艾司摩爾 (ASML) 2023 Q2 法說會逐字稿

Good day, and thank you for standing by. Welcome to the ASML 2023 Second Quarter Financial Results Conference Call on July 19, 2023. (Operator Instructions) Please be advised that today's conference is being recorded. I would now like to hand the conference call over to Mr. Skip Miller. Please go ahead.

Thank you, operator. Welcome, everyone. This is Skip Miller, Vice President of Investor Relations at ASML. Joining me today on the call are ASML's CEO, Peter Wennink; and our CFO, Roger Dassen. The subject of today's call is ASML's 2023 second quarter results. The length of this call will be 60 minutes, and questions will be taken in the order that they are received. The call is also being broadcast live over the Internet at asml.com. A transcript of 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 would 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 second quarter 2023 results conference call. Before we begin the Q&A session, Roger and I would like to provide an overview and some commentary on the second quarter 2023 as well as provide our view on the coming quarters. And Roger will start with a review of our second quarter 2023 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 on our future business outlook.

Roger?

Thank you, Peter, and welcome, everyone. I will first review the second quarter financial accomplishments and then provide guidance on the third quarter of 2023. Let me start with our second quarter accomplishments. Net sales came in at EUR 6.9 billion, which is at the high end of our guidance. We shipped 13 EUV systems and recognized EUR 2 billion revenue from 12 systems this quarter. Net system sales of EUR 5.6 billion, which was mainly driven by Logic at 84%, with the remaining 16% coming from Memory. The net sales value of our fast shipments not yet recognized in revenue in the first half of 2023 amounts to EUR 1.4 billion. Installed Base Management sales for the quarter came in at EUR 1.3 billion as guided.

Gross margin for the quarter came in at 51.3%, which is above our guidance, primarily driven by additional deep UV immersion revenue in the quarter, partly related to starting revenue recognition upon shipment for immersion systems that are fast shipped. On operating expenses, R&D expenses came in at EUR 1 billion, and SG&A expenses came in at EUR 281 million, both basically as guided. Net income in Q2 was EUR 1.9 billion, representing 28.1% of net sales and resulting in an EPS of EUR 4.93.

Turning to the balance sheet. We ended the second quarter with cash, cash equivalents and short-term investments at a level of EUR 6.3 billion.

Moving to the order book. Q2 net system bookings came in at EUR 4.5 billion, which is made up of EUR 1.6 billion for EUV bookings and EUR 2.9 billion for non-EUV bookings. These values also include inflation corrections. Net system bookings in the quarter were driven by Logic with 69% of the bookings, while Memory accounted for the remaining 31%. At the end of Q2, we have around EUR 38 billion in our backlog.

With that, I would like to turn to our expectations for the third quarter of 2023. We expect Q3 net sales to be between EUR 6.5 billion and EUR 7 billion. We expect our Q3 Installed Base Management sales to be around EUR 1.4 billion. Gross margin for Q3 is expected to be around 50%, a little below last quarter due to deep UV mix. The expected R&D expenses for Q3 are around EUR 1 billion and SG&A is expected to be around EUR 285 million.

Our estimated 2023 annualized effective tax rate is expected to be between 15% and 16%. An interim dividend of EUR 1.45 per ordinary share will be made payable on August 10, 2023. In Q2 2023, we purchased around 0.8 million shares for a total amount of around EUR 500 million. As mentioned last quarter, in the current environment, we expect to see ongoing pressure on our free cash flow. As a result, we will be prudent in managing our cash flows and maintaining relatively higher levels of cash.

With that, I would like to turn the call back over to Peter.

Thank you, Roger. As Roger has highlighted, another solid quarter in a dynamic environment. Significant uncertainty remains in the market due to a number of global macro concerns around inflation, rising interest rates, recession and the geopolitical environment, including export controls. Although certain end markets seem to be reaching the bottom of the cycle, the semiconductor industry is running at very high inventory levels, leading customers to moderate wafer output as the supply chain works to reduce and rebalance inventory levels.

In order to limit wafer output, customers continue to run at lower litho tool utilization levels. Customers remain cautious due to the uncertainty around the timing, the shape and the slope of the recovery.

We had an increase in bookings this quarter, resulting in a backlog of around EUR 38 billion exiting the second quarter. In our EUV business, we have seen some shifts in demand timing. The majority of the shifts are due to fab readiness, with some elements of uncertainty around recovery. Deep UV demand still exceeds supply. While we have seen delays in deep UV demand from some customers, it has been compensated by strong demand for tools that mature and mid-critical nodes, particularly in China. The demand fill rate for our Chinese customers over the last 2 years was significantly less than 50%, so they now take the opportunity to receive and install systems in their fabs as the supply of tools becomes available.

Turning to our business, starting with deep UV. We're now planning to ship more than 375 deep UV systems with a mix of over 25% immersion. For immersion systems using the fast shipment process, we have come to an agreement with customers on a reduced acceptance test procedure that allows revenue recognition on shipment. As a result, we now expect additional revenue of around EUR 700 million in 2023. And this, in turn, reduces the amount of delayed revenue out of the year, and we now expect around EUR 2.3 billion of delayed revenue from 2023 into 2024 versus around EUR 3 billion of delayed revenue as previously communicated. This incremental deep UV revenue increased the expected year-over-year growth of our non-EUV business from around 30% as communicated last quarter to around 50%.

In EUV, due primarily to customer adjustments in timing -- in the demand timing related to delays in fab readiness as well as some remaining supply chain issues, we now expect to ship around 52 systems this year, translating to a year-over-year revenue growth for EUV of around 25% versus the previously communicated expectation of around 40%.

For the Installed Base business, with the current utilization rates, market uncertainty as well as timing of recovery, customers are delaying productivity and performance upgrades on their litho systems. Therefore, we now expect our Installed Base business this year to be similar to last year versus a growth of around 5% as previously communicated.

In summary, based on our view today, with higher deep UV revenue, offset somewhat by lower expectations on our EUV and Installed Base business relative to last quarter, we now expect net sales growth for the year to move towards 30% versus the previously articulated expectation of over 25%. We still expect a slight improvement in gross margin compared to 2022. No change relative to what we said last quarter, as the positive margin impact from increased deep UV immersion revenue is expected to be offset by the dilutive impact of lower upgrade revenue in 2023.

On the geopolitical front, as it relates to export control, the final Dutch regulations that were published at the end of last month are basically aligned to our expectations communicated last quarter and published on our website. Due to these export control regulations, ASML will need to apply for export licenses with the Dutch government for all shipments of its most advanced immersion deep UV lithography systems, which means the TWINSCAN NXT:2000i and subsequent immersion systems.

As a reminder, sales of ASML's EUV tools have already been restricted, and the business in China is predominately focused on mature and mid-critical nodes. The new Dutch export regulations will come into effect on September 1, 2023. There were also some reports in the media recently about additional U.S. export controls. Of course, we will and cannot respond to speculation. However, based on our current understanding, we do not expect to change our previously communicated view. Therefore, based on everything we have been made aware of as of today, we do not expect the Dutch and potential additional U.S. measures to have a material impact on our financial outlook for 2023 nor on our longer-term scenarios as communicated during our Investor Day in November last year.

Looking towards next year, our customers across different market segments are currently more cautious due to the continued macroeconomic uncertainties. Based on our view last quarter, customers were expecting a recovery in the second half of this year, but it now seems that this is moving towards -- more towards 2024. Also, the shape and slope of the recovery remains unclear. However, based on a combination of the current firm demand and a strong backlog of around EUR 38 billion, there are clearly still opportunities for growth in 2024. But given the mentioned uncertainties, it's too early to be specific about the forecast for next year. We will continue to follow the market developments and update you on our view of next year in the coming quarters.

Despite the near-term uncertainty, the longer-term megatrends we talked about at our Investor Day are broadening the application space and fueling demand for advanced and mature nodes. Secular growth drivers in semiconductor end markets such as electrification and AI, along with increasing lithography intensity on future technology nodes, are driving demand for our products and services.

In summary, while the current macro environment continues to create significant uncertainty, we are working through a strong backlog and expect growth this year towards 30%. In the near- to medium-term, customers remain cautious as they moderate wafer output to help lower inventory levels in the supply chain and to look to build confidence around the timing and the slope of the recovery. ASML and its supply chain partners are still actively adding and improving capacity to meet future customer demand as we remain confident in our long-term growth opportunity.

And with that, we would be happy to take your questions.

Thank you, Roger and Peter. The operator will instruct you momentarily on the protocol for the Q&A session. Beforehand, I'd like to ask that you kindly limit yourself to one question with one 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) The first question comes from the line of Krish Sankar from TD Cowen.

I had two of them. Peter, I understand you don't want to give an outlook for next year, and I'm not looking for a revenue guidance, but if I look at from a unit standpoint or a system shipment standpoint, do you think deep UV and EUV units would grow in calendar '24 (inaudible) calendar '23?

Well, if I would know this, then I would probably give you some outlook on 2024. But I just refer back to what we call the firm demand from our customers, which -- and the strong backlog. And of course, as you understand, our full 2024 year is not fully covered by POs. So still POs need to come in, but we do have firm demand. Now that is a demand that for 2024, you cannot decouple from the outlook on 2025. And 2025 clearly shows the opening and the first ramp of some significant advanced fabs in the logic space, for instance, the 2-nanometer fabs, 3-nanometer for both the -- all 3 leading customers. That, of course, leads to the firm demand in what we currently see. And that means we see significant opportunities also, like we said, certainly, for growth in 2024. However, we also need to realize that the uncertainties as it relates to macroeconomic developments and particularly, I think, the slope of the recovery.

I think we will very likely as many analysts believe but also customers say, we will probably see, let's say, the [thrust] of this down cycle somewhere this year, and then we see a recovery coming. Then it's all about the slope of the recovery. And that's driven really by the macroeconomic uncertainty. So the extent to which they're going to add more capacity in 2024 due to, let's say, the macroeconomic situation, that's the uncertainty.

I think in 2024, there's a higher level of uncertainty of those fabs that will take those machines because they need to ramp in 2025, the next nodes. That's pretty certain. But it's that uncertainty on the macroeconomic demand that makes us a bit more uncomfortable to give you some specific guidance on next year. So in summary, the order book looks good. The firm demand looks good, but I'd love to see all of that being translated into orders over the next couple of quarters. So this is why we also said we're going to follow this very closely, and we're going to keep you abreast of what we're seeing and what our customers are telling us in the next 1 or 2 quarters to come. And I know I didn't give you a specific answer, but I hope it was specific enough.

No, no, it was. Thanks, Peter. I really appreciate the context and the input. And then a quick follow-up for Roger. Can you give us the composition of the backlog in terms of EUV, deep UV, Memory, Logic, China? And also, what do you expect China as a percentage of sales to be for this year?

Yes. So the backlog in terms of composition, we have around 85% of the backlog is for EUV and immersion combined. So that gives you, I think, a pretty good indication there. EUV in total is around EUR 21 billion out of the total backlog of EUR 38 billion. So then you have, I think, the key parameters in there. We do not distinguish in terms of regions. I think we told you before that China is over 20% in the backlog, and that also drives our expectation for how you're going to see system sales develop in the next period. But I think that's the key composition of the backlog as we see it today.

And your next question comes from the line of Mehdi Hosseini from SIG.

Peter, I understand you don't have a clear picture on '24 outlook. But how are you adjusting your own capacity? Can you give us an update how we should think about DUV and EUV capacity into '24? And I have a follow-up.

Yes. That's a good question. I think this is also what we are, of course, internally discussing. But what we -- the capacity in 2024 is really a function of what we need in 2025. And the good thing about 2025 is when we look at the number of fab openings and the ramp profile of new fabs in '25 across our customer base, which also includes Memory, leads us to believe that we should be very careful in reducing our capacity in 2024. Because if you do that, you won't be able to ship in 2025, given the fact that our lead times in the supply chain are ranging from 12 to 15 to 18 months. So this means we will -- at this moment in time, we don't see any reason to reduce any capacity plans for 2024. Because -- and that's basically driven by our views on the 2025 time frame. So I don't expect any adjustments there, and we're not planning for it.

But perhaps the question has to do with the slope of the capacity ramp like on DUV going from 375 to 600, that requires significant ramp. And I'm just wondering if the ramp would look more like a step function in the latter part of '24 as you prepare for '25?

Well, I mean, it's -- you talk -- it's the difference between the ramp and the capacity. I mean, the capacity is 600 units, but that's about 25% immersion. That's you could call expensive capacity. And the 75% is dry, which is less expensive capacity. We're just going to do that because we are currently -- this year plan to ship more than 375 systems. And I also feel that when we look at the firm demand, of course, for deep UV, we don't have all those orders, but the firm demand, then we actually need more capacity next year. So it's going to be -- capacity are step functions. It is not like a gradual function.

So it means if we want to have 600 units by 2025, 2026, but somewhere by the end of 2024, in 2025, we need to have that step capacity built in the supply chain, whether we're going to put all the orders in, that's dependent on the demand. But I think what we're putting in for 2025, 2026 is therefore the remainder of the decade. So we need to do this anyway because we are strongly convinced, as I said in the prepared remarks, that the long-term view that we have on this market is still very much intact. So you have to distinguish between a ramp as a result, as you know, as a result of the market demand and the capacity ramp because the capacity ramp is a step function and its -- and serves the purpose for the longer term.

Got it. And my follow-up has to do with technology migration, especially on EUV NXE:3800E supposed to be a platform upgrade, which carries a higher ASP. And is my understanding that, that platform could be used for both 3- and 2-nanometer. Where are we with booking for those systems? And would that ASP uplift would provide you something as a cushion against challenging macro environment?

Yes. In terms of bookings, of course, the bookings for the 3800 are coming in because if you look at next year, next year is going to show you a good blend of 3600 and 3800 tools. So obviously, quite some of the bookings for EUV, Mehdi, that are currently coming in are also for the 3800.

The 3800, we promise you that on this call, we would disclose we hit the ASP and the ASP is at least north of EUR 200 million. So that is a clear indication, I think, of how that indeed will also help in terms of revenue. It will also help in terms of gross margin ultimately because even though it's a more expensive machine to make because bear in mind, there are commonality, there is quite some [commonality in parts] between the High-NA tool and the 3800 tool. It's a more expensive tool to make, but it's also a very healthy uptick in terms of ASP. So it will help both on the revenue side and also on the gross margin side, starting in '24, but definitely in '25 when the lion's share of the tools there will be 3800s.

And your next question comes from the line of Stephane Houri from ODDO.

Yes. I would like to speak about the gross margin because you have said basically that despite the changes in the growth rate of different products, you still see slight improvement this year, but you also confirmed the 54% to 56% in 2025. So that's quite an improvement. What does it mean about the ramp of 2024? And can you maybe give us some color on the ingredient for the increase in the gross margin until 2025?

I think you heard our enthusiasm to share numbers on 2024 or lack thereof. So I'm not going to do that. The growth drivers for 2025 in terms of the gross margin, there's a number that I think are significant there. We just talked about one important one, and that's the 3800. Of course, that's an important driver of gross margin improvement, definitely also in 2025. So that's one.

The second one that I think is important in comparison to today. As you know, we are preparing both for capacity expansion on deep UV and Low-NA, but we're also preparing significantly and putting a lot of money into getting everything ready for High-NA, both the manufacturing capacity here, we're building up teams in the field, et cetera, et cetera. That currently is a significant drag on our gross margin as we have it today because all of the costs that we're incurring to prepare for that capacity ramp and for preparing for High-NA everywhere in the entire organization goes straight to the gross margin today.

That effect should be gone by 2025 because at that point in time, you would hope that you're actually going to be in a position to utilize at least a significant part of that incremental capacity that you build. And also by that time, you would see meaningful numbers of High-NA. So those are really important drivers of gross margin. And the other one that I probably would give you is that -- is on the service side.

As you know, we see a continued improvement of the EUV service margin in particular, but also in deep UV. And on both, we are driving to get the service margin up, both as a result of what we're doing on the revenue side, but also in terms of trying to further control the cost. So those are the main drivers why looking at 2025, we believe the scenarios that we gave you there, the 54% to 56% is a tenable and reasonable aspiration for us to have.

Okay. And a quick follow-up, if I may, is about the order book, the Memory now represents 31% of the bookings versus 21% last quarter. Is that the sign of a rebound in Memory? Or is it something special here?

No, I think that's just where we are at this moment. I mean part of it, that's the minority, by the way, is of course, some orders from Chinese Memory customers, but it's the minority. The majority is basically technology transitions out of the leading memory makers. They're just preparing for the next node transition, which is a technology transition, which need, of course, the type of machines and the type of technology that Roger just talked about, like, for instance, the EUV systems, the 3800s. And so this is what it is. It is not -- you shouldn't see this as an immediate addition to the memory output capacity, perhaps except the Chinese ones, but that's -- like we all know, that's bit critical to mature stuff. That's not leading-edge.

And your next question comes from the line of Sandeep Deshpande from JPMorgan.

Can you hear me?

Very good. Loud and clear.

Peter, one question for you. I mean you talked about the challenging macro environment at the moment. How do you see -- I mean, you can see how utilization is doing at your customer base. On average, where do you see utilization is at the moment? Because that will be clearly the driver of when the customers start to get more positive in terms of orders back to you in the next few quarters. And secondly, in terms of China, China clearly is a very strong driver of your sales this year. I mean when we look at utilized revenue, when we hear the data points in the supply chain, at least in the logic companies in China, utilization is as bad if not worse than what we are hearing in other parts in the industry. So maybe to try to understand how sustainable these orders from China are into next year given that the end markets even in China seem to be incredibly weak at this point.

Yes. Good. Basically, the utilization question, good question (inaudible) distinguish between Memory and Logic. I think in Memory, I don't think we see a lot of bottoming out there. Yes, it could be -- you could argue, it's bottoming out, but we don't see a kind of an inflection point. In Logic, though, it's very early, but we could see some of an inflection point today. But that's just over the last -- the short period. So see how sustainable that is. But I would think if you think about that, that's bottoming out and you could even say we've passed an inflection point, although it's still early.

Now on China, how sustainable is -- that's correct. I mean you see the same utilization trends in China as we see in the rest of the world. But you have to realize that the demand in China has 2 elements. One, of course, it needs to fulfill the current demand, and that's what we just talked about. I mean the current demand is, of course, weak. But the most important point is the strategic investments and the fabs are being built for a purpose. When you look at what's been made in China, it's mid-critical to mature semiconductors. And that's the sweet spot. When it comes -- when you look at the big megatrends around the globe where China is leading as a matter of fact, when you think about electrification of mobility, think about the energy transition, the IoT in the industrial space, the rollout of the telecommunication infrastructure, battery technology, that's all -- that's the sweet spot of mid-critical and mature semiconductor.

And that's where China without any exception is leading. Now that means that the Chinese industry, the customers of the semiconductor industry need semiconductors of that kind. And I can just tell you in the discussions that we've had, the concern of many of our Chinese customers is that given the increase of the geopolitical tensions, they do not want to rely on supply that comes out of China. So it's very simple that they're going to build a significant amount of capacity in that space, in the mid-critical to mature semiconductors to actually fuel those megatrends where China is actually leading.

So if you then look at the big [whole] market and their desire because of the fear that they have on the increase in geopolitical tensions, they're going to build all those fabs themselves. And that's what's happening. Those fabs will be built. There are many new fabs and new companies that actually say we're going to provide those type of semiconductors to support these megatrends where China is indeed leading. And that's what's happening today. It's not so much the current macroeconomic or the market situation that drives the demand. It's the strategic investment that drives the demand because it's the dependence that part of the Chinese industry has on imports. And then I think it's very sustainable. That's very sustainable for the next couple of years.

And your next question comes from the line of Sara Russo from Bernstein.

Can you hear me?

Yes.

Great. I was just wondering if you could give us an update on High-NA. So indications are that customers are not delaying the tech transition. So are you still on track for first shipments to customers in 2024? And have you seen any increase in orders as you get closer to those first shipments?

Yes. I think we're still on track for the first shipment in 2024. Yes, we're -- actually, this year, we're starting to ship the first module. So that's on track, and that also means for 2024. Yes, I don't think they're delaying the introduction at all. You're absolutely right. And yes, we are still seeing orders coming in. So both is confirmative with the point made that -- and I think Roger alluded to that, that if there's anything on High-NA, we need to make sure that the supply chain, which, of course, needs to supply us with critical node technology will actually be on time. So our main focus is on the execution in the supply chain, not so much from the demand side. It's really about execution.

Great. And can we -- maybe could you give us a little bit of color on where you stand on High-NA orders in the backlog? So assuming that you now are sort of seeing a good number come in, can you give us a sense of orders in the backlog and timing of those orders?

Yes. We said before that our customers -- given there is only a very limited number of customers for High-NA, our customers really do not want us to disclose PO bookings on High-NA. I mean that's the situation. That's why we're not sharing those data. But this -- for quite a while now, we're looking at double-digit numbers in the backlog, let me put it that way. And that's quite a while back that we started to cross that level.

And it's increasing.

And it's increasing. Yes.

And your next question comes from the line of Francois Bouvignies from UBS.

Can you hear me okay?

Loud and clear.

Perfect. So the first question is, obviously, Peter, you were clear on 2024 uncertainty, at least in terms of units, and you will come back later with a clear picture. And Roger, you started to talk about the ASP for the EUV next year with the E model coming to market, the 3800s. If I understand correctly, an ASP of close to 20% growth versus the older models. Can you help us -- give us some color on the ASP, so something you can have maybe more visibility on to next year for EUV? So you touched upon, but also deep UV with all the moving parts with China, your new models as well of deep UV in the market, the 2100s with a 20% improvement in overlay. You have inflation on top. So just how should we think about the ASP specifically (inaudible), if you like, about your (inaudible) businesses basically?

Yes. Francois, I think I was quite clear, I think, on the ASP for the 3800. So I said north of EUR 200 million. So I think that was clear. When it comes to ASPs in the deep UV landscape, of course, it's very widely distributed. And there, obviously, the mix effect is quite significant and that is true both within the portfolio of KrF of dry tools and also in wet tools. So you're absolutely right. I mean the new models that we're introducing, of course, give significant value to the customer, and therefore, command a significantly higher price than all the models. So that is clearly the case, but it is completely dependent on the mix within the dry business and the immersion business.

And also in the immersion business, you have to also realize that -- what I said in the prepared remarks that we cannot ship our most advanced immersion tools to China, but we can ship our mid-critical immersion tools to China. And that, of course, gives -- even in the immersion scope gives quite a significant spread. So it's very difficult to give you one number for the deep UV numbers. It's basically too heterogeneous.

Okay. The second question is on the Installed Base Management. I mean if we look at the guidance of [flat], again, I understand that the level of upgrade is not as you maybe expect in the current environment. If we look at the guidance of flat , it would imply a decline in H2 year-over-year, at least. So how should we think about the level of -- Peter, you mentioned a small sign of recovery. It's early days, but it's been a small size and the fact that the Installed Base Management, I would imagine would be very close to the demand in terms of recovery or utilization rates picking up. Just trying to reconcile that and how we should think about Installed Base Management into next year with your EUV as well going up and ASP per tool per year, I mean, business model?

Let me first take the question on '23 and then maybe, Peter, you want to expand it further. But as it comes to '23, I think the right frame of reference, of course, is not half year over half year, but is the second half in comparison to the first half. In the first half, we had EUR 2.7 billion and [flat] would mean that we're going to have EUR 3 billion in the second half. So that would point at a recovery. And given the guidance that we've given for Q3, Q3, we indicated EUR 1.4 billion. So it doesn't take a lot of compute power to calculate it. That would mean EUR 1.6 billion for Q4. So that tells you that indeed, we are looking at a recovery there. That would be commensurate with the perspective of the recovery that Peter has been talking about. But that's what we're looking at for this year and the slope of recovery there.

Yes. And I think the slope of recovery is critical and very important because like I said, although it's very early, but you could argue and you look at the utilization graphs, you could think that there is an inflection point for Logic. We've had that. And then -- but it's still pretty early on. But if that would continue, then it's really important to look at the slope because for upgrade business, you -- basically, you could argue you have a relatively short period of time before you hit, again high utilization and then customers say, well, I don't have the time. I don't want to shut down the tool. So I think we will watch this very carefully together with our customers to say, okay, looking at the slope, if the slope accelerates, then we really need to start negotiating with the customer quickly to put in more upgrades. And that could be an upside when the recovery accelerates.

When it's a slower degree slope, they'll probably take a bit more time. But that's also where it's the same reason. We now have time to do upgrades because we don't have a full utilization of the installed base. So there is some upgrade there. But still, customers are currently saying market is not good. It's still CapEx because there are high value upgrades. So they're a bit cautious now. But yes, we have to start being very close to our customers in the next couple of quarters to say if we see an opportunity, let's go because before you know it, they don't have time.

And your next question comes from the line of Aleksander Peterc from Societe Generale.

I just have two. First one would be, you talk about the recovery being pushed out somewhat and you do give a cautious message on 2024. So my question is really, is there a possibility that the significant fab openings you talk about in '25 could be pushed out by 6 months or a year? Is that something that's possible? I mean, if the customers have [idle] capacity for longer when they push out capacity additions as a result? Or are all of those strategic plant openings really strategic that will go ahead regardless of demand pattern? That's the first one. I have a follow-up.

Yes. Yes. I think on this -- on the leading-edge logic fabs, they will happen. I mean, they have -- basically, it's not -- and that's driven by the road maps of the customers of our customers. It's the Apples, the Qualcomms, the NVIDIAs of this world, (inaudible) you have a very clear road map based on the 20- or the 3-nanometer designs, and they want those new products to be introduced at that time. So that's going to happen. We have little doubt there. And I think on the strategic fabs in China, I made that very clear. I think it's just a strategic, very clear focus area that they have because they want to hedge against any negative geopolitical repercussions that could come. So that's also strategic. So I see a little downside in 2025.

Excellent. And then just kind of a technical follow-up on the EUR 700 million [catch-up] in deep UV that are moving out of fast shipments. Did all of that occur in the second quarter that you reported? Or [is it when] they reported in the current quarter? And [is in] what proportions, please? And while we're talking of fast shipments, our discussions on a similar (inaudible) the table for EUV? Or is that on the table now?

So the EUR 700 million is the expectation that we have for the end of the year, right? So of course, there will be a little bit of [flux] during the year, but the EUR 700 million is the expectation that we have for that in the year. Of course, we have some of that also in this quarter, but the EUR 700 million really is the expectation that we see for the full year.

As it comes to EUV, it's based on the conversations that we've had with the customers, they are very happy to take the risk of the tool for the immersion tools upon shipment and based upon a shorter testing program for EUV were not there yet. So the question will be also, based on how next year is going to pan out, I think that we're going to get the question of how much fast shipment are we going to see for EUV next year in comparison to normal shipments? I think that's the primary question that we have on EUV. So if you think about the -- to what extent could we have some tailwind from that -- in that regard, I think it will be heavily dependent on what we're going to do in terms of regular versus fast shipment.

And there are 2 considerations there for next year. One consideration is that as a standard procedure, when we introduce new technology, we want to test them more, right? So the 3800 clearly is a significant development in our EUV shop, and that means that at least for a number of tools, we want to do more testing and more elaborate testing and therefore, at least for a number of the initial tools, we wouldn't fast ship them. So we will do regular shipments and do the full testing program. And secondly, as I mentioned, it will be dependent on the utilization of our capacity, right? Because fast shipment is a way to get the tool earlier to the customer, but it's also a way to optimize our capacity. So it will be driven by those 2 considerations, what we're going to see there next year in terms of type of shipments, and that will tell you whether or not we're going to get any tailwind for EUV revenue as a result of that.

And your next question comes from the line of Alexander Duval from Goldman Sachs.

You spoke about a pushout in demand timing for EUV. I wonder to what extent we should think about this as a one-off push out from '23 to '24 given the customers presumably would still need these tools for their fabs that are still getting built? And their customers, in turn, have product aspiration for '25 that you've just mentioned? Or to what extent would you expect some 2024 units to be subsequently pushed into 2025? And then I've got a quick follow-up.

Good question. We need to realize it. If you look at the reasons, predominantly the (inaudible) had to do with fab readiness. And that was basically driven by construction skills. And you think, well, how can that be? You just hire a couple of construction workers, and you just build a fab. We're just building a EUR 20 billion fab that's going to do a 5- or a 3- or a 2-nanometer product is a skill. And people don't seem to realize that when we start building those fabs across the globe now and now everywhere, that skill is -- has been refined over the last couple of decades in only a few places on the planet, predominantly in Taiwan and in Korea, and a bit in China.

Now having to do that now and accelerate this will lead to all kinds of issues because we are still building those fabs in Korea and in Taiwan, but also in other places on the planet, also in the U.S., for instance. So getting access to the requisite skills and skilled workers to keep the construction plan on time is a challenge as at least what customers tell us. And this is the main reason. So you can easily look at a delay of a couple of months or a quarter.

Now -- and of course, like I mentioned earlier, we need those 2-nanometer fabs or 3-nanometer fabs in 2025. But it also means we need to resolve in, let's say, an 18-month period, yes, some of those skills gaps. And then -- but I think it might easily be a problem also at the end of next year, but let's see how quickly they can skill up the construction industry to help build those fabs. So that's the predominant reason for the timing changes or the demand timing changes. And of course, has also been in this particular year where there's a few supply chain issues that address 1 or 2 systems, but it was -- predominantly, it was just fab readiness and for the reasons that I just mentioned. And I hope they get reskilled quickly and at the end of 2024, we don't have those issues.

And just a quick follow-up. We've seen some news flow on demand for leading-edge ships driven by AI applications. Can you just share your latest views on any growth opportunity from AI in 2024, given that obviously 2023 shipment schedules are full? I think you alluded in your video prepared remarks [that, that] potentially being an incrementally supportive driver of demand. So just curious for any thoughts there.

Yes. I think that's true. But I think we're at the beginning of this, you could say, AI high-power compute wave. So yes, you'll probably see some of that in 2024. But you have to remember that we have some capacity there, which is called the current underutilization. So yes, we will see some of that, but that will be taken up, the particular demand by the installed base. Now -- and that will further accelerate, I'm pretty sure. But that will definitely mean that, that will be, you could say, the shift to customer by 2025. So I don't see that or don't particularly expect that, that will be a big driver for additional shipments in 2024, given the utilization situation that we see today.

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

Yes. One on domestic China demand, you talked about a fill rate that was less than 50%. Do you expect to be caught up to that exiting this year? Or will you still be trying to kind of fulfill that demand looking into '24?

Yes. I think we're still -- like we said, also in the prepared remarks that the demand is still more than we can ship. So that also means that we still have a fill rate that's not 100%. It's still lower than -- of course, it's significantly higher than the -- significantly lower than 50% that we saw in '21 and '22, where we had [screening] customers. So we simply couldn't ship [enough, and] China was one of the real victims. Now of course, today, with the fabs being ready there, (inaudible) being there, anything that doesn't ship to any other country goes to China. But there's still some demand that will move into 2024 because we don't have 100% fill rate today.

Got it. And then just as a follow-up. In the recovery of the Installed Base Management business that you talked about implied for 4Q '23, is that predicated on just logic alone? Or is there also some expectation that you see some memory recovery embedded in that?

I think we don't. Somewhere down the line, there will be a recovery, yes, because that's going to be -- probably when we go through these inflection points in the second half of this year. And then it's all about the slope of the recovery. And this is where we have some uncertainty that we expressed loud and clear, I think. And that's the uncertainty that we get from customers because they don't know either. So I think it's a bit too early.

I think it's fair to assume that the utilization rates on memory are lower than the utilization rates on logic, right? So there, it's reasonable to assume that logic would be ahead of the curve in terms of upgrades.

Yes, also because -- like I said earlier, you could argue when we look at the stats, you could already see an inflection point. But it's -- like I said, it's very early on. So we just have to see how that continues over the next couple of weeks and months on logic.

And your next question comes from the line of C.J. Muse from Evercore ISI.

I guess first question for Roger. I think you fairly clear on the call that no changes to kind of the capacity adds. So curious how we should think about OpEx growth into 2024?

Yes, I think the OpEx that we're currently guiding for the year, I think that's a pretty good estimate, I think, for what we see for the rest of the year. I think in terms of next year, I think it will also be a little bit dependent on how we further see things develop. And that, to a certain extent, will at least drive also the SG&A side of life.

On R&D, as you know, we continue to have really good ideas. And on R&D, we typically try to play this on the longer term. So I think it is realistic to assume that on R&D, you will see some increase, albeit at a slightly lower pace than the very sharp increases that you've seen in the past couple of years.

Very helpful. And then, Peter, I guess as a follow-up, I know that you're actively working with the Dutch government. But curious as to your kind of thoughts around any potential time line from hearing from maybe more restrictive kind of thoughts out of the U.S. government.

Yes. Of course, we have regular discussions with -- now Dutch government, which is inactive because of the political situation here. So [we are] going to prepare for new elections. But I think we just have to wait what comes out of the U.S. now. But the reason why we said -- based on what our understanding is, and I jokingly said here internally, it wasn't even jokingly, I actually meant it. I've been in this business for quite a long time. And my hunch about what the Dutch were finally going to say in the end was about right. So this is why we (inaudible) March. And I also have a kind of a hunch on what's going to happen for the rest of the year and with the new rules. And my just gut feel is based on what we hear, and our understanding is not going to have a material impact. But having said that, we don't know exactly what the content of those new regulations is going to be. But we just have to wait. I think Japan came out, the Dutch came out. I think the U.S. government will probably come out soon. And then we'll know for sure whether my hunch or my gut feel was correct.

All right. We have time for one last question. If you are 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?

We will now take your last question for today. And the question comes from Tammy Qiu from Berenberg.

Firstly, Peter, relating to your China exposure, do you have any format of customer concentration, i.e., does one or few customer accounting for more than, let's say, 50% of the demand from China at all?

No. I think it's the -- the number of customers in China is significantly higher than -- and I just talked about the spread of the customer significantly higher than anywhere on the planet. It has to do with the fact that it also goes back to where Chinese industry, I don't talk about the semiconductor industry, but the industry in general is actually growing. It grows in those areas which are covered by the big megatrends. And that means that specific requirements for semiconductors to support those trends actually ask for very significant and different applications that put the demand on this wide range of mid-critical to mature semiconductors. And there's a lot. And that also means that you see customers -- semiconductor customers now focusing on certain of those areas. It means you have many, many customers, yes? And that's -- so it's pretty widespread, whether it's memory, whether it's logic or foundry, it's almost everything, but many of them and very much focused on specific parts of the industry. So yes, it's on the contrary. I mean it's not specifically focused on 1 or 2 customers. It's a broad based.

Okay. And also, you mentioned that you can actually ship the mid-critical machines to China and still basically allow them to do whatever they want to. So let's say, the mainstream you are shipping to China from an immersion perspective in 1980, if you can only ship something like 1970 or older machine, do you think that can allow them to do what they want to do?

Yes. You have to realize that when you ship an immersion tool and just do the math, which is the wavelength of the light over the numerical aperture of the lens. That's 193 over 1.33 [times a K] factor, which is the process factor, which is an absolute minimum of 0.26 because beyond that, you don't have any contrast. So if you do the math, do it on your calculator, you come to 38-nanometer. So whether it's in 1970 or 1980 or 2000 or 2100, it's 38-nanometer. So how do you get smaller sizes. That is where you start using double patterning, and that's basically determined by your capabilities of materials, which is deposition and etch. So It's, of course, the most advanced have one determining factor in that, it's the precision with which the tool works.

And this is where, if you look at the Dutch regulation, it doesn't mention a [tight] name. It just mentions a technical specification, which focuses on the precision with which the tool works. That's where the cutoff point is. But in terms of feature size, it's the same. But it's really the precision with which you can position the feature size on the wafer. That's where the cutoff point is. And that's determined in the regulation. So it's all deposition and etch.

All right. 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'd appreciate it. Thank you.

Thank you. This concludes the ASML 2023 Second Quarter Financial Results Conference Call. Thank you for participating. You may now disconnect.