Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/asml-reports-2021-euv-shipments-sliding-into-2022.13631/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

ASML reports 2021 EUV shipments sliding into 2022

Fred Chen

Moderator

Peter Wennink -- President and Chief Executive Officer

I think our capacity, our capability internal in the Netherlands, in Veldhoven, to build 50 systems is there in terms of people and square meters. Now we have to build those systems out of modules, which we don't produce. It's in the supply chain. And it's just a reflection of what happened last year in Q2 and Q3 where -- as clearly our key foundry customer came back and said, listen, our key customer for N3 is now blacklisted. So we cannot ship. So we need to adjust our 2021 outlook for EUV systems. Which was followed by another customer and said, well, we're going to delay the roadmap, which also means that this will be pushed back one [Phonetic] year, which actually led to a situation where we actually reduced the number of planned system 2021 for EUV, because customer said these are the two reasons and they were two big customers.

So, what we did, we went to the supply chain and said, sorry, we need those lenses and lasers, very expensive pieces of equipment, we need them actually later. You need to realize that the integral lead time between the installation of an EUV tool and a start of a module production, that is 20 months. So when you at seven, eight months -- then finally customers come to the realization that is not as bad as they've thought and they want those machines, and we have an issue with getting the modules on time and that's the only issue. The only issue is just to resolve, it's a function of the fact that our customers changed their mind in Q2 and Q3 and then rechanged their mind back in Q4 and -- today. There is nothing we can do about it, which actually means that we're all prepared to do 50 units next year, as is in 2022. It will just shift to 2022. So it's -- there -- so as you could say it's supply chain limitations by design, because the customers told us, we don't need them, and then coming back and said, oops! we might have been wrong.


Here is the revenue recognition per quarter over the last three years:
ASML litho units sold per quarter (Q1 2018 - Q4 2020).png
 
How many EUV will be produced in 2021 from ASML? It will be quite possible less than 40 this year.
 
ASML's supply chain constraint leads to another question. Can Intel acquire enough EUV machines in time for their 7nm 2023 rollout? Or will it push Intel's 7nm HVM into 2024?

I'm wondering if Samsung and TSMC's EUV acquisition will negatively delay Intel's product schedule, in addition to all those technical problems Intel has solved or will need to resolve.
 
ASML's supply chain constraint leads to another question. Can Intel acquire enough EUV machines in time for their 7nm 2023 rollout? Or will it push Intel's 7nm HVM into 2024?

I'm wondering if Samsung and TSMC's EUV acquisition will negatively delay Intel's product schedule, in addition to all those technical problems Intel has solved or will need to resolve.
You are assuming that TSMC and Samsung are ahead of Intel in the line. It’s not at all clear from the article who might have to wait. From the article, it’s clear that TSMC canceled some orders because they lost Huawei as a client. What happened then is unclear. They probably came back to ASML with new orders (because of newly acquired Intel business?) but what is the pecking order now?
 
I would speculate Intel's 2023 rollout to be limited. In ASML's earning call, Wennink mentioned that ASML had transferred tools originally designated for "customer A" to "customer B" and "customer C". The big question is who is "customer A", "customer B" and "customer C"? Given TSMC's recent capex expansion and Samsung's recent request to ASML for more EUV machines, it would seem that Intel is "customer A".

That said it may not even matter. Intel seemed to acknowledge that even IF they were to successfully introduce their 7nm node in 2023, it would be one node behind TSMC. If AMD continues to introduces higher-performing CPUs, they will continue taking consumer & HPC market share from Intel. There's a reason why it's nearly IMPOSSIBLE for consumers to purchase Ryzen 5XXX right now.
 
Can a Nikon or cannon be modded to do the same job?

They're not using the euv machines as is the machines are being modded.
 
You are assuming that TSMC and Samsung are ahead of Intel in the line. It’s not at all clear from the article who might have to wait. From the article, it’s clear that TSMC canceled some orders because they lost Huawei as a client. What happened then is unclear. They probably came back to ASML with new orders (because of newly acquired Intel business?) but what is the pecking order now?
No, I didn't assume TSMC and Samsung are getting preferential treatment. I do assume Intel hasn't owned and operated as many EUV machines as TSMC or Samsung do.

I also assume ASML treats these three companies equally.

With only 50+ units of EUV can be produced by ASML each year, will Intel get "enough" units to support their 7nm 2023 production?

Remember, our assumption is ASML treats TSMC, Samsung, and Intel equally. Intel can't just push TSMC or Samsung aside to demand a lot more EUV units. Then what can Intel do? Intel can delay or scale down their 2023 7nm schedule. But it will put Intel in a very bad situation.
 
Last edited:
No, I didn't assume TSMC and Samsung are getting preferential treatment. I do assume Intel hasn't owned and operated as many EUV machines as TSMC or Samsung do.

I also assume ASML treats these three companies equally.

With only 50+ units of EUV can be produced by ASML each year, will Intel get "enough" units to support their 7nm 2023 production?

Remember, our assumption is ASML treats TSMC, Samsung, and Intel equally. Intel can't just push TSMC or Samsung aside to demand a lot more EUV units. Then what can Intel do? Intel can delay or scale down their 2023 7nm schedule. But it will put Intel in a very bad situation.
"Equal treatment" does not mean that they all get the same number of units every year. If TSMC canceled their order for 2021 and Intel ordered, say, 2/3 of available volume, then Intel may get, say, 30 units and TSMC none (for a given year). I do not think ASML can tell Intel: sorry, TSMC changed their mind and we are going to give some of your units to TSMC now. Obviously, we do not know what exactly is going on and who ordered/canceled what and when.
 
Equipment can be recycled and not in the way of being scraped. New lasers, lens, software is added to old equipment all the time.

Questions are does asml's equipment need to be nodded before operation and to what extent? Do they need asml and can they go with someone else?
 
ASML is also changing EUV tool models again, which is slightly self-defeating. The 3400C will be stopped this year, and next year will be the 3600D, so of course, there won't be such a big enthusiastic rush for 3400C (the current best model) anymore this year.

Peter Wennink -- President and Chief Executive Officer:

...And don't forget we have a higher productivity tool coming out, the 3600D in the second half of this year, which has 15% higher productivity. So with 2022 only being Ds, you already get a 15% higher wafer capacity out there...
 
Last edited:
Is there any such thing as a "soft" cancellation? It would be fascinating to see the actual cancellation letter and what lawyers may have added to it.
Also, the difference between models may be mostly source power. It would be tricky to change the optics much I suspect. Anyone have insight on the difference between models.
 
PS As I remembe, Intel ordered the first 7 EUV machines way back when. Don't know if they ever took delivery, since they said they were going with 193i.
 
"listen, our key customer for N3 is now blacklisted. "

So how does this affect N7 ? N7 & N5 are essential the same and N3 represents a quantum jump in technology. There is no specific talk about how the drop in demand for N3 has a Knock-On affect for N7 supply. Am I missing something ?
 
Is there any such thing as a "soft" cancellation? It would be fascinating to see the actual cancellation letter and what lawyers may have added to it.
Also, the difference between models may be mostly source power. It would be tricky to change the optics much I suspect. Anyone have insight on the difference between models.
From https://semiengineering.com/whats-next-for-euv/:

“The 3600 has stage improvements, minor lens improvements and sensor improvements. We’ve made incremental improvements that give it a better overlay and higher productivity, but it’s functionally the same design,” said Michael Lercel, director of product marketing at ASML."
 
"listen, our key customer for N3 is now blacklisted. "

So how does this affect N7 ? N7 & N5 are essential the same and N3 represents a quantum jump in technology. There is no specific talk about how the drop in demand for N3 has a Knock-On affect for N7 supply. Am I missing something ?

It looks like that applies to Samsung, who may or may not be one of the two customers who first pushed out their orders then wanted them back. However, it was implied these two customers were customers B and C.
 

Peter Wennink -- President and Chief Executive Officer

I think our capacity, our capability internal in the Netherlands, in Veldhoven, to build 50 systems is there in terms of people and square meters. Now we have to build those systems out of modules, which we don't produce. It's in the supply chain. And it's just a reflection of what happened last year in Q2 and Q3 where -- as clearly our key foundry customer came back and said, listen, our key customer for N3 is now blacklisted. So we cannot ship. So we need to adjust our 2021 outlook for EUV systems. Which was followed by another customer and said, well, we're going to delay the roadmap, which also means that this will be pushed back one [Phonetic] year, which actually led to a situation where we actually reduced the number of planned system 2021 for EUV, because customer said these are the two reasons and they were two big customers.

So, what we did, we went to the supply chain and said, sorry, we need those lenses and lasers, very expensive pieces of equipment, we need them actually later. You need to realize that the integral lead time between the installation of an EUV tool and a start of a module production, that is 20 months. So when you at seven, eight months -- then finally customers come to the realization that is not as bad as they've thought and they want those machines, and we have an issue with getting the modules on time and that's the only issue. The only issue is just to resolve, it's a function of the fact that our customers changed their mind in Q2 and Q3 and then rechanged their mind back in Q4 and -- today. There is nothing we can do about it, which actually means that we're all prepared to do 50 units next year, as is in 2022. It will just shift to 2022. So it's -- there -- so as you could say it's supply chain limitations by design, because the customers told us, we don't need them, and then coming back and said, oops! we might have been wrong.


Here is the revenue recognition per quarter over the last three years:
View attachment 331
In terms of yearly % increase of EUV capacity worldwide, it is indeed not as high as before, and it should not be a surprise:

yearly increase rate of EUV tools since 2017.png
 
Back
Top