ASM International NV

AEX:ASM

Earnings Call Transcript 2019-Q3

[Operator]

Ladies and gentlemen, thank you for standing by, and welcome to the ASM International Third Quarter 2019 Earnings Conference Call. [Operator Instructions] I must advise you that this conference is being recorded today on the 31st of October 2019.I would now like to hand the conference over to your speaker today, Victor Bareño. Please go ahead, sir.

[Victor Bareño, Director of Investor Relations]

Thank you, operator. ASMI issued its third quarter 2019 results last evening at 6:00 Central European Time. For those of you who have not yet seen the press release, it is accessible on our website, asm.com, along with our latest investor presentation.As always, we remind you that this conference call may contain information relating to ASM's future business and results in addition to historical information. For more information on the risk factors related to such forward-looking statements, please refer to our company's press releases, reports and financial statements, which are available on our website.And with that, I'll 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 2019 results conference call and for your continuing interest in the company. Before the question-and-answer session starts, we start as normally with a review of the operational and financial results. But first, Peter van Bommel, our CFO, will comment on a couple of specific topics. Peter, please go ahead.

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

Yes. Thank you, Chuck. The first topic is the recent announcement that probably all of you have seen that Chuck is going to retire as CEO and President of ASMI. He has been with ASMI since 2001, on the Board since 2006 and CEO since 2008. During this period, under his leadership, ASM has achieved a strong position as a critical and trusted supplier of enabling technologies to the leading customers in our industry. Chuck will stay on until the AGM in May in 2020, and a search for his successor is well underway.The second topic I would like to discuss is the settlement of the arbitration proceedings against Kokusai Electric that we announced earlier this week. As a reminder, last July, we already settled all litigation cases with Kokusai. As part of that settlement, Kokusai agreed to pay us USD 115 million, which positively impacted our second quarter results with an amount of EUR 103 million. Now we have also settled the arbitration dispute that was still pending, and that concerns the breach of a license agreement that we have with them until November 2017. As part of this settlement, Kokusai agreed to pay another USD 61 million to ASM. This will positively impact our orders and sales in the fourth quarter with an amount of approximately EUR 56 million. The cash will be received in the fourth quarter. And with this settlement, all pending disputes between ASM and Kokusai with respect to patent licenses have been resolved.

[Charles Dean del Prado]

Thank you, Peter. Let's now continue with a review of our third quarter results, and I will focus on the comparison of the underlying results, excluding the settlement impact in Q2.Net sales in the quarter increased by 4% compared to the second quarter and increased by 39% from the third quarter of last year. At EUR 271 million, our sales in the quarter were slightly above the top end of our guidance, which was a range between EUR 250 million and EUR 270 million. Equipment sales increased 4% compared to the second quarter. Sales from spares -- spare parts and service accounted for 20% of total sales and were up 5% sequentially. For the first 9 months of the year, spares and service increased by 19%, and equipment sales were up 44%.In terms of product lines, our ALD business was again the key sales driver, but the other product lines also continued to have a solid contribution. In the first 9 months of the year, all of our product lines showed healthy double-digit sales growth.By industry segment, sales were led by logic, followed by foundry and then memory. Combined logic/foundry sales increased sequentially to a new record high driven by advanced node spending and again represented the largest part of our sales. Memory sales increased both sequentially and year-on-year. Key factors behind memory sales in the quarter were the customer-specific projects in the DRAM segment that already drove memory bookings in the second quarter as already discussed in our previous earnings call.The gross margin was 42.4% in the third quarter, slightly down from 42.8% in Q2 due to mix changes. SG&A decreased by 15% from the second quarter mainly explained by a drop in legal expenses. Reported R&D expenses increased by 26% compared to the second quarter. A number of development projects have now moved from the development stage to the high-volume manufacturing stage. And this means that the costs for these projects are no longer capitalized but expensed directly in the P&L, and that amortization of the related capitalized development expenses starts. The increase in amortization, and particularly the decrease in capitalization, was the main reason for the increase in R&D expenses in the third quarter.Operating income increased to EUR 51 million in the quarter. Financing result in the second quarter included a currency translation gain of EUR 14 million compared to a loss of EUR 5 million in the second quarter. We hold a substantial part of our cash balances in U.S. dollars, and the translation difference -- differences are included in the financial results. Net earnings on a normalized basis, dropped to EUR 57 million, down from EUR 125 million in Q2, which was including the litigation settlement. Excluding litigation, net income showed a substantial increase.Let's now briefly look at ASMPT. Our normalized results from investments, which reflect our share of approximately 25% in ASMPT's net earnings, amounted to EUR 7 million in the third quarter, up from EUR 2 million in Q2 but down from $17 million, 1-7, million in the third quarter of last year. In the third quarter, ASMPT had sales of $531 million, up 15%, 1-5 percent, from the second quarter and down 19% year-on-year. Bookings at USD 514 million in the third quarter decreased by 15% compared to Q2 and by 19% year-on-year.Now turning back to ASMI's consolidated operations. We booked orders of EUR 292 million in the third quarter, an increase of 8% compared to the previous quarter and clearly above the range of between EUR 250 million and EUR 270 million that we had guided for. The upside compared to our expectation was driven by order pull-ins in logic/foundry.In terms of industry segments, bookings in this quarter were led by foundry, followed by logic and then memory. Combined logic/foundry orders increased sequentially to a new record high. Foundry bookings were particularly strong, increasing sharply compared to Q2 and also at a new quarterly high for the company. In foundry, bookings were driven by strong investments in the build-out of a 5-nanometer HVM capacity as well as continued demand for 7-nanometer tools.Logic bookings decreased somewhat compared to Q2 but remained at healthy levels, for the largest part, driven by investments in 10-nanometer capacity. In the memory segment, total orders dropped sequentially as DRAM decreased compared to the relatively high level in Q2, and NAND remains at a relatively low level. By product line, orders in the quarter were driven by our ALD business, in fact, fueled by advanced node demand, which had -- we had record-high ALD bookings in the quarter across a multitude of different applications in the logic/foundry sector.So now turning to the balance sheet and cash flow. At the end of September, the cash position increased to EUR 480 million, up from EUR 382 million at the end of June. This was a result of 4 elements: first of all, approximately EUR 51 million settlement payment from Kokusai; secondly, EUR 15 million in dividend payments from ASMPT; thirdly, EUR 14 million positive currency effects; and last but not least, EUR 17 million, 1-7, in free cash flow, excluding settlement payment.During the quarter, we received the first half of the 150 -- USD 115 million, 1-1-5, litigation settlement payments from Kokusai. The other half will be received by the end of Q4 or early next year. In addition, as Peter earlier explained, the payment of the USD 61 million in arbitration settlement that we announced earlier this week will be received in Q4.Free cash flow in the quarter was driven by strong profitability and partly offset by an increase in working capital, excluding the effect of litigation settlement and higher investments in evaluation tools. The underlying increase in working capital was mainly caused by higher accounts receivables. At 57 days, working capital level continues to be at a healthy level. For the first 9 months of 2019, free cash flow, excluding the settlement, amounted to a strong level of EUR 143 million.Cash spent on CapEx was relatively low at EUR 10 million in the third quarter, which is explained by the phasing of investments. We still expect full year CapEx to remain at a higher level in 2019 and in the first part of 2020 as we continue to invest in our new and expanded manufacturing facility in Singapore. It is expected that this building will be ready in the course of the second quarter next year.With the Q3 results, we announce the payment of an interim dividend of EUR 1 per share. In deciding to pay an interim dividend this year, we took into consideration the recent strong increase in the cash position. The final dividend proposal, which will be in line with ASMI's policy to pay a sustainable dividend, will be announced as part of the fourth quarter results publication early next year. The EUR 100 million share buyback program that we announced in July is expected to start this week. Furthermore, during the third quarter, the earlier-announced cancellation of 5 million of treasury shares became effective.So let's now have a closer look at the trends that we see in our markets. The broader semiconductor end market is expected to drop by a low double digit percentage mainly due to sharp correction in the memory markets. Looking at the WFE market, the wafer fab equipment market, it is now generally expected to decline by a mid-teens percentage in the calendar year 2019. Expectations have somewhat improved compared to 3 months ago when spending was still projected to drop by a high-teens percentage. The improved outlook is mainly driven by higher spending in the logic/foundry segment. Advanced logic/foundry spending is projected to show a solid increase this year. New end-market opportunities, particularly in 5G and artificial intelligence, are key drivers for our customers to invest in the most advanced manufacturing capacity.In foundry, spending this year is strongly driven by investments in the 5-nanometer node in preparation of the HVM ramp in 2020, while 7-nanometer capacity are also still being added. In advanced logic, spending is also strong this year driven by 14-nanometer capacity and particularly -- and in particular, 10-nanometer HVM investments. Taking an early view of next year, we expect the strengths that we have seen in advanced logic/foundry throughout 2019 to continue into the first part of next year.If we take a look at the memory segment, spending is still expected to show a substantial double-digit drop in this year as customers in this segment remain focused on balancing supply. While some indicators such as pricing trends in the NAND market seem to bottom out, overall market conditions remain sluggish, and there are no concrete signs that spending -- equipment spending is going to pick up in the short term. Similar to what we said in the last earning calls this year, we do not expect a meaningful recovery in memory demand before year-end. We expect to -- we, ASM, expect to strongly outperform the WFE market this year increasing -- and as a result, increase our WFE and deposition share in the market. The key driver behind our outperformance is our strong growth this year in advanced logic/foundry. The solid increases in this segment spending are supporting our performance given our strong and leading position in these segments.In the first 9 months of this year, nonmemory, of which advanced logic and foundry are the largest part, continued to account for the majority of our business. In addition, as highlighted in previous calls, in the current most advanced nodes compared to the previous nodes, the number of ALD layers have substantially increased, both at the 10-nanometer node in advanced logic and at 5-nanometer in leading foundry, therefore, increasing our single-wafer ALD served available market. We believe we have significantly increased our share of wallet at these nodes.Looking ahead at our customer road map. The road map of our customers then ALD will only become more important enabler of Moore's Law as we move to even smaller geometries and increase device complexity. And increasing device complexity will drive the need for more precision deposition. We are strongly engaged with our key customers in R&D for the next node, for instance, in logic, 7-nanometer node; and in foundry, 3-nanometer. And we expect further increases in our market opportunity at these nodes.If we look at our memory business, on the back of sequential increase in sales in the second half, we expect our memory sales for the full year to hold up reasonably well compared to 2018. And this is primarily driven by customer-specific projects. And it is not an indication of a broader recovery in memory spending for which, as just mentioned, visibility remains limited at this point in time. Our strategic priority in the memory market remains a further increase in our precision step by step. Over the last few years, we had invested and we continue to invest to broaden our served available market in both DRAM and 3D NAND device technologies. In 3D NAND, for instance, the number of single-wafer ALD steps is expected to rise due to increasing complexity at the 96-, 128-layer device generations and beyond.In terms of product lines, our ALD business is on track for a strong year boosted by the strong investments in the advanced nodes and logic/foundry sector. In the third quarter, this was again underscored with record-high ALD bookings. Momentum in the other product lines, non-ALD product lines, also remains healthy. For our epitaxy business, we expect 2019 to be the third consecutive year of solid sales growth and market share gains. Epitaxy has now become our second largest product line after ALD. In the power analog segment, our epi sales are healthy for the year as a whole, although demand is slowing in the second half, as expected and as we shared with you before, and following solid sales in the first half.In the advanced CMOS segment, demand for our Intrepid tools has again strengthened this year. Our target remains to further increase our epi share by broadening our engagements in the advanced CMOS market. In the other product lines, in PECVD and in vertical furnaces, our strategy is unchanged: to invest in target areas and to expand our niche positions in these markets over time.In summary, momentum in the short term continues to be solid. And in the longer term, our company remains strongly positioned to benefit from structural growth in ALD and epi markets.So let's now look at our Q4 guidance as we communicated in our press release. So for Q4, on a currency comparable level, we expect sales between $310 million and EUR 330 million. While bookings, on a currency comparable level, are expected to be in the range between $290 million and EUR 310 million. This guidance for both sales as well as orders does not include the proceeds of USD 61 million related to the settlement of the arbitration proceedings with Kokusai, which will be added to those numbers in the quarter.So that concludes the introduction. At this point, of course, Peter and I are more than happy to answer any questions that you may have.

[Victor Bareño, Director of Investor Relations]

[Operator Instructions] Operator, we are ready for the first question.

[Operator]

The first question comes from the line of Stephane Houri.

[Stephane Houri, Research Analyst]

So in fact, I have 2 questions. The first question is about the visibility that you have because now making -- it makes a few quarters now that you beat the expectation by a wide margin. So the question is are you just too cautious when you give your guidance or maybe your customers are making decisions on a very short visibility mode and investing more than they initially expected. That's the first question, and that's mainly for the logic/foundry business.And the second question is about the memory market because now it's making -- more than 1 year that the largest memory manufacturer has been cutting their CapEx. So I was wondering if you believe that we are now close to a moment where the market will be more balanced between order and demand. And what's your view on 2020 memory market as we speak?

[Charles Dean del Prado]

Okay. Okay, that's clear. Okay. Thank you for your question. So first, are we too cautious on guidance? Well, we try always to be realistic on our guidance and not to overpromise. That is always our focus and to -- as a result of that, maintain our strong credibility that we as a company have built up throughout the years in providing our guidance and then meeting our guidance. So that's what we, first, would like to emphasize. What really happened in recent times is that really pull-ins, significant pull-ins happened in both the logic and foundry segment that -- as a result of which, we were able to adjust our -- even further improve our performance and guidance compared to no more than 3 months ago. So that is the answer on the first question.Then your question on -- yes, on memory, we touched on it briefly during the introduction. And again, there are some signs of early stabilization in the end markets. As we touched on the introduction, for instance, looking at pricing trends in 3D NAND. However, there still are no concrete signs that in terms of equipment spending, the pattern will change in the short term. Customers have not given us that visibility at all at this moment in time. At the same time, we know that in memory, that can change very quickly. Not meaning that we expect that, that will be the case because we have -- we don't have that visibility at this moment in time. And we have, throughout 2019, been very consistent on our guidance with respect to memory. We have always, yes, had a conservative view in that respect. And that basically played out throughout the year.Yes, in general, if any recovery would happen, it's likely also, as you talk to other industry experts in the industry, that people expect that 3D NAND likely will -- may recover first -- is believed to recover first and DRAM later. What we really are focused on is that we are ready. And we are very focused on increasing step by step our served available market in memory. And that's what we have been working on for the last 4 or 5 years that at every node, whether it's DRAM or a NAND, that we increase our served available market. And when you increase your served available market, it means that you are engaged on a broader part of the market in terms of R&D with your customers. That, at some point in time, will pay off. And we have been working on expanding from non -- from patterning into nonpatterning. And step by step, we will see those results will become visible as the recovery comes.And apart from that, the good thing is that in this year, regardless of the weakness in the market that, from a revenue point of view, we were able to keep the memory contribution, in absolute terms, pretty stable this year as a result of a different mix year-on-year, going from '18 to '19. I trust that answered your questions.

[Stephane Houri, Research Analyst]

Right. So it means that you expect memory to be, well, at the moment, stable, not recovering, but you expect good trends in foundry/logic. Is it fair to say that at least for the first half of the year, you expect growth year-on-year?

[Charles Dean del Prado]

Well, that's -- again, that is maybe a little bit to quickly conclude, but we -- again, we continue -- again, we see a solid climate in logic/foundry continuing into the first half of next year. And we expect that there also will be contribution from memory. How sizable the contribution of memory is, that's a little bit too early to tell.

[Operator]

The question comes from the line of Sandeep Deshpande.

[Sandeep Sudhir Deshpande, Research Analyst]

Sandeep Deshpande from JPMorgan. Two questions, if I may. Firstly, regarding ALD, I mean you mentioned in your prepared remarks that, that is the main driver of the very strong orders that you are reporting. I mean how do you see your share overall in ALD trending through this year, in 2019? Given that, in past years, you have lost shares, so do you think that now in logic, ALD, that you are the market leader and which is why the shift has occurred and any previous share losses in ALD were because of some things in memory that you were not exposed to?And then secondly, regarding the -- back again to the question on memory market. I mean you've said that you are fairly stable in memory this year. Could you possibly give us some highlights of how you were able to be stable in memory this year when your peers have talked about a big decline? Is this because of any new exposures in the China memory market or it is because of some particular projects? I'm trying to understand that because that will potentially have implications on future years whenever the memory market recovers.

[Charles Dean del Prado]

Yes. Okay. Okay. Sandeep, thank you for your question. Okay. First of all, on ALD market share development in 2019. Of course, market share in ALD, it's always also a result of the dynamics between -- of spending between the different industry segments. And of course, our position -- our relative position in logic/foundry is the strongest. It's stronger than in memory. So in 2017, '18 when -- especially first half '18 and then all year '17, memory was, by far, the biggest spender. So then the weighting in the market share calculations of memory is bigger than for logic/foundry. And I think starting in the second half of '18 and, for sure, into '19, the weighting of logic/foundry has increased dramatically. And so we expect that, that development is very favorable, has been and will be very favorable, for our market share development in deposition as a whole and within ALD.So that's an answer to your -- and what we, of course, are very focused on is that going forward, as memory recovers by really having been focused on increasing our served available market in memory with regard to all R&D engagements for the last 4 or 5 years, is that also our relative position in memory, in time, will improve so that we can much better defend our market share regardless of the industry segment dynamics in the long run. So that's an answer on your ALD market share question.Then on -- yes, memory. Yes, how come that it in absolute terms has been stable? That's indeed related to, yes, specific customer projects that in '19 were more substantial than in '18 and that some of the big spenders -- yes, and so that we could compensate for less spending of some of the big spenders that basically slowed down in '19.On top of that, yes, you were asking about China. Well, I think as we have addressed in earlier calls, China really has been developing for us. And in the past, we -- in general, overall, regardless of the specific segments, we tripled our China top line contribution from '17 to '18. And we also are increasing the China contribution in absolute terms year-on-year in '19, while the WFE spending of China is likely going down year-on-year. In -- specific to memory China, this year, for the first time, there is, yes, some meaningful contribution of memory in the past. Yes, we've had memory contribution, but in more recent years, apart from this year, it was more logic/foundry-centric. But recently, memory started to contribute also there. It's too early to say how that will develop into 2020. That's a little bit too early.

[Operator]

The next question comes from the line of Peter Olofsen from Kepler.

[Peter Olofsen, Analyst]

Two questions for me, please. First, on the strength that you're seeing in logic/foundry in the second half and presumably also in the first quarter next year, considering the likely backlog at the year-end. How sustainable do you think this demand is for next year? Do you think it's sustainable throughout 2020? Or could there potentially be a pause in spending by some of your customers as they digest some of the equipment purchases before they move to the next node, also considering that in one of your earlier questions, you mentioned the word pull-ins?And then my second question is on the overall revenue mix. Unlike some of your peers, you don't disclose this in your quarterly releases. But given your comments on memory and logic/foundry, could you maybe provide an indication where you expect to end up for the full year in terms of split between logic/foundry versus memory?

[Charles Dean del Prado]

Okay. Okay. Okay, so I'm just -- wrote down your questions. So is -- yes, how will -- you're basically asking for outlook for the logic/foundry segment into 2020. Yes, it's, of course, too early to provide guidance for the full year. And during the introduction, we said that given the visibility we have now that we see so far a very solid start. We believe in a solid start for logic/foundry going into 2020. And that's based on the fact that, yes, we believe that a further 5-nanometer investments in 2020 will happen.And in addition, you see that in general that the drive to go to -- to stay on an aggressive cadence in this industry, in the logic/foundry segment, is really there. So it would also not surprise us if some investments for 3-nanometer would also take place next year. So in that respect -- and that's beside, yes, ongoing capacity expansion also still happening today in 7-nanometer. So you have 7-nanometer; you have 5-nanometer, a further expansion; and possible initial investments into 3-nanometer so -- and all driven by, yes, very strong end-market developments that we touched on earlier: 5G, artificial intelligence, cryptocurrency, we can go on. So -- but again, to provide a full visibility on demand for full year 2020 is way too early.On the logic segment, yes, we believe there is room for further investments in 10-nanometer going into next year. And also there, same as in foundry, we see initial movements. And it shows a strong focus already on the next node, being 7-nanometer there. Our impression is that logic is very determined not to -- yes, to bring back the cadence compared to 14-, 10-nanometer. There was, of course, a lot of years in between there to bring it back to more regular time frame going forward.So yes -- and on top of that, we are -- as we were very successful in gaining share of wallet, node to node, from 7 to 5 in foundry and from 14 to 10 in logic. We are very eager to further increase share of what going to -- in logic from 10 to 7 and in foundry from 5 to 3. And it's more than being eager. We think it's also feasible. It's also feasible based on the engagements that we have ongoing today because, of course, if you're not heavily engaged today then it won't happen, then you are far too late.And yes, I think, in general, so in logic/foundry, there's so much out there, also if you look at more in the distant future, if you look at transitions upcoming from FinFETs to gain all around in the medium to longer term. That will only fuel more demand for technologies like ALD and epi. And that's why we are so strongly focused on everything we do today. Okay?So then -- yes, so then your next -- yes, so then the mix between logic/foundry and memory for 2019. I think we already shared in the introduction that logic/foundry -- let's say, spending there is more than half of the total top line. Yes, it's more than half of the total top line, and it's also more than half of the equipment spending for the full year of 2019.

[Peter Olofsen, Analyst]

Okay. But you don't want to be more precise than that, whether it's 60%, 75%. It's more than half, but we have to do it with that level of disclosure.

[Charles Dean del Prado]

Yes. That's correct. Yes, that's correct.

[Peter Olofsen, Analyst]

Okay. And then maybe on the earlier -- where you see potentially next year some early investment in 3-nanometer foundry and 7-nanometer logic. To what extent have the selections -- equipment selections been made already or what visibility do you have there?

[Charles Dean del Prado]

Could you repeat the question one more time? Sorry.

[Peter Olofsen, Analyst]

Yes. So you mentioned that you are engaged in R&D in 7-nanometer logic and 3-nanometer foundry and that you may already see potentially some initial investments next year. So basically, my question is to what extent have the equipment selections by our customers been made already? Or what visibility do you have there?

[Charles Dean del Prado]

Yes. So when you talk about these investments, so very early pilot investments, and so they provide an indication of your position. But during the general practices, during those early pilot phases, that nothing is fully frozen yet. But it does provide an, yes, initial indication because it does require significant investment on both sides, and nobody wants -- well, we are serving the customer. So customers don't want to waste their time and money and focus on something that has no future. So it is a good, early indication, but it is no 100% guarantee in the early pilot phase.

[Operator]

The next questions come from the line of Quang Le from Crédit Suisse.

[Quang Tung Le, Research Analyst]

The first one, I would ask about 7-nanometer logic. Do you have any visibility on potential reuse of tools similar to what we saw in 7-nanometer foundry, which, well, can potentially reduce customer orders? And I'll have a follow-up.

[Charles Dean del Prado]

Yes. The short answer is we have no visibility on that today.

[Quang Tung Le, Research Analyst]

Got it. And if I were to ask about your OpEx line, specifically, for example, R&D, it was up by -- up quarter-on-quarter driven by the completion of the development you talked about and on SG&A, decreased obviously because you have a lower legal costs. And are these both lines the, let's say, run rate that we can assume going forward or, I don't know, on R&D, it will revert back to what we saw in Q1 and Q2?

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

Let me answer that question. With regard to R&D, there are a few things. We follow IFRS and when you look to our competitors, most of them are following U.S. GAAP. Under U.S. GAAP, you don't capitalize R&D. Under IFRS, you do that. As a consequence of that, on the moment that we have new projects, then at first instance, projects which are related to product development, so basically serve or support once a product moves into high-volume manufacturing, they're not cost that we are capitalizing, but we are capitalizing the product development projects.On the moment that the product development projects are finished, then as from that moment, there could be certain costs still dropping into the organization. And those are immediately taken as cost, and the amortization at that moment of that project starts. In the third quarter is there -- a few of those projects, which are already there for a long period of time, now reached a level that amortization starts. As a consequence of that, you have seen, in the third quarter, that a little bit less projects were capitalized that amortization started to increase a little bit. And going forward, we expect that new projects are coming up, so capitalization will continue, but also amortization of the older projects will remain. We do the amortization in a 5 years time frame. So it's a lengthy answer, but it provides you with some of the mechanics that we are dealing with.So on the one hand, costs that are in for development gradually increasing. And on the other hand, costs, which were partly amortized and that -- and partly also capitalized, that will continue. So we do not expect that the costs will go down to the levels of Q1 and Q2, but we expect that R&D costs, over time, gradually will increase. That's one of the reasons why we published also every quarter, with the different numbers, the amounts that we really spend and the amounts that we take up in our profit and loss account. And the guidance that we are providing of high to -- of low- to mid-teens was always related to the spendings that we are doing. I hope that answers your question about R&D.

[Quang Tung Le, Research Analyst]

Yes. And on SG&A?

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

SG&A, there are some fluctuations indeed. We have seen in the third quarter that those costs are coming down especially due to the fact that we have less legal costs. And it's safe to assume that those costs might increase a little bit over the quarters, but they will not show very substantial increases.

[Operator]

The next question comes from the line of Marc Hesselink from ING.

[Marc Hesselink, Research Analyst]

My first question would be on -- you talked about node transitions. Can you indicate, maybe high level, what is the ALD -- the increase in ALD intensity moving from those technology nodes? So maybe the one that -- and probably that you've seen is going to 7-nanometer from -- to 5 and -- but also what you're already seeing, the early indication going to 3-nanometer and then same for the logic segment.And second question is on your -- the evaluation tools. You had quite a few in the market for quite a while. Is -- are those evaluation tools the reason that you are confident that you'll win back some market share in the memory segment and that you also see some further expansion in, for example, epi?

[Charles Dean del Prado]

Okay. Yes, clear, Marc. So first one is -- question is a tough one. I think it's a very good question. It is a very good question. Yes, we look at it always in terms of the amount of layers, amount of layers that we participate in going from node to node. And as we shared in calls before, going in logic from, let's say, what was it, to 14-nanometer, the transition to 14-nanometer, was way less instrumental for us than going from 14 to 10, and the same applies to 10 to 7 in foundry. It was much less instrumental for us than from 7 to -- 7 to 5. So we really, in a meaningfully amount -- meaningful amount, we increased the number of layers in both areas, really meaningful increases in layers.So that's -- so that means number of applications. But sometimes an application is more than one layer, so it's applications times the amount of layers per application that determines how many layers in total that you gain. And then that's, of course, very interesting for us to look at. And the only thing I can say is that from -- also from -- in foundry going from 5 to 3 and in logic going from 10 to 7, there is meaningful potential in terms of the amount of layers to improve our position. And yes -- but it also depends. And we say that based on indeed, R&D, the engagements that are ongoing.But customers decide at the last moment whether they really will insert certain processes into high-volume because sometimes they have more options. They're looking at integration results and then look at what -- from an integration point of view, is the best path to follow. But linking that to, yes, so to a part of your second question so -- but we -- looking at the intensity of the R&D engagements and, indeed, looking at the intensity of evaluation tools that you have there, that's indeed a good indication of whether the company is moving in the right direction, yes or no. And that provides us the comfort that we are doing our homework well, and it's just depending on the timing of the customer when they insert these new nodes that we then are ready to participate.And then if you link that to memory, your question on memory is -- yes, well, you said win back. I would like to emphasize there, it's not a matter of win back. I don't know if that is the right wording. We really are focused on increasing our served available market from -- because as we shared before, in DRAM, we were -- also from an R&D point of view, until, let's say, 4 years ago, we were primarily engaged in R&D, in patterning space. And we decided we stay engaged in patterning because there are also many developments in patterning in this industry. But we are going to expand aggressively into R&D and nonpatterning. And nonpatterning, for example, in DRAM has been growing faster than the patterning space in DRAM specifically. Also in NAND space, we initially -- now I'm talking about ALD, we were primarily -- we were only engaged probably in terms of served available market within half of that total addressable market in -- for NAND single-wafer ALD. We said we have to aggressively increase that. And that's what we did.And it takes time for -- before those investments are paying off. And one good example was what we shared with you multiple quarters already is that in 96 stack, we -- there was a showcase of a development program with, well, a very strong leader in this memory space in Asia where we penetrated with 96 stack. And that supported our P&L over the last 18 months in a meaningful way. Even though that volumes in memory were low, still we sold systems. And as volume comes back and we gradually further penetrate in new nodes, we expect that, that will pay off. And then it will become easier to defend our market share, a sustainable level in market share going forward regardless of the mix between -- in spending between the segments. Of course, memory still will be less favorable to us for logic/foundry for some time, but the impact may become less over time as we increase our served available market and get results there.

[Marc Hesselink, Research Analyst]

That's clear. And maybe to add one more thing, is that -- what you mean winning that market share, is that a combination of -- is it more deeper penetration within the current client base? Or is it also expanding the client base to get more the smaller clients?

[Charles Dean del Prado]

Yes, our market share, we were talking about share of wallet. But yes, well -- yes, absolutely, we are broadening. We are broadening our engagement with multiple customers in memory.

[Operator]

The next question comes from the line of Wim Gille from ABN AMRO.

[Wim Gille, Head of Research & Equity Research Analyst]

My first question would be on your market share in ALD again. Over 2018, according to the Gartner definition, you had about 40% market share. Based on kind of the results that you've booked year-to-date and the resources you've seen from your key competitors, how many percentage points based on the Gartner definition do you think you've added to your market share in ALD?The second question I had is on your served addressable market. You already alluded just now that you serve about 50% of the applications in NAND today. What's your kind of penetration in DRAM? And what percentage of the gap that you are not covering in your current served addressable market do you expect to close in the years to come based on the R&D that you are currently working on?

[Charles Dean del Prado]

Okay. I got the instruction, since there are a few more people in the queue, that I have to be a little shorter in answering any questions, so I'll try to do that. On our market share in ALD, yes, the year has not closed, we cannot compare it with the overall market numbers, so it's much too early to provide any trends there. In general, we don't provide numbers. We only comment to the trends that are shared in the market by the Gartners of this world, and so we really have to wait for that. But we would be surprised if it not -- would not be favorable to us, of course, this year, given the fact that the logic/foundry weighting -- the weighting of the logic/foundry spending compared to memory has so much increased combined to our share of wallet gains. But to what extent, we don't want to comment to that additional this time.So then on 3D NAND and DRAM served available market, yes, in DRAM, I think that the part that we address is likely -- we're talking about ALD, yes, single-wafer ALD, was -- the nonpatterning part was, in the early years, in '14, '15, was the major -- a very significant part. Now I think the nonpatterning part is lower than 50%, yes. So we are making sure with all the investments that we are doing that we are increasing that. So we're talking about working towards increasing our served available market with hundreds of millions. Served available market, that's not revenue, but that's addressable market in memory in the coming years.

[Wim Gille, Head of Research & Equity Research Analyst]

Very helpful. And then one follow-up question. Regarding the comments that you made on the, let's say, healthy start that you expect in logic/foundry in the first half of 2020, is it -- do you actually require an increase in the memory market in order to grow your revenues yourselves? Or is that not a prerequisite for you to grow in the first half of 2020?

[Charles Dean del Prado]

Yes. I don't fully understand the question.

[Wim Gille, Head of Research & Equity Research Analyst]

You have quite a decent visibility that logic/foundry will be off to a good start. So do you actually need a, let's say, recovery in the memory market in order to grow your own revenues in the first half of 2020? Or would you say, even with another week 2020 in memory, we would still be able to grow our revenues in the first half of 2020 based on the visibility you have in logic/foundry?

[Charles Dean del Prado]

Yes, I understand. Yes, I understand your question. I understand it. I understand it would be valuable to get that full visibility from your point of view. But the only thing we can say at this moment in time, living at the end of October, early November, is that we do see solid start for advanced logic/foundry in 2020, and we don't see a recovery of memory yet. We don't want to go much further at this moment in time. I'm sorry about that. I trust we can provide you better guidance on that in due course. Yes?

[Operator]

The next question comes from the line of Robert Sanders from Deutsche Bank.

[Robert Duncan Cobban Sanders, Director]

I'll keep the questions short, we're running out of time. First question would just be -- you've still got this change of control clause with AMAT. From my memory, this was supposed to expire after about 20 years. So that would suggest, like this year or, give or take, given some patent delays at the time. So when does this actual change of control expire? That would be my first question.My second question would just be on Singapore. I was just wondering what is the fully loaded run rate of revenue you can do when Singapore gets to full capacity in terms of annual revenue. Is it $1.3 billion or something like that? And I just -- related to that, was there any kind of order pull-in given that you're at full capacity? I mean have you seen lead times lengthen a little bit because you are at full capacity and maybe that led to some pull-in? I just wanted to know if there was any relevance.

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

Yes, let me answer that, Robert. First of all, right of first refusal, yes, as what we said in previous calls, the description of this lifetime -- of the lifetime of this right was not exact. Therefore, it's a matter of interpretation. However, it seems reasonable to assume that it's not something which is going on for a very, very long period of time any longer.The second thing is Singapore, the Singapore facility. Yes, it's a good question. It's really an enormous increase in capacity that we have, so we are building the 2 completely new floors. One of the floors we are going to make use of and that floor will enable us to increase our capacity with at least 4x the volumes that we can do in the old factory. And then we still have a comparable space available, which we can use in the foreseeable future on the moment that we have even higher demand.

[Robert Duncan Cobban Sanders, Director]

Sorry, just -- what does that mean 4x the old factory? I assume you don't mean 4x the revenue capability of today. But I mean on a quarterly -- on a revenue basis, what does that actually mean for your total revenue potential?

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

It's, of course, always a matter of mix, but you can assume that it will be at least 3 to 4x the revenue that we can do today in equipment.

[Operator]

The last question comes from the line of Nigel Van Putten from Kempen.

[Nigel Van Putten, Analyst]

Most have been answered, just maybe a cheesy one then. I think you've commented that memory has been strong throughout the year, but your order book is almost -- or at least, on a significant level, filled with logic/foundry. And then reconciling that to your outlook into 2020, do you think you can sustain revenue at a similar level as the current quarter on logic/foundry alone in terms of the strength you're seeing?

[Charles Dean del Prado]

Yes, that's a little bit similar to the question we got a little earlier. Yes, that's a little too -- yes, we don't want to give more guidance than we have given so far. But it's really the case that in most recent quarters, of course, there is still a healthy contribution from memory. What we do -- what we can say is that the memory contribution in the second half is higher than in the first half. But at the same time, the logic/foundry contribution, of course, is really the leading component here, as we earlier discussed, meaningfully more than half of the total.

[Nigel Van Putten, Analyst]

Okay. And then maybe just a small clarification, a follow-up...

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

I guess to add absolute color to it, I mean we will -- when you look to the mid-levels of guidance that we have given, then we will start 2020, of course, at a very, very -- with a very, very healthy order book, so that gives you also a little bit of color in that respect.

[Nigel Van Putten, Analyst]

Yes, exactly. And then the order book, does that carry multiple quarters? Or is that sort of mostly still sort of for the upcoming quarter and we can see -- or we can -- we should expect a similar order intake based on the guidance into the first half of 2020?

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

Order book in general takes into account, of course, how customers are ordering. And while they sometimes order for more than the lead time ahead, most of the order book, let's call it that way, is related to the next quarter.

[Operator]

Dear speakers, there are no questions at this moment. Please, you're more welcome to do your closing remarks.

[Charles Dean del Prado]

Since we ran out of time, we still have many questions and a lot of people. Got a few only who only got the opportunity towards the end of the call. We would like to make sure that any remaining questions that you may have, those will be addressed. That's important to us. So through Victor, of course, we are available to answer more questions that you have. So please contact us if that is the case. And for the rest, thank you for your attendance today and your support of the company, all right? And let's stay in touch, and we will be around. Thank you. Thank you very much. Have a nice day, a nice evening. Bye-bye.

[Operator]

That does conclude our conference for today. Thank you for participating. You may all disconnect. Have a nice day.