Quotes
The week of the competition, a lone ASML technician was already on-site. He decided to fire up the machine, just to check everything was in order before the AMD experts set about their tests. To his horror, he discovered an embarrassing fault. The electric motors were warped, and the test machine was projecting deviant chip patterns. The machine did not stand a chance. Without telling AMD, the teams in Veldhoven set to work. Tension was high: they had only thirty-six hours to come up with a solution. The next day, an ASML engineer flew to Silicon Valley with a modified version of the motor parts in his suitcase. With a ‘borrowed’ access pass in hand, the crew snuck into the chip factory in the middle of the night to screw in the motors. The wily scheme paid off. The machine worked and ASML eventually won the contract, promising the delivery of twenty-five machines by 1987.
Around 150 years prior to this, researcher Carl Zeiss marketed his first microscope from his workshop in the German town of Jena. After the Second World War the city fell under the control of the Russians, however, and would later be demarcated as part of East Germany. Through Zeiss’ supply of lenses to the German army, the Allies had become keenly aware of the company’s strategic importance. US troops occupied the Zeiss factory in April 1945 and spent the following three months seizing its patents and tools. By 1946, the Americans had deported over 70 researchers and craftsmen from Zeiss to Heidenheim, then part of the Allied zone. It was from here that they went to work in nearby Oberkochen. The Russians then moved in; just as aware as the Americans of the importance of good lenses and mirrors for their weapons and spy systems, they looted what remained of the Zeiss plant in Jena.
‘Martin is quite unique, if a little autistic at times. But if you gain his trust, he will accomplish remarkable things.’
The young ASML’er could erupt at any moment if he felt someone was talking nonsense or beating around the bush. You never had to wait long to know exactly what Martin thought. As far as Martin was concerned, engineering consisted of solving problems, not avoiding them. Doing the job properly meant seeking them out so they could be dealt with now rather than left to cause problems later. To this extent, he really was suited to be a physicist. For both physics and engineering the overriding priority is to find out why something does or does not work, even if it means asking difficult questions or breaking with accepted norms. Nothing is off limits: why is a certain number that number? Why not double or half? During technical discussions, Martin would probe the weak spots of every argument. He also had an uncanny sense for when someone was deflecting, never hesitating to confront them and pry them open, regardless of whether this was in front of colleagues, customers or suppliers. And if anyone thought they had a better idea – bring it on. You just had to be prepared for how it would be received if Martin thought otherwise. His second nature was to challenge the world around him, and this brazen and ruthless attitude would become one of the pillars of ASML.
‘That guy should never hold a presentation again, all he did was insult our customers.’
Tempers ran high during discussions in Seoul. The Korean company was marked by an uncompromising ruthlessness – given the exceptionally tight margins of the memory chip game, everything was about efficiency. Best of luck to anyone responsible for a delay, as might happen if a lithography machine jammed. The ensuing meetings were more like interrogations. Passports were taken upon landing in South Korea, after which the ASML’ers were whisked away for ‘negotiations’. In between the yelling, everything from chalk markers, plates, ashtrays and coffee cups flew through the air. Anything within reach was permissible.
Sparks never failed to fly when the technical directors of Zeiss and ASML crossed wires. They both lived for head-on confrontations, as they thought all good engineers should.
He [Eric Meurice] took pride in being able to tell people they were ‘incompetent’ in whatever language they preferred.
‘What’s it like developing machines for ASML? Might as well be developing crack, it’s that addictive,’ says Markus Matthes, former head of D&E. ‘We work so fast that we don’t even have the specs ready when we start to design. Now that’s a thrill.’
But as Frits van Hout explains, it’s not a case of one-size-fits-all when it comes to engineers. ‘A mechanical engineer can create and design structures, whereas an electronic engineer designs circuits. And physicists may “understand” everything, but ask them to actually do something and see how useful they are.’ Spoken as a true physicist.
Unforeseeable complications aren’t limited to the mechanics of the machines. Despite their best efforts, the cleanrooms can’t always keep the outside world at bay. Earthquakes and atmospheric pressure fluctuations due to thunderstorms can easily disrupt the lithography process. Or cows. Intel once faced an inexplicable drop in yield every night for a few hours, with researchers running in circles until they finally realized the cause: cow farts.
But stranger things can happen in the chip factories. About twenty years ago, a lithography system in a fab in Arizona kept experiencing malfunctions. ‘There was no technical explanation,’ according to an engineer from the factory. ‘And it wasn’t just ASML’s gear. All the systems had problems. The factory manager was on the verge of a breakdown, until someone pointed out that the factory was built on an old Native American burial ground. He decided that must be it: the fab was cursed.’ A shaman was summoned. Right away something caught his eye: the concrete obelisks planted by the entrance of the factory resembled tombstones. He gestured; if they wanted to appease the spirits, removing those would be a good place to start. The shaman proceeded, donning a cleanroom suit and inspecting the production line. After a long silence, he turned back to the manager. ‘The spirits are attracted to red,’ he hinted, and quietly left. It was decided. ‘If the spirits want red, that’s what we’ll give them.’ The manager’s eyes fell on a cardboard box, used as packaging for a lens from ASML. It was the brightest of reds. He folded it into a triangle resembling a tipi tent and placed it on the lithography machine. As if by magic, the problems disappeared. The red box stood on the machine for years – no one was allowed near it. When the chip factory later ordered a second machine, they asked the ASML’ers, ‘Could you also put a small red tipi on that second machine?’ Sure.
The Sun’s corona, 93 million miles away, is the closest place you can find extreme ultraviolet light in its natural form. On Earth, however, generating EUV requires extremely sophisticated technology. One way is to shoot a powerful laser beam at a tiny droplet of hot tin. This creates a plasma, an energy charged form of gas, far hotter than the surface of the sun. In this process, invisible light with a wavelength of 13.5 nanometers is emitted. If you can catch this with a mirror, you can direct it into a lithography machine to project a chip pattern. Sounds simple: you just have to build a sun.
Martin was keenly aware of this when he encountered Rob van Aken, ASML’s in-house architect, in 2001. ‘Hey Rob, can you make an entire factory in a vacuum?’ Van Aken was startled: ‘Not really. Don’t people need to work there? Can’t see that happening without oxygen.’ ‘Yeah, damn,’ replied van den Brink, his brow furrowed. ‘Guess we’ll have to figure this one out ourselves.’
Martin was keenly aware of this when he encountered Rob van Aken, ASML’s in-house architect, in 2001. ‘Hey Rob, can you make an entire factory in a vacuum?’ Van Aken was startled: ‘Not really. Don’t people need to work there? Can’t see that happening without oxygen.’ ‘Yeah, damn,’ replied van den Brink, his brow furrowed. ‘Guess we’ll have to figure this one out ourselves.’
‘Moore’s Law is dead!’ jeered the American analysts in their comments. They were celebrating like kids in a schoolyard: you could almost hear the ‘Naah-nah-nah-naah-nah!’ as they wrote the headlines.
He [Van den brink] was never one to count his chickens before they hatched.
‘Martin hates laziness,’ clarifies Jim Koonmen, the head of operations in San Jose. ‘He can’t stand it if you say, “I can’t make sense of the model, so I’ll just train a computer and hope for the best.” He challenges the people here: why do you choose machine learning over modelling that uses the physical principles?’ Overdependence on dark magic is risky: if you do everything with voodoo software, you’ll no longer differ from your competition. Van den Brink sees it as a valuable addition, but only as a final step to speed up reaching a solution. ‘We need to distinguish ourselves. If we don’t include the added value of our physical models, then anyone can do it.’
On January 20, 2017, with his hand on two Bibles, Trump swore to ‘make America great again’. And sure enough, the trade conflict between the US and China rapidly evolved into an all-out tech war, with Huawei as the initial target.
The CEO tried to avoid ideological discussions, but that was not always possible. During a dinner with American diplomats, he was asked about the chip industry’s role in relation to the Uyghurs in China. This Muslim minority is oppressed by the Chinese state and is being monitored with advanced electronics. ‘As the chip industry, we’re not responsible for that. We don’t arrest people,’ Wennink answered. When he bounced the ball back and asked about American school shootings and what responsibilities gun manufacturers might have, an awkward silence followed. Who was he to patronize the US?
Mapper was conceived at the Delft University of Technology, when Pieter Kruit, a professor in charged particle optics, asked his students to design a system for electron beams that could be applied in commercial lithography. Excited by the possibilities of this technology, two of his students, Bert Jan Kampherbeek and Marco Wieland, started their own company in 2000 after they graduated under Kruit. Wieland was the driving force behind the technology, while Kampherbeek was focused on the business end. As always, this meant needing to raise money to keep the business afloat.
To limit the risks, ASML insists that its suppliers get a maximum of 40 percent of their revenue from Veldhoven. This guideline is intended to keep companies from collapsing during a downturn in the chip industry. However, when faced with the chip market’s unrestrained growth and the floods of orders that come with it, keeping to that percentage is easier said than done. In these cases, ASML’s response is usually the same: just figure it out.
Every supplier has its own ‘godfather’ within ASML’s organization. However, if there are serious problems, a higher power will intervene. In 2021, when a supplier failed to resolve a recurring issue with a critical mechatronics component, Peter Wennink insisted the company should change its management. And there was no room for discussion, according to the CEO: ‘The investigation work had already been done by our team of experts. And I’m the judge that delivers the verdict.’ Sometimes ASML will help out its suppliers with an injection of capital, just like it did for Zeiss when it needed support for High NA. The last resort would be an acquisition, which happened with Cymer when that American company was unable to get the light source for the EUV machine to work. On a slightly smaller scale, in 2012 ASML acquired Wijdeven Motion, a company with ninety employees that supplies linear motors and was on the verge of bankruptcy. The supplier Berliner Glas also failed to keep up with ASML’s demands. This German family-owned company with 1600 employees makes the super-flat mirrors on which the wafers are placed. In 2020, ASML demanded that Berliner Glas make an investment of 70 million euros for the next generation of EUV. The Germans were not interested, so ASML bought the company to speed things up. If they needed to do it themselves, they would. You could see the effects at the Berlin location within a year. Berliner Glas was renamed ASML Berlin, surplus business units were sold off, new construction had already been completed, and leases for additional offices had been signed. And the cautiousness of the Germans had been thrown out the window. ‘We don’t make anything here,’ claims Martin van den Brink. It’s a slight exaggeration, but his point is clear: ASML only assembles the lithography machines, and the wide range of suppliers take care of providing the parts. This model has been essential to keeping everything moving. It granted ASML a remarkable ability to recover in the first decades of its existence, while its vertically integrated Japanese competitors were still stuck making most of the components themselves. Just like a lightweight boxer can bounce back to their feet faster than a heavyweight.
ASML accepts that its organization does not always run optimally. Frits van Hout even has his own formula for this: ‘Things you do a thousand times of course need to be standardized. But if eliminating an inefficiency takes up so much time that it stops you from doing other, more important things, then you should just tolerate that inefficiency. As a good Calvinist, you might ask, “Could we not have prevented it?” But we say, just let it go.’
Van den Brink consistently positions himself as the champion of ASML’s free technical culture, which he believes should never be allowed to get stuck by procedures and processes. The mere mention of KPIs, or key performance indicators, during a meeting is enough to send him storming out of the room. He finds it completely trivial – ‘I’ll be in the room next door talking about the actual content. Whoever wants to talk about KPIs can stay here.’
Finding mistakes is the number one pastime at ASML. And it is never just about technology. Peter Baillière notices colleagues even complain about the ‘wrong’ fonts being used in a PowerPoint presentation. He has his own baptism by fire when he takes over the HR department in 2018. During his first presentation to the board of directors about complex European privacy regulations, he is abruptly cut off. He doesn’t fully understand the details of the legislation, and at this level, that is simply not good enough. ‘Here at ASML, you always need to be able to go into depth,’ he is told. This is the big league, not a tea party. The new director of HR walks out of the meeting, convinced he is about to be fired on the spot.
Many ASML’ers have stories of similar near-death experiences. You have to suffer through it, pick yourself up and go on to win a few of those battles. It is a way to see if you are cut from the right cloth: full of substance, and short on ego. At ASML, you are so often confronted with something that you have no idea about, that you never bother to brag about the things you do know about. ‘ASML’ers can spot authenticity a mile away,’ according to Peter Wennink. They value someone who puts the company first, not themselves.
It became an organization ready to leap into action at a moment’s notice: everything needed to be done yesterday, and would be outdated by tomorrow.
As he departed, he whispered some words of wisdom: ‘Gentlemen, persevere.’ And they did.
The people around him realize that Martin’s mind leaves little room for everyday matters. That’s hard when you are thinking fifteen years ahead.
Moore’s Law has run its course. At the current pace, doubling the transistors every two years is simply not sustainable. According to ASML’s director of research Jos Benschop, progress is limited by economics and not by physics. Unhindered, he figures it would still take another forty years before an atom would no longer be able to pass between the lines on a chip. If you believe the marketing departments of semiconductor manufacturers, chip patterns are still rapidly shrinking. The latest craze is to describe their tech in ångströms – a tenth of a nanometer. However, this terminology hides the fact that the chip industry is fast approaching the point where the investments required to scale further down become too large to recoup – that’s the real limit. And if chip manufacturers can no longer add value to Hyper NA machines, ASML stands to gain nothing by producing them. In layman’s terms: it is getting too expensive.
If a small gesture will smooth things over, you never have to think twice about making it, says one manager. ‘You’ll never get in trouble for giving away a lens worth a few million euros. But if that same customer calls Martin because you refused to cooperate, you get your butt kicked.’
‘This is a tough one,’ Martin says, wiping away his tears. He walks to the coffeemaker hidden behind his office. A cappuccino, like always, and a deep breath. ‘For me this was never a job. It never felt like work.’