Ladies and gentlemen, thank you for standing by. Welcome to the ASML 2016 first quarter financial results conference call on April 20th, 2016. Throughout today's introduction, all participants will be in a listen-only mode. After ASML introduction, there'll be an opportunity to ask questions. I would now like to open the Q&A queue. If you'd like to ask a question, please press star one to register. If you'd like to withdraw your question, please press star two at any time during the call. Your questions will be answered in the order 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 DeGroof. Please go ahead, sir.
Thank you, operator, and good afternoon. Good morning, ladies and gentlemen. This is Craig DeGroof, Vice President of Investor Relations at ASML. Joining me today from our headquarters here in Veldhoven, in the Netherlands, is ASML CEO, Peter Wennink, and our CFO, Wolfgang Nickl. The subject of today's call is ASML's first quarter 2016 result. Just as a reminder for this call and for subsequent calls, the Q&A queue starts with the operator's instructions at the opening of the call and not before then, just FYI. As mentioned, questions will be taken in the order that they're received. As another reminder, the length of the call will be 60 minutes. This call is also being broadcast live over the internet at www.asml.com, and a replay of the call will be available on our website.
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 meanings 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. Peter?
Thank you, Craig. Good morning, good afternoon, ladies and gentlemen, and thank you for joining us for our first quarter 2016 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 and provide you our view of the coming quarters. Wolfgang will start with a review of the first quarter financial performance with some added comments on our shorter outlook, and I will complete the introduction with some further comments on the current general business environment and on our future business outlook. Wolfgang?
Thank you, Peter, and welcome everyone. For Q1, our net sales came in at EUR 1.33 billion. This included system sales of EUR 856 million, of which memory represented 42%, with logic representing 58%. Service and field option sales came in at EUR 477 million. Our gross margin for the quarter came in at 42.6%, slightly above our guidance. R&D expenses came in at EUR 275 million, and SG&A expenses came in at EUR 89 million, essentially as guided. Regarding the order book, Q1 system bookings came in at EUR 835 million. As this level is 30% below our prior quarter bookings, I would guess that some listeners might find this confusing relative to our guidance, which includes a 30% increase in Q2 revenues. I would like to remind you that we are not an order-driven company, and that the order patterning varies from customer to customer.
Our bookings are not always a real indication of our near-term business opportunity. However, you can clearly see that our Q1 bookings have changed the complexion of our backlog in a way that supports our Q2 guide of growing strength in logic and flattening in memory. I can also tell you that we expect strong logic bookings in Q2. Just as a further reference in this regard, I would like to draw your attention to our consolidated statement of operations found on slide 21 of our Q1 2016 results presentation, where you can see the lumpy nature of our bookings over the last five quarters as well. Turning to the balance sheet. Quarter-over-quarter cash equivalents, and short-term investments came in at EUR 3.14 billion. Our free cash flow for the quarter was negative EUR 65 million.
This was expected since we received a significant amount of customer prepayments on orders received in Q4, where free cash flow totaled EUR 864 million. We expect free cash flow to return to a more normal level in Q2. Year to date through the 3rd of April, we repurchased 2.7 million shares for EUR 223 million as part of our newly announced EUR 1.5 billion share buyback program for 2016 and 2017 combined. With that, I would like to turn to our expectations and guidance for the second quarter of 2016. We expect Q2 total revenue of approximately EUR 1.7 billion. Due to recent demand forecast increases, we now expect continued stable memory shipments for the rest of this calendar year. Memory shipments for the second half of the year should be roughly equal to shipments for the first half of the year. Logic shipments in Q1 were primarily for the 28 nanometer node.
We expect some level of continued capacity adds throughout the year at this node, as mentioned last quarter, we expect a strong pickup of total logic shipments in Q2 in support of 10 nanometer production ramps. Our current view of combined logic indicates that the second half of the year will be greater than the first half of the year. The extent being determined by ultimate 10 nanometer ramp levels in 2016. Last quarter service and field option sales came in at EUR 477 million and is anticipated to grow throughout the year. We continue to plan on a year-over-year increase of approximately 10% in 2016. This part of our business's growth continues to be driven by strong demand for Holistic Lithography options, high-value upgrades, and a growing installed base. Gross margin for Q2 is expected to come in around 42%.
Gross margin will be significantly influenced by revenue recognition of two EUV systems in the quarter. Our second quarter net sales guidance includes about EUR 110 million for EUV. As previously stated, we can only recognize part of the system revenue, but must recognize the full cost. This, along with an initial low profitability on EUV, will cause a negative impact of approximately five percentage points on the gross margin for Q2, which is included in our guidance. R&D expenses for the second quarter will be about EUR 270 million, and SG&A is expected to come in at about EUR 90 million, both roughly at the same levels as the previous quarter. Peter will talk more about the status of our EUV program, but I would like to mention that we completed the shipment of one EUV system in Q1. This shipment will lead to revenue in 2017.
We expect to ship one additional system in Q2. Finally, at our upcoming annual general meeting of shareholders on April 29th, shareholders will vote on our proposal to increase our dividend by 50% to a level of EUR 1.05 per ordinary share. We fully expect that this proposal will be supported by our shareholders. With that, I'd like to turn the call back over to you, Peter.
Thank you, Wolfgang. As Wolfgang highlighted, our first quarter results were very much in line with expectations and our business is developing along the lines that we communicated over the last two quarters. While Wolfgang gave a qualitative outlook for 2016 with combined memory appearing flattish half over half, with combined logic up in H2 over H1, and with combined services up half over half as well. We do see trends and developments that we believe are worthwhile mentioning. First one, despite a difficult pricing environment in DRAM, our forecast has further strengthened a bit in support of a continued drive by our customers to keep shrinking cost and specifically, for low 20 nanometer and sub 20 nanometer nodes. This has resulted in our current flattish half over half sales view for our combined memory business.
For 3D NAND, shipments continue to new fabs and to fabs preparing for 2D to 3D conversions. Of course, we are watching with interest the developments in the volume introduction of the 3D XPoint architecture, as it will become quite litho intensive in time, push the 3D NAND architecture. Secondly, as mentioned in previous earnings calls and evidenced by our backlog, it is clear that our sales to our combined logic customers will become more important starting in the second quarter. This will continue as we are expecting a continued increase in logic orders in the coming quarter in support of initial 10 nanometer node ramps. As a result, we now forecast the significant increase in combined system and service and field option sales in Q2 to be at the level of EUR 1.7 billion.
Also of note in the first quarter, we saw shipments for 28 nanometer logic capacity additions continue. I would like to make an observation here regarding logic node ramp behavior. Looking at the ramp speed, size, and length of the latest, most advanced nodes, it appears to us that the rollout pattern of such nodes is changing. Previously, node transitions followed each other in a rather predictable pattern throughout our entire logic customer base, whereby new nodes ramp quickly, in turn, ending the capacity ramp of the previous nodes concurrently. What we see today, effectively starting with the 20 nanometer node, is that the initial new node ramp is done by a very limited number of customers, but with greater intensity, meaning speed and initial size of the ramp.
The rest of the node ramp is executed over a prolonged period, whereby the rest of the logic customer base follows the initial customers in phases. Current evidence of this trend is the aforementioned continued shipments for 20 nanometer logic capacity additions. This still continuing more than 4 years after the initial introduction. In discussions with our logic customers, we see similar capacity expansion behavior over time for future nodes. With litho intensity rising significantly node by node, initial node transitions are lengthening to 3 years, with the aforementioned longer tail end of previous nodes. This will likely make shipment and order patterns for a specific node, as well as the ultimate wafer capacities, less transparent over time.
However, based on the input from industry analyst forecasts of end market developments, we believe that our long-term assumption of a 10% node-on-node reduction of the ultimate installed wafer capacity is still appropriate. As you all know, this was one of the pillars underlying the simulation leading to our EUR 10 billion sales target by 2020. As for the 10 nanometer node, its introduction is clearly progressing. The speed and initial size of this ramp can be explained by the value proposition provided by the significant shrink of this node versus the prior node. The ultimate spend levels for logic in 2016 will depend on, amongst other things, both the level of end demand and the rate at which our customers will be able to execute their ramps. This is why it is still a bit too early to predict the overall 2016 spend levels today.
For field options and services, we see continued strength in 2016. This should show growth as previously estimated in the range of 10% over 2015. On the ASML product side, we continued to focus R&D spend on the lithography products that are essential to the ramp of all current and advanced processes. In DUV, addressing the growing litho challenges of complex and lithography intense multi-pass patterning, our recently launched TWINSCAN NXT:1980Di, with significant improvements in all key performance metrics, has rapidly reached productivity of more than 4,000 wafers per day, demonstrating the maturity of our latest NXT platform. In Holistic Lithography, we started rolling out our YieldStar 350E integrated metrology system. YieldStar enables highly accurate metrology for subsequent analysis in ASML's Holistic Lithography software products, which allows customers to control leading-edge production processes for increased yield.
Holistic Lithography products are now extending into EUV processes, with customers evaluating our EUV source mask optimization software for development of seven and five nanometer logic technologies. In EUV, you're all aware that our continued focus has been on improving EUV stability, availability, and productivity, the key performance metrics that drive new technology adoption. Expect no changes in this focus for the foreseeable future. In the past three months, we again demonstrated improved productivity and availability, moving EUV towards manufacturing readiness. By way of example, we achieved productivity of more than 1,350 wafers per day in our factory on an NXE:3350, bringing us closer to our 1,500 wafer per day target for 2016. Separately, three customers showed availability of more than 80% on average for four weeks on the NXE:3300s.
Industry-leading customers presented an abundance of EUV results at the SPIE Advanced Lithography Conference this past quarter that reinforced the need for EUV and demonstrated real and significant progress in key tool performance areas, making increased customer confidence in EUV for manufacturing insertion apparent. ASML's commitment remains to do everything within our capability and power to bring EUV to manufacturing readiness. With that, we would 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. Beforehand, as I always do, I'd like to kindly ask you to limit yourself to one question and one short follow-up if necessary. Of course, this will allow us to get as many callers on to the caller questions on the call as possible. Operator, could we have your instructions and then the first question, please?
Of course, sir. Ladies and gentlemen, this time we'll begin the question and answer session. Again, if you have a question, please press star one to register and star two to withdraw your 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, it is François Meunier from Morgan Stanley. Please go ahead, sir.
Yes, thanks for taking my question. Yes, I think you're making some comments about customers making improvement in terms of availability, which is the key metric for EUV this year. I was wondering, is anyone making, like a pure DRAM maker in those three customers, improving the availability?
There's no pure memory maker. These are customers that are either doing both memory and logic, or they are focused on logic.
Okay. Actually, when in the video you talk about your customers ramping EUV from the end of 2018 in terms of production up to 2020, it is for logic and memory, not just for logic.
Correct.
Okay. Very good. Thank you.
The next question comes from Farhan Ahmad. Please state your company name, followed by your question.
Hi, Farhan Ahmad from Credit Suisse. Thanks for taking my question. My first question is on EUV. There was good progress demonstrated at the SPIE conference, and all the companies appear to be working on it. In terms of the work that you're doing with your customers, can you talk about how many layers of adoption do you expect at the seven nanometer node and also on memory? How should we think about the number of layers progressing for different nodes?
The most relevant information in that sense we have on the node that's upcoming first, that's the seven-nanometer node. The seven-nanometer node for logic, and I have to say that is the seven-nanometer node, not the seven-nanometer node that you need to look at in conjunction with 10. It's what we call seven or five. That can be anywhere between eight and 12 layers. It depends on the customer
Thank you. The second question I have is in terms of your memory shipments. Previously, if I remember correctly, you had indicated that first half shipments would be roughly flat from second half of last year. If I take your first quarter shipment level and I assume that second quarter is flat, maybe down a little bit, it appears the first half of the year is down about 20% from second half of last year, it's going to remain at about that level for rest of the year. Is that a reasonable assumption on how I'm looking at the memory shipments? Also if you could clarify, if there were any push outs on the memory, that led to the somewhat change in the linearity.
For Wolfgang, you're correct. The first half is going to be a little bit lower than the second half of last year, the second half of this year is higher than we originally expected. Overall, there's always like a push and pull, it's actually slightly higher than we thought about last call and also on the call before that.
Got it. Thank you. That's all I have.
You're welcome.
The next question comes from Mehdi Hosseini. Please state your company name followed by your question.
Thank you. Peter, can you please remind us when we're going to learn more about the 3350s that are being installed at the customer site? What do we need to see or track for milestone to better assess the timing of EUV insertion for 7 nm logic? I have a follow-up.
The 3350s, we started shipping 3350s 2 in Q4 of last year, 1 in this quarter. It takes about 6 months to install and to qualify the cluster. Somewhere in this quarter, Q2, you will see the first tools released, which actually drives the partial revenue recognition that we just guided. When that is done, customers start preparing their integration wafers and basically running them in what they call marathon tests of sometimes months, sometimes quarters. We've actually done the marathon test with one of our customers on the 3300, by the way, which involved 13 weeks. That's 1 quarter. This will start in Q2. Somewhere in Q3 we will see the first results. The results will be focusing on the metrics that we just talked about. It's productivity, availability. That's what they will be looking at.
We expect that customers will repeat what we have seen in our own factory, simply because we have the 3350 there since the 2nd half of last year. We've done internal tests also. What you need to track is really data coming out of the customers, and we will support you also with that by the time those marathon tests have been run at the customers and we'll give you the data. Availability and productivity.
Yes. It's hitting the 1,500 throughput target?
I think that's the throughput target that we want to see in a marathon test. At least to see the capability of that tool. Marathon test for customer work could be focusing at different productivity levels, but we would like to see in the install base, the capability shown several times that we can do 1,500 wafers per day by the end of the year. Don't forget that customers don't run 1,500 wafers per day at the current level of the 7 nm rollout. They don't have 1,500 wafers per day for 13 weeks. It's the capability that we need to show on a regular basis plus the agreed-upon results out of marathon tests regarding availability and productivity.
Just one clarification for Wolfgang. I think in your prepared remarks, you said logic bookings will be up in Q2, but you didn't mention memory. Were you referring to booking? If so, any qualitative assessment on memory booking into Q2?
If you look at our total backlog, you see about a quarter is in memory on a total of EUR 3 billion. That makes like EUR 750. With what we shipped already and with our expectations with two equal halves, we will get some more bookings. There will be memory bookings in Q2 as well. You will see again, the bookings weight towards logic because the 10 nanometer ramp is continuing strongly in Q3. You'll see both there, but overall you'll see strong bookings for the second quarter, Mehdi.
Got it. Thanks so much.
The next question comes from Timothy Arcuri. Please state your company name followed by your question.
Cowen and Company. Thanks. I had two. I guess first of all, Wolfgang, I have a question about the customer co-investment program, and what happens when the program ends at the end of next year, and sort of how to think about your development burden. Does the overall R&D spend decline after the end of next year, or does your portion of that burden go up?
Yeah. First of all, yes, the CCIP is about $1.4 billion over five years, and those contributions will end. As you may recall, there is a portion of that recorded through the gross margin. There is a portion in other income, and there is actually a portion that goes straight through the balance sheet. None of it actually is recorded in R&D. Your expectation should be that we are currently running at the EUR 1.1 billion level. We are pretty stable on that, and I would expect that it stays at that level as we continue to invest in EUV, but then also in what prolongs EUV, High NA, Holistic Lithography. For planning purposes, I would just keep it at that level. Just to be safe in our EUR 10 billion model, by 2020, we have modeled 13% that would give us some room to grow and invest here.
For now, I would just keep it flattish at the EUR 1.1, Tim.
Got it. Right. Even though you're being reimbursed through other means, if the development spending isn't going to change, if you're not getting reimbursed anymore, then your burden basically builds up. Is that right?
That's right. We get co-investment, for that co-investment, we made our co-investors also shareholders. When that period is over of the co-investment payments we're getting in, we have to decide by ourselves whether we keep the investment levels, we look at our future business plan and feel it's justified to keep it at that level.
Don't forget, Tim, that the co-investment program was designed to be a kind of a bridge R&D support, until we started shipping EUV. By 2018, you will see the ramp of EUV starting, which also has an impact on the bottom line and will have an impact on the gross margin. By that time, we should be able to stand on our own feet.
Got it. As my second question, do you still plan to ship six to seven EUV systems this year? I think you had said six to seven. Is that still the plan for the full year? Thanks.
That is still the plan.
Okay. Thank you so much.
You're welcome.
The next question comes from C.J. Muse . Please state your company name, followed by your question. Mr. Siche, go ahead, please. The next question comes from Mr. Sandeep Deshpande.
Yeah. Hi. My first question is on the short term, you're talking about this flattish memory trend into the second half, which is not what you were saying earlier. Clearly, things look better than before, Wolfgang. Can you talk about where you're seeing these positive trends in the memory market into the second half of the year, given the difficult memory environment at this point? I have a small follow-up.
Yeah, I'll start and then let Peter chime in. First of all, it's not one customer. We see forecasts at multiple customers. It's also supporting, like in DRAM, the new 18 nanometer nodes to start initial production on that. Then we also see in NAND, continued investments in China with new 3D factories in Singapore and in Japan. It's not one customer, it's for pretty much every customer. It's not like we see overall the same memory levels that we have last year. Memory overall will still be down quite a bit year-over-year, but it came out a little bit better than we expected versus the last two calls.
Thanks, Wolfgang. Following on from that, when you look at the third quarter, when you talked to us in January, you had indicated EUR 1.3 billion in the first quarter, you had indicated that you would see this significant ramp of revenues into the second quarter. How do you see the third quarter and fourth quarter developing at this point in terms of revenue from the second quarter levels?
Yeah, you got to just add up the bits and pieces. In logic, we were pretty clear in both mine and Peter's remarks, logic will be higher in the second half than in the first half. The ramp's not one quarter affair. Q3 is pretty strong. Q4, it remains to be seen just how our customers tape out with their customers and what the yields are and so forth. Memory is stable. Then you see field options and services. We haven't talked about that yet, with a modest start into the year, but every quarter now should be better than the prior quarter, and that will lead us to a 10% increase also year-over-year. It's also quite a bit better in the second half than the first half.
It's relatively safe to assume that the second half is going to be stronger than the first half of the year. As you know, we guide only if we have clarity and certainty, and you also know one of these systems is EUR 50 million, and it can go left or right in a quarter. That's why we rather not do a quantitative guide right now. We're looking pretty good for the year and also for Q3.
Thank you, Wolfgang, for the color.
The next question comes from Jagadish Iyer. Please state your company name followed by your question.
Yeah. Thanks for taking my question. Two questions, Peter. First, one of the things if I look at it, why did your services revenue go down disproportionately when your immersion sales went up quarter-over-quarter in Q1?
I can take this. This is Wolfgang. A little bit of something that we got to consider over time. We actually call this field option and services. That EUR 477 bundle. What you see is that, and we got to quite frankly consider in the future whether we report on this slightly different. The service portion, which is dependent on certain projects like relocations, but mainly on the installed base, is a relatively stable, slowly growing number. What brings the volatility into the combined installed base related revenue are the field options and services. Those include everything from a software option to very, very complex SNAP upgrades that can go into the EUR 20 million or so, or even higher than that. Those vary quarter by quarter. You have seen the exact same thing last year, by the way.
Our Q1 in last year was also, it was just a tad over EUR 400 million, so we're up 20% year-over-year. The service portion is pretty stable, then the options, the field options is the one that brings a little bit of volatility in.
Yeah.
In summary, we look at our forecast and we are very confident that we grow the number year-over-year by about 10% or so.
Okay, great. Thanks for the color. Just a follow-up question. Given how your gross margin is going to be impacted because of the two EUV tools in Q2, and if bulk of the EUV revenue recognition happens next year, what kind of gross margin levels should we be thinking about broadly as we look at in terms of your partial revenue recognition of the six or seven tools that you ship this year? Thank you.
Yeah. First of all, the Q2 dilutive effect from EUV, that's probably the only dilution effect on gross margin that I like because it means that we're actually shipping the product and customers are accepting the product. You're right, the consequence of what we said is, if we have quarters now where we have partial revenue and full cost, we will have in the future, quarters where we have partial revenue and no cost. It will be accretive to a level in a future quarter. It's very difficult for us because it depends on various things, and it varies customer by customer. It's very difficult for us to pinpoint that, and therefore, would rather don't give you numbers. Directionally, it will be accretive in the future of course.
Thank you.
The next question comes from Emmanuel Siche. Please state your company name followed by your question.
Hello, can you hear me?
Yeah.
Sure.
Hey, sorry about that. Yeah, C.J. Muse with Evercore ISI. I guess first question, now that you have your pretty decent visibility into the 10 nanometer ramp, curious how you're thinking about litho intensity relative to 16/14.
Yeah, I think, 10 nanometer, looking at 10 versus 16 nanometer, we think about anywhere between 30%-40% litho intensity. So take an average of 35%.
Okay, great. I guess as a follow-up here, it sounds like, not only a 10 nanometer ramp, but also a nice follow-through of 16/14 spend. If we, I believe, exited last year with about 250,000 wafer starts of equipment shipped, plus what you're expecting here, as well as the EDA guys talking about tape outs of 250, 300 to date. I'm curious, that down 10% node to node that you're talking about, is that something that you actually see at 16/14, or does that start at 10 in your view, given the robust tape outs to date at the 16/14 level?
Well, it actually start in our simulation, in our estimates, it starts at 16 nanometer. Like I said in the early comments that we made, is that, as of 28 nanometer, we see this pattern of this prolonged node. In our 24% of our system sales in the first quarter was 28 nanometer, which is of course, more than four years after its initial introduction is a significant amount of units going to a relatively mature node. When we look at the comments customer currently make about the 20 nm and the 16 nanometer node intensity rising, we expect the same thing. When we look at the 10 nanometer node, we see the initial ramp being relatively speedy by only a very few customers.
That also means that when we talk to the other logic customers, their plans of going to 10 nanometer is a couple of years down the road. The same pattern that we've seen at 28 is going to repeat itself. When we look at the industry analyst reports, we take a couple of analyst reports, and you know the analyst firms, Gartner, VLSI Research and those. We look at what their current expectation is of the end markets that we currently know, and it's all the products that we currently know, like the PCs and the tablets and the servers and the automotive. There's nothing new in there like IoT, because nobody knows what that is.
Let's take what we know and we look at the conservative estimates and the changed estimates based on the most current insight in, for instance, the PC market. We can calculate the number of bits, or the bit growth, going forward. The bit growth we can then translate, look at the roadmap of our customers into square inches, wafers that need to be processed.
We add it all up and we look at the forecast. We just come to a number, that is for those nodes, is about a 10% node reduction. It's based on what we currently know and then based on what we currently see, is based on what customers tell us, which is corroborated by the analyst firms. That is actually what we've used to calculate our EUR 10 billion by 2020 in a number. This is how we do it, and everything that we currently see points into that direction, 10% load on node capacity reduction, but the nodes will be extended. That's going to be the message.
Makes sense. Thank you very much.
Good.
The next question comes from Kai Korschelt. Please state your company name, followed by your question.
Good afternoon. It's Merrill Lynch. My first question was just on the EUV revenue recognition. It's clear that I think you're taking the full cost, only half the revenue, and I think you also mentioned that certainly the EUV tool you're shipping in Q1, you don't expect it to generate any revenues in this year. I'm just wondering how should we think about the balance of this year in terms of revenues? Also looking into 2017, because I think by then you probably would have built up a backlog of, I guess, six or seven EUV tools for which you may have recognized half, possibly less, in a gross margin diluted fashion. I'm just wondering, how should we think about from a phasing perspective about that?
Yeah.
Yeah.
I'll go into that, Kai. First of all, we shipped one system in Q1. We're planning to ship one in Q2. That leaves four to five in Q3, Q4. If you recall, some of these systems are actually the NXE:3300 that customers have already paid for. Some of them are getting some enhancements. Those will lead to quicker revenue, so you can expect some revenue there in the second half. We are also making progress on the maturity of the products and the predictability of the installation process. You see, for instance, we said on these two systems, the 110 we previously said mid-year, and now we're saying it's in Q2. You see we're making some progress, which also makes it easier for us to recognize revenue.
I won't give you a number here, but the additional units that we are shipping, plus some of the performance milestones on the revenue that we already recognized that we may achieve this year, will certainly lead to more EUV revenue this year. Next year, you're right, there's going to be a carryover amount. I mean, we're shipping certain shipments that have no revenue this year. They will have it next year. As Peter mentioned before, we're ready to do a system there on top of that per month. That will be a bit back-end loaded from a shipment perspective, but if you go to a seven-nanometer insertion, or seven-nanometer equivalent, and we talked earlier about DRAM going to be around the same time, and you look at order lead times, people will have to take delivery starting end of 2017 and beginning of 2018.
Yes, you will have a carryover. It's hard to tell you what that number is. We're going to ship more systems next year. The revenue will be plus the predictability of the installation will go up, which in general means we can recognize earlier. The EUV revenue should be quite a bit up next year versus this year.
I think I can't help you any further in that sense, only to mention that 2016 and 2017 will be complex in terms of revenue recognition. That's unfortunately what it is. There's very strict accounting rules for new technology. We have to follow them. On top of that, the first customers that actually place the orders and put certain criteria in there, performance criteria, those are not the same for every customer. The first customer actually gave us the first orders without knowing less than the second customer. Our ability to negotiate terms and conditions and certain performance conditions with the second customer are different than from the first, which will be different from the third. All those performance criterias drive revenue recognition. It is not only what Wolfgang just said, it's also the fact that per customer it is different.
We'll just have to guide you quarter by quarter. It's unfortunately also for us, not always that simple to predict when we do what level of revenue recognition at what margin. I think for the next two years, 2016, 2017, it is what it is.
Okay. Then I just had a quick follow-up because you mentioned it. I believe the 3300s are already prepaid. In terms of cash flow or cash collections on the 3350, how exactly does that work? Does it follow the revenue recognition pattern, or how should we think about the impact on that?
No. We said this also earlier, when we think about our EUV priorities, we think about order first, then shipment, then cash, then revenue recognition. I don't want to sound sloppy, the revenue recognition is what the revenue recognition is for new technology. We have the cash flow much earlier. Again, it depends customer by customer. You can make an average assumption that there is a significant portion of the cash coming at shipment.
Okay. Thank you.
You're welcome, Kai.
The next question comes from Andrew Gardiner. Please state your company name, followed by your question.
Good afternoon. It is from Barclays. Thank you. A bit of a follow-up on that last one in terms of the planning for production next year. You've reiterated your confidence in six to seven EUV tools this year with a gradual ramp in the back half of the year. Clearly, this time last year, we had the sizable multi-year order from Intel. How the conversations are going with the others in the customer base in terms of planning for those production slots. You're clearly planning on increasing the capacity to one a month, as you said. Depending on how people demand the tools or plan for that, you could see some bottlenecks. Can you give us any insight as to how 2017, the planning is firming up?
I think we've also said it on the previous call. I think for next year, we have the capacity to anywhere between 12 and 15 units. Your question, how are the conversations going with the other customers? I think an indication of, you can imagine how that goes. I still like to refer back to what we said at SPIE or what the customer said at SPIE. There was a lot of good data coming out of the customer base that showed a lot of confidence in the fact that EUV will reach manufacturing maturity. You have to remember that those presentations are very often done by the R&D people of our customers, and that the people who have to run the tools and have to commit output to their customers are the people in operations.
The discussions that we're currently having are with the operations people, and they are about availability and levels of productivity. Where they have given us certain targets, and we have given them our internal targets, and they would like to see them. They would like to see them running at the 3350, which is going to be the production tool together with the 3400. This is exactly the phase that we're in, where we're shipping 3350s to the key customers. They're in installation, or in this quarter, some of them will be turned over to the customer. We'll see the first results, and that will drive also the order interaction with us and the customers. To be very honest, I think it's just a matter of time this year that we will see a follow-on.
We're confident that what we see in our factory, we can repeat at the customer side, and that will drive orders, just like you said. If you want to ship 12 to 15 units next year, orders need to come, and they will come.
Okay. Understood. Perhaps just a quick follow-up. Wolfgang, you mentioned that the EUV tool, the 3350 that shipped in the first quarter, you're not going to recognize any revenue on that until 2017. If anything, that seems like a slightly longer timeframe from ship and install to rev rec than you've just planned for with the first two shipments. Is there any reasoning for that?
Andrew, I can't go into much details there because some of this stuff is also customer specific.
Just refer to my answer on the previous question.
Fine. Okay. Thank you, guys.
The next question comes from Gareth Jenkins. Please state your company name, followed by your question.
Thanks. It's Gareth Jenkins from UBS. Couple of good follow-ups. Firstly, I just wondered if you could talk about the applications of the three customers running at 80%, what the sort of intended applications for the EUV process is for those customers. Secondly, I just wonder, Peter, whether you'd expect to step down in availability as you move from an ASML Veldhoven factory environment to a more production-orientated customer site. Then I've got a follow-up on something else. Thanks.
Well, to answer your last question, step down in productivity, no, that's not what we anticipate. We are running these tools here as much as we can under circumstances which are comparable to what the customer will do. That should not be the case. What you're referring to, the 80%, I believe you're referring to availability, Gareth?
Yeah, just the customers in terms of what applications they intend to take EUV to eventually. What are they trialing the 80% availability? What are they running for applications?
Oh, yeah. It's logic and also memory. There's also a memory customer doing this. The 3300s that are currently running are running predominantly in a logic environment, but also in some leading-edge DRAM environments.
Then the last one's just on XPoint. I just wonder whether you could talk about the litho intensity for XPoint that you see, maybe excluding EUV as a factor in the potential step up there.
Yeah, I think the XPoint, it's quite interesting. It's been introduced. If you listen to the customers, it holds great promise in terms of speed and application space. Currently, those XPoint products are made with Deep UV immersion technology, and ultimately, we believe because it's a shrink capability, or at least X and Y and Z direction shrink capability gives us a lot of space for EUV. It's not going to happen before the end of the decade. When you take it all into consideration, and it would ramp in volume towards the end of the decade, then you would think about three times higher intensity for XPoint as compared to 3D NAND.
Thanks.
The next question comes from Douglas Smith. Please state your company name, followed by your question.
Hi, it's Doug Smith from Agency Partners. For Wolfgang, you have continuously guided for around 10% growth in field options and services, it's actually continuously grown much faster than that. I'm not complaining, but since it's around a third of revenues now, can you provide a little more breakdown into that business? I think you hinted that you might want to, or you needed to provide more detail.
You can roughly, in average, think about half of it being service. Like I said, that's pretty stable, and that is growing with the installed base, of course. The other half in average, sometimes it's a bit more than half, sometimes it's less, are various field options in all of our businesses. Will also be the case for EUV. A prominent one is the SNAP upgrade, it stands for System Node Enhancement Package, which is basically a complete open heart surgery on a scanner where you, for instance, can make a 1950, a 1970 or go even beyond that. Those have been introduced last year in volume. We did quite a few there, and that provided for one of the step functions.
We often get asked, "Why is your system revenue flattening out?" Part of the reason is because we are providing a win-win alternative to the customer that helps him with capital intensity, and we're getting an option where the economics for us are acceptable as well. In the past, you may remember, we had several one-time events that gave us growth spurts. For instance, we included Cymer at one year. That's quite a big business as well. Now we're at a pretty decent level. Last year was one third of our business. It's about EUR 2 billion-plus, and we continue to believe that it's growing at least as strong as the rest of our business. That's pretty exciting for us.
Can you say whether the services and field options have higher or lower margins than the systems business?
Service is lower than the corporate average. Field options are higher than the corporate average.
Right. Yeah.
There's a lot of software in there also.
Lots of software components.
Finally, does a lot of that actually come from backlog? I would imagine for field upgrades, it's not like something you would do on the spur of the moment. It must have quite a bit of visibility to it.
That's correct. Just for clarification for everybody on the call, our backlog that we report is just for systems. If we would report that differently, some of the options have longer order lead times. If you think about the SNAP that I mentioned, that's a 6 to 7-week project, and there's limited amount of teams and material, you got to schedule that. There we have more visibility. In some of the software upgrades, we have a little bit shorter lead times. Yeah, there's some visibility on some of the options.
Okay, great.
In the early comments, he said, if that becomes bigger and becomes more relevant, we just need to look at how we're going to report that.
Right. Because if it keeps on growing at the previous rate, it will be EUR 2.5 billion this year or something, just using the trend from the previous years.
Yeah, like Wolfgang said, there are some times these SNAP system upgrades could be anywhere between EUR 20 million to EUR 30 million a piece.
If you have 10 of them, that is EUR 300 million. That is a lot of money.
Yeah.
We just need to look how that is going to develop. If that becomes significant, I think we need to start thinking about a different way to report it so you guys can actually follow it.
Sure. Okay. Thank you very much.
You're welcome.
The next question comes from Amit Harchandani. Please go ahead, sir.
Good afternoon. Amit Harchandani from Citi, and thanks for taking my question. Two quick clarifications, if I may. Firstly, with regards to the topic of equipment reuse, we have again seen some comments from some of your larger customers that have reported Q1 results talking about reusing equipment. Just wanted to confirm if you think the level of reuse in the industry being talked about is still consistent with your longer-term financial model. Also, if you could share any updated thoughts on equipment reuse, and then I have a follow-up. Thank you.
Yeah, I think, in our long-term financial model, as you pointed out, we have included the possibility of our customers making use of the reuse capability. What we're currently seeing is that it's still in line with what we are planning. That is because, like Wolfgang pointed out, these upgrades are open heart surgeries in the field, and they're planned per node. When we discuss a node with a customer, 7 nm nodes, for instance, there is an assumption in there in the discussion with the customer, because of the higher litho intensity, how much of that litho intensity will be split between increase of new systems versus upgrades of existing systems. We have a pretty good view as to how the customers think, and we work that thinking into our long-term planning.
It's pretty consistent, but that is not a surprise because we're executing according to plan.
To add one thing, if I may. We're aware of the comments, but I'd also need to let you know that for us, the 10 and the 7, for instance, in that application, for instance, the litho requirements are the same, and that customer also said that it's actually the same as 16 was to 20. That part, we wouldn't even classify as reuse.
To be honest, when we looked at 28 nanometer, where with some customers, we actually planned reuse of the equipment. Given the strength of the 28 nanometer node, it never got to any reuse. We never got to the point. While with other customers, the 28 nanometer node was not that successful, and we reused it in 14, or in 16, or in 20. It's a bit different per customer, but also the ultimate size of the capacity in a particular node will also determine how much reuse there will be. Other customers will keep using the machine as is.
That's helpful. In terms of the second clarification, if I may, you talked about memory or DRAM in particular shaping up to be better in the second half versus the first half. You talked about the second half being higher than the first half. Would you be willing to also comment and compare it to current market expectations out there for your full-year revenues and say whether you think that that implies an increase in market expectations? Or are there any downward trends that we should be aware of that would deviate away from the market expectation? Thank you.
On the first part, just to clarify, we said memory would be approximately the same in the second half than in the first half. If I would comment on the market expectations, then I would essentially give guidance. We'd better stay away from this and remain, let's just probably repeat it. Logic's going to be up second half. Memory is going to be stable. H1 versus H2, and like we just discussed with Doug, services and field options will go up in the second half.
Thank you.
You're welcome.
I'd like to jump in here momentarily. Ladies and gentlemen, we have time for one last question. If you were unable to get through on this call and still have a question, feel free to contact the ASML investor relations department with your question, and we'll do our very best to get back to you as quickly as we can. Operator, if we can have the last caller, please.
The last question comes from David O'Connor. Please state your company name, followed by your question. Mr. O'Connor?
Operator, let's go ahead to the next-
The next question comes from Amit Daryanani. Please go ahead, sir.
Hi, this is Shaun here for Amit, we're from RBC Capital Markets. One question. In the previous one to two logic node transitions, your bookings were at high levels for about three to four quarters. Given the 10% node-to-node reduction you mentioned and also the quicker ramp at a few customers that you're seeing right now for 10 nanometer, is it fair to assume that the 10 nanometer ramp will be probably one to two quarters shorter than previous node transition, and probably the booking EUR amount will be smaller?
I think because the litho density goes up, it's not likely the booking amount will be smaller. It's not likely. I tried to explain in my introductory statements that these nodes, if you talk about a ramp of a node or a node to a total capacity, it takes a very long time. It has a very long tail end. There will be an initial ramp. Since the pattern is changing, it's very difficult to compare the initial, let's say, first two, three, four quarters of a new node with the nodes with the previous ones. I would summarize it like this. I said the number of customers that over the last couple of years have been able to start an initial new logic node has shrunk. There's only a very few.
They are more aggressive in ramping the first part of that node because they have to make sure that they can provide their key customers with wafers. That's what you will see. Then you have a long tail end. I don't think you can draw any conclusions from that other than that the nodes will be longer and that the litho intensity will go up. Hence, when you also look at the need for EUV, ASML is looking to grow its top line, and we still stick to our simulated number by 2020, EUR 10 billion.
That's helpful. One follow-up. Last year, you had an EUV volume purchase agreement with at least 15 tools with a U.S. logic customer. Given the discussion you have right now with this customer, do you have any update on the number of tools will be shipped under this agreement?
We'll give you the update when we get the orders. This is a volume purchase agreement where purchase orders are issued according to a predetermined pattern, which is a reflection of when the customers need the tools to put them into production. We'll inform you when we get the orders.
Thank you.
You're welcome.
Thanks. On behalf of ASML's Board of Management, I'd now like to thank you for joining us on the call today. Operator, if you could formally conclude the call, I'd appreciate it. Thank you.
Thank you, sir. Ladies and gentlemen, this concludes the ASML 2016 first quarter financial results. Thank you for participating. You may now disconnect your line.