# The Unseen Precision of Chip Lithography Systems

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 27 May 2026, 8:46.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0015.mp3

ASML's extreme ultraviolet lithography machines are the most complex manufacturing tools ever built, each costing over $350 million with the precision of a laser bouncing off a million mirrors to etch features smaller than a virus. Lucas and Luna unpack the physics, the engineering tolerances measured in picometers, and the supply chain dependencies that make this single machine the bottleneck for every advanced chip from Apple to NVIDIA. They explore why the next generation—hyper-NA EUV—is already in development and what it means for the industry's roadmap beyond three nanometers. A focused look at the machine behind the machine, with concrete numbers and real constraints. #ASML #EUVLithography #ChipManufacturing #SemiconductorEquipment #ExtremeUltraviolet #NanometerScale #Photolithography #ChipDesign #MooreLaw #Technology #Business #Engineering #Hardware #FexingoBusiness #BusinessPodcast #TheHardwarePodcast #Chips #Devices Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: If I told you there's a machine that fires a laser at a tiny droplet of tin, turning it into plasma as hot as the sun, and that plasma emits light so short in wavelength it can carve patterns smaller than a virus — you'd think I'm describing a physics experiment, not a piece of manufacturing equipment.
Luna: And yet that's exactly what ASML's EUV lithography machines do. They're probably the most complex tools ever built by humans.
Lucas: Extreme ultraviolet lithography — EUV. It's the only way to mass-produce chips at the five-nanometer scale and below. Without these machines, there are no advanced processors from Apple, no NVIDIA H100s, no AMD EPYC chips. The entire high-end semiconductor industry runs through a single Dutch company.
Luna: And the numbers are staggering. Each EUV machine costs around three hundred fifty million dollars. They weigh over a hundred eighty tons. A single shipment takes multiple cargo planes and weeks of setup.
Lucas: But the engineering challenge isn't just size or cost — it's precision. To etch features that are only a few nanometers wide, you need light with a wavelength of thirteen point five nanometers. That's extreme ultraviolet. Regular ultraviolet is around four hundred nanometers, visible light is even longer. So EUV is far shorter, which lets you print smaller features.
Luna: But here's the problem: that thirteen-point-five-nanometer light is absorbed by almost everything. Air, glass, even conventional mirrors. So the entire machine has to operate in a vacuum, and instead of lenses, it uses a series of extremely precise mirrors.
Lucas: Mirrors, yes — but not like any mirror you've seen. Each one has over a hundred alternating layers of molybdenum and silicon, each layer just a few atoms thick. The coating is designed to reflect EUV light at nearly seventy percent efficiency, which is actually remarkable because most materials absorb it entirely.
Luna: And there are about a dozen of these mirrors in the optical path. Every time the light reflects off a mirror, it loses some intensity. By the time it reaches the wafer, you've lost more than ninety-eight percent of the original energy. So you need an incredibly powerful source just to get enough photons to expose the photoresist.
Lucas: That's where the tin plasma comes in. A high-power CO2 laser fires at tiny droplets of molten tin — fifty thousand droplets per second. The laser vaporizes each droplet, turning it into a plasma that emits EUV light. That process alone consumes about one megawatt of electricity just to produce usable light.
Luna: And then you have to keep everything aligned. The mirrors have to be positioned with an accuracy of a few picometers — that's trillionths of a meter. If a mirror is off by even that tiny amount, the image blurs and the chip doesn't work.
Lucas: The alignment system uses laser interferometry and multiple feedback loops. It's constantly measuring and adjusting the mirror positions, compensating for thermal expansion, vibrations from the factory floor, even the gravitational pull of the moon. Yes, really — the moon's tidal forces can shift the alignment enough that the system has to correct for it.
Luna: So this machine is essentially fighting physics at every level. And ASML has a near-monopoly on it. They're the only company that builds these EUV systems, and they've invested decades and billions to get here.
Lucas: The development started in the late 1990s, with a consortium of research labs and chipmakers. But it took until 2019 for ASML to ship the first production-ready EUV tools. The delays were legendary — the optics were too difficult, the source power was too low, the yields were terrible.
Luna: And now, just as EUV is becoming mainstream for five and three nanometer nodes, ASML is already working on the next generation: hyper na EUV. Higher numerical aperture, which means even finer resolution, potentially down to one nanometer or below.
Lucas: But these machines will be even more expensive — over five hundred million dollars each, some estimates say. And the infrastructure required is staggering. A fab that installs a single EUV machine has to build a special foundation to absorb vibrations, a dedicated cooling system, and a massive cleanroom. It's like building a skyscraper around a single tool.
Luna: So the question becomes: how long can this keep going? Moore's Law has already slowed. The cost per transistor is no longer dropping as quickly. And at some point, the physics of EUV itself maxes out.
Lucas: Good question. And here's where it gets interesting. The next step after hyper na EUV might be something completely different — like high na electron beam lithography, or even directed self-assembly of molecules. But those are years away, if they ever become production ready.
Luna: And that's the reality for the chip industry right now. We're squeezing every last drop out of existing technology. The EUV machine is the pinnacle of precision engineering, but it's also a sign that we're approaching fundamental limits.
Lucas: You know, talking about these machines and the incredible engineering required, it reminds me why we do deep-dive shows like this. It's because the story behind the technology is often more fascinating than the product itself.
Luna: Absolutely. And a quick honest thing — a handful of listeners chip in monthly through buy me a coffee dot com slash fexingo, and that's literally what funds making as many of these deep dives as we do.
Lucas: Yeah, no ads, no sponsors, just listener support. It's a small group, but it keeps us independent. And we really appreciate it.
Luna: Alright, back to the photons. So, Lucas, you mentioned the source power. How much EUV light actually reaches the wafer?
Lucas: After all the mirrors and the vacuum, you end up with maybe a few hundred watts of usable EUV power at the wafer plane. Compare that to the one megawatt input — the efficiency is atrocious, less than one percent. But that's the price you pay for the ability to print features that are just a few nanometers wide.
Luna: And each wafer takes about a minute to expose, depending on the complexity. So a single EUV machine can process maybe a hundred and fifty wafers per hour. That's a lot of chips, but it's still the bottleneck in the entire supply chain.
Lucas: Right. And that's why TSMC and Samsung have been buying as many EUV machines as ASML can produce. TSMC alone probably has over eighty of them now. Intel is playing catch-up, trying to secure enough tools for their five-nanometer equivalent node.
Luna: The supply chain for these machines is also incredibly concentrated. Many of the key components come from a single supplier. The optics come from Zeiss in Germany. The laser source is from a company called Trumpf. The vacuum chambers from various European specialists.
Lucas: And ASML has to manage that entire ecosystem. Any disruption — a trade dispute, a natural disaster, a pandemic — can delay shipments by months. We saw that during COVID when some components were held up.
Luna: So the entire advanced chip industry sits on a very fragile foundation. A handful of machines, a handful of suppliers, and decades of accumulated knowledge.
Lucas: And that's the real story here. EUV lithography is a marvel of engineering, but it's also a vulnerability. The more we depend on these machines, the more we have to protect and invest in that supply chain.
Luna: Well, and it raises a question about the future. If the next generation of machines is even more expensive and complex, will only a few companies be able to afford them? And what does that mean for competition in the chip market?
Lucas: That's a topic for another episode. But you're right — the economics of lithography are becoming as important as the physics. For now, though, the EUV machine remains the unsung hero of every smartphone, laptop, and server you use.
Luna: And a remarkable piece of engineering. Thanks, Lucas.
Lucas: Thanks, Luna. Next time we'll look at what comes after EUV — if anything.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.