Comment by codeulike | Hacker News Reader

Comment by codeulike | Hacker News Reader

Parent Full thread

ASML is so interesting. Its like sci-fi that one firm in the Netherlands knows how to make the most complicated machine ever made, and no one else can do it.

And it is arguably the most complicated machine ever made. 50,000 times a second, the EUV lithography machine hits a 25 micron drop of molten tin that is moving at 70 meters per second with two co-ordinated lasers, the first hit to change the shape of the drop of tin in exactly the right way, the second hit to vaporise it, creating Extreme Ultraviolet Light at the right wavelength to etch chip designs onto silicon at "5nm process" sizes. Some labs can cobble together something similar as a proof of concept, but not well enough to make it feasible for mass production of chips.

Video about the light souce - https://www.youtube.com/watch?v=5Ge2RcvDlgw

No one else in the world is able to make these machines. If you buy one it costs $150m and gets shipped to you in forty containers on specially adapted planes. Very few firms have the resources/know how to even run the machines - which is what makes TSMC so important.

EUV technology was developed in partnership with the US Department of Energy which is why the US can implement export controls (it was an explicit condition of original deal with the DoE). A significant part of the “secret sauce” is manufactured in San Diego.

It’s not really “one firm in the Netherlands”, it’s a global collaboration that goes back to the 1990s. Intel was involved from the beginning, they just dropped the ball.

> it’s a global collaboration

Yep, even on the optics side (IIRC Zeiss, Nikon and Samsung are big players in the optical side...)

I saw a working euv laser at TRW in the early 2000's. https://www.laserfocusworld.com/lasers-sources/article/16550...

EUV LLC is the joint venture between ASML, Intel, etc funded by the Department of Energy. That’s the legal structure that they used to bring the entire deal together. TRW was a partner.

The rest of the machine is no less incredible - the EUV light is reflected off a series of mirrors (the largest weighing, IIRC, somewhere in the region of 300kg, and the largest that Zeiss have ever made) onto a mask of the target pattern, which is rapidly shuttling back and forth, onto another series of mirrors, and finally focused onto the wafer stage, which is also moving precisely in step with the mask, to expose the photoresist on a single die which is on the order of 10mm square.

And each wafer goes through multiple cycles of this, so not only does the machine rapidly create features at the nanonmetre scale with incredible accuracy, but again and again in exactly the same place after removing and returning the wafer.

Oh, and the wafer stage is pretty fun too; it uses an electrostatic chuck that induces a charge in the wafer and therefore holds it firmly in place without needing any kind of suction that could distort the wafer.

Promotional fly through video from ASML here: https://youtu.be/h_zgURwr6nA

Asianometry video about the stage: https://youtu.be/1fOA85xtYxs

Another short video about the light source: https://youtu.be/9VDJMivfhGU

They were a big customer of a former company I worked for. I had a session at our executive briefing center, and myself and everyone else was just floored and geeking out on their tech.

While EUV lithography machines are surely a contender, the most complicated machine every made is likely the Large Hadron Collider (LHC). A 27 km tunnel in which protons (with cross section about 10^-28 square meters) collide head on. Hard to imagine this amount of alignment is possible.

True, perhaps if we call EUV Lithography the most complicated fabricator machine that actually makes something. Whereas LHC is a scientific experiment.

One differentiator is that it's factory produced (not just a one off)

It’s not complex just gigantic, repeated sections. Definitely not downplaying it but definitely euv is damn complex !!

It is nothing compared to the EUV machines and not as complicated either and with a utility value much less than even regular DUV machines.

It appears that ASML alone knows the correct rituals to keep the Machine Spirit cooperative.

> Very few firms have the resources/know how to even run the machines - which is what makes TSMC so important.

I dispute that this is simply what sets TSMC apart. The process of designing transistors and chip "IP" (the term used) such as 2.5D stacking technologies and the like (which is where coordination with EDA companies like Synopsys, Cadence, and etc. comes in) is just one thing. Then to prepare the photomask (I'm sure there are partners, but still), and operate the whole clean room facility around the input and output of wafers and other materials into these ASML/other fabrication machines is another thing.

ASML seems to strike deep at personal identities: it's the only European player in space dominated by American and Asian companies.

Every time anyone mentions ASML, I see comments of adulation. It's a very human tendency to hero-worship, especially fanboys who only recognize the name but don't understand the history or the ecosystem.

It's useful to remember that ASML didn't outcompete rivals through brilliant innovation in a heated market race. Instead it won mostly by being the last one standing. All the other players dropped out. Nikon and Canon made strategic decisions not to pursue EUV because it was too risky and expensive. ASML, a small Phillips spinoff, couldn't do it alone either. It took two decades and billions of investments backed by Intel and TSMC to keep ASML going before the first breakthrough -- it was a lot of persistence and incrementalism. ASML was essentially a side bet/strategic hedge by Intel, TSMC and Samsung.

(all this is covered on Asianometry)

People associate ASML with the Netherlands, but it would not have been possible without the massive contributions of the Americans, the Taiwanese, the Germans, and the Japanese.

It's like Grigori Perelman and the Poincare conjecture -- he didn't accept the Fields Medal for it because he felt that he just happened to the last person to put the pieces together (he was building on work by Thurston, Hamilton, etc.).

We see ASML as this amazing Dutch company, but we forget all the other players were critical to making this singular company possible (only because it happened to be the only one standing, not because no one else is capable -- this is the great misconception).

It seems for some reason that it’s only important to keep these things in mind and mention that it contains American tech, is an international effort, etc, when discussing some successful European company. But when discussing American companies there is never any talk about such things. Why can’t ASML be seen as an amazing Dutch company even if they also use tech from other countries?

The American companies that get talked about most here are mainly software companies. The "supply chain" for software is far less interesting and easily hidden away. Who cares about a litany of Indian outsourcing companies (apart from those who were outsourced i suppose)?

Their supply chains consists of Indian and Chinese Univerities.

Because it's not the case.

ASML is truly an international integration company. This type of integration requires deep expertise and tacit knowledge. But they don't manufacture most of their critical components.

The way most narratives make it sound is that ASML is out of this world. No, it is not. It is unique because it is the only one standing in the market, after a 2 decade period of investment.

It has a compelling moat, but that moat is not entirely technical (though it is substantially technical -- there is a lot of tacit knowledge on integration at that level of precision). It's the supplier network lock in, and sunk cost lockin from chipmakers.

The system is Dutch. Any component of anything is always trivial.

If the Dutch want to design a new EUV light source, they can probably find someone to do it. If they want to find someone to grind mirrors to match those from Zeiss, maybe they can find someone who worked on the ESO or similar, and similarly with everything else. Think of it like how the F-35 is a Boeing project, even though many companies make components.

The integrator is the maker, not the component manufacturers.

I don't think the US could have made something like ASML. Baumol effect due to the successful firms, different attitudes, etc. There's after all a reason why the others dropped out.

My impression is one important strategic factor is that ASML was not build using venture capital -in which the US is dominant - but relied mostly on government subsidies. Once ASML was a strong player they got a lot of investment from American customers (intel et al) to invest, for example in UEV development. This is a funding approach that doesn't work well in most markets but seems to fit well with the deep speculative research typical of these kinds of systems. As some other users have noted, the US was the initial leader in UEV research (in the very early phases), but the technology was deemed too speculative to attract significant market funding.

Yes. Especially Germans can make these kinds of long-terms plans.

I saw one thing that made it obvious to me that they were only people who could possibly develop a certain technology which hasn't been commercialized. Basically, they were doing research on details that they from their analysis could determine were necessary, but which nobody else had even thought about yet. It took me half an hour of a guy who had done his PhD on it explaining until I understood why they needed it. I think his work was paid for by industry though.

ASML is not German, and I'm not sure I understand the rest of your comment.

You’re partly right, but I think you’re downplaying ASML’s achievements, especially when it comes to EUV. The ‘internationalization’ you mentioned was definitely one of the key reasons for their success. They were smart to position themselves as a global system integrator early on, stepping away from the vertical integration that their competitors mostly stuck with. (And let’s not forget, ASML was already a big player before EUV even became a thing.)

That said, this isn’t something unique to ASML. It’s just how global companies operate, especially in hardware. Take Tesla as an example—these companies are inherently multinational. They rely on acquired patents, parts and expertise from around the world, and hire international talent to stay competitive.

As someone else pointed out, it’s interesting how often this comes up in ASML discussions here, especially from American users. Maybe it’s because admitting reliance on another country—particularly those “arrogant Europeans” Americans often feel both superior and insecure about—kind of messes with the U.S. dominance narrative.

> It's useful to remember that ASML didn't outcompete rivals through brilliant innovation in a heated market race. Instead it won mostly by being the last one standing.

They were the last one standing because of their brilliant innovation. It was order of magnitudes harder to get EUV from the lab to an actual working machine than anybody thought in the beginning. There is a reason others gave up.

Fair enough, I can see it was a team effort. But its still interesting that all that work has focused on one place without leaking and it now has global strategic importance.

And being the last one standing must mean something - good at negotiating with funders, tenacious, perceived as a good long term bet etc.

For a long time the Netherlands were somewhat of a tax haven for companies relying heavily on R&D.

Cymer was the best purchase they ever made. But it also means they're totally beholden to US export controls. That light source is made in San Diego after all.

A part of reason No one else can make these machines is because US has tight control on who's allowed to even develop that tech.

There were 9 manufacturers with access to the initial technology, but only one managed to productionize it iiuc.

Reply on news.ycombinator.com