Ladies and gentlemen, thank you all for standing by, and welcome to today's ASM International Q4 2019 earnings call. At this time, all participants will be on a listen-only mode. There will be a presentation followed by a question and answer session. At which time, should you wish to ask a question, you will need to press star and the number one on your telephone keypad and wait for your name to be announced. I must advise you all that this conference is being recorded today, Wednesday the 26th of February, 2020. Without any further delay, I would like to hand the conference over to our first speaker for the day, Mr. Victor Bareño. Please go ahead.
Thank you. ASMI issued its fourth quarter and full year 2019 results last evening at 6:00 P.M. 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. 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. With that, I'll turn the call over to Chuck del Prado, President and CEO of ASMI. Chuck?
Thank you, Victor. Thanks to everyone for attending our fourth quarter and full year 2019 results conference call, and for your continuing interest in the company. Peter van Bommel, our CFO, will discuss the financial impact of the settlement with Kokusai and the movement in our cash position in the fourth quarter. After that, I will continue with a review of our operational results, followed by the normal question and answer session. On the highlights of 2019. 2019 was a highly successful year for ASM. We increased our sales by 57%, including proceeds from the settlement with Kokusai Electric Corporation. Excluding these proceeds, sales increased by 37%, strongly outperforming the WFE market. Our sales growth was driven by solid spending in Logic Foundry on the advanced nodes, where we achieved strong share of wallet gains.
We continued to track well against our strategic targets to further expand our served available markets. We again made significant investments in the future growth of our company, and at the same time, we are returning excess cash to our shareholders. Supported by a record-high backlog, ASM is well-positioned for a healthy performance in 2020. I want to thank all our employees. Without their continued dedication and hard work, ASM's outstanding results in 2019 would not have been possible. Before handing over to Peter, I will also provide you with an update on the implications of the coronavirus situation. The health and safety of the employees working for ASM and our partners, including our customers, is our top priority, and we have taken several measures to minimize the health risks. Far, we have not seen a meaningful impact on general industry demand trends, nor on our business.
Based on the situation today, we expect the potential impact on our sales in Q1 and Q2 to be limited, and we have taken that into account in our guidance. Peter, please go ahead on your part.
Thank you, Chuck. In the fourth quarter, we booked a gain of EUR 56 million related to the settlement of the Kokusai arbitration case as announced on the 29th of October 2019. This gain was included in the order intake and in sales, and also dropped through to gross and operating profit. The EUR 56 million gain in Q4 is an addition to the EUR 103 million gain in Q2 of last year that was related to the settlement of the litigation cases against Kokusai. In terms of the cash impact, we received EUR 108 million in cash from Kokusai during the fourth quarter, consisting of the EUR 56 million from the arbitration settlement, plus the final payment of EUR 52 million from the earlier litigation settlement. For the full year, at a pre-tax level, cash was positively impacted by EUR 159 million from those settlements.
As a reminder, all pending disputes between ASM and Kokusai with respect to patent licenses were resolved last year, and there will be no further impact in 2020. The settlements also affected the tax rates. As at earlier occasions, we guided that the structural tax rate would increase to a mid to high teens percentage once we would have fully utilized the net operating losses. With the gain from the settlement in 2019, combined with the normal operating profitability, we exhausted the remaining net operating losses in the Netherlands. As a consequence, this led to an increase in our tax rate. For the full year 2019, the total effective tax rate amounted to 14%. Let me also explain the moving parts in our cash position in the fourth quarter.
We ended the fourth quarter with a total cash of EUR 498 million. That was up from EUR 482 million at the end of the third quarter. As I just mentioned, we received EUR 108 million in cash from Kokusai during the quarter. We used EUR 52 million for the interim dividends that we paid in November, and almost EUR 100 million for the share buybacks during Q4. We generated a free cash flow from EUR 63 million in the quarter. Despite the strong increase in sales and activity level, our working capital, excluding the settlement effects, decreased by 5% at the end of the fourth quarter compared to the third quarter. For the full year, we generated a strong free cash flow of EUR 206 million, excluding the settlement. This was primarily driven by a solid increase in profitability and a reduction in working capital.
Excluding the settlement effects, working capital dropped to 42 days at the end of 2019, compared to 72 days at the end of 2018. The quality of working capital also improved in 2019, illustrated, for instance, by a relative reduction in the raw materials balance and a continued low level of accounts receivable over dues. The improvement in working capital was supported by structural measures that we have implemented over the last 18 months.
Thank you, Peter. Let's now continue with a review of our fourth quarter and full year operating results, starting with the fourth quarter. I will focus on the results, excluding the impact from the Kokusai settlement. In the fourth quarter, we booked EUR 373 million in new orders, up 28% from the third quarter, and a new record high for ASM. As we already announced on January 14, order intake substantially exceeded the guidance of between EUR 290 million and EUR 310 million that we initially provided with our Q3 results. The upside was largely driven by stronger than expected Logic Foundry demand. ALD continued to be the key driver behind the order intake, but we also recorded strong sequential increases in our other product lines.
By industry segment, equipment orders in the quarter were led by logic and foundry, which were about similar in size, followed at a distance by memory. Logic orders increased strongly compared to Q3, driven by continued 10 nm capacity investments, and also included seven nm development tools. Foundry orders decreased somewhat compared to the record high level in Q3, remained at a very solid level. Key driver behind foundry orders was a continued ramp of five nm. Orders in memory increased slightly from the third quarter, mostly in NAND, but were relatively modest compared to the logic foundry contribution. Net sales in fourth quarter increased strongly to EUR 345 million, up from 27% from the third quarter, and again at a new record high for the company. Sales in the quarter exceeded the guidance of between EUR 310 million and EUR 330 million.
Our global operations team executed in an outstanding way in stepping up the level of shipments to meet the higher customer demand, despite the capacity constraints that we discussed at earlier occasions. Equipment sales increased 34% compared to the third quarter, and spares and service were up 2%. In terms of product lines, sales were led by our ALD business, which achieved very strong record high sales in Q4. Sales of the other product lines were also up in the quarter. By industry segment, the revenue stream in the fourth quarter was led by foundry, followed by logic, and then memory. Foundry sales increased strongly to a new record high. Logic sales decreased slightly compared to Q3, but were still at a very healthy level. Memory sales, both DRAM and NAND, decreased somewhat compared to Q3 and had a relatively modest contribution in the quarter.
The gross margin increased on the back of a better mix to 43.6% in the fourth quarter, which is the highest level in the last 10 quarters. In terms of operating expenses, reported R&D increased 17% sequentially. This is explained by an 8% increase in cash R&D spending and by one-off impairment charges of four million EUR in the fourth quarter. SG&A increased 18% sequentially, which is mainly caused by higher expenses for variable employee compensation on the back of the strong results in 2019. Operating income increased strongly to EUR 75 million in Q4, a new quarterly high. The sequential increase in net profits was the balance of higher operating profitability, plus the settlement gain, and partly offset by currency effects that were EUR 14 million negative in Q4 compared to EUR 14 million positive in the third quarter. Let's now have a closer look at ASMPT.
Normalized results from investments, which reflect our share of the net earnings from ASMPT, amounted to EUR 6 million, approximately flat compared to the third quarter. In the fourth quarter, ASMPT reported sales of $568 million, up 7% compared to Q3 and down 7% year-on-year. Profitability was equal to the third quarter at $29 million. For the full year 2019, ASMPT sales decreased by 19% to $2 billion, and net profits decreased by 72% to $80 million. Turning back to ASMI's consolidated operations, let's now discuss the full-year results. At slightly over EUR 1.1 billion, our net sales in 2019 increased 37% to a new high. Equipment sales grew 44%. Spares and service increased by a solid 15% and represented 19% of total sales. In terms of industry segments, sales for the full year were led by Foundry.
Our sales in the Foundry segment roughly doubled, driven by strong investments throughout the year in the most advanced Foundry nodes, especially the 5 nm node, combined with our share of wallet gains at this node. The second-largest segment was Logic. Following an already strong increase in 2018, sales in this segment grew again by a substantial double-digit percentage in 2019. In Memory, the third-largest segment, our sales were relatively flat in 2019, mainly supported by the benefit of customer-specific projects in the DRAM segment. The Analog segment, which is smaller than the other segments, showed a decent double-digit sales increase for the full year. This was the balance of a strong first half and a weaker second half, as we already discussed and guided for in our previous earnings calls.
By product lines, sales in the full year were led by strong double-digit growth in our ALD product line, which continued to represent more than half of our equipment revenue in 2019. Sales of our combined other product lines also increased by a robust double-digit percentage to a new record high in 2019, led by growth in our EPI business. Gross margin increased from 40.9%- 42.6% in 2019 for the full year. While we still incurred extra costs related to new growth initiatives and product introductions, this was more than offset by better margins on the newly introduced products, which improved as compared to 2018. As always, gross margin can vary from quarter to quarter, depending on the mix. Operating expenses remained under control during the year.
SG&A expenses increased by 23%, mainly due to, first of all, higher legal costs in the first half due to the Kokusai cases, and secondly, due to higher variable compensation, as we touched on earlier. SG&A increased due to investments to strengthen the organization, such as in new IT systems. Total R&D expenses increased by 21%, excluding IFRS effects, and were at a level of 13% of sales compared to 15% in 2018. Operating profit for the year increased strongly by 77%, with the operating margin at 19.5%, up from 15.2% in 2018. Turning to cash flow as part of some balance sheet items. Peter already talked about the drivers behind the increase in the cash position, I will only highlight a couple of points.
Free cash flow strongly increased to EUR 206 million in 2019, excluding the settlement gains and ASMPT dividends on the back of higher profitability and improved working capital. Our capital expenditure, while lower than in 2018, was still relatively high, reflecting ongoing investment in our new manufacturing facility in Singapore. This new state-of-the-art facility progresses well, is expected to be completed end of Q2, and will substantially increase our manufacturing capacity. Under our most recent EUR 100 million share buyback program that we completed last week, we bought back approximately 1 million shares at an average price of EUR 102 per share. During 2019, we spent in total, EUR 200 million on shareholder remuneration, of which almost EUR 100 million on dividends and another EUR 100 million on share buybacks. Over the last three years, we have returned more than EUR 1 billion in cash to our shareholders.
We will propose to the AGM on May 18th, a total dividend of EUR 3 per share over 2019. This consists of a regular dividend of EUR 1.50 per share, a 50% increase compared to the dividend that we paid over 2018. In addition, the total dividend of EUR 3 also includes an extraordinary dividend of EUR 1.50 per share. Following the interim dividend of EUR 1 per share that we paid last November, a final dividend of, in total, EUR 2 per share will be paid in May, following the AGM. The proposed extraordinary dividend and the new share buyback program reflect our continued commitment to use excess cash for the benefit of the shareholders. Our policy regarding regular dividends is unchanged. We aim to pay a sustainable dividend. Furthermore, we also plan to cancel 1.5 million of shares. Let's now look at the strategic achievements of the company over the last calendar year.
Next to our strong financial results, we progress well towards our strategic priorities in 2019. Key pillar of our growth strategy is a further expansion of our served available market in the single wafer ALD space. In logic foundry, we have strongly expanded our position in the most advanced nodes. The number of ALD layers has substantially increased in the logic 10 nm and foundry 5 nm node. This drove solid share of wallet gains and substantial sales increases for our company. We are currently working with our customers on the next nodes, where we expect again, a further increase in the number of ALD layers and applications. In memory, as mentioned, our sales were about flat last year. In DRAM, we benefited last year from a number of specific customer projects where we have strengthened our position.
Over the last few years, we have invested, and we continue to invest to broaden our served available market in memory. Last year in DRAM, after winning a first selection for a new non-patterning ALD application, we won an additional tool selection later in 2019. We expect these tool selections to start contributing in 2020. Non-patterning applications are expected to remain an important part of the single wafer ALD market for DRAM. In 3D NAND, despite the market downturn, we supplied multiple systems as part of a new customer selection that we discussed at earlier occasions in the last calendar year. As the 3D NAND industry prepares for the transition to the next device generations, we expect the number of single wafer ALD applications and ASM's participation in this segment to gradually further increase. Long-term growth prospects for the ALD market continue to be strong.
Conventional deposition runs out of steam for many challenging process steps and needs to be replaced by ALD as the industry moves to smaller geometries and more complex device architectures in the coming years. Next to ALD, we remain focused on expanding our SAM, our served available market, in our other product lines, especially in EPI, which has now become our second-largest product line. In 2019, EPI showed again, strong double-digit growth, and we had our third consecutive year of market share gains. Demand for our Intrepid EPI tool continued to be strong in 2019, driven by our expanded position at the leading foundry five nm node. We booked further progress in R&D customer engagement last year, and we continue to aim further increases in our EPI share over time. In PECVD and furnaces, we invest in selected opportunities to expand our niche positions.
Last year, this was illustrated by the launch of our new A400 DUO vertical furnace tool for growing 200 mm applications, such as in IoT. Let's now have a more detailed look at the trends in our market. WFE spending, wafer fab equipment spending, decreased by a high single-digit percentage in 2019. Better than what was still expected at the time of our Q3 results. Logic Foundry spending was solid throughout the year, driven by spending on the most advanced nodes. In the advanced logic sector, investments in the 10 nm node increased strongly on top of continued 14 nm additions earlier in the year. Foundry spending on the 5 nm node also increased strongly with ongoing demand for 7 nm capacity. Wafer fab equipment spending in memory was down by 30% or more in both DRAM and NAND.
Looking at 2020, WFE spending is expected to be up a high single-digit percentage, with upside to double-digits, depending on the strengths and timing of memory spending recovery. We expect Logic Foundry to continue its strengths as we move into 2020. Based upon this, we expect the wafer fab equipment market in 2020 to be up with a high single-digit percentage, with upside to double-digits, depending on the strengths of the timing of the memory spending recovery. Yeah? Okay. Yeah, I think we had some duplication here in the final version. Apologies for that.
Our customers are investing in advanced node capacity to enable the faster and more powerful chips that will be required for the new end market products that are going to be introduced in 2020 and beyond, in areas such as 5G, high-performance computing, and data centers, which are all expected to be important growth markets for the coming years. Memory spending is expected to increase later in 2020. Supply-demand conditions in the memory sector have further improved in recent months, illustrated by decreasing levels of inventory at certain customers and a recovery in memory prices. 3D NAND spending is first expected to recover, followed by DRAM. The exact timing and magnitude of the recovery in memory spending is, however, not yet clear. Let's look at our guidance for the next quarters as included in our press release that we announced overnight.
For Q1, we expect sales of between EUR 310 million and EUR 330 million, while for Q2, sales of between EUR 330 million and EUR 350 million are expected, both on a currency comparable level. Q1 bookings on a currency comparable level are expected to be in the range of between EUR 310 million and EUR 330 million. After this introduction, as we shared before, we are happy, Peter and I, to take any questions that you may have.
We'd like to ask you to please limit your questions to not more than two at a time, so that everyone has a chance to ask a question. Operator, we are ready for the first question.
Thank you, sir. Once again, should you wish to ask a question, please press star and one on your telephone keypad. Should you wish to cancel it, you may press the pound or hash key. Our first question comes from the line of Krish Sankar. The line is now open. Please ask your question.
Yeah, hi. Thanks for taking my question. It's Krish Sankar from Cowen. First question for Chuck. I'm just curious, you said you're not seeing any impact from the virus, either in terms of your shipments or even demand. Is this still the situation? Is your guidance for Q1 and Q2 sufficiently de-risked, or do you think there could be further downside if the situation gets worse in Korea or elsewhere? Then I had a follow-up.
Our view is that, as we mentioned in the introduction, we reviewed it several times with our teams also, of course, ahead of this earnings release. We really came to the conclusion that there is very limited risk in our current guidance. We took that basically into account in the ranges that we provided. Of course, we never know what tomorrow could hit the market. Nobody knows. Based on today's visibility of the coronavirus, let's say, expansion and the feedback from our customers, this is it.
Got it. That's very helpful. As a follow-up, Chuck, you mentioned about the strength in Epi market share last year, you grew double digits. Is there a way to further quantify it, either in terms of revenue or what your market share number in Epi was in 2019 versus 2018?
Yeah. I think what we can say is that we clearly went from single digits to a meaningful double-digit percentage.
Got it. Thank you very much.
Yes. Okay. By the way, we see, let's say in the coming years, as we said in our introduction, based on the expectation that the Epi content in this industry is going to grow, we see tremendous opportunities for the company in the coming years. To further expand, not only, let's say, our share of wallet, but also our share in the market. As we go to new nodes in both Logic Foundry as well as in memory, we expect that the Epi content is going to increase. Maybe not to the same extent as in Logic Foundry, but it's going to increase there also.
We love that development, and I think as a company, we made inroads at the right time when we took our strategic decisions on Epi in the 2014, 2015 timeframe to reenter the mainstream CMOS part of that market, besides, of course, presence in analog power that we've had for many, many years.
Excellent. Thank you very much.
Yeah, you're welcome.
Thank you for your question. Our next question come s from the line of Achal Sultania. The line is now open. Please ask your question.
Hi, good afternoon. Just trying to understand the size of the memory business now. In terms of, obviously, you're talking about memory being flat last year, clearly versus market being down 25%, 30%. Just trying to understand how is that memory revenue split between NAND and DRAM. Is it more like 50/50, or NAND is much higher in the mix? Can you maybe talk about some more details about these specific projects as to when should we start to see them ramping? I think, the 3D NAND project, with one or two customers, obviously you're talking about DRAM non-patterning ALD projects. How should we think about the revenue opportunity? Is it going to reflect in 2020 or ramp up, or is this more about 2021 and beyond? Thank you.
Okay. So in 2019, it was great for us that regardless of the memory market overall on the WFE level going down in a significant way, that we were able to maintain a year-on-year, basically stable performance in memory. That really was strongly contributed by our improved position, growing, improving position in DRAM. DRAM was the largest of the two segments in terms of revenue contribution in 2019. Also, if you look at 2020, we cannot, of course, oversee how the full year unfolds. We shared with you on several occasions that we are very focused in DRAM, for example, to penetrate more into non-patterning parts of the DRAM market. As we shared in our introduction, we entered two customers with a non-patterning related application, and we expect to see P&L impact of that this year.
As a result of that, it could very well be that although the market, from an overall market point of view, 3D NAND is expected to recover faster than DRAM, that for us, the DRAM performance will be stronger than 3D NAND. We are even pushing for a year-on-year improvement in DRAM from a revenue point of view. Again, it's early in the year, but based on those selections, we aim to achieve that. This is all part of what we shared on several occasions of our strategy to expand our served available in market gradually, especially, of course, in ALD. We talked to DRAM. With respect to NAND, also in the introduction, we touched on an important application with a leading memory player in 3D NAND that we introduced at 96 Tech. We see now that application proliferating into more customers.
Also there, we expect that that will contribute more over time. The increase of applications in 3D NAND, that will go gradually as we go to higher stacks. That will just take time. At each stack, we expect that we will increase the number of applications, but that will go gradually. We have great JDP programs ongoing for years already with several 3D NAND customers to achieve that in the coming years.
Okay. Thanks a lot.
Yeah. You're welcome.
Thank you. Our next question comes from the line of Peter Olofsen. The line is now open. Please ask a question.
Good afternoon. My question is on the Logic Foundry business, which I suppose is the main driver for the strong H1 sales outlook. You have, of course, mentioned that you expect continued strength in Logic Foundry in 2020. Do you have some visibility beyond the first half? Because I think one of your peers, Lam, indicated that they think that the Logic Foundry business in 2020 might be a bit first half loaded. Any thoughts on that will be helpful.
Yeah. Well, that's a little bit too early. It's too early to provide guidance for the full year. The only thing we can say at this moment in time is that we see a very healthy climate for Logic Foundry for the full year. Again, without specifying how exactly the second half towards the first half will be. We definitely don't see any indications that it will fall off a cliff in the second half compared to the first half. We see a very healthy pattern throughout the year based on the visibility today. It's just because there's just a strong activity in both 5 nm foundry as well as 10 nm logic. We see also, as we said in our introduction, early spending now for 7 nm capacity in logic, R&D capacity.
We see the customer being very active in leading Logic Foundry in preparing for, let's say, initial 3-nm capacity in the course of 2021. Today in terms of R&D activity and then trying to get to a point of freezing their processes for that node. That's all we can say at this moment, Peter. Okay?
Okay. Maybe follow up on this ALD business in logic foundry, which has grown strongly in 2019. Well, for a few years already. Just trying to understand to what extent is this driven by an increase in unit shipments? Is there also an ASP element where, if you address new applications, maybe processes becoming more complex, also your ASPs goes up. Is there an ASP element to the growth in the ALD business, or is it predominantly higher unit shipments?
Well, of course, we are working on better gross margins. From that point of view, we are trying to optimize our ASP and our cost. If you look at the top-line development, that's purely a reflection of all the strategic decisions we made a couple of years ago to really aggressively expand our served available market. Especially in 2019, it showed, especially of course, in the Logic Foundry segment. Really working hard to maintain our single wafer ALD leadership and push hard to expand our share of wallet, node to node in foundry, going from seven to five and in logic from 14- 10. That is really what you see. It's nothing different from what we guided for the last two years. The only good thing is now that the customer really ramps that we all together are seeing the results from that. That's it, Peter.
No more, no less.
Okay. Clear. Thank you.
You're welcome.
Thank you. Our next question comes from the line of Marc Hesselink. The line is now open. Please ask question.
Yes. Thanks. The first one is actually on your comment that you made, that you're working with your customers on the next nodes, which also just addressed for 7 nm logic and 3 nm foundry. Are those equipment decisions already made? Did you qualify for doing those processes? My second question is on gross margin. Quite a good step up that we saw in the quarter. How should we read that? Is that because of the factory loading where you are today and relatively the evaluation tools in the field becomes less, that therefore you see the gross margin improve? Are there other stuff and therefore also, what the outlook for the coming quarters averages? Will you further gradually move towards your 45% number or will that take longer time? Thanks.
Okay. Marc, first of all, on the new nodes, decisions have not been made. We cannot make, let's say, we don't know for, let's say 400% where we stand on those new nodes. The only thing we can see is how much opportunity do we really get to enter into joint development programs on potential new applications. You see that in the way wafers are being exchanged between the companies, the amount of evaluation tools, which is not only an expensive exercise for us, for the supplier, but also for the customer. They free up expensive footprint in their very expensive clean rooms. They have to free up people to occupy those tools instead of working on competitive tools.
Based on those activities, the amount of tools on site, the amount of wafers that exchange between the customer and our labs. The amount of wafers that go in a development stage through our tools, those are early indications how well we are doing. Based on that, we feel that we have an opportunity going towards the next node to really grow our presence. Exactly how much? That's too early to tell. In both areas, in logic and foundry, we see an opportunity to further expand going to the next node. That's an answer to that question, Peter, on the gross margin.
The gross margin. Marc, a few remarks to be made. First of all, I think the inefficiency is still not out of our gross margin. You can imagine with the growth that we have seen, especially in the fourth quarter, that we are working very hard to get service people to do the installation qualification. We train a lot of people. There we have inefficiencies. We have mentioned in the past calls that we have, of course, problems with getting everything out of our factory in Singapore because that's fully occupied. On the last moment, when changes are occurring, that leads to a lot of overtime. It leads to a lot of shifts that we have to do. We are still not happy with having reached there the levels that we need to reach.
That will highly become visible more in the second half of the year or in the early next year when the ramp-up of the new factory in Singapore has taken place. Beside that, we still have a lot of eval tools in place. Hopefully, that will remain that way because that's our future business. You might have seen from our numbers or so that in the second half of the year, we spent more money again on eval tools at customers. On the same way, what we starting to see is the more positive contribution, I should say, from new products. Where after the initial phase where we had to do additional costs to make sure that those products were up and running at the customer, those costs gradually are coming down. That has led now to a situation of a gross margin in the 43% range.
What we have said in several calls earlier is that we expect that we would go in the course of 2020, early 2021, more in the direction of the higher end of the range that we have indicated. I think that's still something what we strive for.
Okay. That is clear. Maybe a small follow-up on the first part. When do you expect that decision to be made so that you have really the visibility on how much extra applications you want in the transition in logic and foundry?
Yeah, I think it will take at least until the end of the year. I think it's a little bit, because it's not completely clear yet. That also depends on market dynamics. When our customers really take the final decision to ramp that product, that determines also how much time they have to maybe also delay that moment. It may differ also between foundry and logic. Maybe one area will go a little faster than the other. For sure, not earlier than the end of the year.
Okay. Did the orders come in immediately, or that also takes time, a little bit of a delay?
Could you repeat that?
If they make the decision, does that imply that after they make the decision, the orders come in immediately, or if they're still delayed? In other words, can we see contribution from that transition here in 2020, or is that a 2021 story?
I think ramping new nodes likely will be 2021. Early, expanding your final development environment can still impact this year.
Very clear. Thank you.
Thank you. Our next question comes from the line of Robert Sanders. The line is now open. Please go ahead.
Yeah. Hi, good afternoon. My first question would just be around your ALD long-term growth forecast. We're getting pretty close to that. I'm just wondering when you think you'll give a new, medium to long-term range forecast for single wafer ALD. Would that be after the new CEO takes over? The second question would just be, in that light, is the right way to think about ALD growth from here, quite a lot of the growth is coming at the expense of CVD applications? Like, I don't know, sputtering in the metal gates or whatever it is in the back end of line. Or do you see it mainly in the front end of line of totally brand new applications where you're not replacing steps that today are used by older deposition techniques? Thanks.
Rob, thanks for your question. Starting with your last question, ALD can take markets from both front end of line as from back end of line. That really has potential to take business in both areas. In terms of the ALD market size. Your question maybe is also a little bit where do we stand? Well, we know that the market was
Around EUR 1 billion in 2018. In 2019, there was clearly some growth in the single-wafer ALD market. It's reasonable to assume that in 2020, single-wafer ALD market will not grow less than the WFE market estimates that are there today. We think that EUR 1.5 billion market size that we predicted for 2020/2021, is still absolutely feasible for 2021, more likely than 2020, of course. It's clear. Of course, we learn more and more now as customers go to new nodes. We not only learn more in logic foundry, but also in memory, based on all the activities that we touched on earlier in the call. The conclusion we can take is, it will not stop at 1.5. That's very clear to us. It will not stop there.
That's the best I can share with you at this moment in time, Rob, but, yeah, we take your invitation. As soon as we can, that we provide some guidance beyond 1.5 towards the future. When we can.
Thank you.
Okay. You're welcome.
Thank you. Our next question comes from the line of Sandeep Deshpande. The line is now open. Please ask your question.
Yeah. Hi. If I may, I have a question on ALD. When you started shipping ALD almost a decade ago, you were supplying the metal gate to one of your customers. How many layers of ALD are you now on average supplying in the logic/foundry market, and how do you see that transitioning over the next five years? Thank you.
Well, how many layers? Tens of layers. Many layers. In HKMG, as you know, Sandeep, it was a handful of layers that we were basically, at that moment in time, focused on. That limited, basically, our served available market. That's why, we put R&D to work very hard in the last 10 years. It's now, of course, from a competitive point of view, we don't want to go into too much specifics, but it's tens of layers that we are involved in, Sandeep.
Expect that to continue to increase over the next five years?
Yes.
So what will it be replacing over the next five years? Are you talking about metal deposition or is it replacing chemical wafers? Clearly, it's taking share within the stack, in the semi stack from something else. Is that fair?
Our view is that is, of course, the aim that let's say, the SAM within the served available market for single-wafer ALD within the total deposition TAM is going to grow in the coming years. We saw, of course, maybe with the introduction of 3D NAND, we saw maybe that there was some slowdown in that development. Of course, 3D NAND enabled also big opportunities for layers from PECVD, to a relatively mature, let's say, technology to still meaningfully expand. As we further continue, both in memory and logic, our view is that, if you look at the device architectures and also the material requirements, we'll introduce a higher complexity level, which will drive ALD further.
Thank you.
Okay. Sandeep, you're welcome.
Thank you. The next question comes from the line of Wim Gille. The line is now open, please ask your question.
Yes, a very good afternoon. I got two questions left. First of all, your working capital management has been outstanding in the last few quarters. The days of working capital keeps dropping. Especially given the strong growth, that's quite an achievement. Can you give us a bit more feeling what structural measures you took in order to get here? Also, how much room is left on that end? The second element would be, the timing of the share buyback and the speed at which you expect it to execute. Are the terms of the new share buyback similar as the previous one? As a consequence, do you expect it to be executed as swiftly as the one you just finished in the fourth quarter? Thanks.
First of all, the working capital. As you might recall from mid-2018, there we indicated that we were not very happy with our working capital, especially our inventories. When you look to what we have done in the inventories, is especially addressed the raw materials because we have changed there the systems. We have worked on further reducing the needs that we have.
We always see further opportunities, otherwise it would not be good. The big steps that we have taken in raw materials, that will be not any longer possible. When you look to the other part of the inventories, that's basically the work in progress, and that's a function of the throughput time. That's what we are working on. Most likely we can do there a few other things on the moment that we have for a new facility, that will help us. The rest, spare parts and the finished goods, I think we have taken there also further actions to work on those things. We reach there a level where we can do some small improvements, but not that much. The accounts receivable, that's the other element of it.
We especially look to our overdues there, and they are at a relatively very low level within the industry. I think that you can argue with your customers to reduce the payment terms there a little bit further, but that's it a little bit. I think going forward, and the last one, I think that I should mention, is the accounts payable, and we pay our suppliers on time. We don't want to shorten our balance by leaving unpaid bills there. I think that with the big improvements that we have done, there is not that much improvement. We can still do some improvements, but not the big improvements that we have seen in the second half of 2018 and especially in 2019.
I think with 42 days, we are at a very low level or so when you look to the industry as a whole. Regarding the share buyback, our intention always has been that when we announce a way of distribution capital towards the shareholders, that we want to do that in a short period as possible. It's not our intention to extend the share buyback program for a very long period of time. Having said that, on the other hand, we want to do also the right thing. We don't want to do it in such a short period of time that it will cost us more than what we normally should have paid for it. That's the balance that we will look for, and that we will work that out in the next weeks to come. Perfect. Thank you.
Thank you. The next question comes from the line of Stephane Houri. The line is now open.
Yes, hello. Good afternoon. It's Stephane Houri from Oddo BHF. I have two questions, if I may. I try to reconcile what you are currently saying for the full year and for the first half when you say that the market will grow high single digit and with the guidance that you give for Q1, Q2. Does it mean that if there is no recovery in memory, you think that H2 sales will be below H1? Or is it just that you think that you will outperform significantly the market? I have a follow-up.
That's difficult to answer directly. As we said, it's a combination. Our high single-digit estimate is based on a healthy Logic Foundry segment, throughout the year and let's say a gradual recovery of memory starting to occur in the course of the year. That could go to double digit if the memory recovery would go faster than we are assuming at this moment in time. Your question is really where our high single-digit percentage estimate is based on, right?
Correct, correct. When I do my quarterly model, basically, if I want to make you grow only high single digit, then I have to have a second half, which is supposed to be below the first half. I'll just trying to reconcile if you're just conservative.
We can say something for the full year on the market, but we cannot give guidance for the full year on our sales. The only thing you can work with is if you compare based on the midpoint of the guidance for Q1 and Q2, and you compare first half 2020 with first half 2019, then there's a growth of 30%. Yeah?
The next parameter that you can use is that we forecast for the full year WFE a high single-digit percent. That's the only thing we can share at this moment in time. Yeah. Maybe, Stephane, to add a little bit onto what Chuck already earlier said. We do not expect that the second half will fall off the cliff.
No.
We expect that Logic Foundry will remain healthy.
Okay.
I think that's the distinction that you have. Yeah. Of course, we also have an ambition, like we had the ambition in 2018 and 2019, to still outperform the WFE market.
Okay.
That's, again, of course, also a function of-
The dynamics between the different industry segments. Based on today, that is still our aim.
Okay. Thank you very much. The follow-up is even a tougher question, I would say, because it's about 2021. When you listen to companies like ASML, for instance, you have the feeling that they're even more positive on 2021 than on 2020, partly because of their own project on EUV, but also because they don't know, like you, about the timing of the recovery of memory, but they seem to be sure that in 2021, you will have at the same time logic and at the same time memory. Do you share the same feeling that 2021 is somehow a bit easier to forecast than 2020?
No, it's too early. Some suppliers also deal with higher lead times than others. As a result of that, have maybe different views, develop different views than others.
Okay.
Yeah. Okay?
Okay. Thank you very much.
Yeah. You're welcome.
Thank you. The next question comes from the line of Nigel van Putten. The line is now open. Please go ahead.
Good afternoon. A question on WFE. Some of your peers have been guiding for a strong uptick in China, especially from domestic players there. That may not all be leading edge. How is that opportunity looking for you in this year and maybe also 2021? Is that something that might need some more room to grow from today, or is that already something you're heavily involved in?
Okay, Nigel. Thanks for your question. Good question. Our estimate is, our guess is that total WFE spending by, let's say local Chinese customers, clearly showed an increase in 2019 over 2018. It's very likely that 2020, at least based on visibility today, based on visibility of coronavirus impact today, et cetera, that in 2020, it will show a further increase, their spending in this industry because they are very determined. As we shared before, we have worked hard on improving, strengthening our infrastructure in China, and that resulted into 2018, a year-on-year growth in China. I'm talking primarily about the domestic ones because multinational ones is a separate story. It just follows the pattern of those companies. In 2019, we clearly grew our presence also. Our revenue also meaningfully grew year-on-year in 2019.
Indeed, we see also based on visibility today, a meaningful growth opportunity year-on-year in 2020 in China.
For sure. Very clear. Maybe switching to Peter on OpEx. This year, we've seen plenty of moving parts, from legal costs, I think impairments, higher incentives pay and obviously stronger growth. Could you help us maybe benchmark what should be quarterly OpEx number we should be modeling, at least in the first half of this year? Is there any sort of guidance you can provide on that?
Yeah. What we always have said, that's the reason why we provide also somewhat more detailed information about research and development. You have to look to the research and development expenses there. There you see that the cash expenses are gradually increasing. Q-on-Q, you saw an increase of approximately EUR 3 million. We expect that our R&D expenses are in the low-to-mid teens. Given the fact that we work on a lot of new projects at this moment, they're moving towards more the mid-teens. We have some changing factors there. You have seen in the fourth quarter, as an example, some higher spendings increases compared to previous quarters. Sometimes that has to do with, for EUR 1 million or so, with higher wafers that we receive, blank wafers that we receive.
On the other hand, we have, as part and parcel of the IFRS methods that we are using, we have, of course, the amortization of R&D. Gradually, with the enormous volume of projects that we have started in the past years, you see that the amortization, the depreciation principle on those projects is starting to come in. In the fourth quarter, and we always will have that over time, there was one project or one or two projects where we said, okay, the chances that we have to make this really a commercial successful thing were not there, so we had to reduce that. From a guidance point of view, the best thing is that you look to the cost, and then we have those amortization issues, which can play a certain role.
With regard to SG&A, that's the reason why we have given the guidance this time. The legal costs are out. They are already out of the fourth quarter. The major issue in this case was related to the fact that the fourth quarter was developing that favorable that some of the targets, the internal targets that have been set have been reached, and that led to an extra provision that we had to take on SG&A. That's a one-off.
Okay. Thank you.
Thank you. The next question comes from the line of David O'Connor. The line is now open. Please ask the question.
Great. Thanks for squeezing me in here, guys. Maybe just a quick one for you, Chuck. You mentioned you don't expect Logic Foundry demand to fall off a cliff in mid-2020, is it fair to assume the Logic Foundry is going to decelerate from the current high level, given when you look at 10 nm large capacity additions are slowing from here? Thanks.
I can only reiterate what we said before. We see currently a healthy climate and for Logic Foundry combined, and that's what we are working towards. It's difficult to provide more guidance than we have tried to give so far. I understand that you would like us to provide more color, but that's really difficult to do now. We are extremely well-engaged with both segments at this moment in time, and that's all we can say at this moment. Sorry.
Okay. Fair enough. Maybe just one quick one for Peter. CapEx for 2020. Any indication there, please? Thank you.
Yeah. The 2019 CapEx, as you have seen, was a little bit lower than 2018. We had assumed that the spendings on the new Singapore factory were coming in a little bit faster. We expect that especially the first half of 2020 will show slightly higher spendings for Singapore. For the year as a whole, I expect that our CapEx will be more in line with what we have seen in 2018 than the amounts that we have seen in 2019.
That's helpful. Thank you.
Okay. Thank you for your question.
Thank you. No more questions on the line. Please continue.
Okay. Well, also on behalf of Peter and also Victor Bareño, our investor relations executive, I would like to thank you all for your attendance today. Many questions came up. We were on the call for 70 minutes in total. I can understand maybe you have some follow-up questions still that you were not able to address in the call today. Please feel free to contact Victor. We will make sure to follow up on this in an appropriate way. Thanks again for your loyalty in following the company and have a nice day, nice evening. Thank you again. Bye-bye.
Thank you. That does conclude our conference for the day. Thank you all for participating. You may all disconnect. Have a good day, everyone.