The impossible process of making a computer chip
· Summarized
Key takeaways
- Moore's Law is not a physical law or an inevitable trend; Lewis-Kraus describes it as a collective, self-fulfilling commitment among companies that repeatedly promise to solve problems that may not yet be solvable.
- ASML's EUV lithography machines are indispensable because advanced chips cannot be produced economically without their ability to project extremely short-wavelength light onto wafer designs.
- Chip production is a global network involving thousands of companies and roughly 70 border crossings, so no single country can recreate the industry simply by funding and building a factory.
- TSMC's value depends not only on its machines but also on decades of accumulated engineering knowledge, supplier relationships, schools, and communities that are difficult to transfer to a new location.
- The chip ecosystem works through carefully maintained cooperation among companies such as ASML and Zeiss, even though their margins, secrets, and customer interests are not identical.
- The U.S. reshoring effort should be understood as a generational project requiring suppliers, training, and expertise—not as an overnight response that can be completed within one electoral cycle.
- Dane Cook's early MySpace strategy illustrated how performers could bypass traditional media gatekeepers by building a direct, internet-based relationship with fans.
Summary
Moore's Law is a self-fulfilling industry commitment
Host David Pierce opens by asking why chip-industry veterans dislike Moore's Law, the idea that computing power roughly doubles every two years. Gideon Lewis-Kraus quotes an ASML physicist calling it “just marketing” and saying it is for people who never worked on it. Engineers do not experience progress as an inevitable law; they experience it as repeated struggles against physics. Lewis-Kraus describes Moore's Law instead as a collective delusion and pre-commitment mechanism: companies make road maps for technologies that do not yet exist, trusting suppliers to accomplish equally impossible tasks. The motivation is often professional pride and commitment to peers, not a specific customer demand.
Why Lewis-Kraus followed ASML and TSMC
Lewis-Kraus chose TSMC because it produces about 90% of the world's most advanced chips and occupies a major geopolitical choke point, even though many people know little about the company. He chose ASML because it has repeatedly been described as “the most important company you've never heard of”; ASML's own store sells a hoodie joking about that description. TSMC makes the chips, but ASML makes the specialized machines without which advanced chips cannot be produced. Lewis-Kraus says these two companies provide a way to explain the entire chip system: one is the dominant manufacturer, and the other controls its most irreplaceable production technology.
ASML solved the light problem with extreme ultraviolet lithography
Lithography prints a chip's design by projecting light through a pattern onto silicon, allowing increasingly intricate transistor architectures. As transistor density rose, engineers used shorter wavelengths as a sharper optical “scalpel,” but eventually appeared likely to run out of usable light. Beginning in the early 1990s, Intel, an American consortium, and the U.S. government helped coordinate a search for a solution that took roughly 15 to 25 years to become practical: extreme ultraviolet lithography. ASML is the only company able to make EUV machines at the scale and efficiency required for advanced chips. China has reportedly made progress with older deep-ultraviolet technology, but prototypes producing 30 or 50 wafers per hour remain far behind economical throughput.
A chip is a Tokyo-sized cityscape on a fingernail
To explain chip scale, Lewis-Kraus compares a chip to a city: the transistors form the footprints of buildings, while layers of wiring and interconnections create infrastructure above them. A chip can be understood as a perfect replica of a metropolis such as Tokyo etched onto an area the size of a fingernail; its smallest real-world feature, scaled to Tokyo, would be about the size of a golf ball. Nanometer measurements have become too abstract to compare meaningfully with a human hair. In one example of the precision involved, an ASML researcher said he gave teams a budget of four atoms and made them decide how to divide it.
Inside the five-mile, soccer-stadium-sized factories
At TSMC and ASML, Lewis-Kraus wore full clean-room clothing, passed through air showers, and used special paper that would not shed particles. TSMC's clean rooms are about 1,000 times cleaner than a hospital surgical theater, with yellow lighting, constant air recirculation, and floors venting air like an air-hockey table. A fab he visited had enough clean-room space for 22 simultaneous full-scale soccer games, and his guide's watch recorded five miles of walking inside one building. Automated overhead rails move wafer carriers through thousands of machines, reducing variance rather than merely saving labor; even a slight error can ruin a wafer.
Secrecy protects customers' chip designs
The companies showed Lewis-Kraus machines but concealed details that could reveal customer production lines. At TSMC, paper covers hid machine brands and names because the layout could theoretically identify a customer and product, though Lewis-Kraus could not interpret it. TSMC's founding promise was never to compete with customers or steal their designs. Apple, Nvidia, and Broadcom can use the same company because their intellectual property is kept separate; even visiting customers see only their own production areas. Some process improvements necessarily spread across customers, but TSMC must prevent direct design knowledge from leaking between them.
Why moving chip production to America is not just building a fab
Lewis-Kraus argues that chips are not made in one place: the process crosses borders roughly 70 times and involves thousands of specialized companies. It begins with unusually pure quartz from a mine in western North Carolina, while polysilicon refinement depends on companies in Michigan, Germany, and Japan, among many other links. TSMC's Phoenix operation even had to transport clean-room clothing across the desert because no suitable nearby laundry existed. Mike Schmidt, who led the Biden administration's CHIPS Act implementation, said the goal was a generational project like Morris Chang's founding of TSMC in 1987, requiring decades of suppliers, schools, engineers, and local expertise.
The supply chain depends on cooperation, not simple competition
Pierce and Lewis-Kraus debate whether the United States should replicate the chip ecosystem quickly for resilience, even if the existing global system is an extraordinary cooperative achievement. Lewis-Kraus says economists may identify apparent inefficiencies, such as different companies capturing different margins, but forcing a more competitive structure could break the system. ASML depends on Zeiss for mirrors made from roughly 40 alternating carbon and molybdenum layers, deposited nearly atom by atom; the mirrors take two years and millions of dollars to produce and are extraordinarily flat. ASML and Zeiss employees work so closely that even staff cannot always tell which company someone belongs to. Lewis-Kraus favors building resilience while respecting these relationships.
No one understands the whole chip-making process
Lewis-Kraus says nobody understands exactly how a chip is made—not even specialists who know one part of an ASML machine. Physicists who build the EUV light source in San Diego could explain the reticle stage only in broad terms. He nevertheless left with answers to his practical question: a country cannot simply build a factory, install machines, and become a chip power. Gordon Moore once said that if another industry had advanced as quickly, cars would travel 100,000 miles per hour, get 50,000 miles per gallon, and be cheap enough to discard after one trip; Lewis-Kraus notes that a disposable digital pregnancy test is already more powerful than Apollo's guidance computers.
The closing recommendation: Netflix's Funny Money
Before signing off, Pierce recommends Netflix's documentary Funny Money, about Dane Cook's rise and his brother's theft of money from him. Pierce is especially interested in Cook's early use of MySpace: instead of depending entirely on Comedy Central, late-night television, or traditional bookers, Cook built a direct relationship with fans online and used that audience to fill rooms around the world. Pierce says Cook had early ideas about arena touring and may have influenced how people think about ticketing monopolies, though he does not credit Cook alone for Live Nation and Ticketmaster's structure. He presents the film as an early case study in the creator economy.
Sponsors
- Satelli · Satelli Collar Go is an AI-powered GPS dog tracker that helps locate dogs and identify potential health challenges. The ad offers 35% off with promo code VERGECAST.
- On Investing from Charles Schwab · Hosts Liz Ann Saunders and Colin Martin discuss market developments, portfolio implications, corporate bonds, and sectors on the move.
- Constellation · Constellation describes itself as the largest private-sector energy producer and the nation's largest producer of reliable clean energy, with 55 gigawatts of capacity serving critical infrastructure and more than 2.5 million customer accounts.
- Shopify · Shopify promotes its pre-built themes, AI site builder, mobile store-management app, and Sidekick AI assistant for launching a business before Black Friday. The ad offers a free trial.
- Framer · Framer's AI website builder lets teams collaborate, generate code components and CMS content, optimize SEO, audit sites, and publish production-ready pages. The ad offers 30% off an annual Framer Pro plan.
- Superhuman · Superhuman Go, from the makers of Grammarly, works inside existing tools to summarize threads, prepare for meetings, catch typos, and brainstorm without losing context.
- AWS AI · AWS AI says it helps organizations move AI from possibility to production, including Formula 1's race-day insights, Booking.com's travel personalization, and Siemens' smarter-car production.
- Jira by Atlassian · Jira's teamwork graph connects context from Jira, Confluence, GitHub, Slack, and other tools so teams and AI agents can understand project goals and decisions. It supports agents such as Claude, Cursor, and GitHub Copilot.
- American Express Corporate Cashback Card · The card offers 1.5% cash back on eligible business purchases, spending controls, real-time visibility, and expense management. Terms and caps apply.
- Fred Meyer Pickup · Fred Meyer will review and hand-select online grocery orders, pack them, and load them into the customer's trunk. The ad offers $30 off a first online order and free delivery on the first three orders, with restrictions.
Show notes
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
Follow this show on AudioBrevity and new episodes get summarized for you. Create a free account