Thank you for standing by. Welcome to the ASML 2021 Second Quarter Financial Results Conference Call on July 21, 2021. Throughout today’s introduction, all participants will be in a listen-only mode. After ASML’s introduction, there will be an opportunity to ask questions. I'll now hand the floor to our speakers. [Operator Instructions]. I'd like to now turn the conference call over to Mr. Skip Miller. Please begin your meeting.

All right. Thank you, operator. Welcome, everyone. This is Skip Miller, Vice President of Investor Relations at ASML. Joining me today on the call is ASML's CEO, Peter Wennink; and our CFO, Roger Dassen. The subject of today's call is ASML's 2021 second quarter results. The length of this call will be 60 minutes, and questions will be taken in the order that they are received. This call is also being broadcast live over the Internet at asml.com. A transcript of the management's opening remarks and a replay of the call will be available on our website shortly following the conclusion of this call. Before we begin, I'd like to caution listeners that comments made by management during this conference call will include forward-looking statements within the meaning of the federal securities laws. These forward-looking statements involve material risks and uncertainties. For a discussion of risk factors, I encourage you to review the safe harbor statement contained in today's press release and presentation found on our website at asml.com, and in ASML's annual report on Form 20-F and other documents as filed with the Securities and Exchange Commission. With that, I'd like to turn the call over to Peter Wennink for a brief introduction.

Thank you, Skip. Welcome, everyone, and thank you for joining us for our Q2 2021 results conference call. I hope all of you and your families are healthy and safe. But before we begin the Q&A session, Roger and I would like to provide an overview and some commentary on the second quarter as well as provide our view of the coming quarters. And Roger will start with a review of our Q2 2021 financial performance with added comments on our short-term outlook, and I will complete the introduction with some additional comments on the current business environment and our future business outlook. Roger, if you want.

Thank you, Peter, and welcome, everyone. I will first review the second quarter financial accomplishments and then provide guidance on the third quarter of 2021. Net sales came in within guidance at €4.0 billion. The guided lower revenue was due to a number of systems in the quarter that did not receive factory acceptance testing due to customers' desire to bring systems into production as quickly as possible. Therefore, revenue will be recognized in subsequent quarters after completion of acceptance testing at customer site. We shipped 10 EUV systems and recognized €1.3 billion revenue from 9 systems this quarter. Two EUV systems shipped this quarter without factory acceptance testing so revenue will be recognized in the subsequent quarter after customer site acceptance. For the system we shipped in Q1 without factory acceptance testing, we were able to complete site acceptance test and recognize revenue in Q2. Again, the net result is 9 EUV revenue systems in Q2. Net system sales of €2.9 billion was again more weighted towards Logic at 72%, with the remaining 28% from Memory. The strength in Logic drives both deep UV and EUV revenue. The Memory business is mainly driven by DRAM. Installed Base Management sales for the quarter came in at €1.1 billion, above guidance, due to increased upgrade business as customers continue to pull forward software upgrades that can quickly increase productivity of systems in this high semiconductor demand environment. Gross margin for the quarter was 50.9% and was above guidance primarily due to the additional software upgrade business and one-off revenue accounting releases. On operating expenses, R&D expenses came in at €634 million and SG&A expenses at €172 million, which was slightly lower than our guidance. Net income in Q2 was €1.0 billion, representing 25.8% of net sales and resulting in an EPS…

Thank you, Roger. As Roger has highlighted, we had a good quarter in both sales and profitability. We are seeing continued strong demand from our customers across all market segments, from both advanced and mature nodes, driving demand across our entire product portfolio. Compared to last quarter where we expected an annual sales growth rate towards 30%, we now expect revenue to be up around 35% this year. The higher sales growth comes from our ability to increase output in our factories and in the supply chain as we work to meet the strong customer demand. Looking at the different market segments and changes from last quarter, we now expect stronger growth rates across all markets. In Logic, global demand continues to be strong across a broad application space in both advanced and mature nodes. Compared to last quarter where we expected 2021 Logic revenue to be up 30% year-on-year, we now expect Logic to be up around 35% this year. In Memory, customers see tight supply-demand dynamics continuing into next year. Compared to last quarter where we expected 2021 Memory revenue to be up 50% year on year, we now expect Memory revenue to be up around 60% this year. In our Installed Base business, for the second quarter in a row, our upgrade business has been stronger than guided. Customers are looking to upgrades to provide the fastest path to increase their wafer output capability. Compared to last quarter where we expected 2021 Installed Base revenue to be up 10% year-on-year, we now expect Installed Base revenue to be up around 15% this year. As we continue to strengthen our outlook on the year, the majority of the increase is coming from our deep UV business. We have increased our planned factory output to meet customers growing demand…

All right. Thank you, Peter and Roger. The operator will instruct you momentarily on the protocol for the Q&A session. Beforehand, I would like to ask that you kindly limit yourself to 1 question with 1 short follow-up, if necessary. This will allow us to get to as many callers as possible. Now operator, could we have your final instructions and then the first question, please?

[Operator Instructions]. Our first question comes from the line of Francois Bouvignies of UBS.

Maybe the first one, if I may, is on the we look at your upside for dry DUV, how would you slice this between what ties up to your new leading-edge Logic and Memory capacity? And what relates to new trailing-edge, Logic and analog? Would be interesting to have the color of the two. And the second question I had, maybe, Peter, when we look at the market dynamic, I mean there is obviously a strong demand and, as a consequence, a significant shortage. And on top of that, you have some local capacity concern that you talked about in your video. So what I'm trying to understand is with these 2 factors that probably one concern is kind of inflation of orders, creating some disconnect between the supply and the demand i.e., the shortage and local domestic capacity. So how do you assess this risk? How do you manage this risk of overcapacity when you think about adding capacity in deep UV and EUV?

Very good questions. Let me first answer the upside on the dry deep UV. Whether it's driven by leading edge or, let's say, trailing or mature? I think on the leading edge we have a reasonably good view as to what our customers need in terms of new fabs built, ramp-up plans. And of course, we know the layer stacks. We know the layer composition in terms of dry immersion EUV. So yes, that's growing, but that's more plannable, I would say. That's -- we have more insights. I think what really surprised us is the very strong demand from, let's say, the non-leading-edge customers, which is across the globe. It's in Europe, it's in U.S., it's in Asia. And it has to do with microcontrollers, power, analog, image sensors, it's all over the place. And I think it's also explainable and it's a bit of a lead-in into your next question that if you see where this is going and where the shortages are in automotive and other industrial areas. Even from time to can because we have some good contacts with some semiconductor manufacturers. So we actually see this happening everywhere. You see lead times in household appliances going up, simply because analog, power, sensors, microcontrollers in household appliances are in shortage. So it's basically, it's the rollout of the -- finally what we are seeing is the Internet of Things and 5G, we have the big pipe. So you can actually use the big pipe to actually transfer that data and transport the data. And that's what we're seeing now. So it's -- basically, the big surprise was really in what we would call the mature or the specialty semiconductors, yes, which is just a reflection of the digital transition that we're right in the middle. So…

Our next question comes from the line of Joe Quatrochi of Wells Fargo.

So you talked about the catch-up effect is stretching into 2022. I was curious on the DUV side, your orders remain really strong. So just curious with the capacity increases you're putting in place, is that catch up with your order book, is that more of a first half 2022 dynamic? Or do you see that as continuing into the second half of 2022?

Yes, that's a good question. It depends on the speed with which we can get the -- it depends on the confirmation we can get from our supply chain, because we did ask them and especially for the mature products and dry products to get a significant increase, which is double-digit increase in our capacity or their capacity, I should say. And that could then easily extend into the second half of next year. Like I said earlier, I think this disruption of the supply chain that happened during the global pandemic, I think it's like a traffic jam. You have a traffic jam in 15 minutes, takes an hour to get it resolved. It's basically what's happening also in the industry. It's a global supply chain with many, many key players there. When you start to put locks into this global efficient supply chain effectively and you take out those locks, not at all at the same time, and inventories are depleted before everything starts going again, it takes time. And that's what we see. We actually see that in the supply chain and customers of our customers and customers of the customers' customers. They're basically now reassessing their planning. And they're finding out there are shortages all over the place, whereby the key players in that ecosystem are not fully aligned yet on making this seamless again. And that will take time. So this will lead into 2022 easily. And I think the first half, if our supply chain -- and not only our supply chain but also the supply chain of our peers can follow then, yes, maybe mid next year, we'll see some relief and then you see a tapering off of the order intake. But otherwise, I think we will continue into the second half of next year.

Got it. That's helpful. And then on the increased EUV revenue outlook for this year, from a revenue perspective, it sounds like you're doing some things to maybe improve the manufacturing or your capabilities and maybe get another tool out the door. But just curious, in addition to that, is that reflective of maybe any sort of expectations around mix being a little bit stronger to the 3600D or configurations being a little bit richer than expected when we entered this year?

Yes, I think you got it, Joe. It's a combination of those things. I think what you saw over the quarters -- during the quarter is that indeed the ASP turned out to be stronger than what you saw last year, and I think that was the result as you say of the options that customers asked for. So, they were richer configurations than our gen is, so there you saw on the 3400C you saw an ASP of 145. You see this quarter even a bit higher, but some 3600D in there albeit very, very small. So that's one element and the second element of that, we're doing our utmost to further the decrease cycle time as a result of that crank out with one or two more tools. So that's the reason behind the increase to 35% uptick in comparison to last year, rather than the 30%.

Our next question comes from the line Sandeep Deshpande of JPMorgan.

Peter, I'm just trying to understand -- clearly I mean there is need for capacity both in the DUV as well as EUV. I mean are you -- I mean you have laid out how much capacity you intend to outlay in EUV. Do you intend to outlay this DUV capacity additions over the next few years at your Capital Markets Day or at this point? And my second question is on the margin. Clearly, I mean, the more DUV you ship, it helps your overall gross margin as such, really, and that has helped this year as well. I mean do you see that trend, that shift because of this higher DUV and the sensors and all these other older tools are likely to change your midterm view on the gross margin because of the higher DUV shipments?

Well, I think the second question Roger will answer. On the deep UV capacity, yes, I think it is our intention. If we have -- we feel we're comfortable that giving you that number because we get all the confirmations in from the supply chain, I think we would definitely give you more insight give you more insight into our deep UV capacity on specifically dry, perhaps somewhat on immersion and on, I think more specific on EUV, we will definitely do that by the end of this quarter. But we need some -- so of our key suppliers, we want to have that firm commitment, and we're going to not only get a firm commitment from an e-mail from the CEO, but I think we want to get that confirmed by the people who are actually building that capacity in the customer. So it's going to be quite an in-depth orders that you could say, because we will, based on that capacity if the demand is there, we'll accept orders, and we don't want to disappoint our customers. So it's a process that's been ongoing for the last couple of months. But I think it's definitely our intention to give you all the information that we have at that time and I hope and I expect to be honest, that we will be able to do that to give you that number.

Yes, Sandeep, on the margin side, as you know, particularly with the introduction of the D model, you do see that the deltas between the different products in terms of gross margin becomes smaller. And therefore, the effect that you were talking about becomes smaller. You also know that within deep UV, there are differences between the different products in terms of gross margin, as we said on calls before. What is particularly relevant to look at is immersion, because the immersion is still from a gross margin perspective, a good product for us. And that's the one to watch. And if you, for instance, look at the last quarter, last quarter, you saw that 47% of our system sales was immersion. This year, you -- or this quarter, you see that, that's gone down to 34%. And which is more realistic based, I would say, also for the quarters to come. And of course, that had an impact why the gross margin in Q1 was still very, very strong. why the gross margin in Q2 expectedly was a little bit lower. But I think that's a reset that is important. Immersion percentage for the quarters to come, more or less, mirrors what we had in Q2. And as a result of that, you build up towards the 51% to 52% that we've indicated for Q3.

And on the dry margins, because we were going to ship more dry. Dry margins have generally been somewhat lower than the immersion because there's also more competition there, and there's a cost drive in the mature market, which is also different in the advanced market. So yes, that's -- these are lower-priced tools as you use as a KrF tool. That's a single-digit to a very low double-digit number, but it depends on the configuration with a different profile, which is lower than our energy immersion margins.

Our next question comes from the line of Dominik Olszewski of Morgan Stanley.

The first one I wanted to discuss was the -- on the topic of silicon solvency, as you mentioned. I'm curious in your conversations with policymakers, do you consider there to be any prospects for the U.S. to allow shipments of EUV to China at the point when you're successfully rolling out in tools for research purposes in 2024? Is that the technological and time buffer that you think domestic China gets access to the EUV tools? And then the second question is specifically just on Immersion. Are you seeing momentum, or anticipating momentum to get market share gains further in emerging tools, specifically maybe to your Logic customer base?

Yes. I think on immersion market share gains, I think our market share still is already pretty high. Let's leave it at that. And EUV, I think, yes, is to that, that's a subject that we've discussed many, many times during these calls, I mean, EUV is under export control according to the Wassenaar Arrangement, there's a multilateral agreement between 42 countries. And that requires an export control license from the government of the exporting country, which, in this case, is the Netherlands. I think that's still under review. So -- and I think this is not the place and the time to speculate on what that would mean going forward if we introduce High-NA. I think there's -- what we, of course, do know there are in-depth discussions between governments of different countries to see what they want to do. And I think it's not our role. I think we're in contact, but of course, it's up to the government. And we just wait and see what happens. Now I've said it before, the end demand of leading-edge semiconductor devices is probably not impacted from, not be impacted by where we ship EUV tools. The end market will be determined by the value that's been created by those products and the ability and the willingness of the world to buy these products, which will drive the demand for high-end semiconductors, and that will drive the demand for EUV. And then we will ship EUV where EUV will -- where EUV machines will be made to make advanced semiconductors, whether it is in Korea or in the U.S. or in Europe or wherever, yes, we're going to ship those tools to.

Our next question comes from the line of Aleksander Peterc of SocGen.

The first one will be just on the -- on your speed of the increase of capacity. Are your comments really pertaining pretty much to 2022? Or can you put anything in place in that? I'd just like to understand if we're basically maxed out on what you can do in 2021. And the second follow-up would be just on the phasing of [certain true-ups] in revenue. It seems that you imply what your guidance that Q4 will be down year-on-year quarter-on-quarter. I just want to understand if that's due to the pull-in in that area that you saw in the first and second quarter?

Yes, the increase of capacity, when? I think there are 3 ways in which you can increase capacity. And the first is basically, it's 6 to 9 months. It's just being more efficient, squeezing everything out of your production process, basically what we call the reduction of cycle time. That's what you can do short term. That's what we do today. And I think the second one is you just use the same square meters, but you hire people, you buy machines, which is especially true for our supply chain, which generally has a lead time of 12 to 18 months if you can get the machines. Which, by the way, I made a comment earlier, we also see that in the machine industry also, there's going to be a shortage because of also chip shortages. But that's 12 to 18 months. That brings us into 2022. And it could even be towards the second half of 2022. And that's what we're seeing now in the supply chain. I think everybody works on their cycle time reduction, so we are maxed out. And then if you want to add capacity, people and machines. And then the third one is, well, you kind of suit it in this square inches or square meters or square footage, then you need to build, which has a lead time of 2 to 3 years, which will bring in to 2023, '24. And I think -- so the capacity increase that we're focusing on is within the same square meters, machines, people, cycle time reduction. And that, I think we're maxed out for this year because it's only going to be software time reduction. It has to be more people, more machines, same square meters next year, and that's 2022.

Aleksander, on your second question and your line broke up a little bit, but I understood your question to be the distribution of the installed base revenue over the first half and the second half of the year. And that is a correct observation. So in the first half, we had €2.3 billion revenue for installed base. And with the indication of 50% growth over last year, you would get to €1.9 billion for the second half of the year. And you're quite right. As we also indicated, there has been quite some pull-in of particularly the software-related upgrades by customers who want a relatively easy way to increase capacity without having to give us too much machine time. So both in Q1 and Q2, we actually got more of these software related upgrades than we anticipated. And yes, there is a bit of pull-in there from the second half. So that's the correct observation.

Our next question comes from the line of Stephane Houri of ODDO BHF.

I have a question back on margins. And maybe if you could update us on the evolution of the gross margin at EUV services. And the question linked to that, so the follow-up would be, what is, in your view, your potential for gross margin improvement, if you put together the improvement of EUV services margins, but also the fact that now you're selling the 3600D, which carries, if I'm correct, the same kind of margins that the rest of the tools? Yes, that's basically my question.

Okay. So in terms of EUV service gross margin, as you know, we broke even last year on that. This quarter, we guided to 25% gross margin on EUV service. And I think we've said that within about 4 years' time, we believe that we get that gross margin level to approximately corporate gross margin level. That's what the intent is, and that is primarily by driving down costs, and by helping customers run the machines more efficiently, such that more wafers get produced. And as a result of that, we get more wafers compensated therefore. So that's the model there. And 25% from zero in a couple of quarter's time, of course, is a big uptick. But I think that is an aggressive curve. So the first 25% is, of course, a big development. But then you will see that improvement gradually slow down. In terms of systems gross margin. On the D, the D tool gets us to the corporate gross margin. That's the way to look at it. So still below the deep UV gross margin, but at the corporate gross margin. With the E tool, we hope to get the gross margin to the deep UV level. And at that stage, really have EUV and deep UV be at the same level. That's the intent. That's what we're looking at. And of course, the E will be introduced in a little under 2 years from -- or 2 years from now. And that's the one that should get us to a deep UV type of gross margin levels.

Our next question comes from the line of Rolf Bulk of New Street Research.

You mentioned that around €1 billion of EUV tools are being purchased by DRAM manufacturers this year. And that is primarily for bit capacity for next year and beyond, because your lead times for EUV tools are still very long today. My question is, how should we think about the DRAM EUV business in the context of '22, '23. Do you see a risk of a pullback in DRAM lithography spending as cycle times for EUV toes come down, and manufacturers maybe -- do not need to buy that capacity 1.5 to 2 years in advance anymore.

Yes. I think the -- we won't be -- I think the exact number this year is probably 1.2 billion, so it's a bit higher. So of the -- and you are right, I mean, the 1.2 billion of EUV in our 2021 Memory guidance or DRAM guidance, I mean you need to really look at bit capacity additions from those machines not in 2021, it's going to be later. I think on 2022, 2023, when we look at the road maps, the customer open maps, yes, we will see as a higher number of deep -- of EUV shipments for DRAM, and it's driven basically by the capacity build-outs that they are planning. And yes, when cycle times go down, then also order lead times go down, but so it will have more -- as a more significant effect on when they place the orders and not so much when they need the tools because the tools are based on the fab planning. And as you know, these fab plans are sometimes years out. So we have 2, 3 years of planning visibility. So that will drive the -- let's say, the build capacity, our build capacity and the potential sales and will have an impact on the orders, on the deals, which will be probably coming in a bit later as you’ve actually seen last quarter, I mean, there's a significant number of EUV tools that came in as deal, because there we have long lead times. And actually that stretches our order coverage in terms of our capacity that we have for next year to the point where 80% of our capacity is [now] ordered, yes. I mean that will have -- that will change going forward with lead times going down, but not so much with sales. Sales will be driven by the RAM plans and the capacity plans of our customers. And those are pretty well known because of the big projection that take years.

And those RAM plans are also based on layer count. And that's very clear from the customers that indicate they plan to increase the layer count of OBV in DRAM manufacturing. And just to give you one more data point. So this year, you're looking at approximately 20% of the EUV revenue. So 20% of, let's say, the -- of the 6 billion, 1.2 billion, that's the number. And I think it's fair to assume that the same percentage will apply to next year. So that gives you -- and as you know, we're at least expanding capacity for EUV for next year. So that gives you an indication that we do see continued growth in the EUV insertion into DRAM and with the commensurate growth in revenue from that.

So absolute numbers will grow, because like you said, next year, we'll have 55 units capacity, which will probably very likely we'll sell that. But it also means that from an absolute shipment number point of view, it will keep growing for both Logic and for DRAM.

And our next question comes from the line of Robert Sanders at Deutsche Bank.

I just had one which is about EUV layer count from 3-nanometer to 2 nanometer. That’s when TSMC will be introducing gate-all-around, most observers seem to think there will be close to zero pitch scaling from that transition as TSMC did from 20-nanometer to 14 when the FinFET was introduced. So I was just wondering if that's something you're anticipating in your plan.

No, I don't -- we don't have that view. I think there is -- on the 3 nanometer node and the 2-nanometer node that we look at the tools that we'll be using are different machines on those nodes. And also, there will be an introduction of double patterning EUV, which is basically helping the pitch scaling. So we have a different view.

In our view, the transition from FinFET to gate-all-around will be layer count agnostic. So that's going to be layered.

Yes. On the pitch scale, there will be an introduction of double patterning at that moment. But Roger is correct, I mean in terms of layers, it doesn't matter.

The next question comes from the line of Andrew Gardiner at Barclays.

I wanted to come back to the point you were making around DUV capacity. In particular, for this year, just to try and establish a baseline, can you give us a sense as to how much of the -- of this year's revenue is being driven by the buffer stocks, of the drawdown of that inventory that you've got? If I doing my math right based on your guidance, we're now talking about a low €8 billion level of DUV revenue this year. So yes, how much of that’s coming from the inventory?

Yes. I think let me put it -- let me answer that in a different way. I think this is -- you really have to go deep into the supply chain, because it's in our service inventory, it’s in the suppliers’ inventory. So I would have to really go deeper to give you an exact percentage. But if we look at the 2022 deep UV plans and especially for immersion, not so much for KrF because that's where we can see some improvement in terms of shipment numbers. But I think in terms of immersion 2022, I would currently think that our immersion sales number will be about the same as this year. Whereby this year, we, of course, were helped by this, you could say, onetime depletion of the stocks, is creating them from everywhere that we could. So I wouldn't at this moment in time -- because don't forget, I mean, the immersion numbers this year are quite high. I think from an immersion point you have to go back a long time to look at similar shipment numbers. I think it will probably be the same next year. But next year, we will not have the advantage of being able to deplete the stock. So this is the way that I would look at it. And on KrF, some dry season there, you could see higher numbers next year because that's where we actually need the capacity. And that was also the answer to one of the earlier questions, where do we see it. I think we basically see this dry demand coming out of out of, I would say, the specialty markets or the mature markets, which is basically everywhere. So hopefully, that answers your question, Andrew.

Yes, it does. I mean if I could just follow up with that quickly. I mean you're talking about a significant double-digit increase in capacity to DUV. Clearly, not all of that is going to come online in '22. That presumably is just the starting point. But if we look out over the next couple of years, I mean, significant, perhaps to effect, the obvious is not 10%. And I know you want to save something for September. But I mean it feels like you're talking 20% plus or minus, that kind of a ballpark? Would that be reasonable?

I mean you know us for a long time, so you can assess what significant means in our terms, and that's not 10%, as you’ve pointed out. But what it is, we'll probably be more specific in September, the end of September. But yes, also -- when capacity comes online, there's also a lead time between when the capacity comes online, when we get the models, the part, we can make the tools and we can ship it to the customers. So part of that capacity that will come online, that significant capacity increase, which will be indeed double-digit, that will have an effect in 2023, not in 2022. When it's available, January 1, 2022, yes, then you are right. But that's -- I pointed out, I mean, people and machines and potentially using extra square meters, it takes 12 to 18 months before it's there. And then they need to produce, and then we need to produce and then it needs to be installed. So I think we will see that capacity increase definitely occurring next year. And how much of that we can use for output, that still remains to be seen, and we're figuring that out together with our suppliers.

Our next question comes from the line of Didier Scemama of Bank of America.

I have a first question and a quick follow-up. And maybe, Peter, if you could share your thoughts with us on a sort of a debate in the market that you also touched on your -- one of the 3 long-term drivers. So in the U.S., they've identified effectively a gap between what's being produced in the U.S. and what's being consumed. I think the numbers are 12% and 40%, if I remember correctly. So my question to you is if we were to narrow that gap substantially, how much spending on lease equipment that need to happen, number one? And number two, how long will it take to actually get there realistically? And then I've got a quick follow-up.

I think, you might be surprised, yes, but I think it doesn't matter that much. Because we assume -- I think that's a right assumption, if you look at the expansion plans, which are more -- let's say more concrete in the U.S., but as you know that this discussion was happening in Europe. These expansions will not be made by just a new company coming online, it’d just b -- it will be the -- I would say established players, the leaders that have the capability and the competence to build those types, to manage them and have the process knowledge to copy exact, if I may use that word, those processes from other parts of the world into the U.S. and into Europe. And those companies, they are going to build those fabs to make sure that they can supply the market with the niche for those products. And it's going to be just a few companies, which are going to be rational. I don't think that any administration can go to the CEOs of one of those companies and say, because we want that capacity, you have to build, but then it's going to be idle, it's going to be completely inefficient. They are not going to do that. So it's going to be rational. And only information that we have points to those few companies that have the capability to build those fabs, yes. So I think for my little point of view, yes, there will be some inefficiency, yes, because, you're building a new fab, new operations in a place where you cannot piggyback on your local ecosystem, if you’re do it in different part of the world. So yes, there will be -- there's a ramp up time for those fact, that will be…

Thank you for your answer. As a quick follow-up, I just wanted to come back to DUV. My question to you, Peter, is very simple. Over the years, the semiconductor industry has never managed to effectively exert pricing power with that customer base for all the reasons that we can imagine. But now that -- I'm not saying you should abuse that pricing power, but now that you're in a slightly different position, you're talking about the doubling of market demand over the next 10 years, et cetera. And given the investments that you have to make and particularly shareholders are also worried that, hey, is this the right time to add that much capacity, would it be feasible for ASML to ask your customers to effectively pay in advance for those DUV tools so that you completely derisk your model and completely eliminate the risk for double ordering or triple ordering?

Well, it's always a good question to ask as an entrepreneur to derisk completely their business model. But to be very honest, I mean, that never happens. And I don't think it's the way also we need to deal with our customers, which is only a handful, and there's a handful of equipment players. And on deep UV, yes, I think we will charge our customers the value of those machines. And to give you an example, we will put deep UV KrF on NXT platform, which will significantly increase the productivity for our customers, which also, I think we are entitled to part of that value and which I think they will pay us. So I think the way that we look at increasing prices and is really to provide our customers more value, not to say, well, we're in a squeeze situation, which could be -- last for a couple of years, but then we're back to normal again. And then customers will push back in this traditional customer supply relationship. That's not the way this industry works. And I don't think this industry should work this way. We have to provide value, and we have many opportunities to create value for our customers and thereby asking a higher price, but the customer will get the value. That's how we work. I think on the risk of the capacity increases, I said it before: I think where we've structurally underestimated the growth of this industry, and everything that we're seeing today, I think there are good reasons to believe that the underlying demand of what we see, and especially in dry deep UV and the application space that deep UV is servicing, there is a -- There's a very good reason to add extra capacity because we need that capacity. And we will sell those too. We will make more money, and I think we'll satisfy our shareholders and I think the risk is limited. Short-term, we always have small cycles; but longer term, I don't see the risk.

And in terms of paying in advance, just to remind everyone that of course, paying in advance does happen on the EUV front. So on the EUV front, given the long lead times, we do have prepayment schedules with our customers, which are significant and also clear, I think, from the free cash flow generation of ASML in the past 12 months. So I think in that way, the comment that you made, I think we're doing that. But not to -- we don't do it in this specific circumstance, but just do it as a matter of principle, because of the very long lead times that we have.

That's a very good point, Roger, because I mean, the lead times of deep UV are a lot shorter, which also means that you can manage the supply and demand better. But we will increase the capacity also, not only in base, but also in the supply chain. But again, based on our strong conviction that we need that capacity going forward, because of the market developments that we are seeing. And I think that risk, we believe, is limited.

All right. Thank you. We have time for one last question. If you were unable to get through on this call and still have questions, please feel free to contact the ASML Investor Relations department with your question. Now operator, may we have the last caller, please?

That's the line of C.J. Muse at Evercore ISI.

I guess first question, Peter, I wanted to clarify a comment you made to Andrew. Were you guiding immersion units next year flat, or was that a commentary around supply availability before adding new capacity?

Yes. It also might have been a bit convoluted answer, but what I was trying to do because -- basically giving you 2, I would say, messages. We have -- in 2021, which is this year, we have, you could say, a spurt sales because of the fact that we're depleting everything we can find in terms of inventory. And some would argue that some of our minimum stocking levels for service might be at a real minimum, because we're using everything to make our machines. So that is this onetime step-up, which gives you an immersion number that is high. I think that will be probably the number that we're also looking for, for next year where we don't have that ability to have this onetime step-up. So we're effectively increasing capacity and getting to that same level. But that's basically how I think we should look at next year because that is what I think from a capacity point of view, with what we are seeing, we can do in terms of cycle time reduction in terms of putting some extra people to work we can do. So effectively, in the increasing capacity, but ending up at about the same number.

Okay. Helpful. And then just a quick follow-up, Roger. At your last Analyst Day, you targeted 55-plus percent gross margin. As you think about calendar '22 and exiting this year with EUV margins of 50-plus, EUV service moving higher as they come off warranty and a pretty robust mix from DUV, why wouldn't we be approaching that kind of number in calendar '22?

So first of all, I don't think in Capital Markets Day we mentioned 55. If I recall correctly, we had 50 there, but we did have 2 arrows in front of it, leaving it entirely to your imagination how far you wanted to stretch that. So I think '22, based on the number of the dynamics that I mentioned, certainly from a gross margin perspective, has promise in there for sure, but it's way too early to give any definitive guidance on that. But you would have seen -- if you look at the trajectory over the year, you would see that we're now guiding 50 -- the 51, 52 percentage for next quarter, that's a good basis. That has already quite a bit of D in there. Of course, next year, everything would be D. Next year, you would benefit from the 2050 a bit more than you would this year. So there's a bit of potential there, but it's still a bit too early to give any guidance on what it's going to look like next year.

All right. Before we sign off, I'd like to remind you that our Investor Day is currently planned to be held in London on September 29, 2021, COVID conditions permitting. We will keep you posted on details and hope you'll be able to join us. Now on behalf of ASML, I'd like to thank you all for joining us today. Operator, if you could formally conclude the call, I would appreciate it. Thank you.

Thank you. This now concludes the meeting. Thank you all very much for attending. You may now disconnect your lines.