A Q&A with Tony Yen, VP of Tech Development at ASML

ASML, headquartered in a small Dutch town, is the world's dominant supplier of advanced chip-making equipment
article cover


· 7 min read

ASML is probably the most important hardware company you haven't heard about.

Headquartered in Veldhoven, a 45,000-person town in the Netherlands, the Dutch company is the world’s largest supplier of photolithography systems for chip manufacturers. ASML has a monopoly on extreme ultraviolet (EUV) machines, the most high-end hardware used in high-volume manufacturing by TSMC and Samsung.

We recently interviewed Tony Yen, ASML’s VP of tech development. NB: The following Q&A has been edited for clarity and length.

Starting at a high level, I’d love to hear how you’d describe ASML and what it does. What’s the elevator pitch?

We make a machine, basically, an exposure tool to print chips that everyone needs. TSMC, Samsung, and Intel are the three big customers we sell the bleeding-edge machines to.

What were the ingredients required for ASML to emerge as the dominant leader in this space? Obviously barriers to entry, upfront costs, and technical skills. But why can’t anyone else replicate this?

It’s rooted in the Dutch culture, the determination of the company, the makeup of the engineers, the management, and our customers.

The Netherlands’ population is only 17 million, but its citizens are highly educated. We have lots of PhDs and Master’s degree graduates working at ASML. The company has become very international. You go to the company cafeteria and hear conversations in all sorts of languages: Chinese, Russian, Eastern European, Spanish, Italian, German. They’re all there.

They’re attracted to the company culture, which is very no-nonsense, results-oriented, and technically driven. Several upper-level management executives have PhDs or Masters in physics, and the physicists play an important role at ASML.

We make machinery and the complexity of understanding the science that powers them is very important. Our latest machine is the EUV scanner, which is the engine of the most advanced chipmaking.

When did you start thinking about EUV?

We started in around the year 2000 and made the decision to develop these scanners for high-volume manufacturing. The Japanese didn’t do that, because they saw the systems would be very complex. We weren’t 100% sure it would work.

We have a German partner, ZEISS, that makes optical systems and dates back to 1850. They make optics exclusively for us, because we fund a lot of their development, so they can only sell semiconductor-related exposure products to us. Along the way, we acquired Cymer in San Diego, another critical piece that was needed to make EUV work.

To what extent are you pushing your product out there versus working preemptively with customers, and building around their needs and specs and whatnot?

It’s mutual from the early development stage. They take our early development tools, at low-reliability. Those machines can be down 50% of the time, but are able to print the small features that you need. You can get development wafers to produce devices for your R&D. That’s how we start. At first, it’s never a production machine.

The first EUV machine arrived at Samsung in 2010 and TSMC in 2011. Those were the early machines. We went through a couple generations. The next generation was still a development machine, and then the third generation became a high-volume manufacturing machine.

You have to have faith. Our customers know they’re in it for the long haul. For the bleeding-edge customers, we put the early development tools there and physically co-develop the technology. That’s not how it works with the not-bleeding-edge customers.

Can you give me the state of play for seven nanometers, five, and then three? Can you explain from your end what a migration to a more advanced node process looks like?

The most bleeding-edge technology in high-volume manufacturing is the 5 nanometer node, being manufactured at TSMC and Samsung. Intel fell behind on 5nm. They’re not even doing high-volume manufacturing on the 7nm. But you have to be careful here: Intel has higher design rules, so Intel’s 7nm is like TSMC and Samsung’s 5nm.

I know it must be hard to get that nuance across, because manufacturing processes aren’t pound-for-pound the exact same thing. As for the second part of the question, when a customer is moving on to a more advanced manufacturing process, what does that look like on your end?

Keep up with the innovative tech transforming business

Tech Brew informs business leaders about the latest innovations, automation advances, policy shifts and more to help them make smart decisions.

The development process, from day one, is five years.

Can you say what’s going on at your company over that cycle?

We have development milestones, starting with design. I cannot disclose details, but yeah, a product first needs to be planned and then approved by the board of management.

You’re ordering an EUV tool that is already high-volume manufacturing. If you are ordering from a country that does not require export control, an EUV order has about two years of lead time.

I’m sure you get this question a lot and that it may be a bit aggravating. But while I have you, what are your thoughts on Moore’s Law?

A lot of people say Moore’s law is dead or slowing down. It’s not dead, it hasn’t slowed down. We believe Moore’s law could continue for another 10 years, or even a lot longer than that. It’s two-dimensional scaling—making the chips smaller and smaller—that can go on for at least 10 years.

People are talking about stacking components, putting them on top of each other, which is called a 3D integration. TSMC and Samsung have talked a lot about it. 3D integration is happening, but 2D in-plain shrinkage can continue, we believe, for at least 10 years. And our next-generation EUV machine is being developed—that’s in the pipeline.

When I see ASML mentions, they’re in the financial press or investor reports. I’ll read how buzzwordy technologies like AI and 5G will be tailwinds for you. Are you thinking about, say, new machine learning models or 5G standards and anticipating how that could affect demand? Or is that too far downstream for you?

It’s not too far down. 5G chips are being produced right now, by Apple, Samsung, Qualcomm, and more. These days, Huawei is on the Entity List, so its position is being taken over by Xiamoi and other brands in China. Qualcomm’s highest-end chips being made by Samsung use the 5nm mode. That’s a 5G chip for 5G communications, it’s already in high-volume manufacturing, and it’s part of the reason we’re enjoying good business today.

You’ve hinted at this a little bit so far. How do you navigate geopolitical tensions? I know ASML technology can be classified as dual-use and fall under US export control laws.

I'm not a company spokesman on this, but we definitely want to comply with all the laws. If part of our machine is US-made, and somebody is put on the Entity List, we comply with the US export law.

Brittney Wolff, ASML US corporate communications manager, jumps in: This may be a good one to toss over to me. Obviously it’s something we have to be very conscientious about, because we do business in a lot of countries. We respect and follow the rules and regulations everywhere we do business, and we’ve never had any challenges there.

Anything else either of you would like to add?

Wolff: We joke about being quoted in one outlet as a relatively obscure company, because we’re really not that obscure. We have 26,000 people now worldwide, and started as a spinoff of Phillips. We’ve talked a lot about hardware, but we have US offices—like the Silicon Valley offices where Tony is—focused on the software. You need a lot of software to make these machines work at levels of nanometer precision.

Yen: Absolutely. For the best machines we make, we need software and machine learning to control the machine. We make the machines that make the chips that do machine learning. *chuckles* And we use those chips ourselves for our software, so it’s full-circle.

We think of ourselves as the linchpin in [the tech world]. Without ASML, there wouldn’t be any of the newiPhones today. We take this responsibility very seriously. If we screw up, there are huge consequences. We’re determined, ambitious, and cautious on our roadmap.

Keep up with the innovative tech transforming business

Tech Brew informs business leaders about the latest innovations, automation advances, policy shifts and more to help them make smart decisions.