Ladies and gentlemen, thank you for standing by. Welcome to the ASML third quarter results conference call, October 14, 2015. Throughout today's introduction, all participants will be in listen only mode. After ASML introduction, there'll be an opportunity to ask questions. If any participant has difficulty hearing the conference, please press star zero for operator assistance. I would now like to hand over the conference to Mr. Craig DeYoung. Please go ahead, sir.
Hello, Aaron.
Mr. DeYoung, you can go ahead.
Okay. Thank you, Aaron. Good afternoon and good morning, ladies and gentlemen. This is Craig DeYoung, Vice President of Investor Relations at ASML. Joining me today from our headquarters here in Veldhoven, the Netherlands, is ASML CEO, Peter Wennink, and CFO, Wolfgang Nickl. To remind you, the subject of today's call is ASML's third quarter 2015 results. This call is being broadcast live over the internet at asml.com, and a replay of the call will be available on our website for approximately 90 days. 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, and for a discussion of these 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. The length of the call will be 60 minutes, and now I'd like to turn the call over to Peter Wennink for a brief introduction.
Thank you, Craig. Good morning, good afternoon, ladies and gentlemen, and thank you for joining us for our third quarter 2015 results conference call. Before we begin the question answer session, Wolfgang and I would like to provide an overview and some commentary on the recent quarter and provide also our view of the coming quarters. Wolfgang will start with a review of the third quarter financial performance with some added comments on our short-term outlook. Then I will complete the introduction with some further comments on the current general business environment and our future business outlook. And Wolfgang.
Thank you, Peter, and welcome everyone. For Q3, our net sales came in at EUR 1.55 billion. This includes system sales of EUR 975 million, of which memory represented 56% and logic represented 44%. Service and field option sales came in strong at EUR 574 million. This part of our business' growth continues, driven by strong demand for Holistic Lithography options, high-value upgrades, and a growing installed base. Our gross margin for the quarter came in at 45.4%, slightly above our guidance. R&D expenses came in at EUR 267 million, and SG&A expenses came in at EUR 86 million, both slightly below our guidance, positively affected by slower hiring, more favorable than planned exchange rates, and some one-time effects. Turning to the balance sheet. Quarter-over-quarter cash equivalents, and short-term investments grew to EUR 2.68 billion from EUR 2.52 billion.
During the quarter, we repurchased shares for EUR 142 million, bringing the total for repurchased shares to EUR 426 million year to date. Regarding the order book, Q3 bookings came in at EUR 0.9 billion, slightly below our Q2 non-EUV bookings. Strength in memory bookings continued to be notable. We finished the quarter with a strong overall backlog of approximately EUR 2.9 billion. With that, I would like to turn to our expectations and guidance for the final quarter of 2015. We end the Q4 with a strong and nicely balanced system backlog. With continued strong service and field option sales in the quarter, we expect Q4 revenue to be approximately EUR 1.4 billion. Based on expected customer and product mix, we expect gross margin for Q4 at around 45%. R&D expenses for the fourth quarter will be about EUR 270 million, and SG&A is expected to come in at about EUR 90 million.
Our annualized tax rate is expected to come in at approximately 11%. Q4 is affected by what appears to be a more cautious foundry segment. This is resulting in more limited current foundry node capacity additions, leaving our Q4 revenue guidance below Q3 and removing the potential Q4 upside in this sector we discussed last quarter. However, Q4 foundry equipment shipments will support early next node volume ramps. In addition, given ongoing memory booking strength, we expect that sales to the memory sector will remain healthy in Q4, although slightly below last quarter sales levels. Given these assumptions for the rest of 2015, we can clearly state that this will be a record sales year for ASML. 2015 revenue will have been enabled by another strong memory year, driven by capacity additions in two new DRAM fabs by some additional capacity needs in the one existing 3D NAND fab.
The foundry sector will have increased versus 2014, as volume ramps continue at current and prior nodes. We will also have initial but limited shipments for 10 nanometer. Lastly, 2015 will mark a record sales level in our service and field options business. Peter will talk more about EUV shortly, but I would like to make a few points regarding 2015 EUV shipments. As most listeners are aware, we continue to show great progress in improving on key EUV performance metrics related to productivity and availability, with unchanged targets of improving both further throughout this year and next. Without a doubt, EUV continues to progress upwards along the new technology adoption curve.
We are delighted to have a total of eight purchase orders for our NXE:3350 and NXE:3400 production systems, plus the three remaining EUV development tool orders for NXE:3300 systems, of which two are going to be upgraded. These 11 tools will go to five different customers. You now see that our shipment expectation for 2015 has changed to four systems, down from our prior expectation of seven systems for the year. As a reminder, all top manufacturers have had EUV development tools in-house and in use for some time. Two logic manufacturing leaders have now publicly committed to production insertion of EUV. We expect EUV revenue for 2015 to be limited to the one NXE:3300 that we recognized during Q2.
For the time being, revenue recognition for EUV remains difficult to forecast since it depends, amongst other criteria, on the achievement of performance milestones and differs from customer to customer. Once the performance of EUV technology is matured, we expect to recognize EUV revenue on system shipment as we do it in our DUV business. As we announced on our last call, we will share our expectations regarding revenue recognition as we ship systems. For the NXE:3350 that is being shipped right now, we expect to recognize approximately EUR 60 million of system revenue in the middle of 2016, with the remaining balance deferred to 2017. With that, I'd like to turn the call back over to Peter.
Thank you, Wolfgang. As Wolfgang highlighted, the expectations for the second half of this year have been adjusted due to caution amongst our logic customers as it relates to current 28 nanometer and 16/14 nanometer node ramps. However, with these adjusted numbers, we will still see a record revenue year in 2015. While it's too early to say anything quantitatively about 2016, we do see trends and developments that are worthwhile mentioning. In memory, our customers are currently indicating to us that their system demand will continue at healthy levels throughout the first half of 2016, albeit somewhat below our high Q3 sales levels. We expect, as Wolfgang mentioned, that the two new DRAM fabs will continue to install capacity next year. Meanwhile, in NAND, we expect limited lithography tool shipments to 3D NAND in 2016.
The only existing volume 3D NAND fab will likely be full by the end of this year, and the two new fabs will likely take limited equipment given that 3D NAND is still in its early stages of product introduction, and the new fabs are not fully online. Given this memory landscape, we estimate that our sales to memory customers could be down year-over-year. By how much will depend largely on end demand and the aggressiveness with which customers will fill their fabs. As mentioned earlier, in finishing out this year, logic customers have clearly taken a more cautious stance on capacity spend for 28 and 16 and 14 nanometer. However, the near-term completion of two new foundry fabs are designed to support the next advanced 10 nanometer FinFET ramp. The most recent customer indications are that this technology introduction is progressing well.
This node should provide high value given its significant shrink versus the 20/16/14 nanometer node, hence their continued and clear commitment to ramp 10 nanometer starting in the second quarter of 2016. The speed and ultimate spend levels for logic in 2016 will depend on the rate at which our customers will be able to execute their ramp. Furthermore, field options and services will carry their strength into 2016 and will continue to show growth. On the ASML product side, I would like to highlight the fact that we're launching two new scanners, one under our DUV program and one under our EUV program. This is enabled by an almost equal split in R&D spend on our DUV platform, which is vital in supporting the existing industry needs, and our EUV platform, supporting the industry's future leading-edge lithography needs.
In addition, we continue to enhance our Holistic Lithography product offering, now also moving significantly into the EUV area. Firstly, with respect to our DUV platform, we've now shipped our first NXT 1980 immersion product, which has demonstrated 40% improved focus uniformity 30% improved overlay accuracy versus our prior offering, along with the world's first 275-wafer-per-hour immersion capability. This world-leading capability will help our customers address the increasing and overwhelming cost and complexity of immersion multiple patterning and ever-tightening process control requirements that come with continuing logic and DRAM node shrinks. Regarding our Holistic Lithography products, leading-edge customers are using our full suite of immersion process window enhancement and process control solutions to optimize yield at the one nanometer or the 1X nanometer production nodes.
Holistic Lithography products are now also extending into EUV processes, with customers evaluating our EUV source mask optimization software for development of 7 and 5-nanometer technologies. Finally, EUV. As most of you are aware, our 2015 focus has been on improving EUV stability, availability, and productivity, the key metrics of performance that now drive new lithography technology adoption. In several recent public presentations, our customers have recognized our EUV progress in these areas. On the raw productivity side, we have a system configuration that has demonstrated more than 1,000 wafers exposed in a 24-hour period. In a four-week customer-run manufacturing readiness test at the production conditions, we've seen 15,000 wafers exposed, with comparable results achieved using the same power configuration at multiple customers. Several customers have achieved four-week average system availabilities of greater than 70%.
The overall worldwide average is currently still lower, indicating that performance consistency needs to be further improved. Wolfgang discussed the encouraging state of our EUV orders. Key in our orders are two leading logic manufacturers that are now publicly stating the need for EUV at 7 nanometer. The speed at which these and other customers order and adopt EUV is influenced by multiple factors, but is in any case dependent on the timing of the initial production ramp for these next advanced nodes. In that context, some customers have indicated that with the current challenges of multiple patterning schemes, which drive significantly increased complexity, the timelines for current advanced nodes, for instance, 10 and 14-nanometer logic, have been prolonged. This is forcing them to put priority on their integration efforts on these current nodes.
These combined factors have therefore directly affected EUV tool delivery requirements in 2015, pushing out some of those deliveries to 2016. As mentioned before, the need for EUV for the 7-nanometer node remains unabated. We do not believe that the decisions on timing of EUV shipments will affect the company's stated target of reaching EUR 10 billion of sales by 2020, since, as most of you know, we considered a three-year node cadence when preparing the opportunity analysis that we shared with you last year. Finally, as our customers are collectively saying and consistent with our previous suggestions, EUV remains a question of when, not if. With that, we would be happy to take your questions.
Thanks, Peter and Wolfgang. Ladies and gentlemen, the operator will instruct you momentarily on the protocol for the Q&A session. Beforehand, I'd ask you to kindly limit, as I always do, to one question and one short follow-up if necessary. This will allow us to get to as many callers as possible. Aaron, could we have your instructions and then the first caller, please?
Thank you. Ladies and gentlemen, at this time we'll begin the question and answer session. If you have a question, please press star one to register for a question, and star two to withdraw the question from the queue. Your questions will be answered in the order they are received. If you're using speaker equipment today, please lift the handset before making the selections. One moment please for the first question, and that's from Mr. Sandeep Deshpande. Please state your company name and your question. Go ahead, sir.
Yeah. Hi, Sandeep Deshpande from J.P. Morgan Cazenove. Hi, Peter. I have a question on, clearly your customers are indicating some level of caution in terms of placing orders at the moment for fourth quarter, as well as potentially in the first half of next year. Do you have associated with this likely softer revenue trend into next year, you're not guiding to 2016, but you probably have some views on what revenues in 2016 are going to be. Do you have any views on how your OpEx will trend into 2016, and is there any flexibility on OpEx in the event that there is a decline in revenue next year?
Yeah, I think the flexibility on OpEx, Wolfgang, you can take that. Generally, I think you're saying cautious planning orders. I think it's particularly true for foundry, as I said in the introductory call. I think when you look at foundry, it is very clear in the discussion with our logic customers, it's not only foundry, with our logic customers, we see the 10-nanometer ramp as being very real. I know there have been questions with many people, how real is it? In the discussion we're having with customers and the progress that they are making, which they have shared with us, we believe it's very real, but it will not start before Q2. I think before things turn up in Q2, for logic, things are likely going to be a bit lower than a bit higher.
It will definitely start going up in Q2. On the flexibility of our OpEx, Wolfgang?
Hi, Sandeep. We're right now running on R&D, somewhere just shy of EUR 270 million per quarter and just shy of EUR 90 million per quarter on SG&A. Our intent to keep it around these levels for next year. As we have stated before, we do have some level of flexibility in all of our OpEx. We are not looking at any major downturn that we are right now looking at exercising these options. You should model in your models about flattish OpEx for next year.
In terms of the flexibility, we have about a 20% workforce flexibility, which is labor contracts that are not on the payroll, which is basically outsourced.
Just to follow up on EUV, Peter. There have been multiple milestones reached on EUV. Most of these have been with top-tier customers. Is there any sign of the next tier of customers coming into the program and showing an interest in EUV?
Yes, I think we have discussions with all our major customers, which is about seven. We're all talking very in-depth every time that we meet them. The larger part of our meeting is indeed about EUV. It's not only the, let's say, top-three logic in foundries. Definitely also, very serious discussions on EUV introduction for our memory customers.
Thank you.
The next question comes from Mr. Kursfeld. Please state your company name, followed by your question. Go ahead, sir.
Gents. Thank you for taking the question. It's Merrill Lynch. You may have mentioned some of the smaller customers. I think there was an article yesterday or a couple of days ago that on the DRAM side, EUV adoption could be on the roadmap for 15 and 10 nanometers. I'm just wondering, do you have any more color? If that were the case, in which sort of timeframe could you start to look at orders from those customers? Then the second question was just on the logic side. I think Intel yesterday indicated they'll probably spend more CapEx next year. It also seems that you have TSMC and Samsung, so two foundries ramping 10 nanometers.
You basically have the entire logic camp starting that 10-nanometer ramp, which I don't think has sort of ever really happened in history. I'm just wondering, shouldn't that imply a pretty material snapback in your total revenues from the second quarter onwards? Thank you.
Karl, two questions. First on EUV introduction in memory. In DRAM, you're talking about 15 and 10. Well, we tend to talk about, let's say, around 16 nanometers. Let's start with that. That is, for us, scheduled when we look at the roadmaps, 2018, early 2019. That means it's a year shipment before. That means end of 2017 at the latest shipment. That would mean we need to conclude orders in the course of next year. On the 10-nanometer ramp for logic. Very clear that, yes, you are absolutely right. We see all major logic makers bringing 10-nanometer forward. There is a difference as we see in terms of timing of introduction, timing of start of risk production. Generally, you are right. You see all the leading logic makers moving into the 10-nanometer area next year.
Like I said earlier in the answer to the previous question, we see things turning up in Q2. It depends. Generally, these ramps also depend on how well those 10-nanometer introductions go from, let's say, yield and a production qualification point of view. There, I must say, we have had some good and positive news also from customers that things are going well. Yeah, that might look favorable going, let's say, onward from the second quarter. Like I said, we do not control, unfortunately, those advanced ramps for our customers. That will determine the speed and the size of that opportunity in 2016. Let's keep our fingers crossed that we see the good development that we currently see continuing.
Thank you.
Next question comes from Mr. Hosseini. Please state your company name, followed by your question. Go ahead, sir.
Prior person who asked question. Over the past several years, logic foundry demand has actually turned out to be weaker than expected because Moore's Law is just becoming very expensive. This kind of brings back EUV to the picture. EUV is more like later this decade when it's going to be commercialized. Peter, with that as a background, do you see a scenario where 10 would actually end up disappointing because the cost going up? Isn't this why the 10-nanometer ramp is turning out to be as slow as what we experienced with 20 and 16?
Maybe. It's a bit of a speculative question, which would warrant a speculative answer, which I'm not going to give. What we are seeing on 10 nanometer is a level of cooperation with our logic customers to manage the complexities of 10-nanometer multiple patterning, which should lead, and that's what we're currently seeing, to yields that should provide our customers and their customers with a cost per function, a cost per bit that is still competitive. That is, at least for the larger applications, the application that use the larger volume, still a very attractive node. Don't forget, if you talk about 10 nanometer and you compare the 10 nanometer shrink to the 20 nanometer shrink, because from a little point of view, 16, 14, 20 is about the same. You talk about a significant shrink. That's a real shrink with a real cost benefit.
It's hard to imagine that the end users and the customers of our customers are going to walk away from that benefit. This is also what customers keep telling us. Like I earlier said, the progress on the 10 nanometer development is going well. We currently don't think it's going to be a small node because of that reason. You did mention one other point, which I think is important. You said 10 nanometer is going to be an expensive node, which is true, and also EUV is going to be a solution. To be very frank, we've always planned 10 nanometer to be an EUV node. Now, we're just late, and this unfortunately leads to a situation where customers need to apply complex multiple patterning strategies, which probably means that the 10 nanometer node is going to be prolonged.
Which actually has an impact on EUV, which pushes the EUV need 6 to 12 months back for those guys. Which, by the way, is what we put into our simulation model last year because we anticipated some of that complexity coming. In that context, you need to see 10 nanometer and EUV.
Got it. Thanks so much.
The next question comes from Mr. Srinivasan from Summit Research. Please state your connection.
Hi, this is Srini Srinivasan from Summit Research. I just wanted to know how much DRAM CapEx spending might be going down next year for the full year 2016, if at all. How much of the DRAM spending would be in the form of capacity growth, and how much for shrinks? That's my only question.
Yes. I think the total DRAM CapEx spending is really dependent on the aggressiveness with which our customers want to fill the existing new fab space that has been opened end of last year and in the middle of this year. Those are very sizable fabs. There's a lot of potential capacity there, and I think it will be a function of the end market demand and how much aggressiveness our customers will want to put into the buildup of that capacity, which will, of course, will have an impact on the supply of DRAM. I don't know how they want to deal with that. This is a good question. If you have the answer, I would like to know. We don't have that answer readily available because it depends on the customers. What we do know is that the first half of 2016 is still healthy.
The second half is still a bit too far out. How much of that is shrink? Those new fabs that we're talking about are for the leading-edge nodes. It's shrink and capacity because they're two new fabs, and it will take basically the leading-edge devices. It's both.
Thanks so much.
Next question comes from Mr. Iyer. Please state your company name, followed by your question.
Redstone Research. Two questions, Peter. If I look at the immersion revenues over the last several quarters, starting from the beginning of the year, there has been a steady decline in your immersion revenues. Given that there is significant double patterning happening, why are the immersion revenues going down? I have a follow-up.
This is Wolfgang. I'll take a crack at this. You're right. In general, when you look, for instance, 28 going to 2016, 2014, the litho intensity goes up by some 40%-50%. You're also right in your observation that our systems revenue as a company has been fairly stable over the last couple of quarters. I think there is three reasons for this at least. One is that we have provided upgradeability in our roadmaps to our systems, and you will have seen that if you just take this year versus last year, we went from EUR 1.6 billion to we will be somewhere around EUR 2 billion on field options and services. We do upgrades that don't show up in our systems revenue.
Second reason is that in particular, over the last couple of years, we have provided our customers, in response to the multiple patterning requirements that Peter mentioned, with significant performance upgrades. With 250 wafers per hour, and now we're going to 275 wafers per hour. That's not going to go on forever. We're reaching the physical limits there. Last but not least, as was also already mentioned by Peter, we see the period of time over which customers facilitate certain nodes stretching out. If you take the 2016, 2014, for instance, not all that equipment is being bought in a two-year period. We have started to ship in the second half of 2013. We're shipping in 2014 and 2015. We're actually going to ship into 2016 as well. Those are the three reasons why you're not seeing a big expansion on the system revenue yet.
Okay, thank you. Just as a follow-up, I don't know, Wolfgang, you mentioned in terms of revenue recognition for EUV, did you clarify that for 2016, it will be just probably limited to one system? Is that something that I heard or did I hear it wrong?
I'm glad you asked the question because we want to be crisp on that. Let me just give you a precursor. Revenue recognition is somewhat difficult for us on EUV, and it's in particular difficult to forecast. We have a pretty good view on the volumes, but then the revenue recognition in detail depends really on what customer we ship to, certain performance milestones, and other accounting-related treatments. In the long run, by the way, we want to do this like we do it in DUV at shipment. What we said last quarter is that we are going to give you guidance as we ship systems. Starting with the 3350 that's being shipped right now, we're telling you that of that revenue, EUR 60 million will be recognized by mid-2016, with the rest in 2017. That's just for this particular system that's shipping right now.
As a reminder, we're planning to ship another two systems this quarter. There will be very likely revenue next year as well. We plan to ship somewhere, if I want to pick a range, additional six, seven tools or so next year on top of that. There will be some revenue because remember, there will be some 3300s in there. There will be some revenue. The EUR 60 million I mentioned is by no means the entire EUV revenue, it's just the one that is attributed to the system we have started shipping now. In our January call, we are going to talk about the shipments that we will make this quarter and how that will translate into 2016 revenue. Hopefully, we can give more color around that, even that in January. I hope that clarifies this a little bit for you.
Yeah, thanks for clarification.
You're welcome.
Absolutely clear.
The next question comes from Mr. Sanghvi. Please state your company name followed by your question.
Yes, it's RBC Capital Markets. Peter, question on the memory. You said memory being healthy in the first half, driven more by DRAM new capacity. When you look at the memory next year, you said overall memory could be down. In your downside scenario, are you looking at more downside from NAND or from DRAM?
Well, it's down year-on-year. We've had a particularly strong year in 2015, which was driven by the fact that we had a large DRAM factory coming online the end of last year, just at the end of last year, taking tools throughout this year, and a very large factory coming online middle of the year, taking tools. In that sense, there's a significant DRAM shipment this year. NAND was pretty stable as compared also to the year before, to 2013. When we look at the total memory market, DRAM will be driven by, as I said, the further build-out of the first six months of the year. We have a pretty good visibility on it.
We don't know exactly what the customers are going to do, with what speed and what aggression, you could say, are they going to fill up the remaining capacity in the second half of the year. That's where it becomes a bit more unclear. On NAND is really driven by 3D NAND. On 3D NAND, by the end of this year, the only dedicated 3D NAND fab will be full. That means there will be two new 3D NAND fabs coming online, but they will be limited in terms of the number of capacities that it will take, driven by the fact that it's still a device architecture that in its first phase of its life, it's not going to be a full ramp. Some of that capacity only comes online in the course of the year.
By design, if you then say 3D NAND is the driver next year, there is a limitation in terms of available capacity and available appetite, we think, to go fast. They were probably going to be more 2017. It's those things that we said there could be a likelihood that DRAM will be lower than in 2016 as compared to 2015, largely driven by the strength of 2015.
Okay. That's very helpful. On the 10 nanometer, the volume production is slated for second half of 2017. If that was the timeline, how much before do you really ship the volume to your customer? Is that second half what you're seeing are those pilot lines or mass production tools? What will be the fit with really volume out in the second half of 2017? When does the volume tool go into fabs?
Yeah, I think what you need to understand is that, we mentioned it a couple of times before, that 10 nanometer is highly complex. It's complex because it involves multiple patterning strategies for multiple advanced layers, which actually means that the time between shipping a tool and the qualification for production lengthens. You should not underestimate that time. When on the roadmaps, there is, let's say, a volume ramp starting the second half of 2017. You need to think about a year earlier that you need to put those tools in and to make sure that you can grow to this full ramp, which means that the first pilot ramp will be starting somewhere in the middle of next year, and the first risk production is starting at the beginning of the year. The new fabs will be ready soon.
They will be able to take machines in the second quarter. Those machines will take time to install. It's about a quarter. You are in Q3 of 2016 before they can start working on pilot production. It's logical that the volume production will be in the middle of 2017. These things get more complex, takes more time. In that sense, it all fits.
Okay. That's very helpful. Thank you.
The next question comes from Mr. Arnault. Please state your company name, followed by your question.
Hello, this is Farhan Ahmad from Credit Suisse. My first question, Peter, is on the 3D NAND side. There is a very different commentary which I'm hearing from you guys versus the rest of the industry. We are seeing AMAT orders peaking at all-time high because of orders that they've received from Toshiba. There is a big conversion in Toshiba that's happening next year and also at Micron. I wanted to understand, is there a big difference between a customer that builds a new fab on 3D NAND versus somebody who's converting their fab from planar to 3D? Is there a big difference in the litho intensity?
On the 3D fab is generally as a greenfield fab, we need about 10% more litho in a greenfield fab with the same wafer capacity out. We are basing what we see in 3D NAND based on the customer context that we have and on the shipment patterns that they're asking and on the wafer start capacity that they want to add. We believe it doesn't show a full aggression. You mentioned Toshiba. They will likely be ready with their fab taking tools at the end of this year. We don't foresee it as a very strong ramp. With Micron, that fab will only become available in the course of the year, whereby, of course, the fab needs to ramp up. Those are all elements that make us believe that 2016 will be a limited year for 3D NAND. 2017, definitely.
2017, you see these two fabs ramping. That would probably be a much stronger year. We just see a slower 2016 than what you seem to indicate.
Got it. Thank you. My second question is on the service side. You guys had record high revenues in service. If I look at your service revenues, they've grown from about EUR 400 million a year ago to EUR 575 million, almost like a 40% increase. I just wanted to understand how sustainable is the revenue level on the service side. Usually, service tends to be more stable. Just wanted to get a sense of how we should think about service revenues and quarter-to-quarter variability in that.
Yeah, I'll take that. This is Wolfgang. First of all, you're right. We do in a quarter now that what a few years ago we did in a year. I'd like to just make sure that we are clear on the nomenclature. It's field options and services. It is really driven by options that we make available on our Holistic Lithography software products, but also the upgrades that we enable. If you have a 1950 immersion system today, we can upgrade that two or three levels. Those are major upgrades. They're like EUR 20 million. Peter always calls them heart surgery. It's basically almost replacing the scanner. That is increasing quite a bit. The install base is increasing as well. With that, we have higher service revenue. It's a good business for us.
This year, I mentioned it earlier in a response to another question, I think we'll be around EUR 2 billion, and we see that business continuing to grow. For modeling purposes, I think if you take a 10% increase year-over-year, I think you're in the ballpark. It's a strong business for us, and it continues to grow.
Thank you. That's all I have.
You're welcome.
The next question comes from Mr. Ho. Please state your company name, followed by your question.
Thank you very much. Maybe as just a follow-up, excluding the leading foundry players that are at 16 and 14 nanometers right now, how do you see some of their caution right now potentially impacting their 10-nanometer rollouts?
Yeah. We don't see that caution at all as it relates to 10 nanometer. When we talked about caution, it also had to do with foundries not being, let's say, the leading foundries. We mentioned last quarter that we also saw still some upside in areas of China, areas of Taiwan, not specifically in the leading edge. That has all disappeared. And it had to do with 28 nanometer also. 28 nanometer, where we planned some shipments for the remainder of the year, did not turn out to happen. I think it's largely been in that area, 28 nanometer. Here and there, some 16 and 14 nanometer tools, but those are, you could say two, three tools in total, and two tools are easily EUR 100 million. You have to put it in context.
That caution does not, I would say, affect the eagerness with which customers are now talking about the 10 nanometer ramp. The 10 nanometer ramp, they keep pushing us and telling us that this needs to happen for several reasons. It's very strategic for all the leading foundries to be there with the right yields at the right time for the leading-edge customers, and they're pushing it. And there's no relation to this weakness or caution that we have seen in 28 and 14, 16.
Great. That's helpful. And just a follow-up question on the EUV side, with you now shipping the 3350B systems to your customers, are there different milestones for qualifications on that end that again, will impact the revenue recognition of those systems relative to some of your earlier ones?
Yeah, like we said, it will depend customer by customer. And yes, when we do additional orders with additional customers, it may be that the terms and conditions will differ slightly from what we have on orders right now, and that would impact when we would recognize revenue for these tools.
If I may add to that. The earlier tools, we sold the earlier tools to the R&D folks. The R&D folks had only one qualification requirement, and it said, ship us a tool that can print a wafer. When we can print a wafer, you can do rev rec because that was the only requirement. The 3350 is a production tool. We're selling those tools to production people, to wafer fab bosses. They have different requirements than just to print one wafer. This is where those requirements kick in the deals, and that's why the rev rec is a bit more complicated than the ones that we sold to the R&D folks.
Great. Thank you.
Next question comes from Mr. Jenkins. Please state your company name, followed by your question.
Yeah. Thanks. It's Gareth Jenkins from UBS. A couple of follow-ups, if I could. Sorry to go back to 10 nanometers, but I just wondered if we could talk about the length of the node and maybe the size. Would we be right in assuming that it would be a fairly short node, so maybe one and a half years before seven comes in in logic? Given, I guess, litho intensity, the length of the node and maybe some of the end demand weakness that we're seeing, that the size might be slightly smaller than 28 or 14, in terms of capacity. Then, maybe I'll come up to the other one afterwards. Thanks.
Yeah. Let's make one thing clear. When you talk about 10 and seven nanometer, you probably talk about the specific roadmap that is out there. When we look at it, and let's look at it together. For us, the 10 and seven nanometer is the same lithography tool. Yeah. In that sense, it is not a one and a half year. For us, it's closer to a three year. Yeah. That also coincides with some other comments that were made by other customers about a prolonging of nodes. Yeah. That's what we're seeing. We're seeing it happening at 10 nanometers. The real lithography step that we will see is when we introduce EUV, and whether you call it seven or five or however you want to call it, that is in the 2018, 2019 timeframe.
That means that we need to ship in 2017, start shipping in 2017. Which means, this is exactly in our plan. Node naming can be a bit confusing. We just look at what type of machine, litho tool, is needed when. That means that our advanced DUV immersion tools are needed largely in 2015 and 2016, and a big part in 2017. EUV starts to be needed in 2017 shipment for us and used for our customers in 2018, 2019. This is how we look at it, and I also think this is how you should look at it when you talk about lithography.
Thanks, Peter. Maybe just to follow up on the EUV. We've talked a lot about rev rec, but could we talk about cost rec? I presume you'll be taking cost up front. If so, what's the timing on that? Do you recognize all the cost on the initial EUR 60 million? Can you talk about cash recognition as well, so when you'd actually expect to see cash payments for these? Thanks.
Yeah. Again, you'll not be surprised this will differ customer by customer as well. Yes, this will make it even more complicated. We'll have situations where we have to recognize the cost when we start recognizing some revenue. In that case, we'll get a little bit out of whack on the margin. We'll have to describe that effect to you. Then, the cash, that we put a lot of attention on, you can in general assume that we get the cash earlier than we recognize revenue.
Thanks.
You're welcome.
The next question comes from Mr. Meunier. Please state your company name, followed by your question.
It's Morgan Stanley. I've got two technical questions, actually. The first one is actually on this 3350B, which was, in my mind, supposed to be delivered and installed this summer. Could you maybe explain what are the technical difficulties that you're facing with this tool? I understand it's quite difficult to deliver, but still, I would like to understand what is the delay related to. The second question is around the ecosystem. I'm sure you send people to SPIE Photomask, in Monterey. I was there. It looks like the ecosystem has made loads of progress Regarding the pellicle, regarding the mask manufacturing and with ZEISS. There is still this actinic through the pellicle inspection tool, which is missing for very high volume manufacturing. How do you think this will be solved with ASML? Thank you.
Well, to talk about the first one, the NXE:3350B shipped two, three months later than we thought. That had to do with the fact that we have a new drive laser there. That is the laser that is basically shooting at the thin droplet. It was the stability of that laser we had some issues with, but that's resolved now, so that's why we shipped the tool. This is when you have a big new piece of equipment and you have a planning on the integration, sometimes the integration takes a bit longer. This was the case. On the ecosystem, this question of whether we need or do not need an actinic inspection tool that can actually inspect through the pellicle, it has been a question that has been around for a couple of years.
I think the general consensus is that for the EUV introduction, also for the production introduction, we do not need it. We might in the end need it if feature sizes keep going down and we keep shrinking to 5 nanometer, 3 nanometer. That by that time, we need something like an actinic inspection tool. That's not in sight. That's next decade. Until then, we think we can definitely live with the solution of a removable pellicle. It's up to us and our customers to prove that actually the removable pellicle is such a good solution that we don't need an actinic inspection tool at all. This is where we are today. Whatsoever, no impediment to EUV introduction in production in the, I would say, remainder of this decade.
Okay, a quick question about the new DUV systems in 1980. What's the price increase, versus the previous version, which I think was in 1950?
1980 is a complex tool. It has some price increase, that has to do with the productivity. We have the higher productivity as compared to the previous tool. That provides higher throughput, lower cost per wafer, lower cost per die, lower cost per bit, and that will reflect in the tool pricing. As you know, we had a previous tool that, the 1970, that was able to do 250 wafers per hour. This one, 275.
Okay. Thank you.
Next question comes from . Please state your company name, followed by your question.
I'm ready, yeah.
Oh, hi. Thanks. Amit Daryanani from Citigroup. Thanks for taking my questions. Two, if I may. Firstly, with regards to EUV and the shipment being delayed from this year into next year, if I could go back to that and just try to understand it a little better. That's aimed at risk production. Is there a particular reason why the customers were unwilling to proceed on that as planned in parallel with the focus on the next node, which is the 10 nanometer? I appreciate 10 nanometer is a priority, but is there a particular reason that you think that has changed in the past three months? If you could dig a little bit more into that. As a follow-up, if I may ask you more broadly, strategically, looking at areas outside lithography, you've talked about collaborating with partners on the etching side.
Could you give us a sense of your progress there? Thank you.
On the EUV shipment delay, I think I mentioned, but to be clear, the issue with 10 nanometer is that we are late with EUV. That's the major issue. I mean, we should have had EUV risk production or production ready at 10, which we didn't for all kinds of reasons. It's complex, and we had the Cymer acquisition, and we didn't make enough progress fast enough. I think we're hitting our targets for 2015. We're aiming at hitting our targets in 2016. I think we're on the right track, but we cannot change history. We cannot change the fact that we're late. That meant that we forced our customers into a bridge solution, which is 10 nanometer, doing with multiple patterning immersion. That's highly complex. When we talk to our customers, and you're absolutely right saying that 10 nanometer is the first priority.
It's very strategic. We talk to customers, say, what are they focusing on, and how can they divide their capacity of people that can do the integration for 10 nanometer and do the development work for 7 nanometer? That capacity is limited. They have virtually all their people working on the 10 nanometer introduction because that is what the first priority is. With that fact that 10 nanometer is also stretching out, the lifetime, which you could say the timing of that node is getting longer. It is natural that also they need the EUV tools later. They need it later because EUV is getting introduced a bit later, but also they need all their resources to do the 10 nanometer development.
Those are the reasons why they say, okay, we can now take the EUR 95 million or close to EUR 100 million tool today, or we can take it when we actually need it, and we have the resources to deal with it. This is where we are today. It's a very practical solution that customers are now discussing with us, finding out that 10 nanometer is not as simple as we thought, and that their initial desire to stay on a two-year cadence is getting more complex. You move on, and you learn more. This is what has happened, and this happened throughout the year. I think it has become clearer to us over the last three months that this is where we are, and this is the situation, and we're going to help our customers deal with it. Okay, the second question.
On the etch, I think, one of the issues is when you keep shrinking and you do multiple patterning, that one thing is overlay. Overlay, we actually have a metrology solution that is integrated called YieldStar. You also have a CD, the critical dimension, how you want to control the critical dimension. The CD uniformity is not only determined after the imaging process but also after the etch process. We are working together with an etch company, but also with an applied research company called imec, to see what we, as a combination of tool suppliers, can do to help our customers to control that CD uniformity when we keep shrinking to 10 to seven to five. That's work that's ongoing, but I would say that's ongoing more on a research and development level, where we work together with those companies.
That's very helpful. Thank you, Peter.
You're welcome.
Let me jump in here for a moment. I think we have time for one last question and, of course, the associated follow-up. If you were unable to get through on the call and still have questions post-call, feel free to contact ASML Investor Relations Department with your question, and we'll get back to you on that as soon as we can. Operator, could we have the last caller, please?
The last question comes from Mr. Menon. Please state your company name, followed by your question. Go ahead, sir.
Hi, it's Liberum. Thanks for taking the question. Just going back to the 7 and 10 nanometer nodes in the context of trends in immersion revenue. I think on this Q2 conference call, TSMC made a comment that they expect total wafer starts at 7 and 10 together, to be lower than at 20 and 16. Given that outlook, and if that were true, and given that you're seeing an increase in litho intensity clearly at 10 and potentially at 7 as well, would you expect that your foundry immersion revenue trends will inflect at 10 nanometer and into 7 nanometer? Given that there is also the situation where we're seeing some weakness in the smartphone market, that you would see more a sort of a continuation of the trend that you've seen in the last one or two years? I have a brief follow-up.
Okay, Janardhan. Thanks. If you don't mind, I will abstain from commenting on the smartphone market because it's a bit too far away. We just listen to the customers. The assumption that the wafer capacity might be lower in, let's say, at 10, 7 than in 2016, I think is also something that we have assumed when we talked to you last year, when we gave you an outlook till the end of the decade. We have actually seen a 10% node after node reduction of the wafer capacity needed.
That has to do with, indeed, the increased cost, and especially the increased initial cost, because the market for those customers that can afford that cost and can create value with that cost, that market in terms of number of players, although those players are very large, but that market is of course smaller than when we had a 28 nanometer and above. That is an assumption that we also follow. However, you mentioned it, the litho intensity node by node goes up with about 40%. Especially if we go to 10 and 7, we see very intense multiple patterning strategies. Wolfgang alluded to that part of it. We are going to help our customers by upgrading machines in the field, which will be an impetus for our service and field option sales.
Next to that, when we look at the multiple patterning strategies and the number of layers that are affected, we also see upside in the number of units that they need from an immersion point of view. Now, not everything in life is that rosy. It means that costs will go up and complexities will go up, it will force customers to, after this node, to move into EUV. I think it's inevitable. I don't think it has a major negative impact on us because of the reasons that I just mentioned. I think the immersion business will stay very healthy for the years to come.
Got it. Just a very brief follow-up to Wolfgang. You said that R&D spending on EUV and DUV is roughly similar. When do you expect the DUV part to start falling off? In your current planning, when would you start dropping that, and sort of increase the EUV spend from your overall R&D point of view?
Let me put the framework out first. We're just shy of EUR 1.1 billion per year right now, and our target is by 2020, at EUR 10 billion revenue to be about 13%. We're seeing a little bit of a growth there. The split is a little bit less than half EUV, and the other half is not only DUV, but it's also Holistic Lithography and it's fundamental research. Fundamental research, I would suggest, will be flat to slightly up. Holistic Lithography will go up year-over-year because it's where a lot of additional value comes in through the software options. DUV over time will go down. We've just put major innovation, in the 1980. We see one platform after that. I don't think that you see a major inflection before the end of 2017, beginning 2018.
That's when we will ramp more in EUV, when we go into technologies that prolong the life of EUV, such as High NA, that some of our customers are now starting to mention publicly as well. Overall, R&D will grow at a slower slope than the revenues. The percentage will come down and, somewhere in the 2017, 2018 timeframe, you'll see DUV come a bit down and EUV extension and high-end applications will go up in the piece of the pie.
Got it. Thank you very much.
Welcome.
With that, on behalf of the ASML Board of Management, I'd like to thank everybody for joining us today, and I'd also like to thank the operator and ask him if he would formally conclude the call, please. Thank you.
Ladies and gentlemen, this concludes the ASML Third Quarter 2015 Results Conference Call. Thank you for participating. You may disconnect your line.