Thank you for standing by. Welcome to the ASML 2019 fourth quarter and full year financial results conference call on January 22nd, 2020. Throughout today's introductions, all participants will be in listen-only mode. After ASML's introduction, there will be an opportunity to ask questions. I would now like to open the question and answer queue. If you would like to ask a question, please press star one to register. If you would like to withdraw a question, please press star two at any time during the call. Your questions will be answered in the order that they are received. If any participant has difficulty hearing the conference, please press star zero for operator assistance. I would now like to turn the conference call over to Skip Miller. Please go ahead, sir.
Thank you, operator. Welcome everyone. This is Skip Miller, Vice President of Investor Relations at ASML. Joining me today from ASML headquarters in Veldhoven in the Netherlands is our ASML Chief Executive Officer, Peter Wennink, and our Chief Financial Officer, Roger Dassen. The subject of today's call is ASML's 2019 fourth quarter and full year results. The length of this call will be 60 minutes, and questions will be taken in the order they are received. The call is also broadcasting 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'd like to turn the call over to Peter Wennink for a brief introduction.
Thank you, Skip. Welcome everyone. Thank you for joining us for our Q4 and full year 2019 annual results conference call. Before we begin the Q&A session, Roger and I would like to provide an overview and some commentary on the fourth quarter and the full year 2019, as well as provide our view of the coming quarters. Roger will start with a review of our Q4 and full year 2019 financial performance, with other 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. Thank you. Roger.
Thank you, Peter, and welcome everyone. I will first highlight some of the fourth quarter and full year financial accomplishments, then provide our guidance for the first quarter of 2020. Q4 results were basically in line with our guidance. Net sales come in at EUR 4 billion. Net system sales of EUR 3.1 billion was heavily weighted towards logic at 83%, with the remaining 17% from memory, clearly showing the continued strength of logic business, as well as the ongoing digestion phase of the memory business. We reported EUV system sales of EUR 922 million from eight shipments. Installed base management sales for the quarter came in at EUR 906 million. Gross margin for the quarter was 48.1%. Overall operating expenses came in above our guidance, with R&D expenses at EUR 516 million and SG&A expenses at EUR 148 million.
Higher-than-guided SG&A is due to additional employee benefit costs and costs related to our IT implementation. Turning to the balance sheet, EUR 186 million worth of shares were repurchased in Q4. We ended last quarter with cash equivalents, and short-term investments at a level of EUR 4.7 billion. This amount is significantly higher than anticipated, with most of the cash coming in in the December period. Moving to the order book, Q4 system bookings came in at EUR 2.4 billion, including EUR 1.1 billion for nine EUV systems. Logic order intake was 79% of the total value, with the remaining 21% from memory, again reflecting the continued strong logic demand for leading-edge lithography. Net income in Q4 was EUR 1.134 billion, representing 28.1% of net sales and resulting in an EPS of EUR 2.70. For the full year, net sales grew 8% to EUR 11.8 billion.
The installed base management sales was EUR 2.8 billion, which was a small increase compared to previous year. In 2019, we booked EUR 6.2 billion of EUV orders, which is more than 50% of the total booking value for the year, reflecting customer strong demand for EUV technology. We continue to invest in the future of ASML and increased R&D spend to EUR 2 billion in 2019. The increase was primarily driven by the acceleration of our EUV roadmap, low-NA and High-NA program. Overall, R&D investments as a percentage of 2019 sales was about 17%. SG&A was about 4% of sales. In addition, ASML invested EUR 886 million in CapEx, supporting our long-term growth opportunities, primarily around High-NA capacity and infrastructure.
Net income for the full year was EUR 2.6 billion, resulting in 21.9% of net sales and an EPS of EUR 6.16. With that, I would like to turn to our expectations for the first quarter of 2020. We expect Q1 total net sales between EUR 3.1 billion and EUR 3.3 billion. We expect our Q1 installed base management sales to be around EUR 950 million, driven by strong demand for field upgrades, especially EUV. Gross margin for Q1 is expected to be between 46% and 47%. The lower gross margin relative to the strong Q4 number is primarily due to Deep UV mix effect, fewer immersion, and more dry systems, with some positive EUV mix effect. The expected R&D expenses for Q1 are around EUR 550 million, and SG&A is expected to come in at around EUR 140 million. Our estimated 2020 annualized effective tax rate is around 13%.
Regarding our capital return, ASML paid total dividends of EUR 1.3 billion, made up of the 2018 dividend and 2019 interim dividend, and purchased EUR 410 million worth of shares in 2019. Through December 31st, 2019, ASML acquired 9 million shares under the 2018/2019 program for a total amount of EUR 1.6 billion. Supported by our long-term business plan, ASML will submit a proposal at the 2020 Annual General Meeting of Shareholders to declare a dividend for 2019 of EUR 2.40 per ordinary share. Recognizing the interim dividend of EUR 1.05 paid in November 2019, this leads to a final dividend of EUR 1.35 to be paid in the second quarter. This is a 14% increase compared to the 2018 dividend.
The 2020 Annual General Meeting of Shareholders will take place on April 22nd in Veldhoven. ASML announces a three-year share buyback program of up to EUR 6 billion to be executed in 2020 through 2022.
ASML intends to cancel these shares after we purchase, with the exception of up to 0.4 million shares, which will be used to cover employee share plans. With that, I'd like to turn the call back over to Peter.
Thank you, Roger. As Roger highlighted, we had a very strong quarter resulting in another solid year of growth, driven by logic and EUV. We were able to achieve an 8% top-line growth despite an overall industry decline of around 10% due to a weak memory market. This is always a reflection of our logic customers' drive to continue to innovate and invest in technology for future nodes. For 2020, we currently expect a year of double-digit growth in both sales and profitability, primarily driven by EUV and installed base business. Major innovation drivers such as artificial intelligence, 5G, high-performance compute, autonomous driving, and big data are creating new end-user applications.
These applications require more high-performance logic, fueling increased demand for leading-edge nodes, this is evident in several customer announcements regarding ramp plans for their 7 and 5 nm nodes, which will drive another strong logic year and an increased demand for EUV. In the memory market, customers have indicated they're seeing signs of demand recovery in some market channels, and improvements in memory chip pricing also support this view. As customers have lowered litho tool utilization to reduce wafer output throughout the weak memory demand period, they will first use this underutilization to return to normal supply levels, which will take some time. Subsequently, this will also trigger equipment demand, albeit a bit later than the supply/demand recovery for memory devices.
Taking the slope of the recovery of our litho equipment utilization as a proxy, it seems likely that we will see stronger litho equipment demand for memory in the second half of the year. We expect significant growth in our install base business. Service business will continue to scale as our install base grows, and we'll also see EUV contribute to service revenue as these systems start running wafers in volume manufacturing now. We expect significant demand for upgrades, particularly in EUV, as customers utilize upgrades as a quick way to increase capacity. In EUV, it was a breakthrough year with the technology now starting in high-volume production and producing consumer products that are already available in the market.
As we continue to execute on our accelerated EUV roadmap, we were able to ship our first NXE:3400C in 2019, which provides high productivity translating to increased customer value, delivering higher ASPs and improved gross margins. We shipped six 3400C systems in Q4 of the eight EUV systems total we shipped in the quarter, bringing the total to 26 EUV systems and a full-year sales of around EUR 2.8 billion in 2019. Increase in customer confidence in EUV is translating into more layers in logic production, as well as expanding to new markets with the adoption in memory. For full year 2020, we plan for EUV sales of around EUR 4.5 billion on 35 systems. We continue to see demand building for next year's shipments and expect a healthy order flow to continue.
In order to fulfill the expected strong demand increase, we're working on the cycle time reduction to enable a capacity of 45 to 50 systems next year. 2021 is shaping up to be a very busy year. Regarding our current outlook for the year, we expect 2020 to be another growth year as mentioned before. Although it's too early to provide quantitative expectations, let me make a few qualitative comments. Major innovation drivers and applications that require high-performance logic are driving increased demand at the advanced nodes. Logic demand is currently strong, we expect that this demand to remain healthy, primarily driven by EUV. As previously communicated, we expect the sales of EUR 4.5 billion on 35 systems this year, which translates to EUV sales growth of approximately 60%.
Memory is showing an early signs of recovery, although there's still uncertainty around exact timing of the recovery, it is likely we will see stronger demand in the second half of the year. Taking this into account, we expect a stronger second half with strengthening sales throughout the year. Summer 2019 was another great year with continued positive momentum in EUV, as well as solid demand across our entire product portfolio. We expect another growth year supported by healthy logic demand and likely recovery of the memory market, with increased sales from our installed base business as well as demand for EUV. The positive industry momentum around innovation and expanding new markets further strengthens our confidence in 2021 outlook and our 2025 growth scenarios. With that, we'd be happy to take your questions.
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 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?
Yes. Thank you. At this time, we will begin the question and answer session. Again, if you have a question, please press star one to register for your question and star two to withdraw a question from the queue. If you are using speaker equipment today, please lift the handset before making your selections. One moment please for the first question. The first question comes from Mehdi Hosseini. Please state your company name followed by your question.
Yes. Thanks for taking my question. My question has to do with your comment regarding EUV manufacturing capacity of 50 system by 2021. Peter, do you think your backlog would reflect that capacity, as we progress through the year? In other words, would you be able to have a full commitment from your customer for full capacity? My short follow-up has to do with the multi-beam EUV wafer inspection. Are we still on target for the first shipment in the first half? How should we think about the time it would take for your customers to evaluate the tool?
Okay, Mehdi. Thanks. On the EUV capacity, yes. I think the backlog will reflect this. I think the order intake on EUV is looking very healthy. I have little doubt that we'll have the backlog filled this year to support the capacity that we have lined out now for 2021. I think the issue here is really, we need to reduce the cycle time, which we have good plans for. We see first progress. I think the focus is on cycle time reduction in this case. MB, multi-beam inspection, yes, we'll ship in the first half. I think the customers will probably take throughout this year to evaluate the tool, so that we can start shipping next year in higher volume.
Great. Thank you.
The next question is from Mr. David Mulholland. Please state your company name, followed by your question.
Hi. It's David Mulholland from UBS. Just to follow up on some of the comments you made around memory. Obviously, there are some indicators things are improving. I think some of the checks we've been doing through the supply chain are certainly pointing to improved capacity plans, particularly potentially from Q2. When do you think you could start seeing that in orders? I guess in some respects, why haven't we already seen some of it in Q4? I'll come back with a follow-up.
Thanks, David. It's a good question. I wish I had a definite answer, because that would make things easier. I'm basically saying, when I said in my prepared remarks that we see the utilization of our tools going up and we just extrapolate the slope of utilization increase, I think it would mean that in the first half of this year, we will see likely a return to a normal supply demand balance for our customers. They see that also, and taking into account the order delivery times, the order lead times, I would expect orders that have to come in somewhere in Q2 in order for us to make sure that in the second half of the year, we could see an increase of our memory business.
We're in this business for quite a long time. In my experience, memory always comes back with a vengeance. When it comes, it always comes quick. We'll just have to wait. I think it makes a big difference whether orders come back in Q2 or they come back in Q1 or in early Q3. It makes a big difference for the year. We'll have to wait and see. I wish I had a final and definite answer.
Just one second follow-up on the install base management business. Obviously, very strong run rate in Q1 at EUR 950 million. How do you think about this on a full year basis? In the past, you'd had a target, I think, of EUR 3.7 billion, but been slightly more conservative run rate through the last couple of years. What's driving the pickup in Q1? How should we think about that on a full year basis?
David, I think if you look at Q1 and also if you look at Q4, you see that the momentum was already building up in Q4, where we were already slightly over 900, 950 for the quarter.
We expect that we will not be able to sustain it at this level for the entire year. I think a good way to go would be to say that we're probably going to see a 20% increase over last year annualized. That would get you to approximately EUR 3.4 billion for the year. That's where we would see it for the full year.
That's great. Thanks very much.
The next question comes from Mr. CJ Muse. Please state your company name followed by your question.
CJ Muse with Evercore ISI. Thanks for taking the question. I guess first was hoping to hit on gross margins. Can you walk through where we exited on EUV in the fourth quarter, and then how you're seeing the trajectory for overall gross margins through the year as you contemplate increased EUV shipments as well as likely higher immersion shipments in the back half for DRAM?
Yeah. Thank you, CJ Let me first talk about gross margin for EUV and then give you the wider picture on gross margin. On EUV, what we said, as you know, last year on the systems side, systems gross margin for EUV, we're looking at about 30%. This year we're looking at about 40%. That's what we had in the plan, and that's what we're executing for, and that's also in our models for this year. On the wider picture for gross margin, and I know that many of you are looking at the Capital Markets Day. At the Capital Markets Day, we mentioned 50%. Let me start there.
Let me start at the 50%. Let's look at what the circumstances were at the time, what the circumstances are today. Then we will start talking about how we see this further unfold and what the potential is that we see for this year. Back in November at the Capital Markets Day in November 2018, I think there are three things that we should bear in mind. First off, at that stage, what we modeled at that stage for you, and that had the 50% in there, was what we called a mid-market growth scenario. I think in all likelihood, if we look at the circumstances today, if we look at the circumstances of the memory market today, I think it will be hard to say that for the full year, we're looking at a mid-market scenario. It's definitely not what we're looking at today.
Depending on when it's going to come back, and as Peter said, how it's going to come back, you could still on average see a mid-market scenario. At this stage, I think it's hard to say that we're looking at a mid-market scenario for that. At this stage, the memory market in these months is fairly flat. That's one important circumstance, I think, to recognize. The second thing that I think changed from November 2018, I think Peter already responded to that in the first question, is the multi-beam. The delay in the multi-beam, where I think we're going to see a commercial application and commercial sales of multi-beam only in 2021. Where at the Capital Markets Day as you know and as we already told you also last year, we were still looking for 2020 as commercial application of multi-beam.
That's in essence shifted with a little under one year. That's the second circumstance to bear in mind. The third thing that deviates a little bit from what we told you in the Capital Markets Day November 2018 is a bit of accounting issue, and that has to do with High-NA. We are preparing for High-NA, not just on the R&D side, but we're also preparing for High-NA on the manufacturing side and on the supply chain side. We are incurring cost there that we cannot capitalize and have to run through cost of sales, which is a bit weird because we're not selling High-NA, but nonetheless, that's what the accounting rules dictate you to do. That represents a little short of 1% alone in gross margin.
Those are three things to bear in mind that might be different from the perspective and the model that existed in November 2018. Now let's look at this year, and let's look at the 46%-47% that we have for Q1, and let's look at the potential for the rest of the year. I think there's a number of drivers in there that I think could further drive the gross margin up. The first one obviously is related to the situation in the memory market. That is pretty important, not just for the top line, but it's also very important for gross margin because if we see a solid recovery in the second half of the memory market, that will have a significant impact on the sale of immersion tools, which as all of you know, comes with pretty high gross margin.
Also with a good recovery in our voltage contrast business, which is, as you know, is also very much tied to the memory business. Again, that is a very high-margin product that we have. If we get the recovery in the memory market, that will also have a significant impact on the gross margin on those two elements alone. The second element that is expected to further drive up the gross margin throughout the year is the EUV service. EUV service for the full year is still expected to have negative gross margin. Quarter-over-quarter, you will see a sustained improvement in the gross margin that we have on EUV service for two reasons.
First off, because as you know, with a number of customers, we have the pay per wafer model, and to the extent that EUV continues to go into high volume manufacturing, obviously we get more revenue. Secondly, the cost that we have per EUV machine goes down because we get more efficient in doing it, and we also get scale effects. As a result of that. Quarter-over-quarter, you will see that the EUV service margin will improve. For the full year, it's still negative, at least in Q4, maybe even a little before that, we will see that it starts to become positive. A third improvement that we expect to occur in the course of this year will be the introduction of our new immersion tool, the TWINSCAN NXT:2050i, which again comes with a good improvement of our gross margin.
Those are three significant drivers that we have that we believe give us a good shot at achieving a significant improvement of our gross margin, particularly in the second half, and a good shot at the 50%. This 50%, we would be able to then sustain further into 2021, which, as Peter already alluded to, we think is going to be a very busy year. A very busy year at that stage would also come with a number of scale benefits in our gross margin, better fixed cost coverage. Also in 2021, we would see the introduction of a successor tool on EUV that would also again come with gross margin improvement. We think the momentum that would be created in the second half around reaching the 50% would then be further sustained and elaborated on in the 2021 timeframe.
Very helpful. If I may follow up, I think the buyback announcement, as we're moving into high volume manufacturing EUV, seems to be a bit of an inflection here for ASML moving into cash cow mode. Curious if there are metrics that we should be looking at, whether it's free cash flow margins or working on working capital or perhaps CapEx intensity coming down as you've invested in EUV capacity, High-NA multi-beam. Curious if there's kind of metric we should be looking at to gauge free cash flow in the coming years.
In the short term, you will see that with the increase in EUV and EUV becoming increasingly important for the company, given the cycle time of EUV is significantly longer than DUV. In the short term, you may expect that inventory levels will continue to go up a little bit. That's the dynamic on the one hand. Also in the short term, as I think we've said before, you might expect that the CapEx level that you've seen for 2019, that will be above the CapEx level that you might expect for this year and for next year. Around EUR 900 million to EUR 1 billion is the CapEx number that we think is likely for this year and for next year. Those are dynamics in, I would say, the short term.
In the years thereafter, I expect CapEx level to level off and actually go down because then, the significant preparation for our future, I think CapEx-wise, I think will have been done and will be able to go down. Also, I would expect that in one to two-year timeframe, you might expect inventory levels to go down for a number of reasons. First off, at that stage, we talk about 2021. There is a significant buildup of our capacity for EUV. Once you're there, and once you are at those levels, then the further buildup of inventory will no longer have an impact on your inventory levels.
Secondly, as we mentioned before, the way we believe we will be able to get an increase in our capacity from the 35-ish that we have for this year to the 45-50 that we talked about for 2021, will primarily be the reduction of cycle time. Of course, the reduction of cycle time will then kick in and further reduce working capital requirements. In the short term, I think the burden on working capital will still be there. In the longer run, so let's say one and a half, two years, you will see that we will be able to get it under control better.
One final dynamic as far as that is concerned, we talked to you in previous calls on the introduction of down payments for EUV, which we're pushing and where we had the initial accomplishments on that in 2019. We'll continue to drive that. That will also be a mitigating factor, if you like, in our working capital burden. Long-winding answer, just to tell you that in spite of all of that, in spite of this preparation for the growth, we're still looking at a pretty healthy free cash flow development both this year and next year. The years thereafter, I think the free cash flow that we generate will further increase substantially. All of that taken together give us more than enough comfort to introduce this EUR 6 billion program for the next three years.
Okay, next question, Krish Sankar. Please state your company name followed by your question.
Hi, it's Krish Sankar from Cowen, Roger, thanks for the detailed comments on gross margin. I have two questions. First one on EUV. Peter, it looks like your commentary on 2020 EUV of 35 systems at EUR 4.5 billion revenue, and next year about 45-50 systems capacity is similar to about three months ago. I was under the impression that over the last three months at the margin, there was more incremental demand from DRAM for EUV. I'm just wondering, is it an issue that you're still capacity constrained and that's why you cannot ship more, or are you being conservative? My second question is, how to think about DUV units this year relative to 2019. Would it be similar levels, lower, higher? Any color would be helpful.
Yeah.
On your first question, actually, I tried to answer that in an earlier question, that the issue with 2021 is to make sure that we can reduce the cycle time so much that we can create a capacity between 45 and 50 units, which is a capacity issue, not a demand issue. Yes, DRAM will be there. We just have to make sure that we can squeeze as many EUV systems out of our available square meters so that we can fulfill the customer demand. It's not a demand issue, it's a capacity issue. In that sense, that's why the commentary is similar to what we did last quarter, because the capacity lead time, unfortunately, is a lot longer than just the customer order lead time. Now, on the DUV units, good question. Very much hinges on, and Roger said it, on the recovery of the memory market.
Memory is still very much driven by immersion. Of course, we are seeing with the increased number of EUV systems in logic, some cannibalization because of multiple patterning schemes that will actually move on to single patterning EUV schemes. If we would not see a recovery of the memory business, which I do not expect, because we do expect a recovery, clearly the DUV units would be down. It's really the timing of the memory recovery that will determine how much of the DUV units we're going to see this year. It is really hinging on the timing of the memory recovery.
Got it. Thank you very much, Peter.
The next question, Alexander Duval. Please state your company name, followed by your question.
Yes. Hi there. Alexander from Goldman Sachs. I just wanted to ask or clarify on the extra R&D and SG&A for the first quarter that you've guided to versus where the street was. I'm just wondered if you could help decompose a bit the most important drivers and sort of what underpins them. For example, to what extent is this more about investing in faster cycle times for those 50 or 45 to 50 units for 2020 to 2021? To what extent is it about increasing functionality of future EUV versions as we move beyond the 3,400C? To what extent does it hinge on any other key factors? Many thanks.
On R&D, last quarter, we were at EUR 516. This quarter, we're guiding EUR 550. I think the vast majority of that increase is labor cost increase. As we mentioned before, at this stage, we have about a capacity that we think we need in order to accomplish the R&D objectives that we have, which are all of the things that you just mentioned, but primarily focused on the low-NA, High-NA roadmap, multi-beam, but also a number of developments in DUV, obviously. We think we have the capacity that we need in order to get that done, and that's why we said on previous calls, expect the going-out rate for the fourth quarter to be the basis, and then obviously it needs to be what we call inflation-adjusted, which is obviously linked to wage increases for that. That's what you see.
The 6% increase from EUR 516 to EUR 550 really is primarily the wage increase on the R&D department. In terms of SG&A, I think is modeled at EUR 140, which is, I think, very much in line with what you saw in previous quarters. Q4 had a little bit of a spike, and there was some accounting adjustments in there in Q4 that had a little impact. Also we're in the process of implementing a new IT system, which has some impact on the SG&A number. Those are small things. The guidance of EUR 140, I think, is pretty much in line with where we were in previous quarters, and obviously there too, the wage impact.
Yeah. I think on the wage, in fact, I don't think we give our people 10% wage increase because that will probably raise a lot of questions when people start listening to this call. It's the combination of the normally inflationary wage increase and the fact that, of course, we added people in 2019, so you see the full-year effect now of that growth in R&D in 2020, which actually happened throughout 2019. Now you see the full-year wage effect.
Very clear. Many thanks.
Next question is from Joe Quatrochi. Please state your company name, followed by your question.
Yeah, thanks. It's Wells Fargo. I had a question on the memory side. I know that ASML has historically been more tied to DRAM than NAND. I was hoping you could kind of help us kind of parse out the comments that you've made in terms of the recovery and seeing potentially improved bookings kind of looking into 2Q. Is that more of a NAND flash comment, or should we think about that from a DRAM perspective?
Yeah. I think if you split the DRAM and NAND, and like I said, the proxy that we have is basically looking at some utilization data. I think it is first noticeable in 3D NAND. I also think the slope of the recovery is a bit more aggressive in NAND than in DRAM. They are both there. Having said that, I also mentioned in my prepared remarks that customers were creating underutilization in order to make sure that they could rebalance the supply and the demand in the memory space sooner. That underutilization correction was also deeper in 3D NAND. I was also not able. Big surprise that, of course, the return slope back up is also a bit steeper. That's where we are. I think if we had started a bit earlier, it's a bit steeper slope, but both are trending in the upward direction.
Okay, that's helpful. Just for the March quarter guide, I was wondering, in the past, you guys have given us the EUV shipment and revenue expectation for the quarter. I was curious if you could give us that for the March quarter, because I know that there's four systems that are included from 2019. Maybe just any thoughts on the cadence for 2020 just now, given that the 3400C is available for the full year.
Yeah, I think we've indicated in the past that as soon as we really see that EUV is in essence going into high volume manufacturing, and also at a point where we can take revenue upon shipment, that would be the time where we're no longer going to give separate guidance on EUV shipments. That is clearly the case by now. That's the reason why starting this year, we no longer do that. What we do, however, is continue to indicate, at least for 2020, for the full year, we give you an indication of the euro value and also the number of units for EUV for the full year. Of course, we will report quarter by quarter. We will report to you what the euro number and the unit number of that will be.
As it relates to your second question on how is it distributed over the year. It's not completely evenly distributed. It's a little bit tilted towards the second half, but modestly so. It's not as exacerbated, if you like, as we had it in 2019. There's a better balance, but it is still a little bit tilted towards the second half of 2020.
Thank you.
Next question, Janardan Menon. Please state your company name followed by your question.
Hi, good afternoon. It's Janardan Menon from Liberum. I just have two follow-ups on the logic side, especially on the EUV front. Sorry, on the DUV front. If I take your growth number for this year, just calculating EUV revenues alone, you're going from EUR 2.8 billion to EUR 4.5 billion. If I put that EUR 1.7 billion of additional revenue, which on the logic revenue of EUR 6.6 billion last year, that's about 25% of additional growth in logic for this year. I'm just wondering, given that DUV shipments will probably come down during the year, what kind of a decline in DUV are we looking at? Is it three or four units? In which case you will still be growing your logic revenues at about 20% or higher?
Will the DUV drop you a bit more than that, and you could be sort of in the 10%-20% range of growth on your logic side? A short follow-up is on the EUV capacity. Given that you're seeing so much demand for EUV right now, and you're saying that in 2020, it's more a capacity issue, otherwise you probably could ship 50 units. What can you do to add capacity further into 2022? If this kind of strength, and especially if the memory market, the DRAM market comes in more strongly for EUV in 2022 and the logic strength continues. Is there a scope to further reduce cycle time to take your unit shipments above 50? Can you use some of the bays in your High-NA new facility for low-NA systems if that were to be required?
Is there a way that you can go above 50 units by 2022?
Let me take the first question, and then Peter can go into the second question. If you piece together the data points that we gave you, I think you can find sort of an answer to your first question. As Peter said in the video, we're looking at a double-digit growth. If you do that math, then you have a number that you arrive at. Then we gave you two other important components. On the one hand, indeed, as you mentioned, the EUR 4.5 billion for EUV. Also on this call, gave you the 20% increase over the installed base, which would get you to approximately EUR 3.4. Then you can sort of calculate where DUV and apps combined, where they would land for the year. I think that's what you're looking at.
Of course, as Peter said, it will be dependent upon the timing and the extent of the recovery. We believe that the number that you derive in that way is a safe number to go by with some potential, obviously, if the memory recovery is significant and timely.
Yeah.
Peter, second question on the-
Yeah, second question on when you look at the EUV capacity that we currently have, and like I said, it will be driven by cycle time reduction, because the lead time reduction for capacity adds is longer. Yes, for beyond 2021, we are looking to bring the output capability above 50. Now, which is what we're doing today, is really looking at how much should that be. If we have to go over 60, we probably need to extend square meters of production capacity at ASML and at suppliers, which could be, like you said, using production facilities that we're currently building for High-NA, use that for temporary use in a low-NA. Although we'd like to prevent that because you have been here, Gennady, and these bays are different.
The high-NA bays are different than the low-NA bays, so that would mean some extra cost, but if push comes to shove, you could probably do that. I think with our cycle time program, we can go over 50. I think it will be very difficult to go over 60. If we have to go over 60, then we probably need to quickly add some manufacturing capacity at ASML and at the supply chain.
Understood. Thank you very much.
Next question is Amit Harchandani. Please state your company name followed by your question.
Hello, everyone. Amit Harchandani from Citi. Thanks for letting me on. A couple if I may. The first question goes back to the demand for EUV. You've talked about, obviously, the capacity for 2021. In terms of the drivers of demand from your customers, what really do you think is changing or accelerating from a customer standpoint? Is it the number of layers of adoption? Is it the pace of cadence down the nodes? Is it their end customers push? Could you give us a sense for what really do you think is driving this optimism and acceleration towards the number that you've talked about for 2021 from a customer standpoint? I have a follow-up.
Yeah. Amit, the easy answer is all of the above. We are seeing increase of layer counts. The fact that EUV works also gives the customer the confidence on their roadmap. When they get confidence on the roadmap, you see a cadence change. There's a push, maybe a pull in. Also, I think the discussion that we're having with customers and customers having with us, although without going to very specific customer details, we will never do that. It's very clear that the number of tape outs and the requests from their customers, our customers' customers, on different types of end applications is going up. It is basically the combination of those three things. You've mentioned them all. That's what it is.
I think we are responding to what our customers are asking us based on those three drivers, and this is why we come to the problem of more the capacity issue in 2021 than a land issue.
Thank you, Peter. Secondly, if I may, could you maybe give us your latest thoughts on how you're thinking about demand from indigenous customers in China? There's obviously news flow around the EUV tool, and you've commented on that very clearly. More broadly, as you think of your 2020 guidance and 2021, things have changed a bit since November 2018. Any clarity on demand from indigenous China, the various end markets, and what are baked into your assumptions right now?
Yeah, I think things have, of course, changed. The market's changed since November 2018. I would say on China, we're pretty much on target in terms of the strategic rollout, especially in the memory space. In logic, it's not so much the leading-edge logic, it's more the mature logic systems. I think our current assessment of the market is a bit higher than it was at 2018. If anything, at the leading edge, we're on plan. On the trailing edge, in a more mature technology, we see an upside. No downside.
Okay. Thank you.
Next question, Achal Sultania. Please state your company name followed by your question.
Hi, good afternoon. It's Achal from Credit Suisse. Roger, maybe on EUV services gross margins. Just trying to understand how much of a headwind it has been at a group level in 2019. Obviously, you made a comment that it will still be loss-making this year, but basically, we have improvement through the year. I'm just trying to understand how much of it has been already a headwind in 2019, and then how quickly can that business ramp up towards a 40%-45% services gross margins like you have most likely in DUV? Does it take two years, three years? Any color around that would be helpful.
Secondly, on the mix, when you talk about this 45-50 capacity for EUV in 2021, can you help us understand, obviously foundry is a big part of that number, but what are you hearing from logic and DRAM customers in terms of the unit breakdown of that 45-50 unit number? Thank you.
Thank you. On the gross margin impact of EUV service, that was around 2% for 2019. That's the gross margin impact of that in 2019. As I mentioned to you, we do see it coming to a positive number in the course of this year. I expect it to be at least Q4, maybe even before that it will turn positive. Before we have EUV service gross margin at the corporate gross margin level, I think we're probably two, three years away from that. The aspiration clearly is there to have it at that stage in that timeframe.
Yeah. On the split, the 45-50 unit split. I cannot give you any details, but the majority, and I mean it's above the 50%, is going to go to the logic space. It's still dominated by logic, but clearly, the 2021 numbers for DRAM will go up. It's logical. In the logic space, we have several customers. In the DRAM space, we still have one. That's also an issue that will drive the division between logic and between DRAM. It is going to be significantly above 50%, it's going to be logic.
Okay. Thank you, Peter and Roger.
Next question, Adithya Metuku. Please state your company name followed by your question.
Yeah. Good afternoon, guys. It's Bank of America. I had two questions. Firstly, just thinking about the memory demand as we go through this year and into next year. Obviously, when we look at the last two years going into 2018, we had a very strong increase. We had a doubling in 2017 and then another 50% increase in revenues from memory customers in 2018. Now, when you look at the next two years, obviously, things may be a little different. I know it's very difficult to give a pinpoint number, but I just wondered, what do you have in your scenarios internally? If memory were supposed to come back in 1Q, where do you see memory revenues for this year? If it were to come back in 3Q, where do you see that coming?
Any color that you can give us to help us get a rough sense of where we might end up would be very helpful. Secondly, just a question for Roger, just on the OpEx. Just wondered if you could confirm whether the OpEx annualizing 1 Q number would be a good proxy for the full year or whether there is anything else we need to think about. Thank you.
Yeah. Well, I think on the question on the memory demand, I did give you some indication. I can repeat myself. When I look at the best proxy we have is to just look at how our machines are being used. That's why I said, I do believe that in the first half, and it's a bit difficult to understand the exact timing or to gauge the exact timing, but we will be in the first half. I think our customers will come back to this more healthy supply-demand balance, and they will see this coming, so they will place orders. For us, it's likely going to be a second-half event. Having said that, 2019 was, of course, a weak memory in our market for us. Also the first half of 2020 could be not very strong, as Roger indicated.
When they come back the second half, we will have an impact on our business also on our financial performance. When the memory comes back, then I think you really need to look into 2021. Like I said earlier, 2021, we do see strong EUV demand. I see no reason why the demand of our customers for leading-edge products in the memory space, 5 nanometer will kick in then will go down. When the memory recovery starts in the second half of the year for us, it will extend into 2021. You would then have had effectively a year and a half of a memory downturn, which I think historically is not long, but not short either. It all seems to fit. This is the kind of call that I can give you based on top of what I already said.
Adithya, on your questions on OpEx, indeed, I can confirm that what we have, so the EUR 550 for R&D and the EUR 140 for SG&A, those are good run rates for the quarters in this year.
Okay. Thank you, guys.
Next question, Sandeep Deshpande. Please state your company name followed by your question.
Hi, Sandeep Deshpande at JPMorgan. Most of my questions have been answered, but just actually a clarification, Peter. Firstly, whenever this memory recovery occurs, you can see how the customer utilization is doing. Your exposure to NAND is lower than your exposure in DRAM. Do you see that NAND utilization is rising faster than DRAM or vice versa at this point? Because that will determine the timing of when your orders come in. Then secondly, regarding the multi-beam tools that you're working on, you think at this point is the view that these will begin shipping in 2021, and thus that there could be even further accretion to the margin in 2021? Thank you.
I think I will do the memory recovery. Like I said earlier, the slope of the utilization recovery is a bit faster, a bit steeper for 3D NAND, but it also come from a deeper point. In that sense, and yes, we have, as you mentioned, less exposure to the 3D NAND market, but the 3D NAND market needs a lot of exposures. This is what we're counting, and this is what we're seeing. I think NAND probably rising a bit faster. Yeah. I think that could be the conclusion. Sorry, the second part was?
The second part was on the gross margin and multi-beam tools.
Oh, it was on multi-beam tools. Okay.
Yeah.
Yeah. Indeed, multi-beam, as I mentioned, multi-beam is expected to be a high-margin product. To the extent that when it will go into commercial application, and that is expected for 2021, we do believe that it will be accretive to our gross margin.
It is not only if you look at an e-beam tool it's a machine, but it is very much compute power, so it's a lot of software also. That's why margins are generally higher in e-beam space than in the lithography space.
Understood. Thank you.
Next question, Alexander Peterc. Please state your company name, followed by your question.
Yes, good afternoon. Thank you for taking my question. This is Alexander from Societe Generale. I'd just like to understand, as you now contemplate 45 to 50 EUV units in 2021, does this in any meaningful way accelerate your path to higher gross margins, closer to DUV for your EUV business overall? There's nothing change in terms of your gross margin scenario for this business unit? Just briefly, your comment regarding the timing of free cash flow generation. It looks like over the next three years, you will have an acceleration in free cash flow generation.
Yeah. Let's first talk about the gross margin for EUV. I do believe a further increase in numbers will improve our gross margin for three reasons. One is obvious, right? To the extent that given the capacity that we have, we have a higher output. Of course, your fixed cost coverage will improve. As we mentioned, the increase in capacity will be achieved by reducing cycle time and will not be achieved by further CapEx. That's why more units will result in gross margin improvement. That's one element. The second element is to the extent that we have more of these tools in the field, that also means that our service margin will, in all likelihood, improve because the more tools we have in one location, the more efficiency we have in having our service crews there. The efficiency per tool will further increase.
Thirdly, not necessarily related to the number of units, but since you talked specifically to 2021, as I mentioned, that we'll also see the introduction of the successor to the 3,400C, which again, we are hopeful will bring such value to our customers that the gross margin will benefit from that. Yeah, a number of reasons why we believe gross margin for EUV will continue to further improve through 2021 and beyond. On the free cash flow question, I think that is right as well. I mentioned to you that in the very short term, the working capital burden that we have from further growing to, let's say, this capacity level that we talked about will be there, because given the cycle time, it will mean that we have to take significantly more inventory on board to get it done.
At a certain stage, you will see the offsetting factor of the fact that cycle times get reduced. I think in this one-year window, you will see a spike and then a leveling off as a result of the reduction of the cycle time. As I also mentioned, we do want to get down payments more as the default in our commercial model, and that should also, at some stage, lead to an offset in the working capital burden.
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. Now, Operator, may we have the last caller, please?
Yes, sir. Mitch Steves, please state your company name followed by your question.
Hey, thanks. RBC Capital Markets. Most of my questions are answered, but I just want to clarify a couple small points. First of all, based on the tone of this call, it sounds like 2021 will probably be an accelerated growth year relative to 2020. I want to make sure that that's a reasonable assumption. Then secondly, and I realize you guys can't time the exact recovery of memory, but from a historical perspective, when you look at when the memory market recovers, what type of sequential growth do you expect from the initial first batch of that recovery, regarding looking at your Q-over-Q number?
Well, I think the second question is almost impossible to answer. The memory business is quite different than it was a couple of years ago. six players in 3D NAND, three in DRAM. Those patterns will also be a function of the composition of that market and the individual position of those companies in that market. That's really difficult to use historical rates as a proxy for what's going to happen now. I think, yes, 2021 could be an accelerated growth year. However, Roger said it, even without the assumption on the recovery of the memory market and the growth of the memory market, we already see a double-digit growth this year based on logic and on this base management. On top of that, we could see a recovery of the memory market. I think you could see an acceleration this year also.
That's not for the full year, granted, because we do expect it in the second half. It will definitely continue, like I said earlier. Once this memory market recovers, it doesn't recover for two quarters. It recovers for a longer period like it always does, on top of the logic market. Yeah, I think we're looking forward to some acceleration.
Interesting. Thank you.
All right. Thanks. On behalf of ASML, I'd like to thank you for joining us today. Operator, if you could formally conclude the call, I'd appreciate it. Thank you.
This concludes the ASML 2019 fourth quarter and full year financial results conference call. Thank you for participating. You may disconnect.