main microchip rival. So the industry turned to ASML. ASML had been doing its own research into extreme ultraviolet, and in the early 2000s, it took on the challenge of making a profitable commercial machine. At the time, there was still quite a bit of work left to do. The prototype was a nice proof of concept, but in order to etch a single wafer of microchips,

this prototype might have taken a whole day. A commercially viable machine needs to make hundreds of wafers an hour. Otherwise it just wasn't worth it. That's how brutal the economics of the microchip industry are. You have to have a machine that keeps on running day in, day out without too many troubles, without too many errors. So it's like crazy science and crazy economics in one machine. In the beginning, ASML was pretty sure they could get the science and the economics to

work. They thought they could make a machine ready to go into factories by 2006. Not quite on the floor in 2004, but close enough. The main problem they had to tackle was gathering enough extreme ultraviolet light. To etch microchips, you need a lot of light. The more light you have the faster the machine can go. But for the longest time ASML couldn't get that extreme ultraviolet light bright enough. And once you get to see one of their machines up close you start to understand why.

Yes, yes please. That is Alex Schaffgans. He is the head of engineering at ASML's San Diego Lab, which Jeff went to visit. He says there just isn't any simple way to generate extreme ultraviolet light. Remember, this is the part that sounds like science fiction. First, you needed a giant laser. Alex takes me to one of their laser rooms. It has rows and rows of these like six foot tall beige cabinets. This entire room powers just one laser.

Can I ask why that red light is rotating? Looks like an alarm light. The laser is armed and can fire. Oh, good to know. But this laser itself did not produce extreme ultraviolet light. It was just a regular laser, albeit supersized. But then the laser had to hit that droplet of molten tin, create that superhot plasma, and that would give off a flash of extreme ultraviolet light.

And in order to get enough extreme ultraviolet light, they needed these tin plasma explosions to happen 50,000 times a second. Yeah, Alex takes me to a test chamber. Looks like a big metal sewer pipe with a plexiglass window on one side. We're about to see something that very few people have seen. Inside is a nozzle spitting out these molten tin droplets. And we've lit up the chamber with a flashlight so that you can visually see this spiderweb looking string of droplets.

Holy wait. Oh, it does. It looks like a very thin spiderweb. It took years to figure out how to make that thin, silvery spiderweb. The engineers had to come up with clever ways to vibrate the nozzle so every droplet would be the exact same size and shape. There were hundreds and hundreds of engineering challenges like this. And every time they solved one problem, another one would pop up. By 2006, their original deadline, they had a machine,

but it was too slow and it broke down too often. ASML told the industry, its customers, its investors, just give us another year, one more year. They said that year after year after year. By 2011, ASML's Japanese rivals, Canon and Nikon, had given up on extreme ultraviolet technology. And some of the execs at ASML were wondering if they should give up, too. They went to their customers, chipmakers like Intel, and asked them, are you absolutely sure that you still want these machines?

It's going to take more time, more money. Maybe we should just call it now. But the chip makers are like, no, no, no, no, no, we still want these machines. Because without them, progress will slow and we're not really sure how to make faster chips. So ASML forged ahead. And in 2017, they did it. Back at the lab, I suit up to see the final product. Alex leads me into what looks like a cavern or a cathedral. We turn the corner and…

This is the plasma vessel. Ahead of us is this stainless steel sphere. It's the size of a car. Inside this sphere is where the tin droplet meets the giant laser. The plasma from the explosion gets 40 times hotter than the surface of the sun. A nearby screen shows what's happening inside. I can see a steady purple glow. This light source is currently making plasma.

So we... This is running right now? It's running right now. You tell me that! It's running right now. Standing there was kind of terrifying. All those violent explosions happening 50,000 times a second inside that chamber. You're seeing just the glow of this hot plasma.

Oh my gosh. Yep, we're making plasma. What we were looking at, that purple glowing plasma, it was the culmination of almost 40 years of research. Started in the 1980s with people like Andy dreaming of this new way of making microchips. It was nurtured through the 90s by this partnership between the US microchip industry and US nuclear weapons laboratories.

And then for almost 20 years, it was in the hands of ASML who finally brought it to market. And that final chapter tends to be the hardest and the most expensive part in the life of a new technology. In this case, the early research and prototype had cost U.S. taxpayers and U.S. companies maybe $300-400 million. But ASML says after they picked up the baton, they spent about 15 times that amount. More than $6 billion.

It's funny, ASML execs say if they had known how much it cost, how long it'd take, they probably wouldn't have taken a bet on this technology. But their gamble paid off. ASML now controls maybe the most valuable technology in the world. Its latest extreme ultraviolet machines go for about $380 million each. They are some of the most complicated objects that humans have ever built. And they've become indispensable. They're used by companies like Intel, TSMC, and Samsung.

They're producing the chips powering the most advanced AI models. And you know, advanced chip technology like this, it kind of feels like inevitable. It feels like one of the most basic underlying facts of our modern world. But this stuff almost didn't happen. This episode was produced by Willa Rubin and edited by Jess Jang. It was fact-checked by Danya Suleyman and engineered by Patrick Murray. Alex Goldmark is our executive producer. Special thanks to John Carruthers, Danny Brown, Andrei Litvak, and Georgi Vashenko. And if you're interested in Mark's book about ASML, it's called Focus the ASML Way.

I'm Jeff Guoh. And I'm Sally Helm. This is NPR. Thanks for listening. On the TED Radio Hour, clinical psychologists John and Julie Gottman are marriage experts. And after studying thousands of couples, they have found... Couples who are successful had a really different way of talking to one another when there was a disagreement or a conflict. How to be brave in our relationships.

That's on the TED Radio Hour podcast from NPR. Hey everyone, I'm B.A. Parker, a host of the podcast Code Switch. And on my show, I get to dig into all of the facets of being a black woman, from honoring my ancestors to exploring representation in reality TV. Code Switch is a place where I think out loud about how race and identity are connected. Join me on the Code Switch podcast from NPR. Some of our favorite planets aren't even real,

but could they be? Here on Shortwave, we journey to other planets, distant galaxies in our universe and in our favorite works of science fiction. Listen now to the Shortwave podcast from NPR.