Siltronic AG
XETRA:WAF
Utilize notes to systematically review your investment decisions. By reflecting on past outcomes, you can discern effective strategies and identify those that underperformed. This continuous feedback loop enables you to adapt and refine your approach, optimizing for future success.
Each note serves as a learning point, offering insights into your decision-making processes. Over time, you'll accumulate a personalized database of knowledge, enhancing your ability to make informed decisions quickly and effectively.
With a comprehensive record of your investment history at your fingertips, you can compare current opportunities against past experiences. This not only bolsters your confidence but also ensures that each decision is grounded in a well-documented rationale.
Do you really want to delete this note?
This action cannot be undone.
52 Week Range |
46.42
92.75
|
Price Target |
|
We'll email you a reminder when the closing price reaches EUR.
Choose the stock you wish to monitor with a price alert.
This alert will be permanently deleted.
Hello, everyone, and welcome to Siltronic's conference call on its Q3 2020 results. Please note that this call is being recorded and streamed on Siltronic's website. The call will be available as an on-demand version later today. Your participation on this call implies your consent with this. Also at this time, I would like to turn the conference over to Petra Müller, Head of Investor Relations and Communications of Siltronic AG.
Thank you, operator. Welcome to our Q3 results presentation. Joining me on today's call are Siltronic CEO, Dr. Christoph von Plotho; and Siltronic CFO, Rainer Irle. Following our usual procedure, Chris will start with some remarks on group results. Rainer will then comment on key financials, followed by Chris again, updating you on our guidance and current market developments. After the introduction, we will be happy to take your questions.Please note management's comments during this call will include forward-looking statements, which involve risks and uncertainties. For discussions of risk factors, I encourage you to review the safe harbor statement contained in today's press release and presentations and in our annual report. All documents relating to our Q3 2020 reporting are available on our website. And now I would like to turn the call over to Chris for introductory remarks.
Thank you, Petra. Welcome, everyone, and thanks for joining us for our Q3 call results today. I hope all of you and your families are healthy and safe. Before we start with the Q&A session, Rainer and I would like to provide you with an overview and some commentary on Q3 as well as our outlook. Let's start with the key financials. Our Q3 sales were down by 7% to EUR 299 million, below the good Q2 performance. On the positive side, wafer area sold was slightly above Q2. But we had to digest headwinds from a negative product mix and from the FX side, especially the strong euro, which was up to the U.S. dollar from 1.10 to 1.17. Our EBITDA came in at EUR 80 million, EBIT was down to EUR 44 million. CapEx was EUR 35 million after EUR 48 million in the second quarter 2020 and was mainly related to capability and capacity expansion projects. Net cash flow was slightly up to EUR 29 million compared to EUR 27 million in the quarter before. Our net financial assets increased to EUR 519 million at the end of September. Now let's have a look at silicon demand and end markets in Q3. In Q3, the logic and foundry business was strong. Servers performed from home office and streaming. Inventory levels were okay. Memory business, on the other hand, was still a challenge. Raw wafer inventory levels were slightly up and we see partly elevated NAND inventories in the value chain, for example, for server components. And there is still overall uncertainty due to the U.S.-China trade tensions. Silicon demand for automotive saw an uneven recovering in China up, but U.S. and Europe only flattish. The next phase of recovery will depend on the corona-driven uncertainty will develop. We assume that overall, silicon demand was more or less flattish compared to the second quarter and the 300-millimeter shipments were slightly up. In Q3, our 300-millimeter polished loading slightly increased and 300-millimeter EPI stayed very strong. 200-millimeter was down, but SD was rather stable. As expected, we saw a flattish price trend quarter-on-quarter. However, ASP for the whole year 2020 will be down versus last year as prices started to fall during the year 2019. Rainer will now explain in more detail our financial performance in the third quarter. After that, I will give some overview on what we expect to see in the short and mid-term. Rainer?
Yes. Thank you, and good morning, everybody. Sales in Q3 were weaker quarter-on-quarter. On the positive side, we were able to sell more wafer area than we had expected. But as Chris already mentioned, we had headwinds from the negative product mix and the strong euro. Overall, sales reached EUR 299 million, down 7% quarter-on-quarter. Without the FX effect, sales would have been around EUR 16 million higher. As ASP started to come down in 2019 already, this had a negative impact on our sales in 2020. After 9 months, we have generated sales of EUR 923 million, a minus of 4.5% year-on-year. Quarter-on-quarter, prices were stable, but ASP was down due to mix and exchange rate. Cost performance was excellent in Q3. Almost all production lines had a record productivity. Depreciation went up. FX was helpful. But the positive impact on cost was obviously much lower than the negative impact on sales. Overall, cost of sales went down 1.3% quarter-on-quarter to EUR 221 million, despite higher scheduled depreciation and slightly higher volumes sold. In the 9-month period, COGS were EUR 656 million, up 8% versus prior year. Main reasons are the higher wafer area sold and the increase in depreciation. Excluding depreciation, COGS per wafer area would have been down compared to last year. Our gross profit fell to EUR 78 million in Q3. Gross margin was down from 31% to 26%. Reasons are again exchange rate and product mix. Gross profit for 9 months was EUR 267 million, 26% below last year. Gross margin was 29% compared to 37% last year. Our selling, R&D and admin expenses of EUR 34 million were on a stable level compared to the previous quarter. As already pointed out, we saw a clearly negative effect from currency in Q3. An unchanged U.S. dollar would have resulted in EUR 16 million more revenues in Q3 and about EUR 12 million more profit in that quarter. FX had no major impact quarter-on-quarter on other operating income and expenses. In the 9-month comparison, we see a positive development of currency effects. While we recorded expenses of EUR 23 million last year, expenses this year only added up to EUR 2 million. As you know, we have changed our hedging strategy, and we now use cylinder options in addition to forwards. U.S. dollar forward pricing has improved now, given the lower prime rates in the U.S. As the U.S. dollar seems to stay at around 1.17, 1.18, we will see the small positive results from hedging this year. In line with our lower gross profit, EBITDA was down to EUR 80 million in Q3, a 20% decrease quarter-on-quarter. EBITDA margin was 27% after 31% in Q2. This was triggered by negative FX effects on sales and the negative product mix with positive contributions from cost reductions. The EBITDA after 9 months was EUR 265 million, which is 17% below last year. EBIT in Q3 came in at EUR 44 million with an EBIT margin of 15% compared to 21% in Q2. Depreciation increased slightly quarter-on-quarter by EUR 3 million. In the 9-month period, EBIT was EUR 164 million, down EUR 78 million year-on-year. Apart from negative FX and product mix [Audio Gap] in Q3 was further down from 12% to 9%. In the 9 months, overall, the tax rate was 8% versus 12% last year. Net profit were EUR 39 million in Q3 versus EUR 61 million in Q2. Earnings per share came in at EUR 1.08. Working capital went up in Q3 to EUR 218 million, driven by an increase in trade receivables and a decline in trade liabilities, while inventories were roughly stable. It increased. But it is really just normal fluctuation and it remains in the usual corridor. Looking at our balance sheet. Equity came down to EUR 833 million at the end of September. Equity ratio was 44%. The decrease is based on the profit minus the dividend payment, the strong increase in pension obligations as well as negative currency effects. The pension provision in Germany was discounted at only 0.95% as of December -- as of September versus 1.24% as of December last year. Also in the U.S., the interest rate came down from 2.98% to 2.41%. Net financial assets came down by EUR 70 million to EUR 519 million, even though we paid EUR 90 million and refunded EUR 49 million of customer prepayments. As of September, pension provisions went up EUR 140 million due to lower interest rates, as said. In comparison, if we used a 3% interest rate to calculate the benefit obligation, it would be EUR 870 million and not EUR 1.256 billion and then the pension provision would come down to EUR 220 million, which is almost EUR 400 million less than we have to state in our balance sheet. CapEx in Q3 was down to EUR 35 million. Most of it was related to capability projects and the 300-millimeter capacity expansion and projects to increase EPI capacities to capture the growth opportunities in the logic market. Our CapEx guidance of roughly EUR 200 million remains unchanged. Hence, we will see some backloaded CapEx in Q4 due to some delays of equipment, which had been scheduled to happen more in the first half of the year. Our operating cash flow in Q3 was slightly up to EUR 61 million, falling EUR 59 million in Q2. We refunded EUR 10.5 million of net customer prepayments in Q3. Net cash flow was EUR 29 million. The expected higher CapEx in Q4, and typically on payments, will impact our net cash flow, which we expect to be negative in the last quarter of the year. And with that, I would like to hand over to Chris.
Yes. Well, thank you, Rainer. Let's come to the outlook. Our outlook remains unchanged for the full year 2020 despite corona and macro-related uncertainties. Q4 is expected to show the usual seasonal fluctuations with lower wafer area sold, however, higher than we expected 3 months ago. In total, wafer areas sold in the second half of the year will be slightly above the quantity in the first half of the year. Sales and EBITDA will be negatively impacted by the headwind from the strong euro and the negative product mix in the second half of 2020. We expect that we will be able to finish almost all of our capability and capacity projects as planned. Due to corona and the related uncertainty about measures taken by individual countries, there is a risk that we might have to shift some of the projects set into next year. Our CapEx guidance for 2020 remains unchanged. Now let's have a deeper look into market expectations for the fourth quarter. We had a good Q3, which was driven by healthy wafer demand. We expect Q4 wafer volumes to be slightly better than expected 3 months ago. That said, please be aware that Q4 wafer volumes will still be down quarter-on-quarter. The uptick in wafer orders is not only driven by strong 300-millimeter EPI demand but also by some spot orders for 200-millimeter and smaller diameters. We assume some of the demand is driven by U.S.-China trade tensions and some inventory buildup to counter possible supply chain disruptions. According to what we read and hear, the automotive industry starts to see some light at the end of the tunnel and place some spot orders. However, more for legacy than for high-end products. Hence, the negative product mix effect, which we have already seen in Q3, will prevail in Q4, together with a strong euro impact our profitability. Foundry and logic customers continue to see strong demand. Growth drivers, such as 5G, artificial intelligence and high-performance computing, drive the demand for advanced nodes. In the past, digital transformation was already on top of the agenda, but the corona pandemic accelerated this topic even further. We expect demand for 300-millimeter EPI to stay healthy. Memory was driven by -- to a large extent, by smartphones being the biggest consumer of silicon area. With every new smartphone generation, we see increasing silicon content, mainly driven by more memory. But also the increasing demand in servers, especially hyperscale data centers, will add to silicon demand going forward. Customers expect higher bit growth for next year. The automotive industry is currently on a slow recovery part. The picture is very mixed for various regions. But taking aside the short-term development, the structural driver, more advanced automated data assistance systems per car is intact. And the transition to hybrid and electric cars will further accelerate silicon demand. Maybe one word to the U.S.-China trade tensions. While we do not comment on individual customers, we aim to serve and support all our customers worldwide. And this includes China. We are compliant with all laws and regulation and are aware of the requirements defined by the U.S. government. However, so far, we are not impacted by any regulations and can ship to all customers no matter where they are located. It is too early to provide today a guidance for 2021. There is still significant -- there are still significant uncertainties. The most important question will be if we will be able to keep COVID-19 under control and how the pandemic will impact the global economy and end markets going forward. Currently, infection rates are raising all over in Europe, and we see first regional lockdowns already. Additional uncertainties will persist with the geopolitical uncertainties, mainly created by the U.S. trade tensions. And of course, with our customers and further up in the value chain, the question of inventories will have an impact. In case inventories will normalize in the short term, we could see a stronger bit growth in 2021. If this is not the case, growth might be further down the road. Design rule development is another big topic for the semiconductor industry. Customers' technology road maps are continuously evolving. The capability of our customers to go on the next node and the respective node ramp will determine wafer demand. To summarize, there's a lot of external factors, which will have an influence on our business. But one thing is for sure, digitalization won't stop. Demand drivers, like 5G, servers, artificial intelligence, will contribute to the demand of advanced technology. We at Siltronic, being one of the technology leaders, are very well positioned to serve this demand. The whole industry is currently expecting near-term uncertainty, but the long-term demand drivers are intact and provides a bright future growth outlook. In their October press release, the SEMI organization was forecasting wafer shipments to grow by 5% in '21 compared to the actual year 2020. With this, we close our presentation and are now available for your questions. Operator, please open the Q&A session.
[Operator Instructions] And the first question is from Francois Bouvignies, UBS.
The first question I had is on your comments on the product mix. So you mentioned that the product mix is negative this quarter, impacting the profitability. Can you give us a bit more color on what product mix exactly is dragging? And what is the outlook for the product mix in Q4, i.e., is it going to improve or not according to your orders? The second question I had is on your comments on inventories and the outlook for Q4. I don't quite understand when you say that inventories will be in digestion in servers. But there is uncertainty with maybe likely to elevated inventories due to U.S.-China trade tensions. So it would be very interesting to have your view of how do you think, net-net, the inventory will trend, taking into account these 2 aspects. And more importantly, do you think it's going to happen in Q4? Or is it going to last a bit longer? And the last question I had is on your long-term agreements. So could you confirm what percentage of LTAs this year you're going to end most likely? And more importantly, for 2021, how should we think about the LTA proportion for '21?
Okay. Well, thank you for your three questions. Let's try to start to find an answer for question number one, which was on the product mix. You remember that we mentioned in our last call, the reason for that -- this negative impact on product mix, in Q1, corona was, to a large extent, a Chinese problem. And consequently, some of the players in the market were not convinced that they might get from China what they need in the second quarter. So they ordered additional quantities, which helped us to have an excellent Q2. But at the same time then in Q2, what happened? Industrial production and automotive production was significantly down. So these additional quantities, which were ordered by customers, were not consumed. And therefore, inventory went up. And this additional inventory will be consumed in the second half of this year, so not only Q3 but also Q4. Why does this have an impact on, let's say, revenue and ASP? Typically, these products are specialties. And specialties have a higher ASP. And if the percentage of specialties goes down, this then has a negative impact on the overall ASP. We said that this is -- this will take some time to use up the inventory. We believe that this will be done by the end of the year, so its effect on the second half of this year. We do not expect similar negative impacts going forward into 2021. Now I hope this helps for question number one. Let's try to do a good job also on question number two, which is regarding the inventory. So the inventory that we see, raw wafer inventory, is, I would call it okay, maybe slightly higher. But in this field of uncertainty, customers typically want to be a little bit more on the safe side, so slightly higher inventory is okay. What is a little bit the concern is the inventory in finished goods. And this is NAND shifts up to what we saw from analysts. This is not a major concern, but it's somehow up. And whether there will be a digestion of this inventory, yes or no, is something that we simply don't know, but -- because it's not up to us to work on it. I do believe that, in general, trade tensions and uncertainties typically lead customers in the direction to have slightly higher inventories in order to make sure that they have enough material to produce what they need to produce to fulfill the orders of their customers. So last but not least, I come to the LTA situation. The LTA situation this year, it was not significant -- we didn't see significant changes compared to prior year 2019. And as far as I remember, we always mentioned a figure of around 60%. But at the same time, we said this is the average of Siltronic. And like also mentioned by competitors, their share for higher diameters, like 300-millimeter, is above that average and smaller diameters is probably below. I do not see any significant changes going 2021. Basically, when I look at the total picture of 300-millimeter, even customers which actually are on shorter terms, so on contractual obligations that we, with our definition, wouldn't call LTAs, I think when it comes to renegotiate these quantities during '21, they will consider an LTA because we expect that the overall shipments will be around 6.6 million 300-millimeter wafers compared to a possible share capacity of 7.2 million. This is only a gap of 10% and the average growth rate of 300-millimeter in the past was typically around 5%. So theoretically, this will be eaten up in 2 years, may last 3 -- may take 3 years, maybe also shorter. But I think customers in the meantime are aware that sometimes in the future there is a shortage ahead of us. And by the way, due to that, I believe that this scenario will give us and also other players in the market the opportunity to consider price increases for everything, which will be renegotiated in '21 for the period after '21.
The next question is from Achal Sultania, Crédit Suisse.
Just coming back to that gross margin comment that you made. So if I look at your Q3 margins, it's about 400, 450 basis points below what we saw in Q1, Q2. Can you maybe help us quantify like how much of that was from product mix versus FX, just rough breakdown of that decline?And secondly, if I look at, going forward, what you're saying about mix, clearly auto demand has started to pick up. And I presume a lot of weakness was predominantly on the auto side. And my sense was auto was only about 10% of your total revenues or probably even less than that. So is mix still going to be a meaningful driver for gross margin headwind going into Q4 or a lot of that is already kind of behind us in Q3? And secondly, on the FX impact, I thought like the math that we were working on FX on sales was about $0.01 having a EUR 6 million per year impact. So using that math, it typically -- like I was expecting around EUR 10 million to EUR 11 million of sequential headwind from FX on revenue. You mentioned it was around EUR 16 million. So just trying to understand like whether the -- like what's changed on that FX impact calculation in the quarter.
Well, let me try to find an answer first to the second question, which was also related, you said auto picks up. And your question was whether we will see this mix effect also in Q4. I think I answered that already with the first question, but I'm willing to repeat it. Yes, we will also see some negative impact related to auto and industrial applications in the fourth quarter. But we are convinced that this is an effect. And for the second half of the current year, we do not expect any negative product mix impacts going into 2021. And for the other two parts of your question, FX-related and in gross margin, Rainer, would you please answer?
Sure, Achal, yes. So gross margin drivers, I think, the FX impact on profit is roughly EUR 12 million and the impact of the product mix also is not only a top line, it's also a bottom line impact because certain products tend to have a bit higher margins. The third one is depreciation, which is up EUR 3 million. And then that already explains more than gross margin actually went down. And then on the other hand, we had positive contributions from excellent cost performance. Your calculation on the FX headwind is basically correct, it's EUR 6 million per $0.01 in a year. So if you talk a quarter, 1/4 of that times $0.07. But please consider that it is not only the U.S. dollar, it's also the yen that changed. We have the updated numbers in the presentation, what their sensitivity is, again also contributed -- and to give you a better, a clearer view on how much it really, was we disclosed the EUR 16 million for the top line and the EUR 12 million for the bottom line as the impact of FX.
And the next question is from Amit Harchandani, Citigroup.
Amit Harchandani from Citi. A couple of questions, if I may. My first question goes back to the tone of discussions that you're having with your customers, please as they think about demand into 2021 and potential implications from the pandemic and the lockdowns. On one hand, clearly, it's not great that lockdowns are coming back and impacting some of the end demand. But on the other hand, you could argue, this continues a greater emphasis on remote working, potentially also drives demand for other areas in tech. So I'm just trying to understand how are your customers thinking in terms of what they need from you, depending on the various scenarios, if you could give us a feel for '21. I appreciate you can't give us a clear guidance. But any steer in terms of tone of conversations and potential scenarios would be helpful. And second question, if I may, with regards to the LTAs. My understanding was you were in conversations with your customers and there was an LTA to be signed potentially on the horizon in Q4. Could you give us a sense for where the conversations are in terms of timing? And what could we expect or what should we look forward to in Q4 and Q1?
Well, first of all, the tone of discussions that we have with customers, Amit, it's a very, very interesting question. I would like to know, too. Because like we said, in general, there's a lot of uncertainty. And uncertainty, unfortunately, is linked to be in the human brain to be negative, although I always say uncertainty only means that you don't know, might also be better than people tend to believe. There is a lot of more short-term thinking than anything else. I do not remember any report from any significant customer where we had an outlook for the total year. So basically, I can't comment on the tone of discussions that we have. It's driven by uncertainty. But please keep in mind, uncertainty is not, by principle, negative. We also try to find out what is the realistic outlook into next year. And what we are doing, we are not looking so much at the customer. We are a little bit looking at the end markets. And when you take these 2 end markets, one is automotive and the other one is smartphone, both together sum up to, let's say, 1/3 of wafer supply into the market. I do believe that it's fair to expect that automotive will be down this year somehow around 20%, smartphone, probably down around 10%. I'm pretty sure that we will see in these 2 end markets a positive effect going into '21. On the other hand, I'm convinced that they will not go back to the level of 2019. But let's assume that they will make up half of what they lost in 2020. This would translate into a 2.5% area growth overall for this market, not considering any content change. This is only based on pieces per use and pieces sold. We are convinced that on top of this 2.5%, we will also see some additional area coming from content. So we assume that these two markets will lead to a 3% area growth into next year. And on the other hand, we do not see any other significant end market to have a negative impact. So therefore, I think we can't guarantee anything. But like I said, SEMI talks about 5% area increase going into '21. And I think that with all the uncertainties that we have, I think that's a good figure. Now with the LTAs, we talk about LTAs when they are done and not when they are under discussion. So -- and as there are things under discussion, we will not disclose the details of discussion. But I'm pretty sure it will have a positive outcome for the customer and also for us. And I already made some comments regarding LTAs, which will be discussed in '21 for the period after '21. And there, expect opportunities to increase prices.
And if I could just clarify, please, on response to the first question. You talked about smartphones and automotive. That's very helpful. And just so that I understand with respect to the other end markets, even for data centers, you don't expect anything to be either positive or negative into '21, just to be sure. And actually, on an unrelated note, a clarification to what you said earlier about shell capacity. Is that shell capacity a reflection of the latest competitor dynamics, including some of the latest additions we have seen in Asia? Can you please confirm that?
Well, I don't know what you're referring to in Asia. I think it considers some additions. It contains the additions, which are supposed to have the right level of technology development. There are other capacity additions, which happened in '21 that we do not really consider as a prime capacity. We consider that more as a test wafer capacity. And for data center, well, I would summarize that everything, which is around the computer, laptop and data center, all this area together. And from that, I do not expect a negative impact. And the reason is we saw this year that there was a strong tailwind coming from digitalization. It was home office. It was home schooling and other things. And I do not believe that this effect will slow down. In Germany, obviously, there is a discussion that every teacher should have or must have a laptop. There are 800,000 teachers in Germany. And I think home schooling is not a German challenge. It's all over the place, including many countries in Europe and probably also North America. So I think from that, I expect some positive effects going into 2021.
The next question is from Florian Treisch, Commerzbank.
Also a follow-up on the LTA discussions, taking an initial kind of agreed volumes with the client as a starting point and looking at the actual volumes, which are simply below what you have initially agreed, is that turning out to be something like a trump card for you in renegotiation, the prolongation of the LTA, i.e., that you are giving the customer a go that he has not to take up the volumes, but as a kind of positive for you, they are willing to prolong LTAs here? And the second angle would be, again also a bit on the recovery path, as you have outlined a bit, you sound quite optimistic about auto recovery, smartphone recovery going into 2021. Assuming or taking your historic comment as a starting point that you always see these turning points with a 1, 2 quarter gap, you should have rather high confidence that H1 should be a growth half year for you, assuming that the STMicros of these world are all right in assuming that a better-than-feared Q2 will be followed by a better-than-feared Q4? Or what can really crash the party for H1 for you? Is it memory? Is it FX? Is it pricing, to get a better feeling what can really go wrong for an H1 recovery for you?
Well, what could go wrong in H1, I think it's for sure not FX. Although also on FX, we might see some fluctuations. But I do not expect any major change. It's, I think, the trade tensions. And I don't know whether trade tension is still the right wording. It's much more about discussions who is the technology leader in the world. This is uncertainty. And then yesterday, many European countries announced the so-called, I would call it, lockdown light. Whether lockdown light is sufficient to counter the development of COVID-19, we will see that in 2 or 3 weeks. I hope it is. This lockdown light from our perspective will have a negative impact on, unfortunately, on many people, restaurants, bars, cinemas and so forth. But it's not related to industrial and to production. And therefore, I do not expect that this lockdown light will have a direct impact on us. And yes, you said listening to my comments, whether it's fair to assume to see a positive development in H1 compared to H2, I think that's a fair assumption. But like I said, there are uncertainties, uncertainties. We might see positive surprises, but I cannot exclude negative ones either. And the LTAs, there is -- like we said before, there is one which is supposed to be renewed. I think this will be -- there will be an LTA which will replace an LTA, which expires towards the end of the year. But while the discussions with the customer is ongoing, we will not talk about too many details of that development. And then -- and as far as I remember, we do not have LTAs, which will run out during the year '21. But we have at least one which expires end of '21.
And the next question is from Gustav Froberg, Berenberg.
I have a couple of quick ones. Firstly, just on capacity, you mentioned that you gave an overview of the shell capacity that you're seeing in 300-millimeter and indicated that there is some oversupply in the market. But it seems like volumes are improving or at least developing quite nicely. And given that capacity expansion does take some time, I'm wondering if it's too early still for you to be talking about potential additions a couple of years down the line or if this is something that you've now started to think a little bit more about.My second question is on cost reduction. You talked about cost reduction being a positive for your gross margin in the quarter. I was just wondering if you could help us understand just how much more potential is there for cost reduction in the business. And then finally, again tied to volumes, given that volume development seems to be a little bit better, are there any positive impulses that you're seeing in ASPs or on pricing of your wafers coming from a pure volume side to maybe offset some of the negatives you're seeing on mix?
Well, let's talk a little bit about the capacity and potential, let's say, greenfield thoughts. You asked whether it's too early to talk about it. Yes, it is too early to talk about it. But this does not mean that it is too early to think about it. Of course, now we see the share capacity, 7.2 million. We see the shipments of 6.6 million, somehow 10% below. And even if you assume that, for whatever reason, the next year's average growth rate will be only 3%, then we are done with the existing capacity by 2024. And beginning 2024 is only 3 years out and a greenfield investment needs some time. On the other hand, we also are under the impression that we need to see some commitment from customers related to such a huge investment. And this is something where, from today's point of view, I would argue, we are not yet there. But this does not exclude that this might change in the next 6 to 9 months. We are certainly, from our perspective, approaching a phase where it's really time to make a decision. And that's true for the industry and that's probably also true for Siltronic. Cost savings, we should really differentiate between the different periods. I think when you look at the track record of Siltronic, we had excellent years with the excellent cost reduction performance. But of course, from year-to-year, it becomes more and more difficult. And in 2020, we did set very, very aggressive targets for productivity increases. And the people in the production lines and also in technology, they really had an outstanding performance. Like I said, the targets are very, very ambitious in basically all lines around the world. We exceed this very aggressive target. I'm pretty convinced that we will have additional cost savings coming out of productivity going into 2021. But it would be unrealistic to expect a comparable performance that we had in this year. And then cost savings, you need also to consider that sometimes we have so-called onetime effects. I think we mentioned in one of our calls that we had a project called carve out to separate completely from Wacker regarding the IT. And there is some significant additional cost, onetime costs, that we had to accept in the year 2020. But these additional onetime costs, of course, they will not be repeated in 2021. So if you compare then total cost this year to total cost next year, we will have a reduction, not as much as we saw this year on productivity. But this onetime event from IT cost will be, I would call it, significant.
And the next question is from Jürgen Wagner, MainFirst Bank.
Your customer, Samsung, this morning indicated a memory recovery for the first half next year, when they reported. And when would you see that recovery in your numbers, given what you said on Slide 3, that NAND inventory -- raw wafer inventory levels are high? And the second question would be on auto. You indicated 10% used to be the run rate. What was that in Q3?
So first of all, Samsung, of course, we will see it sometimes. When they are positive, at the end of the day, it will be positive for us. Whether it will be the first quarter or the second quarter, we don't know. But we expect with Samsung a positive effect on our memory business. That's for sure. And then on auto, I'm sorry, I didn't get your question.
Yes. You said it used to be 10% of revenues and now you had a negative product mix in Q3. What was the auto share in Q3 in terms of shipments?
Do you want an honest answer? I think so. I don't know. But I would assume that it was significantly down to -- from 10%. I would argue somewhere maybe between 6% and 7%.
And the next question is from Rob Sanders, Deutsche Bank.
I've got a couple of questions. The first one is you mentioned there's a gap of 10% between demand and shell capacity next year in 300-millimeter. I was just wondering what you thought that percentage gap might be in 200 and 150. The second question is given the potential you said for price increases in 2022 and beyond in 300, can you remind us what pricing above current levels you would need to justify commencing a new cleanroom build? And the last question would just be on CapEx into '21 that you've given, is it going to revert back to depreciation?
So yes, pricing, no, we said for brownfield, we need to see 30%. This was reached. This was even overreached. We never made a statement how much do we need to see for greenfield. I know that one competitor once mentioned, I think, double of that, which was something like 50% to 60%. And the only comment that we made was, "This is not a bad figure." But at the end of the day, it's -- we need to see contractual obligations from customers. And specifically, when you talk about greenfield, greenfield is a little bit different than brownfield. Because brownfield, you add a certain amount -- you only add depreciation based on additional quantities. But in a greenfield investment, you also add depreciation on, let's say, quite a quite big number, which is not related to -- directly to capacity because it's building infrastructure and so forth. And therefore, at the beginning, there is significant importance to be loaded. So let's put it that way, so -- and we will -- we have some expectations. But we will not be as transparent as we were on the 30% for brownfield investment. And then you had a question on CapEx. I forgot about that one, sorry for that.
Sorry, yes. So the other two questions were, is CapEx going to revert down to depreciation in '21? And then the last one was about the gap between demand and shell capacity in 200 and 150 in next year.
I think -- well, 150, it's very difficult to say because it's close to impossible to really find out what the installed capacity is in Mainland China. And the installed capacity in Mainland China is in k wafers. But these k wafers do not have the same capability as the ones which are produced by the, let's say, top 5 players. So that's very difficult to say. And 200-millimeter, 200-millimeter, of course, suffered a little bit from industrial development and automotive development this year. But like I said before, I expect some positive development over there. And we will have a good loading in 200-millimeter next year. And this is, I would say, not only true for Siltronic, it's true for the total market. Well -- and then CapEx in '21, we have our Supervisory Board meeting next -- end of November. We are not even yet done preparing the budget that we will present -- or preparing the amount of money that we will present. So this is something that we will make some comments in the next call that we will have, I think, end of January, early February on the preliminaries for 2020.
And the next question is from Constantin Hesse, Jefferies.
So most of them have actually been answered. But if I may just quickly follow up on LTAs, do you see a potential for maybe price increases in the LTAs that you are renegotiating at the end of this year and the end of '21? Second question is maybe a little bit more on the macro side. Where are you in the current discussions with regards to next design rules? And then just to confirm, you said that in terms of the amount of price increases for brownfield, it would have to be around 30% from current levels. And a competitor mentioned that it would have to be double that of the current levels for greenfield. Is that correct?
Well, let me try first to answer the last question on price increases. And I'm sorry, you misunderstood me. We said for brownfield, we need to see 30% price increase. And this was based on the historic data. We said that in -- towards the end of '16 and the base was Q4 '16. So we said we will add brownfield capacity if prices go up by at least 30% based on Q4 '16. And this was something that we saw realized in September, October '17. And this was also the reason for us to announce at that time brownfield investment. Afterwards, we are now -- we talked about additional brownfield investments without disclosing details. And the reason for that was that we were very detailed on the first announcement, but we saw that competitors, they simply do expansions without talking about it. And I'm willing to give you all the information which are helpful for you. But at the same time, I have to realize that everything which goes to you goes directly also to competitors. And we are not the Siltronic information channel for competitive intelligence at our competitors. So -- and price increases, you are referring to the contract that we are just now under negotiation. And during negotiation, we will not talk about details. And I think you will also never hear that one LTA was completed with a customer and we realized price increases. We will talk about on average but not by contract. I hope that you understand this. And then design rules, design rules, I think we are capable of leading edge. We supply leading edge to the logic and the foundry business. We have a significant share in this leading edge. We know that for leading edge, not all suppliers are qualified. We know that, for example, at one player for leading edge, there are 3 qualified suppliers and one is Siltronic. So we have good reasons to believe that we have the capability to deliver leading edge to whoever is asking for it.
And the next question is from Holger Schmidt, Metzler Capital Markets.
I have one question left from my side. A couple of days ago, Intel agreed to sell its non-flash memory business to SK hynix. Do you expect this to have some kind of negative impact on your future wafer volume shipments?
We are pretty sure it will have no impact on us at all. Because when you go into the details, Intel will continue to run the facility in Dalian and Intel will also continue to be in charge of purchasing. So we do not expect any change there.
And the next question is from Martin Jungfleisch, Kepler Cheuvreux.
Just a quick follow-up on the supply-demand balance for the next couple of quarters. I understood that your peer, GlobalWafers, is ramping up some 150,000 wafers per month in their new location. You also commented raw wafer inventory levels remain elevated with the server market also seeing softening. So the question is do you see a risk on 300-millimeter polished pricing in the short term also in non-LTA contracts, given the potentially widening supply-demand balance? Or would you say that, given your comments earlier on the high LTA exposure and potentially even higher LTA demand in 2021, the risk is -- or the risk or exposure is rather limited?
Well, I think the risk is very limited. I'm not the person to talk about the details of GlobalWafers. But if I remember right, when GlobalWafers was announced its investment in 150,000 in Korea, they also said that this is completely covered with additional LTAs. So I do not expect that this ramp, which will happen obviously in Q4 or maybe early Q1, will have an impact on pricing because I think it's related to existing LTAs with fixed prices.
And the next question is from Francois Bouvignies, UBS.
Just one quick follow-up, sorry about that. I just want to talk about the Chinese manufacturing -- I mean, capabilities. Given the trade tension is accelerating probably the surge from the Chinese local capabilities, especially for 200 and 200-millimeter, so two quick question to that. Do you see an acceleration of the investment into this area? And if we assume that there is a shortage, let's say, in 2022, 2023, or you may even mention 2024 for greenfield, is there a possibility that the Chinese local could capture the demand extra by that time? Or is it something you rule out completely?
Well, that's a very interesting question. And future will tell. So -- but from today's perspective, there is still a relatively big gap in capability. Yes, this trade tension, of course, even brought more focus within the country of China to concentrate on be self-sufficient. I think they reach, in the meantime, adequate test wafer level. But test wafer level is something which is around 45-nanometer design rule. I'm not saying that they don't have any capability on 28 or 22. But we didn't see it yet. So -- and I think they -- when you catch up, you might be a little bit faster than in the original development. But there is no step function when you do design rule. I think you need the experience that you learn when you go from 28 to 22 in order to create the capability to go to 16. So from our perspective, it's close to impossible to skip one design rule. And based on that, we do not have any reason to change the wording of what we say since years that we are convinced that the Chinese player will not be a significant player for prime wafer in the next 5 years.
And the next question is again from Amit Harchandani, Citigroup.
Just a couple of follow-ons, please, if I may, Chris. My first question is a simple clarification. You talked about the difference between shell capacity and installed capacity in the industry. And if I remember correctly, in the past, well, at the 2017 Capital Markets Day as well, you had given some data around the shell capacity at Siltronic. Is it fair for me to interpret that even for you, the difference between installed and shell capacity is 10%? Or is it higher or lower? If you can comment on that. And my second question is with regards to EBITDA margin. I mean we saw a peak at slightly more than 40% back in 2018. And what I'm trying to understand is in your view, more on a medium-term basis, what do you think is the right level of EBITDA margin for the business that you run? I understand that near term, there is uncertainty. But if you take a longer-term view, what are the puts and takes? And what should be the right level of EBITDA margin for a business like Siltronic through the cycle?
Well, a very good question. For the second one, I would argue the higher, the better. But we thought in the last process, in the last process, I remember when we went into the first internal thoughts about increasing prices, we were we were not pessimistic, but we were cautious. We were questioning whether 5% or 6% or 7% is the right figure. And at the end of the first contract that we concluded was between 12% and 15%. So a shortage will create possibilities, then don't try any limit. The limit is in the market that you cannot be completely different from the other players. But let's say, in a shortage that you see a 10% price increase on shortage products, it's something that wouldn't surprise me. Now let's come to your question related to shell capacity. And you are perfectly right. Because what I use today as wording is slightly different from what we did in the past. In the past, we always talked about installed capacity. And this time, I didn't talk about installed capacity, I talked about the shipments which were made, which is minimum the installed capacity and the shell capacity. And this was 6.6 million compared to 7.2 million. I do believe that in EPI, the industry is fully utilized. And I do also believe that we are not 100% utilized in polished, which leads to the conclusion if the quantity shipped is 6.6 million, then probably installed capacity is somewhere between the 6.6 million and the 7.2 million, most likely somewhere in the middle. But it's difficult to say because I do believe that not only at Siltronic, also at competitors, there was a negative impact on completing existing projects due to the COVID restrictions. On the other hand, I'm pretty sure that the installed capacity will be at shell capacity level by end of next year.
That's very helpful, Chris. And just so that I understood your answer to the first question, did I understand that you think there's a 10 percentage point more possible? Or sorry if I did not understand the answer to the first question.
Price increases is something to predict is impossible. And the biggest impact on EBITDA is price increases. So we were at 40% in a period of shortage. Today, we are below 30%. But in a shortage, we might go back to 40%. Why not?
And there are currently no further questions. [Operator Instructions] And we've received another question from Holger Schmidt, Metzler Capital Markets.
I have one follow-up with regard to the greenfield investment. If you would decide today for a greenfield investment, when do you think you would be able to ship the first wafers? How long does it take?
I think 3 years is a good assumption.
And then how long would it take to come to a really efficient production process, another 6 months or 9 months?
What does that mean? Is the background of your question how much time does it take to ramp the fab?
Yes.
To ramp the fab, I think it's difficult to ramp the fab with more than 150,000 to 180,000 per year. I think that's difficult. It's not impossible, but it's difficult. But once you have this -- created the shell capacity, adding capacity beyond the first step is a brownfield investment. And so the brown -- in the brownfield, you do it based on customer demand. So therefore, the first investment is you need to do the infrastructure, you need to the building, the cleanroom and then you install, let's say, 20% of the capacity. And this 20% will be very efficient, very competitive based on running costs. But these 20% have to suffer from the full depreciation of the infrastructure measures. But going then to full load depends on customer demand. And when we look at SSW, which is today our most profitable activity in 300 millimeter, we decided to do it in 2006, 2008, the first wafers came out and fully build-out will be done by end of next year. So it was never the target that it will take that long. But taking that long does not mean that it's bad. We had a negative period with disappointing results. At that time, we were using, let's say, I don't know, between 40% and 50% of the shell capacity. But the problem was not using 40% or 50%. The problem was the price. And therefore, it's important not to ramp too fast. As soon as you create capacity, which is not covered from customer demand, there is a risk of a negative impact on pricing. And this is something that you -- that everybody should try to avoid.
And there are no further questions at this point. So I hand it back to the speakers for closing remarks.
Okay. Thank you, operator. So this concludes our Q&A session for today. Thank you for joining us, and we will hope you will join us again on our full year release in March 2021. Goodbye, and stay safe and healthy.
Ladies and gentlemen, thank you for your attendance. This call has been concluded. You may disconnect.