ASM International NV

AEX:ASM

Earnings Call Transcript 2018-Q3

[Operator]

Good day, and welcome to the SAM (sic) [ASM] International Q3 2018 Earnings Call. Today's conference is being recorded.At this time I would like to turn the conference over to Mr. Victor Bareño. Please go ahead, sir.

[Victor Bareño, Director of Investor Relations]

Thank you, Anna. ASMI issued its 2018 third quarter results last evening. For those of you who have not yet seen the press release, it, along with our latest investor presentation, is accessible on our website, ASM.com.As always, we remind you that this conference call may contain information relating to ASM's future business or results in addition to historical information. For more information on risk factors related to such forward-looking statements, please refer to the company's press releases, reports, and financial statements, which are available on our website.And with that, I will turn the call over to Chuck del Prado, President and CEO of ASMI. Chuck?

[Charles Dean del Prado]

Thank you, Victor, and thanks to everyone for attending our third quarter 2018 results conference call and also for your continuing interest in the company. After a review of operations, Peter van Bommel, our CFO, will join me in answering any questions that you may have.So let's start with a review of the third quarter financial results. Net sales in the quarter decreased by 6% compared to the second quarter and increased by 10% from the third quarter of last year. At a level of EUR 196 million, third quarter sales were towards the high end, higher end of our guidance, which was a range between EUR 180 million and EUR 200 million.Service and spare parts counted for 24% of total sales in the third quarter and were up 14% year-on-year. Equipment sales were led by our ALD business, but we also recorded healthy sales levels in our other product lines.By industry segment, revenue in the quarter was led by foundry, followed by memory and then logic. Within memory the larger part of sales was related to DRAM investments, while NAND sales decreased compared to the second quarter, as expected. The gross margin amounted to 41% in the third quarter, a decrease compared to 42% in Q2 and an increase from 39% in the year-ago period. The decrease compared to the second quarter is largely explained by the higher cost on evaluation tools, reflecting an increase in customer engagements for new products and applications.As earlier indicated, while remaining within the target range of low to mid-40s, our gross margin is expected to show a somewhat higher volatility in the next few quarters than we have seen in previous years. This expected higher volatility is mainly explained by a healthy pipeline of new products that we are bringing to the market, which initially might lead to some extra costs.In addition, from quarter-to-quarter the gross margin may be impacted by changes in the sales mix, as well as costs associated with our preparations for expansion. As discussed at earlier occasions, these preparations for expansion include a step-up of our service organization as well as the construction of two new facilities. In October this year we completed our new facility in Dongtan, Korea. This new facility includes an expanded state-of-the-art R&D lab and is much closer located to our key customer in Korea. Preparations for the construction of a new manufacturing facility in Singapore are ongoing.If we then look at the operating expenses, SG&A increased by 2% from second quarter, while R&D increased by 10%, mainly due to lower capitalization. The operating result decreased to EUR 28 million, down from EUR 38 million in the second quarter, which is for the largest part explained by the lower revenue. Operating results increased with EUR 7 million, compared to EUR 21 million in the third quarter of last year.Financing results in the third quarter included a currency translation gain of EUR 1 million, and this compares to translation gain of EUR 8 million in the second quarter and a translation loss of EUR 8 million in the third quarter of last year.Net earnings on a normalized basis amounted to EUR 42 million, down from EUR 20 million from the second quarter and about unchanged compared to the third quarter of last year.Looking at ASMPT, normalized results from investments, which reflect our share of approximately 25% in the net earnings of ASMPT, decreased to EUR 17 million for the quarter, down from EUR 22 million in the second quarter. In the third quarter ASMPT had sales of $658 million, down 2% from the second quarter and up 1% year-on-year.Bookings in the third quarter decreased 15% compared to Q2 but increased 8% year-on-year for ASMPT. The third quarter bookings are [indiscernible] are reflecting the seasonal correction, having said that they were at a historic high -- historical high third quarter level.So now turning back to ASMI's consolidated operations, we booked orders of EUR 258 million in the third quarter, which is 25% above our previous highest level in the history of the company and well above the range of EUR 200 million to EUR 230 million that we had guided for. Orders increased strongly by 47% compared to Q2 and by 61% compared to the third quarter of last year. The upsides compared to our earlier expectations was driven by higher than expected orders, and particularly in the logic and analog segments.If we look at the ranking by industry segment, bookings in the quarter were led by logic, followed by foundry and then analog. Logic bookings increased sharply compared to Q2, and foundry bookings were solid and also increased compared to Q2, driven by further spending on the 7-nanometer node as well as early investments in the 5-nanometer node. Also, in analog, bookings jumped sharply compared to Q2. In the memory segment total orders dropped compared to the previous quarter, driven by a sequential decrease in DRAM, while NAND remained at a relatively low level. By product line, orders in the quarter were driven by our ALD business. In epitaxy we had high bookings, driven by strong demand in analog power, and vertical furnace bookings were also healthy, driven by logic and analog.So now turning to the balance sheet and cash flow, at the end of September the cash position dropped to EUR 266 million, down from EUR 651 million at the end of June. This decrease is fully explained by a total of EUR 383 million that we returned to shareholders during the quarter in the form of share buybacks and a capital repayment.During the quarter we also received EUR 15 million in dividends from ASMPT, while the operating free cash flow was negative EUR 17 million. The letter is mainly caused by an increase in our working capital to EUR 225 million at the end of September, up from EUR 213 million at the end of June. This in turn is mainly explained by inventories, which were higher at the end of the third quarter in preparation of expected higher sales in the fourth quarter and because we got early receipts from suppliers, as tight conditions in the supply chain loosened somewhat in the third quarter.If we look back in more detail at the share buyback during the third quarter, we spent EUR 167 million to buy back 3.7 million of our own shares at an average share price of about EUR 45. This was part of, as you well know, of the 250 million share buyback program that we started last June. As a reminder, this share buyback program was funded with parts of the proceeds of the 9% stake sale in ASMPT that we executed in November of last year.We used the other part, a bit over EUR 200 million, for the tax-efficient capital repayment of EUR 4 per share. This was paid out to shareholders on August 10 of this year.Furthermore, during the quarter the earlier announced cancellation of 6 million of our Treasury shares became effective. The number of outstanding shares at the end of September net of Treasury shares amounted to approximately 50 million shares. This is a decrease of 14% compared to the share account at the end of Q3 last year.Following the end of the third quarter, on October 15 we completed the EUR 250 million buyback program. In total, we purchased slightly over 5.4 million shares at an average price of almost EUR 46 under this program. As communicated before, we intend to cancel the shares that we have repurchased under this program.In 2018 year-to-date and including the share buybacks in October, we have returned a total of EUR 607 million to our shareholders in the form of annual dividends, the tax-efficient capital return, and share buybacks. This is up from a total of EUR 285 million that we returned to shareholders last year.So let's now take a closer look at the trends that we see in our markets. The semiconductor end markets have shown solid growth so far this year and are expected to grow a low double-digit percentage in 2018. However, against a backdrop of rising inventories and increased price pressure in primarily the memory segments and the macroeconomic risks related to trade frictions, market research firms are forecasting growth of the semiconductor end markets to slow down in the coming quarters.Looking at the WFE market, the wafer fab equipment market, we expect growth in 2018 to be in the mid to high single digits. Drivers behind this growth of WFE are still expected to be the logic and DRAM segments, while our view is that the spending in 3D NAND will be down year-over-year. Spending in the second half of 2018 is expected to be lower than in the first half of the year for the WFE market as a whole due to a drop in memory. This is unchanged compared to our expectation that we shared with you 3 months ago.For ASM, we expect our sales for the second quarter to increase strongly compared to the first half. The key drivers behind this sequential increase are logic, foundry, analog, and, to a lesser extent, DRAM. Our NAND sales are expected to lower in the second half, which is fully in line with our previous expectations and guidance to the market. We still expect to outperform WFE this year.An important factor that supports our expected outperformance compared to WFE in the second half of the year is our strong exposure to the logic and foundry sector. The sector exposure to logic foundry has substantially increased over the last couple of years. As we highlighted on previous occasions, with the transition to the multi-advanced node our relevant SAM served available market. Analogic and foundry has significantly increased, including a meaningful increase in the number of ALD layers.This has helped us to achieve a substantial share of wallet gains with key customers in the logic and foundry sector. To further illustrate this point, in the first three quarters of 2018 and also in the full year of 2017 our non-memory markets, for which foundry and logic are the biggest parts, accounted for clearly more than half of our total sales.Looking in more detail at WFE in the logic sector, while the mix of the different nodes has somewhat changed during the year, spending in the advanced logic segment is clearly higher in 2018 compared to last year. Looking at foundry, spending is still expected to be lower this year, albeit still at a higher -- at a healthy level, and is driven by further 7-nanometer investments as well as the first investments in 5-nanometer. Investments for the 5-nanometer node are likely to further increase in the course of next year.We repeat our expectation that the 5-nanometer foundry node is going to be an important transition for our company, with good opportunities to expand our position with multiple new ALD applications/layers and to increase our overall share of wallet in foundry. Hence, foundry has a potential to be a strong contributor to the company's 2019 revenue.The DRAM segment is still expected to show solid growth within the WFE market in '18, in 2018. For ASM, DRAM is also one of the drivers behind higher sales this year. We remain strongly focused on broadening our position in DRAM to new ALD solutions beyond our traditional position in [indiscernible], and we aim to achieve the first contribution of these new solutions once customers start to invest more meaningfully in the next nodes.Now turning to 3D NAND, as we already discussed in our Q1 and Q2 earnings call, WFE spending in the NAND segment is coming down significantly in the second half of this year. Also, for the full year of 2018 NAND spending is expected to be lower, in our view. This correction is not unexpected. Customers made strong investments in both new capacity and 3D NAND conversions over the last few years, including the first half of '18, and are currently digesting these investments.A recovery in investments in the NAND segment will likely be driven by the transition to the next device generations of 90-plus layers, which customers are planning to start ramping no earlier than the second half of next year. Our focus is to continue working with customers on these next device generations that are expected to need an increased number of advanced single wave for ALD applications. We still plan to increase the part of the 3D NAND ALD market that we address.This quarter we also highlighted the record high order intake in the analog segment. While this segment is clearly smaller than logic or foundry, it has a meaningful contribution to the increase in our bookings in the quarter. Demand in analog was driven by a broad base of customers across the globe for products such as power devices and IoT-related products. In terms of product lines, the strong demand in analog contributed to higher bookings in vertical furnaces and in our Epi business in the quarter. For the full year 2018, we expect a marked step-up in our analog sales compared to relatively steady levels in the previous years.The structural outlook for the ALD market remains strong. ALD continues to be a key enabler of Moore's Law. An increasing number of ALD steps will be needed to help create the smaller geometries, new materials, and more complex device architectures that are on our customers' roadmap. In addition, we remain focused on expanding our served available market within the total ALD market, including parts of the market that we have so far not yet addressed.Our new Synergis product that we launched this past summer at SEMICON West is an important next step in this growth strategy. By the end of the year we expect to have both notable orders of this new ALD tool, with a further increase foreseen in 2019.Looking at the solid momentum in our other product lines in recent quarters, Epi and furnaces, we are strengthened in our belief that the combined revenue of the other product lines will have a structurally higher contribution for the company in the coming years.So let's now take a short look at Q4 guidance as we communicated in our press release, in our overnight press release. So for Q4 on a currency comparable level, we expect sales of between EUR 220 million and EUR 250 million euro and an order intake between EUR 250 million and EUR 260 million. And, yes, we added the comment that Q4 still reflects some uncertainty around the exact timing of individual tools.So that concludes our introduction, including our outlook. At this point Peter and I are more than happy to answer any questions that you may have.

[Petrus Antonius Maria van Bommel, CFO & Member of the Management Board]

We'd like to ask you to please limit your questions to not more than two at a time so that everybody has a chance to ask a question. All right, Anna, we are ready for the first question.

[Operator]

Thank you.[Operator Instructions]We will now take our first question from Peter Olofsen from Kepler Cheuvreux.

[Peter Olofsen, Analyst]

My first question is on the order guidance for Q4, which looks pretty strong. I understand there was a solid contribution from foundry in Q3. To what extent will you see further contribution from the 7-nanometer buildout to the Q4 order intake, or is that buildout nearing its completion? And is it fair to assume that some of the opportunities that you talked about in NAND and DRAM won't yet have a big contribution to the Q4 order intake and thus may be more for next year? And then I have a follow-up, please.

[Charles Dean del Prado]

Okay, Peter, thanks for your question. So, yes, pretty straightforward answers to your questions. So first of all will foundry have -- is expected to have a solid contribution also to Q4 bookings. The answer to that is clearly yes. Secondly on new NAND and DRAM contribution to Q4 bookings, we do foresee memory -- yes, it's very well possible that memory contribution to bookings will go up Q on Q, but more in the DRAM space, maybe, than in the NAND space, but more focused on existing applications. New applications really are more related to, let's say, a ramping of new nodes, and that's not to be expected before the new year kicks in.

[Peter Olofsen, Analyst]

Okay, and just to clarify, the foundry will still be mainly 7-nanometer related?

[Charles Dean del Prado]

I think more and more you are seeing that the spending in foundry is shifting to 5.

[Peter Olofsen, Analyst]

Okay.

[Charles Dean del Prado]

So earlier in the year let's say the ratio 7 to 5 was more in favor of 7, but in the course of the year it shifted more towards 5, and that's to be, certainly for bookings Q4, so it will be more related to 5.

[Peter Olofsen, Analyst]

Okay. Then a more strategic question. Some of your competitors can combine different competencies because they are active in deposition but also in etch, clean, and inspection. With chip making getting more and more complex, would that put you potentially at a disadvantage, or is it something you may try to solve by wanting to get a way for other players in the industry?

[Charles Dean del Prado]

Our growth strategy takes very well into account that we don't have -- we are not active, let's say, in core etch, and we don't see a limitation, let's say, in the next couple of years in any way for us to go.

[Operator]

We take our next question from Nigel Van Putten from Kempen.

[Nigel Van Putten, Analyst]

It's great to see analog is picking up strongly. You already provided some context. It's broad-based. Could you also provide some indication of is this a sort of one off, this is a temporary spike? Do you expect this to continue well into the next year? And what will be driving it? That's my first question. Thanks.

[Charles Dean del Prado]

Yes, Nigel, first of all, we absolutely don't think this is an incidental thing. Will there be some cycles, at certain period of time it will be stronger than other periods of time? That for sure will be the case like in every industry segment. But we really see -- if you look at the underlying end markets, such as industrial and automotive, those are really long-term drivers of growth of this analog market. So we see a pretty bright future in these segments in the coming years. And we've worked hard to position ourselves well in that market, and we are continuing to adjust some of our configurations in several product lines to even more tune those to the needs of these markets. And specifically we see -- we've seen bookings significantly increase in the analog market Q on Q in Q3, and our expectation is that at least in the short term the strengths will stay. How it will develop in the course of 2019, that's of course too early to say. But if you look in the short term we have pretty good visibility on a solid climate there based on the feedback from our customers today and the way the backlog has built in this segment.

[Nigel Van Putten, Analyst]

Great. Thanks. And then indeed on visibility also in the foundry and logic segment, I think logic's strong. Maybe we've seen a CapEx bump, but I think IDM helping this quarter, maybe also the coming quarters. And also to what extent do you see that maybe going into 2019? I know it's early days, but is that also mostly first half related, or could it be longer in the year? Then also on foundry, we've seen reuse to equipment from 10 to 7 and now moving from 7 to 7-plus. The new flag, obviously, the investment in 5-nanometer picking up quite strongly. So how should we see that in terms of on the foundry side will that be enough to offset maybe reuse, and in logic specifically do you expect that spend to be maybe a bit more 1H weighted versus the rest of the year, or is that too early to call still?

[Charles Dean del Prado]

Yes, of course, our visibility is best for the first half at this moment in time. But looking at the full year, so far logic in general for the market in general is expected to be reasonably healthy, assuming that the 10-nanometer ramp will continue to get the focus that has developed, that has been developed in advanced logic, let's say, in the course of this year. So under the assumption that 10-nanometer continues to get focus, then '19 could be a very healthy year for logic. For foundry, yes, it's really determined by the spending plans for 5-nanometer, and, yes, so far the feedback from customers, the foundry customers, is that there's a very strong commitment to build initial capacity in the 5-nanometer node, and in the course of 2019 we of course will see the level of wafer starts that really are needed by end markets in the foundry space. But the initial commitment to get the ramp going in 5-nanometers we have seen to be the commitment we have recognize to be very, very strong. And that's what we are serving at this moment in time. And we view also there's a very strong opportunity for ASM, this transition from 7 to 5. We view it as a much stronger opportunity to build, to increase our share of wallet, than the transition from 10 to 7. As you may remember, 16 to 10 was a tremendous opportunity for us to increase the amount of layers, increase share of wallet, and the transition from 7 to 5 we perceive as a similar opportunity. So we're very happy with the developments so far. And, yes, in terms of reuse, we have seen some reuse in the 7-nanometer nodes, coming from 10 to 7. Somehow also given the fact that we are engaged with more applications now, we trust that there is a -- that's a bigger opportunity to build a healthy revenue level in 5-nanometer than in 7.

[Operator]

We will take our next question from Quang Le from Credit Suisse.

[Quang Tung Le, Research Analyst]

Thank you for taking my question. My first question will be regarding your gross margin. Obviously you said it was down, quote, unquote, in Q3, and you are still seeing that could be some volatility in terms of higher number of evaluation tools. So in Q4 I'm not sure what is your visibility right now, but do you see gross margin still in your low 40s, and is it going to be up or down from Q3?

[Petrus Antonius Maria van Bommel, CFO & Member of the Management Board]

Yes, let me answer that, Quang Le. What we have seen, we always have said that the gross margin will be in low to mid-40s. So we have seen somewhat more volatility this year due to the introduction of new products. We have explained also in the previous quarters because there were a few things which were jumping, especially the introduction of the intrepid and PECVD. That is now -- those margins are now more or less on the normal levels. So we are back in that low to mid-40s on a more structural way. As we have seen now in Q3, the impact was mostly of the [indiscernible] things. The mix in general in the third quarter of products was more at the lower side than on the upper side than what we have seen in previous quarters. And we would be disappointed a little bit when the margin in the fourth quarter would not be a little bit higher than what we have seen in the third quarter.

[Quang Tung Le, Research Analyst]

Oh, so you expect Q4 to be higher, and otherwise you would be disappointed. That's what I understand. Another question of mine will be regarding your outlook, whether it's WFE or on your site only for 2019. We saw -- we see your outlook for 2018. But do you have any visibility beyond that?

[Charles Dean del Prado]

Yes, okay, okay, yes. So we gave some comment based on the former question on '19, but I assume you are looking at the market as a whole. Again, we have -- on 2018 we have been -- first on 2018, we have been very consistent in our guidance to the market from the beginning of the year. Initially at the early start of the year we guided toward let's say high single digit growth in '18. Then we adjusted it to mid- to high single digit and we stuck to that guidance throughout the year, and it now seems to really show that that was the right guidance on the market as a whole. We also commented early on in the year when the projections for next year were aggressively negative, in the 20s, minus 20% negative, we thought that that was too much given the fact that some of the slowdown, especially in memory, already was happening in the course of this year. And you saw also industry analysts in the course of the year adjusting their number for '19 downwards, in a way, less negative. And now the common denominator is somewhere in the high single digits negative. So that's between 8% and 10% negative or so for next year. And at this moment in time with the visibility we have we tend to concur with that projection. And then with the assumption again that logic will stay reasonably healthy in 2019, again, assuming that 10-nanometer RAM will get continued focus. Foundry expected to be up, driven by 5-nanometer. DRAM to, yes, to maybe year-on-year drop somewhat after two very, very strong years of strong spending, but still, although there's a year-on-year drop, still a decent contribution throughout the year on a quarterly basis. And then for NAND we expect that the first half will still be weak of next year in line with also somewhat weaker second half of this year. And then possibly in the second half of next year the NAND will start to spend overall a little bit more in the 96 stack, in the 90-plus stacks and beyond. That's the color we can give based on the visibility we have from our customers today.

[Operator]

[Operator Instructions]We will take our next question from Robert Sanders from Iberia Bank.

[Robert Duncan Cobban Sanders, Director]

Yes, it's Deutsche Bank, not Iberia Bank. Just related to the previous question, if logic and foundry again are going to be relatively strong next year and that's where you have your highest market share in ALD and where you are the primary source to the two largest players, does that have any effect on your mix and your gross margin as we look into 2019? And I have a follow-up. Thanks.

[Petrus Antonius Maria van Bommel, CFO & Member of the Management Board]

No, Robert, the gross margin for -- we have no big differences in gross margins with the different customers that we have.

[Robert Duncan Cobban Sanders, Director]

Okay, got it. And just related to your commentary about 5-nanometer and 10-nanometer and the big logic -- I'm sorry, go ahead?

[Petrus Antonius Maria van Bommel, CFO & Member of the Management Board]

The gross margin mix is more related to certain applications that we have. That can differ per customer, of course, the sort of applications that you have. But not in the customer as a whole.

[Robert Duncan Cobban Sanders, Director]

Got it. And just one thing that struck me is that you say that your large foundry customer is starting 5-nanometer quite aggressively and the large IDM is planning to do 10-nanometer next year. I'm just surprised the order book has swollen so much so soon given you have a 3-month lead time. So what explains -- a lot of people might look at these numbers and say this is very lumpy and it's not likely to repeat, and this is kind of orders that have been pulled in. But what explains the kind of earlier than expected timing for these orders that you've seen in Q3 and Q4 relative to what most people talk about being the ranked schedules, which is obviously 5-nanometer foundry in 2020 and the big IDM 10-nano next year? Thanks.

[Charles Dean del Prado]

Yes. Yes, okay, so first to clarify that, because we have not been completely clear on that, we did not state that the foundry is aggressively ramping. We said that the investment levels for foundries in the course of the year shifted more from 7 to 5. But it is really early, let's say, pilot production that is being built now, and real ramp preparations will need to happen in the course of '19 and that preparing for solid wafer start-outs, high-volume wafer starts out in 2020. So this is really an early start, but we are happily engaged in that early start in the foundry space. So that's what you are seeing now. Ten-nanometer, you saw that in the course of the year, that's our assessment, that in the course of the year in logic, in advanced logic, more of the bookings started to shift from 14 to 10. So and we are not about the exact timing of our customers in ramping their wafers. But they are at least more and more building capacity now based on the order book that we have for 10 than on 14. That's what we are seeing. Just clarifying some of that. And then, yes, you said if this is a one-time bump for us. I think maybe you can read it a little differently. As you very well know that when you started also the coverage on the company, which we are very happy with, you said you know that we have a relative strength in logic foundry versus memory. So that is a very important data point that you have to take into account in looking at ASMI. And secondly, it's very important to know that we have shared with you and with the market over the last couple of quarters that we are aggressively working on increasing our share of wallet, going to new nodes, so in foundry going to 5, in logic going to 10, and that we, besides that, also are aggressively working on expanding our source available market, not only in ALD but also in the other product lines. And, yes, if you combine that with the fact that logic and foundry in general are spending more in the second half than in the first half, as we have also guided the market for since the beginning of the year, while memory is slowing down, especially NAND, that that explains also, that explains for a big part that as a result of that our Q3 and potentially Q4 orders are strong contributors there, our logic foundry and to a lesser extent analog. That's basically the story.

[Operator]

We will take our next question from David O'Connor, Exane BNP Paribas.

[David O'Connor, Analyst of IT Hardware and Semiconductors]

Maybe firstly on the linearity of orders in the advanced logic, so the strength we're seeing in logic in Q3 and again in Q4, the order bookings, in the past what has been the duration of the logic order strength when a new node ramps? That's my first question and I have a follow-up.

[Charles Dean del Prado]

Can you repeat one more time? Sorry. David, can you repeat one more time your first question?

[David O'Connor, Analyst of IT Hardware and Semiconductors]

Yes, sure. I just want to get a sense for how the duration of strength that we could see in the advanced logic at the 10-nanometer ramp, just wondering in the past, for instance, at the 14-nanometer, how long does the initial ramp-up last? Is it a 1 to 2-quarter kind of bump, or do you see like 3 to 4 quarters and it tails off after that? Just trying to get a sense of the duration of this order strength on the advanced logic side.

[Charles Dean del Prado]

Yes, that's very difficult to assess, David. We would love to give you a little bit more guidance on that, but the only thing we can say is that in logic this node was basically -- had an early announcement some time ago in '16, 2016, and the customer really took the time to prepare the ramp. And now we really see a meaningful momentum developing like we have not seen any time before since they made that announcement in '16. But how long it will last, that's not for us to comment on. The only thing we are focused on is to position ourselves well with multiple product lines. And, yes, and again, we saw it already earlier on. We basically built the foundation for this ramp. This is not something -- the bookings development now is not something of today. It's something we worked on over the last 3, 4 years, to really increase our share of wallet in this node, so which would become visible as soon as the customer would ramp. And that's basically what's happening now.

[David O'Connor, Analyst of IT Hardware and Semiconductors]

Okay, got it. Just to clarify, my question is more on the previous node, so looking back at the 14-nanometer node, do all those orders come in 1 quarter to 2 quarters, or is it more spread out over a couple of quarters?

[Charles Dean del Prado]

That's spread out. That has been spread out. If you look at the history for the year, that has been spread out.

[David O'Connor, Analyst of IT Hardware and Semiconductors]

Okay, got it. And then maybe a follow-up. When you look at the current, so this current 10-nanometer advanced logic, and compare that to 5-nanometer foundry, how would you compare them in terms of ALD intensity? Thanks.

[Charles Dean del Prado]

So you are -- within foundry, you said?

[David O'Connor, Analyst of IT Hardware and Semiconductors]

Well, look, between the 5-nanometer foundry and what you see in advanced logic 10-nanometer. How would you kind of -- could you give us some kind of sense of how they compare in terms of ALD intensity?

[Charles Dean del Prado]

Well, what we can say is that in both instances, in logic the intensity, the ALD intensity, is increased, going from 16 to 10, in a significant way in terms of amount of layers. And the same is happening going from 7 to 5 in the foundry space. And then of course combined with that you have the amount of layers with which you are penetrated, and then for the rest it's how many wafer starts the customer is going to make, which determines ultimately how much revenue you will generate. That's as far as I would like to go in commenting in this forum, David.

[Operator]

We will take our next question from Marc Hesselink from ING.

[Marc Hesselink, Research Analyst]

The first question may be a little follow-up still on the last one. If you see the shift from 7-nanometer to 5-nanometer, what does it do for the ALD intensity, and then also taking into account that you -- that they won't be using EUV in that node. If you're talking about that's going to be a big node, does it have to do with the fact that there will be more wafer starts, or is it really the ALD intensity that's also really a step-up there?

[Charles Dean del Prado]

What's relevant to us is both. So it's not that -- so if your question is our engagement in terms of ALD layers going down, going from 7 to 5 as a result of EUV, then we have to compensate that with wafer starts, then the answer is no.

[Marc Hesselink, Research Analyst]

Okay, clear. And then the other question is on the evaluation tools. What does it mean for going forward? Is that -- does it mean that there are some big opportunities that works with new applications, new clients, going forward because there are so many evaluation tools in the field at this stage?

[Petrus Antonius Maria van Bommel, CFO & Member of the Management Board]

What we normally do is when we introduce a lot of new products, well, we introduce new products in the field, and for a period of time, mostly for 6 to 12 months, we place a so-called eval tool with the customer. Then he can align his processes with our tool, and when that works out well, then he buys the tool and afterwards he will buy equipment in high-volume manufacturing. So we have seen this year, you see that also on the balance sheet how that was developing. I mean, we saw that we have nearly 20 million more of eval tools at this moment outstanding until Christmas. And that's representing the volume of new activities that we have started up in the past period, especially for complete new tools for new applications. We deliver those sort of eval tools. So we are on this moment happily working together with customers. That is also one of the underpinning factors why we expect that order intake is what it is in Q3 and is also the basis of the further guidance that we have given for the fourth quarter.

[Marc Hesselink, Research Analyst]

Okay, clear. And then maybe as a final question. With -- yes, we see the weakness in 3D NAND, but given the fact that you’re still trying to get more applications in 3D NAND, what should it imply for you? Could it be that even if the market is down that because of the fact that you're getting more applications that you're still going to grow your 3D NAND revenues?

[Charles Dean del Prado]

So your question is if we can grow our NAND revenue regardless of the climate in the NAND space. That was your question?

[Marc Hesselink, Research Analyst]

Yes, actually, given the fact that you didn't have all the applications in [indiscernible].

[Charles Dean del Prado]

Yes, okay, I understand your question. That makes sense, your question, indeed. But, yes, again, the opportunities we are focused on to expand the number of applications in NAND are in, let's say, the more advanced sets. And the investments in the advanced sets are low at this moment in time. So that is -- so in that respect we are also depending for that expansion on our customers reengaging on their -- in their commitment on ramping the newer nodes.

[Operator]

We will take our next question from Edwin de Jong from NIBC.

[Edwin de Jong, Analyst]

A few questions left on analog. In your prepared remarks you said that could be down a little bit more significant for the top line of ASMI. Could you maybe give some more color? What percent of sales it is now more or less and then what percent that you see it going to in the next 2 years? That's my first question, and then I have a follow-up.

[Charles Dean del Prado]

Yes. That's a tough one. Let's put it this way, that coming from a low base, let's say the bookings much more than doubled Q on Q going from Q2 to Q3. But we are not going to give percentages. And as we guided for, it's the number 3 segment behind logic and foundry in Q3 in terms of bookings, let's say above memory in Q3. So that should give you some color also.

[Edwin de Jong, Analyst]

Okay. And then maybe on the working capital, also going for [indiscernible] had high step up there to EUR 185 million in Q3. And can you give some color on how that could develop in [indiscernible] that you are currently seeing in Q4 and maybe Q1.

[Petrus Antonius Maria van Bommel, CFO & Member of the Management Board]

Yes, working capital of EUR 185 million eventually -- indeed [indiscernible] too high Edwin, so let's mention that also. There are a few elements in it which I think are important in this moment. First of all, they look to the different elements of the inventory. You should have to look to on the one hand the work in progress. The work in progress is a function and principle from the throughput time and volume of orders that we have. That looks very healthy. That's also -- there's no issue [indiscernible] in that area. When you look to past, then you see that revenue increasing. And that's also in a lot of cases in preparation of the new activities. When you have a lot of new products with a lot of new customers you need to have more space, so that's also developing and aligning with the activity level that we see. We have seen some finished goods from this moment increasing in the past, especially the third quarter, where that's normally that's very stable and a very low level because in principle we do not have much finished goods. That's more a short-term issue. Customer desires not to get the orders of the equipment delivered at the end of that quarter but early in the next quarter, so that's also something that shall be phasing out rather quickly. That has an impact of approximately 10 million inventories in the third quarter. And the most important one is the raw materials, and the raw materials are in principle a function of the activities that you are going to do. And here are the issues that we have on this moment. On the one hand, of course, we need more raw materials at the beginning of the quarter, at the end of the quarter for the start of the new quarter. So then you can imagine when we give a guidance which is substantially above the level before that we need some more raw materials, so that's a healthy part. But on the other hand we have been confronted in the past periods that the full supply chain was heavily loaded. As a consequence of that, the lead times of our suppliers were increasing, and that has led to the situation that we ordered more in advance than what we have done before. And now that that supply chain has less orders, even cancellations, you see now that our suppliers are delivering within the lead time or even sometimes much shorter, and that’s caused to certain overstock situation at the end of this quarter. And that is something where we are working on with the organization, because that has to reduce, and highly likely you will see the first results of that ready by the end of this quarter.

[Edwin de Jong, Analyst]

And this is also let’s say EUR 5 million to 10 million?

[Petrus Antonius Maria van Bommel, CFO & Member of the Management Board]

That is more than EUR 5 million to EUR 10 million.

[Operator]

We will take our next question from Nigel Van Putten from Kempen.

[Nigel Van Putten, Analyst]

Just a couple of quick clarifications. First off, on your comment saying NAND memory is well over half of the revenue in this year and recent years, that's over the entire products portfolio, right? I mean, do we look specifically to ALD? That could be the other way around. Is that correct?

[Charles Dean del Prado]

That's all products.

[Nigel Van Putten, Analyst]

Yes, okay. Clear. And then on R&D capitalization, amortization, still quite high level of capitalization in the quarter. I think next year at least amortization is probably going to increase quite a lot. Is capitalization expected to stay at the current level, or is it also coming down? So is it going to be higher amortization, lower capitalization, or is there effect of canceling the two out?

[Petrus Antonius Maria van Bommel, CFO & Member of the Management Board]

I expect that capitalization will be slightly higher than what we have seen in the past, so I don't expect that 2018 that that's -- 2019 will show a level what we have seen before in 2017. But I do not expect 2019 the capitalization to be as high as it was in 2018. It's very slowly dependent on so all the development projects that we have, when they exactly finalize that, because what I explained to previous, in previous calls is that the more that you reach the end level of a development project and then that looks rather successful, then according to [indiscernible] we have to capitalize, and then we are also capitalizing that. And we have seen those impacts in principle peaking in the second quarter. You saw a small reduction in the third quarter. I expect in the fourth quarter that the capitalization will not be higher than what we have seen in the third quarter. Will it be much lower? That is, I think, too early to judge from this moment.

[Operator]

We will take our last question from Peter Olofsen from Kepler Cheuvreux.

[Peter Olofsen, Analyst]

Sorry, I was on mute. Yes, I had a follow-up on epitaxy in foundry, where you had a breakthrough, I think, in 2017, the 7-nanometer. With the foundry customer to go to 5-nanometer, will that have a positive impact in terms of number of layers, number of applications that you may address with your epi product?

[Charles Dean del Prado]

We expect to improve our position going from 7 to 5.

[Peter Olofsen, Analyst]

That is somewhat similar to what you see in ALD, because the number of applications is increasing and the number of layers is going up.

[Charles Dean del Prado]

Yes, we really intend to gradually really strengthen our position in mainstream CMOS, and let's say in the industry logic foundry memory, and we see indeed the shift from 7 to 5 as an opportunity to make a step there also. And we trust we will see the results on that in epi foundry also in 2019.

[Peter Olofsen, Analyst]

Okay, and the progress so far, is it still --

[Charles Dean del Prado]

Of course, you cannot compare it with ALD. ALD, of course, has a much broader application space in number of layers. But we do expect to expand our engagements. So the answer there is yes.

[Peter Olofsen, Analyst]

And the progress so far, is it still with this initial foundry customer, or is it also with other foundry logic customers already?

[Charles Dean del Prado]

Our engagement in the epi space has been already for let's say couple of years broader than just this, let's say, this HVM engagement, this high-volume engagement. And we're working very hard to see also -- to also get, let's say, broader traction in , '19, '20, beyond this engagement. But it's just too early to claim victory there, Peter. But it's a very relevant question, and it also -- your question also ties -- fits completely with our ambition, with the Epi strategy that we set out as a company a few years ago.

[Operator]

Thank you. It appears there are no further questions at this time. I would like to turn the conference over to CEO Mr. del Prado for any closing remarks.

[Charles Dean del Prado]

Yes, Anna, thank you. Thank you, everybody, for attending the call today on this day I understood with a few other conference calls. So thank you that you still were all able to make it. And of course feel free to contact us directly or through Victor with any further questions you may have on this overnight release. I'm looking forward to staying in touch with you on future roadshows or any other event that this week has scheduled. Thanks again, and have a very good day. Goodbye.

[Operator]

Ladies and gentlemen, this concludes today's conference call. Thank you for your participation. You may now disconnect.