Welcome to TSMC's third quarter 2012 earnings conference and conference call. This is Elizabeth Sun, TSMC's Director of Corporate Communications, and your host for today. Starting last quarter, we have combined the quarterly earnings conference with the conference call, and the event is webcast live via TSMC's website at www.tsmc.com. If you are joining us through the conference call, your dial-in lines are in listen-only mode. As this conference is being viewed by investors around the world, we will conduct the event in English only. The format for today's event will be as follows. First, TSMC's Senior Vice President and CFO, Ms. Lora Ho, will summarize our operations in the third quarter and give you our guidance for the next quarter. Afterwards, TSMC's Chairman and CEO, Dr. Morris Chang, will provide his general remark and a couple of key messages. We will open the floor to questions.
For those participants on the call, if you do not yet have a copy of the press release, you may download it now from TSMC's website at www.tsmc.com. Please also download the summary slides in relation to today's earnings conference presentation. Before we begin, I would like to remind everybody that today's discussions may contain forward-looking statements that are subject to significant risks and uncertainties, which could cause actual results to differ materially from those contained in the forward-looking statements. Please refer to the safe harbor notice that appears in our press release. Now, I would like to turn the podium to TSMC CFO, Ms. Lora Ho.
Good afternoon, good evening, and good morning, everyone. Thank you for joining us this afternoon. Today, my presentation will start with the financial highlights for the third quarter, and followed by the outlook of the fourth quarter. I am pleased to report a record quarter for TSMC in both revenue and net income, thanks to customers' strong demand from mobile computing applications and our leadership in technology. Third quarter revenue increased 33% year-over-year, and 10% sequentially, to reach TWD 141 billion, exceeding our guidance. The better than expected revenue was due to higher shipments to support China smartphone demand, and higher mask and backend revenue. On the profitability side, third quarter gross margin was 48.8%. Operating margin was 37.2%. Both are better than our guidance. The higher margin was due to higher utilization, and a better product mix, and a higher yield.
Total operating expenses dollars increased slightly in the third quarter. However, as a percent of revenue decreased to 11.6% on a larger revenue base. Overall, our third quarter EPS was TWD 1.9. ROE for the single quarter was 30.3%. Let's move to revenue by product segment. During the third quarter, wafer demand for mobile computing devices continued to be strong, leading to double-digit growth in communication and industrial related segments. While demand for computer and consumer related products were relatively soft during this quarter. We are happy to see the good progress of our 28 nanometer. 28 nanometer revenue and shipments more than doubled during the third quarter due to solid customer demand and excellent execution. The contribution to total wafer revenue has increased from 7% in the second quarter to 13% in the third quarter.
We expect 28 nanometer revenue will exceed 20% of our total wafer revenue in the fourth quarter, and will be more than 10% for the whole year. Taking a look at the balance sheet. Cash and marketable securities ended the quarter at NT$148 billion, down NT$40 billion from the second quarter. Current liability decreased by NT$93 billion, mainly due to the payment of NT$78 billion in cash dividend in July. Taking advantage of the low interest environment, we raised NT$40.6 billion in corporate bonds at an average interest rate of 1.33%. Our total long-term debt, interest-bearing debt, has increased to NT$78 billion. On the cash flow side, we generated NT$77 billion from operation in the third quarter, invested NT$78 billion in capital expenditure, paid NT$78 billion for dividend, and raised NT$41 billion through corporate bonds.
Overall, our cash balance decreased NT$40 billion to NT$139 billion at the end of the third quarter. Due to more capital expenditure, the free cash flow in the third quarter was a negative NT$1.7 billion. Let me make some comments on our capacity plan. As we continue adding capacity for 28 nanometer, our total capacity has increased 5% to 3.8 million wafers in the third quarter. We expect 28 nanometer will be fully utilized. For the full year, our 12-inch capacity is expected to increase 21% in 2012, and total annual capacity for the company will increase 14% to reach 15 million 8-inch equivalent wafers. Regarding capital expenditure for this year, up to the third quarter, we have spent US$6.2 billion, representing 75% of our total year budget. I have finished my report on the financial highlights. Let me turn to the fourth quarter outlook.
Based on our current expectation and the forecast exchange rate of 29.47, we expect our revenue to be between NT$129 billion and NT$131 billion. In terms of margins, we expect the fourth quarter growth margin to be between 45%-47%, and operating margin to be between 33%-35%. This concludes my remarks. Let me turn the podium to our chairman and CEO, Dr. Morris Chang.
Good afternoon, ladies and gentlemen. Third quarter, we had a record quarter in both revenue and net income, as Lora has already reported. Fourth quarter, as guided, is going to be a modest dip, about 8% in revenue. We also think that this will be followed by another modest dip in the first quarter, and we expect a rebound in the second quarter. All this we actually predicted three months ago. The fourth quarter dip and the first quarter dip is caused by supply chain inventory adjustment. Right now, the days of inventory, DOI, is about 13 days above seasonal. We expect that it will be adjusted to seven days above seasonal at the end of the year. Then it will become normal in the second quarter of next year. That is consistent with our prediction of two dips caused by supply chain inventory adjustment.
Both our third quarter and our fourth quarter are better than our expectations three months ago. Third quarter, we did exceed guidance, even though fourth quarter will have a dip, as we forecast now, but it's a dip from a higher level than we forecasted three months ago. Why? We think the reason is the strength of mobile product demand. Mobile IC demand is indeed very strong, stronger than we expected even three months ago. TSMC is foundry leader in mobile IC. Together with our partners, we are the technology leader in mobile IC. This is a leadership that we and our partners will continue, will maintain. TSMC value added increases as smartphones get smarter and feature phones get smart. We expect mobile products to fuel TSMC growth for a number of years. Next, I will talk about a few product segments. 28 nanometer.
In the third quarter, it was 13% of our revenue. In the fourth quarter, it will be higher than 20% of our revenue. For the whole year, next year, we expect it to be more than 30% of our revenue. Yields have continued to improve, the gross margin of the 28 nanometer will be in the low 40s, low 40% in the fourth quarter. In 2013, it will be at corporate average. Another product segment, CMOS image sensor. We have made significant improvements in optical performance, dual cameras and higher resolution sensors have been introduced in smartphones and tablets as an example of value-added increase in mobile products. Another example of value-added increase in mobile products, TSMC value-added increase in mobile products, that's the fingerprint authentication and Near Field Communication. 20 SoC, 20 nanometer SoC.
Our 112 megabit SRAM yield has progressed significantly, we are now accepting customers' test chips. We have already accepted several customers' test chips and are accepting more. 16 FinFET, we plan to accept test chips in first quarter, next quarter, first quarter '13, Q1 '13, next quarter. In June of next year, we plan to have the first CyberShuttle. In November of next year, we plan to start risk production. This is a somewhat faster cadence than our previous generations. One big reason for that is that basically 20 and 14 are quite alike. The interconnects are very alike. Of course, 16 is FinFET. That's the big difference. Other than that, there are many similar aspects. So we will be able to introduce 16 quicker, a lot faster than normally we introduce a new generation.
Another point that's worthy of note, along with our development of 20 SoC and 16 FinFET, that's that our Open Innovation Platform, OIP, which is our design ecosystem, it is becoming a more important competitive advantage as the technology advances to 20 nanometer. At 20 nanometer, our competitive advantage in our design ecosystem is greater than in 20 nanometer or earlier generation. Our customers rely on us for design exploration, solution development, and design validation. Next item I want to report is that we did invest in ASML to the tune of approximately EUR 1.1 billion. Of the EUR 1.1 billion, a quarter is in R&D sharing, three-quarters are investment in their stocks. The R&D sharing will be spread over a five-year period, starting in 2013.
This investment is important to us because of EUV and 450-millimeter lithography, both of which are important to us at 10 nanometers and beyond. Next item I want to report is that we have purchased 14 hectares of land in Zhunan. It's about 20 minutes of driving from the science park in Hsinchu, so it's really very nearby. The reason it has to be nearby is that we're using the land for our new advanced R&D fab. That R&D fab is going to be for 450 millimeters and seven nanometers development. Next, we expect 2013, 2014, 2015, and 2016, those four years, to be either growth years or strong growth years. We believe that the strategy we adopted a few years ago of vastly expanding our R&D and our capital investment is beginning to pay off. Actually, you can already see it from the last couple years' results.
More is yet to come. The best is yet to come, and we'll continue the R&D expansion. Now, on capital expenditure, I will make separate comments. Capital expenditure will be about TWD 8.3 billion this year, which is right in the middle of the range that we guided a few months ago. We said TWD 8 billion to TWD 8.5 billion, and it now appears that it will be about TWD 8.3 billion this year. Of course, the resulting cash flow stream from this capital expenditure is going to be much greater than the capital investment. I will give you a ballpark number for next year's capital expenditure now. It will be the same ballpark as this year's capital expenditure. Same ballpark. Just to preempt the question of how big is your ballpark. Well, my answer is it's neither very small nor very big, okay? It's in the same ballpark.
I think that you can use this common sense. I'm a very commonsensical person. I'm not trying to play any tricks on you. We plan to do more borrowing via corporate bonds. Actually, our net cash flow from operations in the three years, 2012, 2013, 2014, I'm talking about a three-year period. Our net cash flow from operations in those three years is sufficient to support our CapEx in those same three years. I'm going to let that sink in a little bit because that defines another boundary for our CapEx. Why do we borrow money? We borrow money to pay dividend, which we will keep at TWD 3 per share in this period. That concludes my comments. Thank you very much.
Thank you, Chairman. This concludes our prepared statements. Before we begin the Q&A session, I remind everybody, please, to limit the number of questions that you have to two at a time so that other participants have opportunities to ask questions to the management. Questions will be taken both from the floor and from the call. Should you wish to make your question in Chinese, I will translate it to English before our CEO or CFO answers your question. For those of you on the call, if you would like to ask a question, please press the star, then one on your telephone keypad now. Questions will be taken in the order in which they are received. If at any time you would like to remove yourself from the questioning queue, please press the pound or the hash key. Let's begin the Q&A session.
Our first question comes from Citigroup, Roland Shu.
Would you translate that into English, Elizabeth?
Roland's question is about the near-term outlook, where the chairman had once said last quarter that the second quarter next year we will see a rebound. He would like to know a little bit more color in terms of the strength of that rebound for the second quarter next year.
Well, it is like my comment earlier that the third quarter and fourth quarter are actually better than we forecast three months ago, even though we're still forecasting a dip in the fourth quarter. It's better than we forecast three months ago because we are dipping from a higher level now. To your current question, I think that your present question is, I said three months ago that the second quarter will be a strong rebound. You are a very alert person. You had noticed that I did not use the word strong this time. Is that the origin of the question? Well, my answer, however, is that I did not use the word strong because we are now talking about a higher level in the first quarter. It will be a rebound.
In fact, in some people's eyes, I think it will be viewed as a strong rebound. I decided not to be too aggressive on these sort of things. I say it will be a rebound. We're still talking. It's still two quarters away, frankly. Even though I think our visibility is, in general, quite good. Two quarters away, we can't be sure exactly what the strength of the rebound is. I know, I am 100% certain that it will be a pretty good rebound.
Thank you. I think basically you mean that you have a rebound from a relatively higher base than what you thought three months ago.
Higher revenue base than we predicted-
Yeah
Yeah, that was my point.
Okay. Thank you. I think my second question is regarding the 20 nanometer and the 16 nanometer. A lot of your clients talk about the 20 nanometer cost, the cost saving for chipmaker cost, actually is still not reached the economy scale or reached the sweet spot. Actually, Broadcom also had the same comment during yesterday's earning conference. My question is, since a lot of the customer actually, they are not afraid to get into a 20 nanometer so far, however, listen to your comment, you are very aggressive. You are talking about 20 nanometer and even 16 nanometer in FinFET. My question is this 20 nanometer or 16 nanometer driven by customer specific demand, or is it just overall demand? Thank you.
Well, on 20 nanometers performance, I want to give you the correct numbers that our technical people have given to me. The performance gain from 28 to 20, that's your question, right?
From 28 to 20, the performance gain is 15%-20% at the same total power. The power reduction is 20%-25% at the same speed. You got that? 15%-20% performance gain at the same power, and 20%-25% power reduction at the same speed. I know that some customers want even more. One phenomenon that we have faced is that customers are consolidating. The bigger customers are bigger than the other customers, than big customers used to be bigger than other customers. All right?
Recently, I was rereading a passage of Michael Porter's classic competitive strategy just the other night. He said that there are five things that every company has to cope with. One of those things is customers' bargaining power. Another thing is suppliers' bargaining power. I'm glad to report that after looking over those five things and thinking about each of them in relation to us, we have come out, I think, pretty good overall. Your question reminds me of customers' bargaining power and also customers' relative importance.
All right. Here, there may be a larger number of customers that want more performance or less power than there are customers who are already satisfied with them. It turns out that those that are already satisfied with them buy more than the ones that want more. Okay.
Thank you very much.
Yeah. Mm-hmm.
Okay, next question goes to Deutsche Bank, Michael Chou.
Hi, Chairman. When do you expect the mass production for 16 nanometer FinFET? You said it before, it should be second half of 2016. Do you expect a earlier mass production for 16 nanometer, given that you have a earlier risk production of 16 nanometer?
The question is.
Right. Michael's question is, given that we begin risk production of 16 nanometer FinFET at November 2013. All right. Risk production of 16 nanometer starts November 2013. When will we start mass production for 16?
I think it will be approximately a year later.
My second question, sir, could you give any update for your CoWoS? Do you expect broad-based adoption for your CoWoS in 20-nanometer product?
Yes. CoWoS will be used with the 20-nanometer product. In fact, it is already being used with the 28-nanometer product. The revenue is still relatively small. As I said, technically, it is progressing fine. We are trying to reduce the costs more. There will not be significant revenue until 2015, 2016.
Next question goes to Credit Suisse, Randy Abrams.
The question, to follow up on the 20-nanometer and the 16-nanometer, if you could talk about how customers are weighing choosing one versus the other. If you could talk about the benefit. Also, do you expect customers to choose one versus the other, or a number of customers to migrate to both?
We talk to every one of them. Every major one, explain to them painstakingly and exhaustively. The smaller customers, we of course do it in the technical symposium that we hold three or four of every year. Yeah. Is that your question? Do we explain to them the pros and cons of 20 and 16?
More just the few, usually it's two years between each node.
Yeah.
It's one year. Are customers choosing one versus the other? Do you expect customers to actually design year after year for both, and will that add cost?
I think that there will be customers that will go light on one and heavy on the other. There may even be customers that will skip one, skip 20 to get to 16. Right now, we're still in the discussion stage with a number of customers, with a pretty large number of customers. Again, going back to my earlier point, we would not be making this kind of investment if we didn't have some very big customers already in view.
My follow-up question, you mentioned CapEx has been the same ballpark for next year. Could you talk about the financial, how we should factor it in, if CapEx is in that ballpark, just depreciation and expenses next year, reviewing structural profitability in light of that?
Lora, would you answer that?
We have not finalized next year's CapEx, although Chairman talk about will be in the range of the ballpark. You can see this year, the depreciation has increased by about 20%, maybe a little bit more than 20. With the ballpark that Chairman was just describing, I believe the depreciation will also increase at maybe 20% level for next year. I don't want you to be scared by those depreciation increases, because depreciation increase is not a problem, as long as the capacity that we invest is getting utilized. Those products have the very similar or better structural profitability. You will not see the deterioration factor to the overall margin.
It is our strategy to have 100% utilization at the leading edge.
All right. We will now take our next question from the call. Operator, please proceed with the first caller on the line.
Your next question is from Mehdi Hosseini from Susquehanna International.
Thanks for taking my question. First one has to do with your commentary on Q1. You refer to it as a dip. Is the dip similar to Q4 level?
Whether or not the magnitude of the dip in Q1.
Whether or not magnitude of the dip in Q1 will be similar to the dip in Q4.
It will be in the same ballpark. Well, look, Normally, we don't really guide two quarters ahead of us. The last time I did give a rough indication of Q1. I really would hate to make a habit of it, since you asked, now I will honestly, I try to answer you as honestly as I can. I think it's about the same ballpark. What we forecast now to be an 8% dip in the fourth quarter, Q1, I think, will be in the same ballpark. It's really a bit early for us to tell.
And then.
Yeah.
The second question has to do with the 20 nanometer and the tape out. How should we think about the number of tape outs or the level of customer interest as you are installing the pilot line?
The question is how many tape outs on 20 nanometer that we have already received.
On 20, do we have a number? Lora?
I think currently the 20 nanometer tape out is roughly maybe around 50 tape outs, which is maybe one fifth of our 28-nanometer tape out.
The pilot line, should I assume that pilot line will be constructed and ready by summer of next year?
I can't hear you very clearly. Can you repeat that again, please, Mehdi?
Just to be clear, the 20-nanometer pilot line should be constructed and ready to go by the summer of next year. Is that correct?
The 20-nanometer pilot line is Yeah. Lora?
We have started to spend money on 20 since this year. The real production will not start until 2014. We do have an engineering line per se, being built up this year to next year.
Great. Thank you.
Now we're coming back to the floor. The next question goes to Bank of America Merrill Lynch, Dan Heyler.
Thank you, Elizabeth. Thank you, Chairman and Lora, for the great introduction, comments, and color on your products. I wanted to follow up on a question I had last quarter, a bit on the idea of perhaps building more of a focused fab approach for some of your larger customers, versus the very diversified fabs that you currently run. As you look at some of the largest mobile chip users in the world, you look at their product cycles can be quite volatile, with some huge swings on new product cycles back and forth. I'm wondering, as you manage those fabs, should we anticipate perhaps more volatility in those fabs versus your very diversified fabs? You're getting new business, so perhaps it's good that you're generating more revenue, albeit perhaps with more volatility.
Do you think you can achieve kind of the same level of returns on a focused fab versus a broad-based diversified fab? Thank you.
Well, to summarize Dan's very long question is the comment Chairman made last quarter about dedicated fabs for customers. Chairman made the dedicated fab comments. Dan's question is, given single product life cycle, product life cycle could be more volatile than a diversified pool of products. If we have fabs that are dedicated to one or two single products, we have to be able to manage higher volatility. Chairman thinks that the volatility will be about the same. Volatility. More volatile.
Specifically big new product cycles that
Right
take two quarters to ramp and then go away. Thank you.
Well, all right. Let me try to figure out what you really want to ask. You appear to be concerned that if we dedicate the fabs to a certain customer, then that customer may leave the fab because he migrates to a more advanced technology. Is that right? Mm-hmm.
Yeah. A group of products that are dedicated to a fab. Indeed. Thanks.
I think it's very important for us to be able to convert to a new technology, to convert the fab to a new technology with as little loss as possible. I think that's an art that the DRAM companies actually have learned. We have also done some of that, we are going to do the same thing the DRAM companies have done in the past. We're trying to make the product, the equipment, as commonly usable as early as possible. Of course, we do have additional advantage because we do have a very large second wave of users. We have always had that. In the past, the second wave users have always begun to take over the fabs that the first-wave users leave behind, and we'll continue to have that.
I do want to clarify this business of dedicating fabs. I don't think that one should take that literally in many cases. There may be, or there will be very rare cases when we would dedicate something to a customer.
Okay.
That's not going to be a very common practice.
Mm-hmm. I guess I'll take a follow-up to that. I guess the distinction being that your second wave adopters in the past were quite mainstream and quite broad-based. If you're pushing the frontier technology quite a bit ahead, I'm wondering if there's a concern that may be N-minus-one group may come later, and hence, perhaps the returns on the most advanced-
You're asking how in the second wave of-
Whether the second wave adoption is sufficient, as in the past, to generate the higher returns.
Generate higher returns? Well Yeah. We generate a lot of returns from second wave and third wave and fourth wave. I mean it. Fourth and fifth wave. Yeah. Look, we're still running Fab 2, and it's full more frequently than it's not.
Okay.
By the way, Fab 2, in case you have forgotten, is 60-inch, half micron. Half micron.
Okay. You anticipate no change in the timeline between the early adopters and the second wave of adoption, even though you're putting your technology very leading edge, that second wave will continue to be quite strong, even with concentrated customer base and even with a very mobile focus? Okay. Thank you very much.
All right. Next question goes to Goldman Sachs, Donald Lu.
Chang, Lora, I have two questions. First is on the structural profitability. You commented earlier that 28 nanometer margin will improve in the first half next year. If we assume the structural profitability, i.e., gross margin will reach 50% at 100% utilization target next year?
Well, we hope that it will be 50%, but 50% is something that we have very rarely done, frankly. To assume that the corporate average is Well, I think the 100% utilization is 50%. Yeah, right. We're talking about corporate average next year. Next year, as I already said, we're going to start off with a not very good quarter. Next year's corporate average utilization is not going to be 100%.
Sure. I meant in assuming the utilization is 50%, in Q3, I think utilization is a little bit over 100%, growth margin was 49 and 48.6. Next year will be improving some.
Your question is about 28 nanometer, right? Is that right?
Yes. I mean, 28 would be the key-
Yeah
factor impacting the-
Well, actually, I really have a simple answer for you. If you're asking about our total, the whole company's structural profitability, a very simple answer is, it is a challenge every year, but we have so far managed to at least maintain it. We actually have eked out a little progress in the last few years. You can tell from the results, we have eked out a little progress, but it's a challenge every year. We consider it to be our major challenge, a major task. We are going to at least maintain this structural profitability. It cannot be allowed to deteriorate.
A follow-up question on this is, after 20 nanometer, presumably we will do double patterning, the cost will increase. If that adds 14 nanometer, probably even more it will be. Who is going to pay the bill going forward? If TSMC wants to maintain profitable profit, will someone else have to pay for it, or are you going to-
We reduce the cost. I don't think it's a question of who pays. It's never a zero-sum game. We reduce the cost. As you may or may not have noticed, that our average rate of decline of price has slowed down in the last three years. It's still going down, but the rate of decline has slowed down. I really don't want to tell you any more.
Okay.
Yeah.
I think we will go back to the call and have the next question come from the call. Operator, please proceed.
The next question is from Brett Simpson from Arete Research. Please go.
Thanks very much. I had a question for Lora, just on the gross margin guidance for Q4. Can you maybe talk a little bit about how you see depreciation moving Q on Q, and whether there's any changes ahead in other manufacturing costs per wafer? The guidance seems to imply there's some savings in cost of sales, so I just wanted to understand that more. Maybe the second question for Dr. Chang. Regarding the trend towards CoWoS and TSV, when we think about your wireless customers moving to these new packaging technologies, how do you engage with memory as part of this roadmap? Is there a need for a more strategic partnership with DRAM makers? Given there's so few mobile DRAM makers out there, how do you really manage this with your leading-edge wireless customers going forward? Thanks very much.
Brett, I think your first question is for Lora, and your question is with respect to depreciation quarter-over-quarter increase and how that is impacting the cost. Is that correct?
Yeah, also whether there's any changes in other manufacturing costs per wafer.
All right. How we can change the other manufacturing costs for the wafer manufacturing cost.
Other manufacturing costs.
Yes. Okay. Second question, Brett, I think you need to clarify. You asked chairman about TSV, and you're talking about-
Yes
3D IC packaging solution using TSV and with a memory partner, right?
Yes, exactly.
Okay, we'll have Lora answer your question first.
Okay, your first question asking about quarter-over-quarter change of depreciation.
Yes.
Now, we're in the third quarter now. The third quarter depreciation for the whole quarter is around 35 billion TWD. We estimate the next quarter, fourth quarter depreciation will be a little more than 36 billion TWD. Each quarter will increase slightly. You also asked about how do we reduce the non-depreciation cost.
Yes.
To be frank with you, we have take a significant effort in the past two months, trying to reduce the all other costs, including the material purchase price, the indirect material used for the wafer, productivity and so on and so forth. That has been actually one of the continual effort in TSMC, especially when we see the continual two quarter in the downturn, the management team has put a lot of effort trying to reduce that. So far, we have been quite successful in doing that.
Great. Thank you.
Brett, could you please repeat your second question a little bit slow?
Okay. When we think about the move to 3D packaging, particularly from a wireless perspective, how does TSMC engage with memory as part of this roadmap? Is there a need for a more strategic partnership with mobile DRAM makers, and how do you manage this with leading-edge wireless customers?
Okay. Your question is, when we move to 3D IC packaging, how do we engage with memory makers to form a strategic alliance, and how we manage that?
Yes.
Is that right?
That's right.
Okay.
Yes.
We have been in active collaboration with two memory makers. Hynix and Micron. Actually, we started discussions with Elpida, then they went bankrupt. We have been in active collaboration and discussion with Hynix and Micron.
Now we are coming back to the floor, the next question goes to You are in Daiwa now, right? Daiwa's Eric Chen.
Daiwa. Actually, my first question probably goes to Lora. Once you talk about the depreciation expenses for next year, probably jump by 20% year-over-year. What kind of CapEx you assume? It's a flat CapEx trend or up the CapEx? How many percent?
You're asking the question I cannot answer. It's in the ballpark of what the Chairman just mentioned.
I see. Okay.
Well, of course, this is very preliminary because we are still in 2012. The number may change, so I just give you a range based on a range of bulk, is in the 20% range. I can only say that. It can be a little bit higher, it can be a little bit less.
Okay. Let's talk about consensus of the CapEx, for the TSMC next year, probably around $10 billion. If that's the case, let's talk about all the EBITDA for next year, I guess probably around $11 billion-$12 billion. We look at the cash dividend, we probably have to give like $2.6 billion. If that's the case, can I assume that you are going to borrow like $2 billion US dollars as a corporate bond?
Chairman just mentioned we borrow to pay dividends. Every year, we pay TWD 2.6 billion dividend, so three years is 2.6 times three. We are planning to do that much of borrowing in three-year timeframe.
Are you talking about the borrowing for the coming three years?
Including this year.
Thank you. My second question, probably I need the answer from the Chairman. If I'm right, I remember you talked about 2013 to 2016 will be very strong. Strong for the semiconductor, strong for the.
Either strong, those four years will be either growth or strong growth.
Okay, what kind of strong growth you talk about?
You tell me.
Okay. I will say 20%.
20% is strong. Yeah. Okay.
Oh.
Yeah. I more or less agree with that.
Okay.
Those four years will be either growth or strong. Yeah.
Okay, I see. Actually, I would like to get an idea, what's your logic behind? What kind of product, what kind of trend you see the TSMC on the very good position and to catch up this kind of high growth?
We're talking about mobile products.
Okay.
Basically, I think it should be clear by now that our emphasis is on mobile product market segment and also leading-edge technology. Whatever goes with leading-edge technology like 3D IC and whatnot.
Okay. I see. Okay. Thank you.
Next question goes to JP Morgan, Rick Hsu.
Thank you. Hi, Chairman. Just one question from me. Can you talk a little bit about next year's outlook on a macro basis? For example, your macro economy forecast and the global semiconductor industry forecast, global foundry forecast, and also how does it perform relative to the global foundry average?
This year, the semiconductor actually is a negative growth, about 2%. Based on what we have seen, the next year forecast, semiconductor will grow about 3% for next year. What's your other question?
The foundry growth, also how TSMC would perform relative to the foundry average.
Okay. We currently expect foundry will grow around 7% in 2013, and TSMC will outgrow foundry in 2013 as well.
Well, semiconductor growth, I know that you are interested in it, and we also keep track of it, but frankly, the total semiconductor growth is becoming less and less relevant to our growth plan. Yeah. Actually, the total foundry growth While still relevant, but also is becoming less and less so. We feel that we are on a level in foundries by ourselves. Thank you.
Next question goes to Barclays, Andrew Lu.
Dr. Chang, Lora. Special congratulations to Dr. Chang, TSMC to get the most important order from your competitor. Some structural question I have.
I'm sorry, I didn't hear you. Come again?
I say congratulations to Dr. Chang and TSMC to get the most important order from your competitor. My first question, in the past 25 years, TSMC keep its value proposition to keep most of the time ROE over 20% and showing maintain good margin. While you take a lower margin order to below the corporate average ROI in the next 1 or 2 years, but actually think that TSMC's strategic position in the next 5 to 10 years, which means some of these orders we might consider short-term, we are giving up some margin. For 5 to 10-year position, we are willing to take this kind of low margin order in the beginning time. That's my first question.
I'm sorry, I didn't really hear. I didn't understand it. Anyways.
Right.
Yeah.
Andrew's comment was that we, in the past, said that we want to keep above 20% ROE. His question is, will we be willing to take lower margin business for a short time, like 2 to 3 years, in order to improve our strategic position in the 5 to 10 years?
Do we take-
Lower margin
low margin products in the first few years in order to improve our later position? Are you asking a philosophical question? My own philosophy? No, I do not do it, because I have to live the next few years first, okay? Yeah. I do not. That's a personal philosophical answer. All right. Mm-hmm.
Thank you. The second question I have, if your mobile customers are not adopting CoWoS in year 2014, because earlier, Dr. Chiang mentioned the CoWoS won't ramp up until year 2015 or 2016 to have a revenue contribution. Are you raising in-house packaging on packaging, or you need to partner with some other packaging house to do the PoP? Thank you.
I think that we are going to be partnering, or we are going to look at both possibilities. Of course, we prefer to do it ourselves. I think that we have become more flexible in partnering with OSAT. Is it OSAT? Yeah.
Yeah, OSAT. Yeah.
I think Dan will have a follow-up question. Dan Heyler from Bank of America Merrill Lynch.
Just very few, very short-
I thought you asked this question.
This is the follow-up.
Very short housekeeping question for Elizabeth. You can take a break, and I'll keep it short. Could you walk through your industrial products? What's in that category? I think I heard you say earlier communications was strong, and I think I heard some of the communications or mobile-related strength was also helping drive industrial too. I wanted to understand, what's in industrial that's growing so strongly? Thanks.
Lora, will you answer what is included in industrial or you
That's growing strongly.
Oh.
Many things included in industrial, such as power IC, MCU, data converter, flash controller, POD, MEMS, and smart card.
I guess you're not putting the power management in communications?
No.
Okay.
This is generic. It's standard.
Okay, got it. The second classification, I think perhaps I wanted to ask on your definition of foundry. Obviously, the lines are getting blurred now. I think most of the capital equipment companies include Samsung in their concept of foundry. I know that it's both an IDM as well as a foundry, as well as an equipment company. So perhaps in your definition going forward, whether or not you could talk about foundry growth incorporating your largest competitor, Samsung, when you're talking about forecast. Is that something that you would entertain? Thanks. You're frequently talking about a semiconductor forecast, which in fact is not relevant to your forecast. What is important is your perception of the foundry growth industry, but your forecasts are excluding Samsung. I'm wondering, going forward, if we could have your forecast of the foundry growth including Samsung. Thank you.
I don't think our forecast excludes Samsung. I think for Samsung's semiconductor part, well, for semiconductor, we even include Samsung's memory. For the semiconductor ex-memory, we include Samsung's System LSI in our forecast.
Thank you.
Is his question whether our foundry forecast includes Samsung?
I think it does.
Yes, we do.
Yeah, we do.
Yeah, it does.
Just Samsung's foundry business or their logic business overall?
Their what?
Because they have-
Their logic business.
Overall. Okay.
Their logic business. Well, I don't think they do any memory foundry, do they?
They have their own products in logic. They also sell products into the merchant market. They also have a dedicated foundry business.
I think our numbers include Samsung's. Our foundry numbers include Samsung's foundry numbers. If they make the products for themselves, then we don't include that.
That's right.
Perfect. Thank you.
Next question goes to Citigroup's Roland Shu.
Thank you. I think my question is for Lora. Lora, can you talk about the photomask business, since that was one of the trends in your 3Q? How about the momentum in 4Q, and what's the application for this new photomask in 3Q? Thank you.
I think this question coming from, we have said third quarter revenue was exceeding guidance partly because we have higher mask revenue. Looking to fourth quarter, mask revenue overall will be less than third quarter, will be a decline in mask. However, the new tape-out mask will not decrease. It's the repeatedly tape-out will reduce a little bit. Mask business has been a very good business for TSMC, and it has very good margin as well.
Another follow-up question coming from the floor that goes to Goldman Sachs, Donald Lu.
I'll take the opportunity to maybe ask a more long-term question. As Moore's Law going forward to 16 nanometer, from your discussion with your customers, do you think the absolute growth of the high-end revenue, the leading-edge revenue, will continue to grow at the same pace because of mobile computing, or we are going to see people are more hesitant to migrate after 16 nanometer? Thank you.
I think yes, the leading-edge revenue for each generation leading edge will continue to grow as Moore's Law progresses. For instance, our 28-nanometer revenue is, while it's leading edge, it is now ahead of our 40 nanometer at the same stage.
I anticipate that 20 nanometer will be the same way. I think that will continue. Of course, nobody asked about how important EUV is, how important 450 millimeters is. They are very important. EUV, I think, is the only economic way of doing, I won't say ten, but surely it will be the only economic way of doing seven. Even at ten, if we have a good EUV, a high throughput EUV, I think that our cost will be in good shape, will be in better shape. Of course, we can use double patterning, triple patterning, quadruple patterning, but those we would like to avoid.
Sure.
To avoid them depends on getting ASML to succeed with their high throughput EUV. Likewise, 450 millimeters, that's going to be another impulse on cost reduction, I think. I think that will cut in at, 450 millimeters will cut in at 7 nanometer. It may cut in earlier. I think Intel is talking about cutting in at 10 nanometer. My feeling is that we're cutting at seven.
Donald? Okay. In the interest of time, I think we will just allow one last caller, and we will give the venue to the call. Operator, please proceed.
The next question is from Steven Pelayo from HSBC.
Great. Thank you. Just a question on 28 nanometer. TSMC is much larger than its competitors. It's going to have 20% of revenues in the fourth quarter from 28 nanometer, when many of your smaller competitors are still struggling to get their first 5%. For now, you're just dominating this space. The competition is trying to get more aggressive there. I'm curious what you think about the total capacity out there and if the competitive environment's going to be more intense, maybe by mid-next year at 28 nanometer. Do you have any thoughts on that?
The total foundry industry capacity on 28? Well, I think it's quite large, but frankly, I know that in oxynitride, which is equivalent to our 28 LP. Our 28 LP is oxynitride. I think that our competitors also have produced pretty good yields, not nearly as good as ours, but good enough for them to sell. I think so. On the high-K metal gate, which is our 28 HP and our 28 HPM, I think that we will be the only one that has effective capacity. Effective capacity, I mean capacity that you can sell, that the customers will accept. I think that we will be the only one on 28 nanometer high-K metal gate for quite a long time. By that, I mean two years, maybe.
Wow, okay.
Yeah.
Then one follow-up question, which is, I'm curious about your customer discussions going on now. The perception is that Samsung's probably going to have some free capacity available, maybe second half of 2013, 2014. Maybe they're starting to engage more of your traditional customer base. Are you hearing that back from your customers, the potential that they may be being offered capacity elsewhere?
Question is, Samsung may have extra or free capacity at 28 nanometer starting 2013 and 2014. Have we heard anything from our customers that Samsung has approached them and want to engage them?
They will use that excess capacity to invade our customers? Is that your question?
That's my concern, yes.
If I was Samsung, I would certainly do it. It's a question of whether we can prevent them from successfully doing it. Certainly, if you want me to predict the outcome, I will predict to you, but you have to wait for a while to see the final outcome. I have no doubt what the final outcome will be.
Steve, this is a very cheerful answer to your question, and this is a very good ending for our today's conference and conference call.
Thank you.
Thank you very much for attending our session, we'll see you next quarter. Thank you and goodbye.