Ladies and gentlemen, thank you for standing by. Welcome to ASML 2017 second quarter financial results conference call on July 19, 2017. Throughout today's introduction, all participants will be in a listen-only mode. After ASML's introduction, there will be an opportunity to ask questions. I would now like to open the question-and-answer queue. If you have 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 Mr. Craig DeYoung. Please go ahead, sir.
Thank you, Peter. Good afternoon and good morning, ladies and gentlemen. This is Craig DeYoung, Vice President of Investor Relations here at ASML. Joining me today, as always, from our headquarters here in Veldhoven, the Netherlands, is our CEO, Peter Wennink, and our CFO, Wolfgang Nickl. As a reminder, the subject of today's call is ASML's Q2 2017 results. The length of the call will be 60 minutes, and questions will be taken in the order that they are received. This call is also being broadcast live over the Internet at asml.com. A transcript of management's opening remarks and a replay of the call will be available on our website shortly following the conclusion of the 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.
Good morning, good afternoon, ladies and gentlemen, thank you for joining us for our Q2 results conference call. Before we begin the question and answer session, Wolfgang and I would like to provide an overview and some commentary on the recent quarter, as well as provide our view of the coming quarters. Wolfgang will start with a review of our second quarter financial performance with added comments on our short-term outlook. I will complete the introduction with some additional comments on the current business environment and our future business outlook. Wolfgang, if you will.
Thank you, Peter, and welcome everyone. I would like to first highlight some of the second quarter financial accomplishments and then provide our view for the coming quarter. Q2 net sales came in at EUR 2.1 billion. Net system sales accounted for EUR 1.38 billion, showing another quarter of increasing memory business, which is now at 54% of net system sales, and also a strong quarter of logic sales, which represented the remaining 46%. Installed base revenue for the quarter came in stronger than expected at a level of EUR 717 million, driven by major DUV and Holistic Lithography upgrades. For the first half of the year, our total installed base revenue is already at EUR 1.45 billion compared to a full-year sales of EUR 2.12 billion in 2016.
Our gross margin for the quarter came in at 45%, slightly higher than guided, driven by a higher top line and a favorable mix. Overall, OpEx came in as guided, although R&D expenses came in slightly lower at EUR 313 million, and SG&A expenses came in slightly higher at EUR 102 million, driven by litigation expenses. Turning to the balance sheet. Quarter-over-quarter cash equivalents, and short-term investments came in at EUR 2.51 billion. As a reminder, in Q2, we had several extraordinary cash outflows, which have brought the overall cash balance back to our target level. We paid a dividend of EUR 1.20 per ordinary share, or approximately EUR 517 million in total to our shareholders.
We also have repaid a maturing bond with an outstanding balance of EUR 238 million. Lastly, we have closed the acquisition of a 24.9% interest in Carl Zeiss SMT during the quarter for EUR 1 billion. Moving on to the order book. Q2 system bookings came in at EUR 2.37 billion, including orders for eight 3400 EUV systems from two customers. Six of the EUV orders came from one customer for use in both logic and DRAM. Total bookings were almost EUR 500 million higher than in the previous quarter. The breakdown of the bookings of 60% logic and 40% memory is the same as in the previous quarter. The strong bookings in the logic sector are in support of the 10-nanometer ramps and in support of the EUV insertion at the 7-nanometer node.
Memory bookings, mainly in DUV, strengthened further from its strong Q1 level, supporting an expected 50% year-on-year revenue growth in the memory sector in 2017. The continuing order flow for EUV systems increases our EUV backlog to 27 systems valued at EUR 2.8 billion. Our overall systems backlog now stands at a record EUR 5.35 billion. After two strong quarters in 2017, in combination with a record order book, we are now expecting full year's net sales, which are up approximately 25% from our previous record revenue of EUR 6.8 billion in 2016. 2017 revenue is driven by continued strong demand for our entire portfolio, driven by both logic and memory. With that, I would like to turn to our expectations and guidance for the third quarter of 2017.
We expect continuing sales strength in Q3 with total net sales of around EUR 2.2 billion, including an estimated EUR 300 million of EUV revenue. We plan to ship three NXE 3400 in the September quarter. Our EUV shipment plan for the full year remains at 12 systems and is back-end loaded. We expect our Q3 install base revenue to come in around EUR 600 million, driven by continued demand for Holistic Lithography options, high-value upgrades, and our growing install base. For the full year, we expect our install base revenue to be up by approximately 20% versus the 2016 levels. Gross margin for Q3 is expected to be around 43%. Excluding EUV systems, gross margin is approximately going to be at the same levels as in Q1 and in Q2.
R&D expenses for Q3 will be around EUR 315 million, and SG&A is expected to come in at about EUR 105 million. SG&A includes expected increases in legal expenses. Finally, ASML will resume share buybacks in Q3. As a reminder, we had paused our share buyback program for about one year to acquire HMI and a minority share of 24.9% in Carl Zeiss SMT. We have EUR 1.1 billion remaining for 2017 from our previously announced share buyback program. We do not expect to execute the entire remainder of the program in Q3 and Q4. In line with our policy, we will return excess cash to our shareholders through share buybacks, and we will make announcements on future share buyback programs when appropriate. With that, I'd like to turn the call back over to you, Peter.
Thank you, Wolfgang. As Wolfgang has highlighted, our business continues to perform very well. We expect our positive momentum to continue throughout the year based on the market environment and the related strong demand for our products. This should deliver another record year with net sales growth expected at about 25%, representing one of the strongest gains in annual revenues in our history. While Wolfgang reviewed our current quarter performance and outlook for the coming quarter, I would like to provide some commentary on the longer-term outlook of our market drivers, followed by an update on the progress and plans for our product groups. Let me start by making some comments on the market drivers and their impact on litho demand. As we have moved further into the year, the demand for memory has continued to strengthen, especially noticeable in DRAM.
We are on track to see our memory revenue grow by around 50% year-on-year, creating the highest memory demand in history of ASML. We need to remember that this growth in spend is coming off a year and a half of wafer capacity reduction due to significant underspend in 2016, and relocations of leading-edge tools to 3D NAND, that combined with a significant end market demand growth this year. In 3D NAND, the industry continues to witness a number of greenfield fabs that are ramping, which is driving very strong lithography growth. Logic demand for our tools is expected to grow around 15% year-on-year, driven by the continued ramp of 10 nanometer, as well as the start of the 7-nanometer node, which is particularly driving the logic growth this year as it concentrates on the planned EUV adoption.
With regards to China, we've been doing business in this region for over 25 years, and we currently have over 600 employees in 11 cities, supporting an installed base of more than 400 lithography systems. We also have two R&D centers in China and are working to deepen our relationship with Chinese semiconductor industry customers by collaborating with industry consortia. We signed a memorandum of understanding with the Shanghai Integrated Circuit Research and Development Center, ICRD, a public research consortium dedicated to the advancements of the semiconductor industry in China, to set up a jointly owned world-class applications and training center in Shanghai. We have seen continued revenue growth for China, from China over the last five years by both domestic Chinese as well as non-domestic companies, and we see a lot of opportunity for growth in this region going forward.
However, as we mentioned on earlier occasions, the speed with which this growth will translate into sales and earnings is dependent on the ability of our new Chinese semiconductor customers to effectively bring qualified and competitive products in volume, to the market. This might take some time. We are currently in discussions with five domestic logic and memory customers, which per their published fab plans, translates into a lithography opportunity of more than EUR 3 billion. This opportunity led last quarter to bookings from a new Chinese domestic memory customer for shipments later this year. In summary, we will see significant growth in memory demand versus prior year, and logic will build further on the healthy demand levels seen in 2016, largely driven by EUV, with China providing a meaningful medium-term growth opportunity.
Installed base revenue continues to grow at an even greater rate than last year, driven by broad-based adoption of high-value field options and upgrades. Finally, demand will be further accelerated with the EUV adoption as customers start ramping this technology in volume production. Our current view is that the positive business trends that we're seeing in 2017 are likely to continue as we enter 2018. On the ASML product side, let me start with an update on our EUV business. In EUV, we continue to make progress as planned. We now have demonstrated all key performance specifications on our NXE 3400 system. This includes a throughput of 125 wafers per hour. We also demonstrated 250 watts of source power, enabling productivity improvements beyond 125 wafers per hour.
Availability continues to make progress towards the 90%+ target with continued focus on reducing the variability. We now have a system configuration that provides all of the agreed product specifications, which will enable us to now focus our work on executing on the planned availability improvements that will drive broad-based EUV insertion in mass production. In addition, clear progress on the ecosystem continues as communicated by many of our customers. We have produced zero defect pellicles, and our customers continue to make progress on photoresist sensitivity, enabling higher wafer per hour productivity. Based on this progress, customers are now more and more confident in inserting EUV technology in manufacturing, as clearly indicated by the continued order flow.
Our Deep UV business is expected to grow this year off a record revenue in 2016, fueled by the demand for our immersion and KrF products in both logic and memory. We announced our latest TWINSCAN NXT:2000i immersion system at SEMICON this past week. This new Deep UV immersion system features several hardware innovations that deliver improved imaging and overlay performance in support of aggressive lithography requirements on future nodes, including mix-and-match with EUV. We're also seeing exceptional demand for our KrF products, notably in 3D NAND. In Holistic Lithography, where we bring together scanner, metrology, and software to provide high-value process control solutions for our customers, we expect sales to grow about 50% from last year.
We have announced our latest metrology system, the YieldStar 375F, featuring new optics technology that generates more accurate data at a higher speed, providing increased quality data to feed the process control systems. In addition to YieldStar metrology systems, we're also shipping HMI e-beam systems in support of 3D NAND voltage contrast and defect inspection applications at both memory and logic customers. Product integration of HMI is progressing well with pattern fidelity metrology e-beam tools being evaluated by customers, which enables pattern fidelity control capability in support of the 7-nanometer node. To further drive productivity improvements in the e-beam area, we're in the process of developing a multi-beam system that combines leading-edge e-beam technology with ASML's unique stage and computational lithography technology. Finally, we also closed the acquisition of a 24.9% interest in Carl Zeiss SMT.
The main objective of this agreement is to strengthen our longstanding partnership with Carl Zeiss and facilitate the development of the next-generation EUV lithography system, which we call High NA, due in the first few years of the next decade. This technology should enable the semiconductor industry to produce much higher performance microchips at lower costs, supporting customer roadmaps throughout the next decade. In summary, great first half of the year with strong industry demand across all market segments, translating to very strong growth across our complete product and service portfolio for 2017. In previous quarters, we mentioned how we felt we passed an EUV inflection point. We now see volume orders from all segments of the industry, clearly marking an increased rate of adoption, with order flow expected to continue providing significant EUV growth in the coming years.
As mentioned earlier, our current view is that the positive business trends that we're seeing in 2017 are likely to continue as we enter into 2018. With that, we'll be happy to take your questions.
Thank you, Peter. Ladies and gentlemen, 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, Peter, operator Peter, could you we have the final instructions and then the first question, please.
Of course, sir. Thank you. Ladies and gentlemen, at this time, we will begin the question-and-answer session. Again, if you have a question, please press star one to register for a question and star two to withdraw a question from the queue. If you're using speaker equipment today, please lift the handset before making your selections. One moment please for the first question. The first question comes from David Mulholland. Please state your company name followed by your question.
Hi, thanks very much. It's David from UBS. Just first question just on EUV. Obviously, very good progress in the quarter, and you put a slide in the presentation showing potentially for 15 or 16 layers that you could see EUV adoption out for 7 nanometers. I just wonder if you could let us know, you know, have there been any changes on your base case assumption for how many layers you could see EUV adoption at 7 and maybe even 5 nanometer as well, in logic? Then I have a follow-up.
Okay. I don't think we currently have any reason to change our base case, you know, assumptions.
We put that in there because the, the additional five or six layers that you have identified on top of the 10 that has been communicated earlier are really layers that in the discussion with the customers we have identified as potential additional layers. Now, that really depends on the speed with which we will be able to mature our EUV systems and drive the productivity up and the availability. 'Cause it's all a matter of cost. You know, I think the initial 10 are driven, you could say, by the lithographic needs. You just need to use EUV. The additional five to six will be a function of the, you know, productivity and, you know, maturity of the tool.
That's great. Thanks very much. Secondly, just one on the commentary on the installed base and field option sales. You know, you've obviously had very strong growth in H1, and given the commentary that you've given for the full year, maybe my math's wrong, but it was just a bit of a slowdown in the absolute level in the second half. I wonder if you could just comment on why you feel that's the case, or maybe correct if I've got something wrong on that.
No, David, this is Wolfgang. First of all, you're right. I mean, we had an extremely strong start into the year. We have had EUR 1.45 billion versus EUR 2.1 billion for the total last year. It was an incredibly strong start. We think we'll be up 20%. Be reminded last year we said we think we grow that business by approximately at a, the run rate of 10%, which we were very close to last year, but this year it's 20%. Within that, you have approximately half service and half options. The service piece is pretty stable with an upward trend because it's a function of the installed base to a large degree. The upgrades are a little bit more volatile.
That has largely to do with when our customers can afford to take the upgrade. Because you got to remember, some of these upgrades take their machines down for five weeks or so. When they're firing on all cylinders, they, even though they see the great impact of the upgrade down the road, in the short run, for their years, they can simply not afford to take the machines down. That is why in the second half, the options piece is coming down a little bit. Having said that, we are having a fantastic year there, 20% up year-over-year. I think the trend of continuous growth there will also go into next year.
That's great. Thanks very much.
Good.
The next question comes from Mr. C.J. Muse. Please state your company name, followed by your question.
Yeah, good afternoon. Thank you for taking my question. I guess first question, you know, as you look at the building EUV backlog, clearly confidence is rising with your customers. We'd love to hear from you as to, you know, what kind of improvements you're showing, particularly on the reliability and uptime side, which I think is the clear factor that is causing these guys to come in. We'd love to hear, you know, over the last three months, what kind of data you've seen.
I think the most important factor that drives the confidence of the customer is actually in the key performance requirements that they need. It's in 250 W, which we showed, 125 wafers per hour. Of course, we make gradual progress in the availability and in the maturity of that system. Like we said on earlier occasions, that is going to bring us to a situation where we have that when they ramp in volume in 2019. They know that will take some time, just like the development of the ecosystem for them with photomask and with photoresist will take some time also.
The most important part that actually drove their, you know, confidence goes down to meeting all the key lithographic performance criteria. That drove, you know, the confidence. We now have a system that actually has all, you know, that performance in it. We just have to make it, you know, a bit more reliable. That will take a bit of time, and we all know it. There's a whole program driving it. It's as simple as that.
Very helpful. I guess as my follow-up, in terms of your commentary around memory, in particular DRAM, can you specify, you know, how much of that strength you're seeing across the board on shrinks versus 2D NAND upgrades over to DRAM? Thank you.
Difficult question to answer. I'm not going to guess in this particular case. What we are seeing, it's both a combination of filling up available spots, open spots in current DRAM fabs because of the market demand and the technology transitions that are happening across the customer base. You know, some customers are leading in that sense, and other customers are followers, but all do technology migration. It's across the base. It's a mix whereby open spots in, you know, open pedestals in factories are currently being filled because of the strong end market demand in DRAM, which is particularly driven by the data centers. It's a combination of, you know, both.
With DRAM that strong, there are not that many 3D NAND relocations happening. That happened in the past. That happened over the last 18 months, which created this space to backfill with leading-edge lithography systems to address the rising demand in the DRAM market.
Very helpful. Thank you.
The next question comes from Mr. Andrew Gardiner. Please state your company name, followed by your question.
Good afternoon. Andrew Gardiner with Barclays. Thanks for taking the question. I was just wondering if we could revisit sort of the 2020 model, particularly in light of the strength you're now seeing in 2017 and your expectation heading into next year as well. If I go back to the scenarios you guys outlined at the capital markets day last year, you gave us sort of four scenarios that supported the EUR 10 billion revenue pre-HMI. Given what we've seen, some of the lower sort of end scenarios that you had there seem particularly unlikely. I mean, first of all, you guys have talked about the sort of layer count initial layer count expectations of EUV. You're now seeing sort of 10 layers.
At the time, we were still between six and 10, so you know, six layers, quite a bit lower at the low end of that range. Similarly, in terms of the end market assumptions you are using for your lower demand scenario, a 20% sort of node-on-node decline in wafer start seems particularly cautious at this point, given what we're hearing in terms of new compute applications and the greenfield build-outs you've highlighted in China. Given that, I mean, do you not think it's more fair to look at the two higher-end scenarios that you've outlined there? We should sort of minimize the likelihood of the two weaker ones.
Well, you did a good job almost providing the answer along with your question. I think you're exactly right. I mean, we were not there standing there in New York last year telling you exactly where to pinpoint the revenue in 2020. We gave you a bit of a model with the key sensitivities and as we have explained, the two major ones when we go through all of our sensitivities are end market demand.
There we have shown a little bit of sensitivity, and you see it's not been that much of an upside and a downside compared to the intensity of the EUV insertion, where we have seen between a high insertion and a low insertion there was quite a bit of a difference. I think the market assumptions provided EUR 1 billion to the upside and then there was another EUR 2 billion or so in the EUV insertion.
I think what we can say is that we were certainly more confident that we're not going to the bottom end of the scenarios because you're absolutely right with everything that you're hearing from the end markets, be it strength in memory, being it new applications, being it like autonomous driving or a move towards autonomous driving, data centers, big data, analytics. Whatever you hear is pointing that the market is not going to be soft. With the recent accomplishments on EUV, if we follow through, which is our intent, to now that we have the specs met, that we get the availability up.
To Peter's earlier point, right now we've highlighted ten layers, for instance, in logic, that will be initial layers. As we move this further up, we have an opportunity to overachieve that number, but it's too early to pinpoint a new number. That's why we have given you these sensitivities. We feel pretty comfortable right now.
Okay. Understood. Thank you. Just a quick follow-up. Wolfgang, in terms of EUV Rev Rec this year, so earlier in the year, you were thinking, if I recall, sort of 1.1.2. Is that still a reasonable number given sort of shipments and tool performance in the field?
Yeah. A good question. I mean, at this point, I would say that it's going to be around EUR 1 billion or so. I mean, it has to do with a whole bunch of things. like, you know, since last year, we can recognize majority of revenue with shipments, but there are still a few things that we need to defer. For instance, if the installation with a little bit if we provide a longer warranty, we have to defer a little bit there. We also got to really look at whether the configuration at the factory test is exactly the same as when we retest the system at the customer site.
Which in a time when you're struggling to get all your material together can sometimes a small pieces are different at the final test at the customer. there we have to defer again by a quarter or so. I'd be careful there and say even though we ship 12 systems and even though we get some deferred revenue in from the past we will still have to defer a bit of the revenue in into 2018. if you want to be on the safe side I would count on a billion right now. Of course this has the effect now that we have a stable configuration next year we are going to recognize in the same year plus we then have a catch-up.
I mean, we'll have a deferred revenue balance at the end of this year. Next year, not only will the shipments go up from 12 to 20+ for upgrades, but we're going to get EUR 200 million on deferred revenue. I would plan on EUR 1 billion right now, Andrew.
Okay. Thank you, guys.
The next question comes from Mr. Alex Duval. Please state your company name, followed by your question.
Yes. Hi, everyone. Alex Duval from Goldman Sachs, and congrats on the strong quarter. Just a quick question on HMI. You've talked a bit more about the innovations ASML is driving to combine E-beam with your computational lithography. Wondered if you could give a bit more color on the key technical aspects you're working on. What's the feedback been from customers you've been discussing with, and anything else important on the solution so far? Many thanks.
Yeah. Thank you, Alex. The development we're seeing there is that we're executing as planned as part of the acquisition on the combination of the ASML lithographic, let's say the holistic lithographic computational competence with the e-beam competence that HMI has in the field. The combination of those two, the first product will be shipped in the second half of this year and will be evaluated by three customers. With positive evaluation, we will then start to ship that product in 2018, which will be a single beam tool combined with the ASML computational lithography competence. Now that will be an intermediate solution and will be focused on defect inspection or wafer inspection.
That will follow by a more, let's say, economic, solution, which will involve multi-beam. A multi-beam, which actually has the advantage of being able to inspect the wafer much faster, which will bring the cost of inspection down. Area or the surface that you can inspect will also go up, which will also have, which could have a very positive impact on the customer yields. That is a product that is then scheduled for, the year thereafter, so after next year. Whereby, it's not only the multi-beam column, because if you have a, capability to move the wafer faster, you also need faster stages.
Lo and behold, there is one company on this planet that's very good at fast stages, and that's ASML. I think fast stage technology combined with good computational lithography, combined with a multi e-beam solution will give a very powerful solution for wafer inspection going forward.
Very helpful. Thank you.
The next question is coming from Mr. Mehdi Hosseini. Please state your company name, followed by your question.
Yes, yeah. Thanks for taking my question. Just going back to EUV, revenue recognition, booking, how should we think, or how should we model deferred EUV revenue by year-end 2017? As shipment starts in 2018, should we think of, revenue recognition on these EUV system, happening, at a earlier time? In other words, would the deferred revenue start to go down? I have a follow-up.
Yeah. Yeah, Mehdi. Hi. Yeah, you're right. For next year on the shipments, you should clearly think of the revenue in the same period as the shipment. Then we currently think that we'll have a deferred revenue balance of up to EUR 500 million at the end of this year. Again, next year you will very likely see our revenue be higher than the number of shipments times the price, because all the deferred revenue is coming in next year.
Great. Then follow-up regarding the DUV business. Your booking for the June quarter was down about 4% excluding the EUV. In that context, should I assume that the DUV shipment in the second half of the year would be flat to down compared to the first half?
Yeah.
Yeah. I think, I mean, you can do the math. If you take what we said with the 25%, and then you take the EUV shipments and within install base revenue a little bit down in the second half. I mean, the end of the story is DUV in the second half is not going to be too different from what it was in the first half. I think we said also on the call that this business was already very good last year and we're up another 15% or so. I wouldn't put too much weight, albeit it gives you some info on the structure of the backlog, but I wouldn't put too much on the details on the bookings there. Although we get asked a lot, well, is the party over next year? We see trends continuing in 2018. DUV is also going to be strong next year.
Yeah. To add to that, I mean, a 4% down quarter-over-quarter, I mean, you know how our quarterly bookings vary. We have only a few customers and they tend to send those orders in by, you know, batches and then few percent, low single-digit percentages, in my mind are completely meaningless.
Got it. Thank you.
The next question comes from Mr. Farhan Ahmad. Please state your company name, followed by your question.
Hi, this is Farhan Ahmad from Credit Suisse. Thanks for taking my question. My first question is on memory. The memory shipments are up 50% year-over-year. Peter, can you just talk about how much of the growth is driven by increase in capital intensity and EUV? How much of it is really going to drive the demand higher this year? Also, if you could touch on what % of bit growth do you expect in NAND DRAM market this year based on the shipments that you're supporting? That would be helpful.
Yeah. Could you re-repeat the first question? Because I thought you talk about memory and then growth in capital intensity. I thought you said EUV?
50% growth that you have in memory-
Yeah.
Shipment, there is some EUV component in that as well, I'm imagining.
No.
Please correct me if I'm... Okay.
No. I think when you think about the growth in capital intensity or the 50% growth is really Deep UV. I mean, we do take some EUV memory orders, but that is not for capacity additions. That's really for, you know, ramping up and qualifying the product. The ramp will be at the end of the second half of 2018 going into 2019. That is not driven by any EUV shipments. It's really Deep UV and a particularly strong in DRAM. That is, like I said in the prepared remarks also, but I'd like to repeat it. It's not a major surprise either if you think about it.
I mean, the, you know, DRAM supply and demand curves are. I talk about the end product, yeah? The, you know, DRAM device, are really driven by the capacity situation and the end demand. In a situation like we had in 2014 going into 2015, where two major fabs in Korea, M14 and L17 came online, started to take, as many tools as they could to fill up the fab. You get a big step up in capacity, where of course the end demand doesn't follow that step curve.
You have a period in which there is a supply and demand imbalance, and it leads to lower DRAM prices, which actually at that moment in time, memory makers that could also see strong demand in 3D NAND started to relocate capacity out of DRAM into 3D NAND. Now, that end, you know, demand driven by strong data center demand, you know, when that goes up, will catch up. That has actually happened the second half of last year, leading to increased DRAM prices. That is not a surprise then that those empty pedestals where previously there were little tools that are now in the 3D NAND are being backfilled with the technology transitions on top of that.
That is creating the strong demand this year. Also driven by a strong end, you know, market where the data center demand is very strong for leading-edge DRAM. Now, what does that mean for bit growth? For bit growth, 26% is what market research firms say. You know, listening to customers, they might come up with some different numbers. You know, trying to predict those numbers is very dangerous because last year, you know, we saw market research firms talking about DRAM growth rates, and I'm talking about the end demand growth rates of lower than 20%. Well, it ended up more, well, in the high twenties. Difficult to predict. I would suggest we keep looking at the DRAM prices.
The DRAM price, it is a commodity, is a reflection of the supply and demand balance. Currently, you know, there is some under supply, and that's for sure.
I think for NAND, we expect around 40%.
Yeah, around 40%. Well, we expect nothing. We just repeat what other people are saying, yeah? Which is around 40%. You know, who knows?
Got it. On the EUV, I saw in the presentation that you posted online that there is a mention of unidirectional and bidirectional design. Just based on the industry chatter also, it seems at least one of the customers has picked up the activity on doing bidirectional design. I just wanted to ask you in terms of the insertion of EUV how does it affect the opportunity for you, whether it's bidirectional or unidirectional? Is it fair to think that if it's bidirectional, then there will be a lot fewer steps that are needed for deposition and etch?
No, I think it doesn't really matter. When I can refer to slide 17 and that you're what you're referring to is that 3D patterning, one EUV exposure. One patterning, one EUV exposure. That's the same. It doesn't really matter, yeah? Well, I'm not going to comment on deposition and etch because we're not experts on it. From an EUV point of view, it's no issue.
Got it. Thank you. That's all I had.
Next question comes from Mr. Douglas Smith. Please state your company name followed by your question.
Hi, it's Doug Smith from Agency Partners. From my tracking of the industry, it looks as though the percentage of litho spend versus total wafer fab equipment has dropped a little bit the last maybe two years or so. First of all, do you agree with that assessment? Second, do you imagine that going forward that litho intensity might go back up again?
Yeah, Doug, this is Wolfgang. I gotta disappoint you there a little bit. We are not big trackers of this metric. We're solely focused on our EUR 11 billion and the potential upsides to that. The reason why we're not focusing on this too much is twofold. Number one, historically, the customer CapEx has not shown a strong correlation to our own revenue in a given year. But more importantly, it is something that is very difficult to normalize because you gotta just think about FX, for instance.
I mean, two and a half years ago, the exchange rate only was 1.40, then it went down to 1.10 dollar to euro, all of a sudden, the litho spend looks lower, but it's nothing to do with our business. It's just a different exchange rate. Secondly, as you heard earlier in the call, we have a very specific strategy to provide upgrades to our customers that we have in field options and services, and that they don't necessarily count in the different people's calculation on adding up these numbers. I think from that perspective, we are not looking at it. We're not worried about our share on the overalls.
We're worried about executing our roadmaps and getting to our revenue. Then it will be whatever share of whatever somebody puts together.
Got it. Perhaps I can ask another question, like a more technical question, and that is, we're seeing a lot of chips made these days using full reticle sizes, 800 plus millimeters. Is that gonna be a problem for the current generation of EUV to print such large chips? How about for High-NA where you're using anamorphic projection?
Yeah, I think not for the current EUV generations. Potentially for the High-NA, but we don't think it's going to be a major issue. In the discussions we've had with customers, so that's being addressed. It also is a matter of design parameters that you can take into consideration. That's all part of the equation of why High-NA is a economic solution or not, and everything that we currently calculate. I think our customers agree with us that the High-NA specification, which includes indeed a different mask size or it's very likely points into the direction that it's highly economical to do it.
Mm-hmm. Okay, great. Thanks.
The next question comes from Mr. Sandeep Deshpande. Please state your company name followed by your question.
Yeah, hi. Thanks. My question, Peter, is on 2019. I mean, clearly now you have orders in your book for 18 of the 20 tools that you're going to build next year. How do you see 2019? I mean, are you already talking to your customers about 2019? Because some of your customers have indicated seven nanometers, starting with as much as eight plus layers. Some of that will have to be shipping in the first half of 2019, as well as some of the memory customers. Maybe you can make a comment on 2019 conversations you're having with customers. I have a quick follow-up after.
Yeah. Well, I think, you're absolutely right. I mean, 27 systems in the order book with the shipping pattern that we gave you means that we're virtually done for 2017 and 2018, so everything moves into 2019. With the order lead times that we're currently having, yes, you're absolutely right. The order discussion that we're currently having are about 2009, you know, 19. That's the case. Yeah.
People are giving you indications in terms of what they might be building at this point?
Oh, yeah. I think when you add it all up, you see 2019 and what we ship in 2018, you know, I think we're currently seeing it. We're currently added up. It's around 30 units. Could be a bit more than that they need for 2019. Which by the way, you know, it could be almost similar to what we're currently looking at, what we can ramp in terms of our own capacity. We've said before we will ramp our capacity double from 2018 to 2019. That is true for our build capacity here in Veldhoven, but we're seeing one or two quarter delays in that capacity in the supply chain.
That means that capacity buildup of, let's say, 45 systems capacity is really there in 2020. In 2019, we see one or two quarters, you know, delay. That, you know, around 30 demand, if I add it all up today, could probably nicely fit what our capacity is. Yeah, could be a bit more than 30, but it's a bit too early to give you a final number. You, since you asked me the question, I added up today, and that's what the number is.
Okay. Thanks, Peter. Just a quick follow-up for Wolfgang. I mean, Wolfgang, in terms of the numbers that you've given at the Analyst Day last year, I mean, the moderate case was EUR 11 billion in sales and EUR 9 plus in EPS. If you actually just roll out the numbers that you've printed in 2Q, you're well ahead that on a full-year basis already. Would you need to give new guidance at this point, or you think that you are at a much higher level in terms of demand or something else has changed in terms of demand at this point where we are in the cycle?
I'm not sure whether you were on early on the call. We tried to address that. I mean, we're not changing the 11 billion right now, but we pointed towards the sensitivities we showed last year in New York. I think you can at least at this point conclude that the bottom end of these scenarios, both on the demand side and on the number of litho layers, seems more and more unlikely. If there's any bias, I think everybody would say now that we have an upside opportunity. Let's just get this year under our belt and next year and at the appropriate time we'll give you a formal update on the model with new sensitivities.
Like I said earlier, we're feeling pretty comfortable about those levels at this point.
Understood, and sorry about that question. Thanks.
No worries. No worries.
The next question comes from Mr. Amit Harchandani. Please state your company name followed by your question.
Good morning, and good afternoon, everyone. I'm Amit Harchandani from Citi, and thanks for taking me on. Two questions, if I may. The first one really is a clarification to what was said earlier. If I understood correctly, Wolfgang, did you indicate that the deferred revenue for next year would be around EUR 500 million, and also the shipments next year would see revenues being fully recognized? If I assume, say, EUR 100 million per tool, does that imply 20 into 100 plus 500, we are looking at a revenue of EUR 2.5 billion? Could you just tell me-
Your math reflects what I said.
Okay. Thank you. That's helpful. Also, given that there is EUR 500 million of deferred revenue, wouldn't that have a one-off implication for the gross margin? Because most of that would be at a fairly high gross margin.
Yeah. Well, also there, it certainly helps, but we knew that. I mean, we have said, we're coming on a journey from minus 75% last year to 40%. Now you can argue whether it makes the break-even a little bit tougher this year and the 20% a bit easier. In general, it has a little bit of a relocation left and right, but it doesn't change our trajectory. What we have in mind is really 40%. We need that to get to our greater 50%. We always said the biggest variable there is the volume. The second biggest variable is the serviceability of the tool.
We have, of course, the learning curve and avoiding E&O. With the tool now performing to the specs, and that enabling us to, quote-unquote, "freeze the spec" and really work on availability, that makes us really much more comfortable that also on the financial side, we can deliver on that 40%.
Thank you.
You're welcome.
Just the second question, with respect to the supply chain or the comment I think Peter referred to earlier, just wanted to understand, are you looking at potentially helping your suppliers or supply chain build up capacity to give you the flexibility to potentially ship more EUV tools by 2019? In other words, is that already starting to emerge as a constraint in your view?
Well, I think the ramp up. It's a good question. I think the order flow is also a good evidence of it, and it also has ignited a lot of activity in the supply chain. However, going from a very low level to then ramping it up, here and there, you know, it doesn't always go as smooth as we would want. That's why I alluded to a one- or two-quarter delay to the capacity ramp to, let's say, 40-45 units. That's where we are. I think it's not so much a matter of money. Money doesn't always help, you know.
It's just a matter of, can you get the people on time, train them, can you get the materials on time with long lead times? You could argue money will help you to put to get people in faster, but the learning curve is the learning curve. These are the kind of things that we are, that we're driving together with our suppliers. This is what the current status is. Now, rest assured, you know, we'll do anything, you know, to get more out. If the demand is higher, then we'll just push everything that we can. I'm afraid money alone will not do it. It has to be people, knowledge, and how far we can push it is too early to say.
Thank you, Peter. Thank you for the color.
The next question comes from Mr. Jagadish Iyer. Please state your company name, followed by your question.
Yeah, thanks for taking my question. Two questions. First, Peter, one of the things that some of the companies who are involved in multi-patterning have been stating is that EUV insertion will initially happen for vias and cuts. Looking at the cartoon on slide 17, it looks like your insertion is going to be for metal lines and spaces. Just want to understand the disconnect where we are on that, in terms of that.
Well, you should ask the other companies.
Okay. Okay. The second follow-up question I wanted to ask you is that you did mention in your prepared remarks about China, where you said there are about five fabs. How would you characterize between the need for leading edge versus the trailing edge there? Thank you.
Oh, I think that's a good question, an interesting question. You know, we. It is not the case that all five of those customers are all leading edge. You have to define what is trailing edge. If you would say in logic, 28 nanometer and below, you qualify as leading edge, then all the logic is leading edge. For memory, at two DRAM initiatives there are definitely, I would call them leading edge if I look at the nodes that they are focusing on. In 3D NAND, what's leading edge is the number of layers, yeah?
I think that is where they will start on the learning curve and will not be immediately at 64 or 72, you know, layers, but it's going to be a learning curve. Now, you have to put that into perspective of why China has taken this step. China's taken this step from a very strategic point of view. In the discussion we have had, it became very clear that the reason why they're stepping up this, let's say, investment in capacity, in leading edge capacity is because of the dependence that they currently have on non-Chinese companies to provide China with the right technology.
The geopolitical situation has not become more stable or more reliable or more trustworthy from their point of view. China has, you know, decided that leading technology and leading-edge technology should also be local. That fits perfectly in the five-year plan. If you read their five-year plan, that's what they want. I don't think that the investment money will be allocated to those companies that are focusing on trailing edge technology. That's not going to happen. It only is going to be on leading edge.
Well, thanks for the clarification, Peter.
The next question comes from Mr. Robert Sanders. Please state your company name followed by your question.
Yeah, hi. It's Deutsche Bank. Quick follow-up for Wolfgang on this deferral question. How should we think about Q4 gross margin given the deferred revenue into 2018? I guess the flip of that is what? How should we think about 2018 gross margin given this rather, you know, skewing effect? I have a follow-up. Thanks.
Yeah, I think it's a little bit early to hash out, I mean, given revenue guidance for Q4, and I think I'm not going to start with gross margin guidance. In general, it's clear next year, even if we I think I said before, if we defer our revenue, it in general makes it a little bit more difficult to get to the break even, but it makes it more simple to get over the 20% next year. There's a bit of a positive effect. Next year it just depends on the. We gotta just see the overall business, and it depends on the rest of the business, right?
If I think Amit did the math before. If you're going from 1 billion to over 2 billion, and even if you go from around break even to 20%, it's still far below the average of the rest of the business. Without a specific number, which I'm not going to give today on the non-EUV business for 2018, I can't answer the gross margin question either. I think the most important thing for us is what I said earlier. Our confidence in EUV volume and EUV gross margin is growing. Our gross margin in DUV, in CLS, in HMI, in applications, it is healthy. It's exactly where we need it to be.
Therefore, we feel very comfortable that we get over 50% in 2020. It's too early to talk about Q4 and next year specifics, Robert.
Got it. Got it. Fair enough. I just had a follow-up on the 250 W demo. Looks like you're gonna get that in the field by 2019. Given that that's beyond the performance spec, how will you monetize that, I assume, through a software upgrade? How should we think about the value of that software upgrade?
Yeah.
In terms of when you look out to 2019? Thanks.
Yeah, I think this is a good question. One correction. I think 250 W is not going to be 2019. We have 250 W now. We have the modules that will be inserted in our tool shipments by the end of the year, which will provide our customer with 250 W. So will be available as of next year. So one. Two, the value of our tool is really driven by the ability of our customers to keep cutting costs, the cost per, you know, wafer. That the biggest driver there is productivity.
if going forward, we can improve, for instance, the transmission of the lens, we can improve the transmission of the pellicle, which all takes way currently takes away a bit of light. Like I said in the prepared, you know, remarks, customers are progressing on the sensitivity of the photoresist. These are all things that are actually helping to get more light on the wafer. when you get more light on the wafer, you move the wafer faster. When you move the wafer beyond 125 wafers per hour, and you can guarantee that, then the tool provides more value. that's exactly how we're going to do this. this is a general concept which customers accept.
You know, when we give them instead of 125, 145 wafers per hour, 145, then we'll charge a higher price because we basically split the value of that 20 extra, you know, wafers. 50% for the customer, 50% for ASML. That will also mean a higher sales price for the EUV system. That happens after we can guarantee the over 125 wafer per hour performance, which is the function of the things that I just mentioned.
Got it. Thank you.
Ladies and gentlemen, I'm afraid we've run out of time today. If you were unable to get through onto the call and still have a question, feel free to contact ASML's Investor Relations Department, and we'll get back to you as quickly as we can to answer your questions. Now, on behalf of ASML's Board of Management, I'd like to thank you all for joining the call today. Peter, if we could formally conclude the call, that'd be great. Thank you very much.
Of course, sir. Ladies and gentlemen, this concludes the ASML 2017 second quarter financial results conference call. Thank you for participating. You may now disconnect.