Discover more from The Asianometry Newsletter
The ASML First Half 2021 Mega-Review
If you want to watch the video first, you can below:
Anyone interested in a video about some of these competitors? Let me know.
If you would like to support the channel, consider subscribing to the Patreon. Members of the Early Access tier get to watch videos first before they released to the public. I just recently released a video on Nvidia’s rise in the graphics card industry.
2020 was a crazy year for ASML. Between COVID outbreaks and worldwide economic instability, ASML found itself juggling whipsawing customer demand.
TSMC lost Huawei as a flagship customer. As a result, they did not need so many EUV machines and they cut their allocation.
Then, Intel delayed their roadmap by 6-12 months, and the EUV order book shrank yet further.
Suddenly, ASML found itself making more EUV machines than their customers needed. So they cut their 2021 budget. They told their entire supply chain to hunker down.
Well, you know what happened next. The economy came roaring back and everyone needs chips now.
EUV is back on the menu, boys. And everything else, for that matter. If it can make chips, then the customers need it. ASML goes into 2021 trying to figure out how to make it all happen.
In this video, we holistically review the Dutch tech giant - as of this writing one of Europe's top five most valuable companies - and how they are handling the year 2021 so far: their financial results, customers, competitors, product roadmap, and more.
The Big Story of 2021
2021 continues the big economic trends driving the semiconductor industry forward. A combination of broad macroeconomic factors like the work from home trend means that demand vastly outpaced supply even as company management was busy moving up those demand expectations.
To help customers meet that demand, ASML has been digging up ways to help their customers be more productive. The start of that involved pushing out new software updates, while at the same ramping up the supply chain.
ASML also pulled out a number of stops to get machines out to their top customers. They drew down on their inventory to speed up their factory output. And they rejiggered machines where they were available, buying back two machines from the unnamed customer who delayed their plans - likely Intel - and resold them.
Regardless thus far, the company has sold about 9-10 EUV machines per quarter. Their output this year limited by the pace that critical suppliers like Carl Zeiss can go at. With the time between EUV module production start and installation being over a year, not much can be done.
For 2022, ASML plans to make and sell 55 EUV machines. Many of these will work better than the 50 the company plans to ship in 2021. They are NXE:3600D's, which can process 15-20% more wafers than the previous generation NXE:3400C's. The first ones shipped in the second quarter. Looking into 2023, ASML wants to have enough capacity for over 60 EUV machines.
But just as important as the sales of EUV machines has been the sales of DUV. They are nowhere as expensive or as flashy as EUV. But as many media reports note, the ongoing chip shortage is worst in chips made with lower process nodes. And those nodes use DUV machines.
For whatever reason, this surge in DUV demand - which management sees as being "stronger for longer" - has seemed to catch the Dutch giant off-guard. And it opens an opportunity for its competitors to jump in. More on this later.
ASML earns two types of revenue. The first type represents the money that ASML earns from selling a system. Mostly lithography systems.
These machines are sold to two types of customers: Makers of logic chips - like the microprocessors you find in your phone or desktop - and memory chips.
The second type of revenue is system service and upgrade revenue. These are pretty complicated machines, and require a lot of maintenance. Sadly enough, this 24/7 engineer effort is not free.
Company estimates find that service revenue after the sale can cumulatively total over 50% of the original system's sale price. This is measured over the lifespan of the system, usually 20 years. It is quite significant.
Anyway. Prior to 2020, the proportion of systems sales revenue to service revenue was about 75-25. Since the start of the chip shortage however, that number has looked to be more like 70-30.
As I explained earlier, this is due to clients and customers spending more for software and service upgrades to make their existing machines more productive.
ASML machines are more than just hardware. Because the whole thing has to be automated from start to end, the software is critical. And also very complicated, with over 45 million lines for the EUV machines.
ASML customers can buy and install these upgrades to get fast productivity improvements without having to stop the line to install, integrate and inspect new hardware. And they did just that.
After everyone updated and upgraded, the proportion fell back to the more normal 75-25 in the second quarter of 2021. From then on, the adjustments to speed up factory output kicked in, delivering more revenue than expected.
Finally, the thing that really caught my eye is the surge in net bookings of lithography systems. In the first half of 2020, the company had 107 systems ordered for future delivery. A year later, that number has swelled to 287, worth 13 billion euro. 167 of those bookings came in the second quarter alone. It’s indicative of just how much demand the company has to fulfill in the coming years for their customers.
ASML makes a product that few others can make. But it is also true that there are only a few customers out there who can make use of it.
In 2020, ASML's largest customer - which goes unnamed in the financial statements - represented 31.4% of their total net sales. In 2019, their largest customer represented 39.7% of total net sales. Like I said, the financial statements don’t say who those largest customers were for any single year, but we can make an educated guess.
The company has been working on spreading out its customer lineup. In the second quarter, I noticed that the proportion of gross shipments going to Taiwan and South Korea shrank from 87% to 75%. With those points going to China, the US, and the EMEA area.
A potential EUV sale to leading Chinese foundry SMIC fell through due to licensing and national security concerns.
But other customers have stepped up to fill the void. For the first time, major memory makers like SK Hynix and Micron have begun integrating EUV into their fabs.
SK Hynix first with their fourth-generation 10nm process.
And with Micron planning on joining a few years later in 2024. With the memory market growing faster than anticipated, EUV revenue from memory should expand as quickly, outpacing the rest of the order book.
ASML's closest competitors in the lithography space are the Japanese tech companies Nikon and Canon.
Only ASML has successfully shipped an economically viable EUV machine. We all know this. But EUV is not needed for process nodes greater than 7nm. For those, manufacturers use DUV machines, DUV standing for Deep Ultra Violet. These products use the 198nm wavelength or higher to etch wafers.
And it is here where ASML encounters more competition. For all the attention that EUV gets from the media - and this channel - the majority of ASML's revenue so far this year has come from DUV technologies.
The DUV space is split into two sub-categories: Dry and Wet, or better known as immersion lithography.
Nikon offers wet DUV, while Canon offers dry DUV. Immersion DUV systems are more advanced than dry ones, so Nikon's offerings are considered more "high end".
Nikon has been struggling as of late. Last year, they cut 2,000 jobs or 10% of their workforce. This is largely due to downturns in the camera business.
But with the job cuts, their semiconductor division is also trying to cut costs so that they can sell more of their immersion DUV machines at a disruptive price, with a special focus on China. But the pandemic has made this difficult thus far.
Canon is in better spirits. Their dry DUV technology is far from the leading edge, but it costs a lot less and right now is in the right space at the right time. And it is not a bad product too - it can process more wafers per hour than ASML's current dry DUV offerings.
Thus, the demand for this technology is quite strong. Revenue is expected to grow 36% year over year. Canon expects to ship 134 DUV machines in 2021, up from what was shipped in 2020 and 2019.
As I said earlier, there is some indication that Canon's insurgent burst of sales growth caught the Dutch giant somewhat off-guard. ASML prides itself on having a superior product in the market - and that is not the case right now.
They recently announced that they are prepping a dry DUV product offering capable of processing over 300 wafers per hour. That's more than Canon's offerings, but it won’t be out until late next year. Like they say, the best ability is availability. So until then, the company is missing out on sales in this niche.
I don’t really think there’s a real possibility of disruption from the low end right now in this sector. But it does makes you wonder if there is a place for up and coming low end lithography players. Canon especially, but also the emerging players coming out of China.
The most formidable of these Chinese lithography players is Shanghai Micro Electronics Equipment, or SMEE.
Currently their systems are for the 90nm, 110nm, and 280nm processes - but they are developing an immersion DUV system said to be capable of 28nm. Assuming that thing ships - supposedly by the end of the year - and actually works, then it would be a huge boon for leading Chinese semi makers like SMIC, Grace Semiconductor, and Yangtze Memory.
ASML and its supply partners caper are working on their next breakthrough to keep Moore's Law chugging full steam ahead.
And that something is called "High Numerical Aperture EUV".
Okay Shakespeare probably didn’t write that last part. At least not in the manuscripts I’ve seen. But it is a real thing. A detailed discussion of how this new technology works is beyond the scope of this little video. It probably belongs to a Zeiss update. But here is the gist of why it is such a big deal for ASML and its customers.
The attainable structure size in photolithography - as in how small it can etch things onto wafers - depends on the resolution provided by the tool's optics. This resolution, sometimes also referred to as the critical dimension, is calculated using the famous Rayleigh formula.
Named after John William Strutt, 3rd Baron Rayleigh because of his work in setting telescopes.
And just want to add. My man Rayleigh literally has a Wikipedia page of things named after him. What a baller. Anyway, Rayleigh's formula. It goes like this.
CD = k1 x λ / NA
Where k1 is a process factor determined by the actual optics system. What illumination techniques are used and the system's stability. It has a physical limit of 0.25.
Lambda (λ) stands for the light wavelength. Visible light has a wavelength of about 400-700 nm. Ultraviolet light has shorter wavelengths.
And finally. NA, or "numerical aperture", a measure of how much light an optics system can collect and focus.
EUV shrank the critical dimension by shrinking the wavelength from the 193 nanometers used in DUV to the 13 nanometers used in EUV.
The next step to continue shrinking the critical dimension would be to raise the numerical aperture. Currently, the optics systems in today's EUV machines have an NA of 0.33. ASML and Zeiss are working on a new optics system with an NA of 0.55. This new system would allow the next generation of EUV to have a critical dimension of just 8 nanometers.
The company is continuing to put a lot of work into this new product. And they seem pretty excited about it. They say it will be ready by 2024 but who knows.
It is kind of funny. I remember when TSMC started to get into the news, all the chip hipsters started peppering us with comments saying that ASML is even more important than TSMC. They were the One True Monopoly.
And now that ASML has started to gain broad recognition with the general public, the chip hipsters are going deeper. ASML might be delivering these big machines but wait ... now it's actually Carl Zeiss who's doing the real heavy lifting! That made me chuckle.
Anyway, I want to close with something I noticed at one of the investor calls. They were talking about the major trends driving the semiconductor manufacturing equipment market upwards. The first two are the COVID up-cycle and the explosion of digital innovation.
The third is the unstable geopolitical situation and the national sovereignty drive towards chip self-sufficiency. Governments in the US, Asia, and Europe talking all about this. Here is the quote:
"Well, that will lead to higher capital intensity because it’s decoupling as a worldwide eco-system. But it also leads to some capital inefficiency. Well there is a beneficiary of that capital inefficiency and that’s us."
ASML is basically saying that governments are going to splash out a lot of money to chase self-sufficiency. And some of that money is going to be wastefully spent. Wastefully spent on ASML. And they painted that as a good thing for the company.
The statement must have freaked someone out because they kind of walked it back in the latest earnings call. Saying that they believe "this potential inefficiency will be managed rationally by a few very large manufacturers which are crucial in building this additional infrastructure". But it is an interesting thing to admit, and it makes me wonder who will end up bearing the final costs of those inefficiencies.