ASML Holding N.V. (ASML) Earnings Call Transcript & Summary

So the -- on the cycle time reduction, yes, we're making good progress there. I'll give you the numbers. We started the year with 24 weeks cycle time in our cabins. Sorry, that's incorrect. 24 is what we had at the end of Q2. We started above 30 with the year. 24 is what we have right now at the end of Q2. We're driving it down towards 20 by the end of this year, and we actually -- we think we can get it down to 17 weeks in the first half of 2021. So that's on track, and that should allow us to get to the capacity of 45 to 50 that you were referencing. So yes, we are on track to achieve that. With regard to your second question, well, of course, that's dependent on some of the things that we were just talking about, right? So it's a bit too early to talk about 2020, 2021. And the shape of the ramp in Logic and also the speed at which we really see Memory coming back and then, of course, the EUV insertion into that, those are the 2 main drivers. Next couple of months are going to give us more -- are going to shed more light on that. So then we will see whether the capacity is -- whether that capacity is going to be fully utilized or not for next year shipments. So a couple of months' time, we will have more clarity on that front.

Okay. I must confess we do talk a lot about EUV shipments into 2021. But something which I think the community of investors talks a little bit less about is some of the other moving parts into 2021, say, particularly around average selling prices. And I do know you've already had orders coming through for the 3600D tools. Could you give us a flavor for how should we think about ASPs into 2021? What about the EUV Installed Base business, the upside in services?

Yes. So you're right. I mean the 3600D development is on track, and we will introduce that model and ship that model starting the second half of next year. And also, I think it's fair to assume that if you look at the composition of sales of next year, my expectation would be that there is a healthy 3600D contingent in that sales pattern for next year. So it's going to be more than a handful, let me put it that way. In terms of the ASP, the D model has improvement both on -- on all fronts but primarily on throughput and overlay. Those are 2 important elements. If you look at throughput you're looking at an improvement of over 15%. So in terms of ASP development, I think conservative estimate would be somewhere between 10% and 15%, I think, would be a good estimate for the ASP development for the D model versus the C model. So indeed, as you say, given it's a -- there's quite a few tools -- we've got a few D tools that are going to be sold next year and shipped next year, that will have an impact on ASP but definitely also on the gross margin for us. On the service and the upgrade business, sort of -- let's start with the upgrade business. Upgrade business is going very well this year, and we also have quite a few upgrades available to our customers next year as well. So I think that will continue to develop very positively. And typically, the upgrade business for us is a good and healthy business also because in many instances, there is a software element in it where typically the gross margin is very high. On the service business, I think I -- in the past, I've given you the different dynamics around the service margin. So roughly speaking on the revenue side, with a number of customers we have contracts where the service fee is based on wafer output, right? So you get a fixed fee per wafer. And of course, with EUV going into high -- or having gone into high-volume manufacturing but also seeing throughput increasing and increasing every month, of course, as a result of that, that means that revenue per tool is going up. And also, of course, the installed base is going up. Whilst on the cost side, we are making very good progress both in reducing demand-to-machine hours, which is a function both of the fact that we get smarter at this, of course our experience increases, but also with more and more tools getting into one site. Of course, that's -- the -- of course there, you will see that efficiency increases, utilization of service crew increase. As a result of that, the man-to-machine hour drops. And also, in terms of parts usage, we are improving and making good progress. So all in all, I think we're really making good progress on the cost front as well. I expect this quarter already, Q3, to become gross margin positive for EUV service, which would be the first quarter where that would happen. And also for the full year, I expect at least a breakeven but hopefully even a little bit of a positive gross margin on the EUV service. So that's what we're looking at on the EUV service side. So yes, you're right, if you look at all the 3 main components in EUV, both on the ASP, on the upgrades and on the service margin, there are positive dynamics for increasing gross margin.

Thank you, Roger, for walking us through all of those. Another one, if I may, on EUV is with regards to its adoption in DRAM. "So obviously, Logic is the big driver. But ultimately, DRAM would also be significant. The critical determinant there is, of course, the number of layers. So do your conversations point to a growing appetite for number of layers? But do you sense a bit of a pause as customers reassess their longer-term road maps and think about alternatives to EUV? I guess you could also talk about a bit of traction outside your main customer in DRAM."

Yes. So on -- the conversations that we're having with customers on the EUV insertion into DRAM are pretty consistent. And what they -- what a leading customer tells us is a couple of things. First off -- and they made a public comment about the insertion of EUV into the current nodes, right, and also the 1y 1z node that they're also talking about, the insertion of it into 1alpha, 1a. There has been some -- there have been some public comments there as well on number of layers. And the official statement which they made on that was that the number of layers was a meaningful increase over the current number. So that's the -- that gets you to the interesting guessing game of what is a meaningful increase over 1, and we have reason to believe that it's probably -- that you're probably looking at 4 layers of EUV. I think that's in all likelihood what you might get to. The reason why they want this is not just for cost, right, not just for cost reasons but also that they have determined the device characteristics. That's the public statement that they may made, device characteristics are actually superior to the device characteristics, including yield, of DRAM modules being made on -- being made solely on DUV. So that's what's driving it. And all the comments that they make and all the interactions that we have with them are consistent with that. And quite frankly, they have been saying this for quite a while. And I see them -- we see them executing on that in a consistent way. The other 2 players, 1 player operating in the same country as the customer that I was just talking about, I think, follows nicely the pattern that is laid out there typically with a bit of a delay. And maybe with -- not with the same level of -- with the same insertion, if you like, of EUV in terms of number of layers, so they might be a little bit less aggressive, if you want to use that word, but definitely are following that track. And the other large DRAM manufacturer demonstrates a healthy interest in EUV as well. So it's definitely having EUV inserted into its road map. It's not explicit on the timing of that but definitely has EUV interest and has that as part of their road map thinking as well. But publicly not yet clear on the timing of the insertion.

Thank you, Roger. I guess switching sides to Logic. A question that's come in today and also one that I had on my list is, "To what extent do you think logic customers are concerned about some of the technological constraints which are cited for EUV at the industry 5-nanometer node? Stock of resist, dosage levels, defect. I mean I won't go into all the technical jargon. But clearly, there is some sense of concern around these constraints. It's been in the past you've overcome them. Is that your philosophy looking at some of these defects as well?

Yes. I think you're right. I mean with every -- so as long as we're talking about EUV, these concerns are being mentioned. So it's not new. And I think that -- the level of concern, I think, if anything, has gone down quite significantly. And I think the proof point for that is probably the SPIE conference that was there in February. And I think there was a pretty good review at that point in time of the entire ecosystem. So I think -- and we actually came away from that pretty comfortable about the progress that was being made. I think if you look at resist, for instance, I think good progress was reported there in terms of sensitivity, improvements in sensitivity, and the improved performance that with the same dose could be achieved there. There were good discussions about dry resist. I think on the pellicle front, I think both on the durability but also the transmission quality of pellicles, I think good progress was being reported. Pellicle generation and, therefore, defects, if you like, on the reticle, I think very good progress is being made there. Of course, we are one of the key players there. But in general, I think good progress is being made there. And also, the actinic inspection, which some customers value very much because of the ability to do that, too, and, in that way, do an inspection through pellicle. Also, good progress was made there. So we came away from the SPIE meeting really saying that the ecosystem has fully embraced EUV as the technology to go forward and also is fully embracing it in order to enable continued progress on the node cadence from 7 to 5 to 3 and all the way forward. You will always see that the real progress of some of these players is only visible not too long before the introduction. And I guess resist is one that comes to mind, right? So the -- typically, only when you get into volume, that's where resist is ready. But so far, they've always been ready. And given the progress report that we received on that on this conference, no reason to believe it will be any different here. That's what we see. Frankly, it's also what our customers see. So actually, we do get these presentations, and they are public, of some of our customers on how they look at the ecosystem, and I can tell you that the color-coding of that in the past 2 years has really shifted from the amber to red side to the green side quite a bit. So I think -- and that's also what we see, obviously, that the comfort level has gotten to a much higher level than even 1.5 years ago.

Thank you, Roger. Last one on EUV, at least for now, is "Can you give us an update on the High-NA EUV road map? To what extent are your customers pushing you to accelerate it?"

Yes. So of course, the customers are very interested in it and want to use it, and they're more advanced nodes, because customers want to avoid multi-patterning at certain node transitions. So that's why they're clearly interested in it. And I think the fact that customers were making -- have been making orders and down payments on it many years before it actually kicked in, I think, is a proof point of that. So we're on track. We're making good progress there on all the relevant parts, particularly on the stage on -- first and foremost, obviously, on the optical cycle, making very good and strong progress together with ZEISS. And we are currently looking at being able to ship the final modules to customer by the end of '22. That's what we're currently looking at.

Thank you, Roger. Let's now move on to geopolitics. Now I have 2 questions, 2 names and -- instead of asking the questions. I guess they are pretty obvious. So let's start with the first name, Huawei. You've talked in the past about short-term and medium-term impact being limited. Is that still the view? Just give us the puts and takes around what the Huawei legislation means for ASML.

Yes. So directly, the implications for us are very, very moderate. Of course, it could impact some of our customers, of course. But as we said before, and I really think that, that's also what's going to happen, it will lead -- what it will do is it will lead to a shift in market shares. So market shares will change as a result of this in the different markets that Huawei is a player. So to the extent that Huawei really would be impacted by this in their ability to continue to produce at the innovative levels that they want to, of course it will have an impact on their market share. But to us, that means that you're probably going to -- that we're probably going to see a bit of a shift from one customer to the other. But I don't think that the overall size of any of the end markets that Huawei serves is going to be impacted by this. And for us, that means that we don't believe that, that will have an impact on the number of wafers that will be in demand. And as a result of that, we think it will only have an impact on the customers that we ship to. That's the only implication it has for us. So it's not a direct source of concern for us and will not have a big impact, we think, on our shipment schedules.

So let's maybe move on to something that could potentially be a source of direct concern for you, which is, of course, the news around a potential ban against SMIC. Firstly, could you give us a sense for what's your current level of exposure to domestic customers or indigenous Chinese customers? Potentially, what you were thinking about as you thought about '21 and towards the 2025 target? What does a potential ban for SMIC mean for ASML?

So if you look at the sales to indigenous customers in China in 2019, that was around EUR 800 million. And we forecasted that for this year, there would be a bit of growth, and we actually said we believe it was going to be about EUR 1 billion this year. So that's what you're looking at. And of course, that's a mixed bag of all the indigenous Chinese customers. I have to say that it's probably a little bit skewed towards Logic, that number. And then, of course, within Logic, the customer that you were just referencing, of course, is a big customer in -- relatively speaking, in China. If you talk about '21 through '25, of course we were projecting continued growth of -- and we are projecting continued growth of indigenous Chinese customers. But that really isn't the way that we make our models because as you know, the way we model the revenue for us in those years is really based on the global demand for wafers. And whether the wafers will be produced in country x versus country y doesn't necessarily have a big impact on us. And definitely for '25, that's the case. For, let's say, the very short term, of course we do recognize that Chinese customers might not necessarily already produce for the world, right? So that might be part of -- the equipment that they take is for their learning curve, is for the journey that they're on to increase on their learning curve. So you might say that some of the Chinese demand is a little bit incremental to the global demand for wafers that we calculate. And therefore, there's always a little bit of the icing on the cake, if you like, if you take that model as a basis, as I just articulated. So that's what you're looking at. So it's not like this is impacting in any way our 2025 calculations because that really is based on the global demand for wafers. I would also say, particularly on the customer that you were just talking about, of course we've seen the articles there, but we also recognize that there is still a lot of uncertainty and there's a lot of speculation. And I think we will have to wait until there is a lot more clarity on what exactly it means, what exactly it means if anything like this will occur. Also for us, for manufacturers, let's say, outside of the U.S., there are still so many uncertainties. It's very, very hard and speculative to really determine what the impact is going to be on a go-forward basis.

Right. I'm just cognizant of time, so we'll maybe try to fuse some of the questions that we still have to run through. Firstly, on the financial side, "Could you give us an update on capital allocation? You have paused the buyback. You've talked about supporting your supply chain. Your supply chain seems to be holding up well. So can we expect a resumption of the buyback?"

Yes. So there are 2 main drivers of buyback. One is the generation of free cash flow. And then the other one, is there a specific circumstance as a result of which we have to hold it back, for instance, because of the issues in the supply chain? So on the generation of free cash flow, that has been modest in the first half of this year, and we went through the reasons for that on the call. A very important one there is the fact that down payments on EUV are still not the norm. More and more, we're working on that and are actually now working with all of our customers to get there. So that's an important area of work. But that's still work and is not something that, based on the original contract, is something that we automatically get. And on the flip side, in a number of these older contracts that were drafted like 3 years ago -- so you know that we're now shipping based on contracts that, in a number of instances, were drafted 3 years ago. Three years ago, our world was slightly different than the world that we are in today. And part of that is that we also granted certain customers extended payment terms. So this year really is a year where, on the one hand, you're not always getting down payments. And in fact, in a number of instances, you have to work with extended payment terms. So that's why there is a bit of modesty in the free cash flow generation. And also, because of the way we are growing and also because of the COVID situation, we had to take in quite a bit more inventory just to be safe. So all of those dynamics have led to a rather weak cash flow generation in the first half. Second half will be far better. And then in 2021, we should really see the benefits of some of the upside of the things that I was just talking about. So one is the free cash flow generation is really starting to get traction. And again, that's what we see and expect for the remainder of this year starting. And also, the conditions are better, the conditions in the ecosystem, and, therefore, our judgment about whether or not we have to come to the rescue of our supply chain. And I agree with you, I think that, that is trending in the right direction. And so far, not a lot of support has been needed. Then I think that the share buyback can resume, and that could resume in -- starting Q4.

All right. Maybe to round off the discussion, a couple of big-picture questions, if I may, Roger. We published a report last week, a bit of a detailed one on ASML. And one of the ideas we have put forth is that you would be looking to evolve into a computational platform of fabrication over a period of time as you leverage your position in lithography and try to expand further into process control and look at expanding your share of wallet. Strategically, would you agree that's the direction the company could potentially look to go further into?

Yes. I mean I think that has a lot of merit. And in fact, I would even claim that it's consistent with what we said at the Capital Markets Day in 2018. And I think you spotted that nicely in that material. And I think the way you think about it and the way you lay it out makes good sense to us. I think when we started talking about holistic lithography for a start, holistic lithography really was to have process control, inspection tools, metrology, et cetera, have that primarily for the lithography steps, right? So using process control, e-beam -- and e-beam, overlay, et cetera, all that primarily driven towards the process control of the lithography steps. I think what you see more and more is that we're now looking at other process steps within the fab to see to what extent the e-beam, multi-beam technology, the overlay technology with the YieldStar, et cetera, to what extent we can also use that to look at other process steps and introduce process control technology there, of course added by the computational software technology that we're offering there. So it is that combination of tools, software -- computational software, overlaid technology and e-beam multi-beam tools, it's that combination of tools that more and more allows us to play a bigger role in the process control within the fab. And take overlay, for instance, overlay started strictly looking at the lithography side. But as you know, we now also have overlay and YieldStar technology that is actually looking at -- after etch, so -- right? And so -- also, and the deposition and etch phase of the fab plays a role. So yes, you're absolutely right that we are expanding our role in that -- in the process technology side in the entire fab.

All right. And maybe a final question to round off our discussion, Roger. And maybe a question on sustainability, please. It's becoming ever increasingly important. Help us understand, when it comes to sustainability, what's the level of emphasis in ASML? And would you say, for example, that management compensation is being increasingly tied to sustainability?

It already is, and it has been for a few years now. So sustainability is a part of the long-term incentive plan for senior management. So it's already there on our -- our achievements there. And if you look at our integrated report or integrated annual report, you will see that the emphasis we put on both the target setting and also the monitoring of the progress that we're making on the ESG goals and ESG targets, that level of attention is going up significantly. And more and more, it really becomes part of our normal way of working and part of the DNA of the company. And if you look at the annual report, you'll see that we have identified a number of ESG topics that are very important to us. One is anything around people. One is around innovation ecosystems, so what is it that you do within your entire ecosystem to advance that ecosystem. One clearly is on the responsible supply chain. One is on waste. So what actions are we taking to reduce the waste within that we as a company generate. And one very big one is on anything associated with climate and energy, which we have -- where we have targets at different levels. So we have targets at the level of the energy consumption in our own manufacturing but also ideas around how can we reduce the energy consumption of our tools, which I can promise you is pretty significant. So anything we can do to reduce the energy consumption of our tools. So that's big time on our development agenda. And also, how can we contribute to better energy usage and higher economics there, better economics there. And for instance, data centers, by helping to produce microchips that are ultimately conducive to better energy efficiency. So we have goals and ideas around all of those levels and have set those goals for next year but also for 2025 and are working towards that. So it's a big element, an increasing element of emphasis, I would say, in the company.

All right, Roger. Well, we can talk lots more, but we are out of time. Thank you so much for taking all the time to join us from a busy schedule. We appreciate it.

My pleasure. My line is always...

Thank you to all of the listeners and thank you. All the best. Stay safe, stay healthy. Bye.

Bye-bye.