Taiwan Semiconductor Manufacturing Company - Earnings Call - Q4 2011

Transcript

Speaker 9

Ladies and gentlemen, thank you for standing by, and welcome to TSMC's 4Q11 Results Webcast conference call. This conference call is being webcast live via the TSMC website at www.tsmc.com and only in audio mode. Your dial-in mics are also in listen-only mode. I would now like to turn the conference over to Dr. Elizabeth Sun, TSMC's Head of Investor Relations. Thank you. Please go ahead, Dr. Sun.

Speaker 0

Thank you. Good morning and good evening, everyone. Welcome to TSMC's fourth quarter 2011 conference call. Joining us on the call are Dr. Morris Chang, our Chairman and Chief Executive Officer, Ms. Lora Ho, our Senior Vice President and Chief Financial Officer, and Dr. C.C. Wei, TSMC's Senior Vice President of Business Development. The format for today's conference call will be as follows. First, Lora will summarize our operations in the fourth quarter and give you our guidance for the next quarter. Afterwards, TSMC's Chairman Dr. Chang will provide his general remarks on the business outlook and a couple of key messages. We will open the floor to questions. For those participants who do not yet have a copy of the press release, you may download it from TSMC's website at www.tsmc.com. Please also download the summary slides in relation to today's quarterly review presentation.

I would like to remind all listeners that the following 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. Information as to those factors that could cause actual results to differ materially from TSMC's forward-looking statements may be found in TSMC's annual report on Form 20-F filed with the U.S. Securities and Exchange Commission on April 15, 2011, and such other documents as TSMC may file with or submit to the SEC from time to time. Except as required by law, we undertake no obligation to update any forward-looking statement, whether as a result of new information, future events, or otherwise. I would like to turn the call over to Lora.

Speaker 3

Thank you, Elizabeth. Good morning and good evening to everyone. Welcome to our 2011 fourth quarter earnings conference call. Before I start, I want to send my best wishes for the Year of the Dragon, as it is only a few days before the Chinese Lunar New Year. During today's call, I will start with the financial highlights in the fourth quarter, followed by a recap of 2011. Then we will move on to the outlook for the first quarter 2012. All dollar figures are in NT dollars unless otherwise stated. Fourth quarter revenue decreased 1.7% sequentially to $104.7 billion. In U.S. dollar terms, revenue decreased 5.4% compared with the third quarter. Our wafer shipment decreased 8% to 2.9 million 8-inch equivalent wafers in this quarter. This customer inventory adjustment continued to affect TSMC's wafer demand.

By applications, computer, consumer, and industrial-related revenue were more affected and decreased by 13%, 10%, and 14% respectively. Wireless communication increased by 6% thanks to the demand for smartphones. Overall, communication accounted for 53% of our total wafer sales in this quarter, while computer, consumer, and industrial accounted for 20%, 9%, and 18% of our wafer sales respectively. By technology, contribution from 28-nanometer process technology already represented 2% of the total wafer sales. We believe 28-nanometer contribution will be more than 10% in the second half of 2012, also for the whole year. 40-nanometer contribution remained at 27% in the fourth quarter, while 65-nanometer increased 3 percentage points to 30% of total wafer sales. Combined contribution from 65-nanometer and below represented 59% of total wafer sales in this quarter, which is 5 percentage points higher than the previous quarter.

Gross margin was 44.7%, up 2.7 percentage points from the third quarter, mainly attributed to a favorable exchange rate and a higher utilization rate. Operating margin was 31.4%, up 1.7 percentage points. Operating expenses increased 5% from the previous quarter, mainly due to higher opening expense for Fab 15 in Taichung. Overall, our fourth quarter's net margin arrived at 30.2%. EPS was $1.22. During this quarter, we further reduced our inventory level by 2 days to 43 days, and we believe this inventory level is healthy. Cash flow from operations totaled $73 billion, up $18 billion from the third quarter, primarily due to higher other operating sources. Also, our capital expenditure decreased by $8 billion in the quarter. These two factors helped us help our free cash flow increase from $17 billion to $43 billion during the fourth quarter.

Overall, our cash and short-term investments increased $30 billion to $151 billion. Our capital expenditure was $1 billion in the fourth quarter. Total capital expenditures for the whole year were $7.3 billion. Our total capacity slightly declined in the fourth quarter as a result of capacity conversion in 12-inch fabs and annual maintenance in certain 8-inch fabs. Full-year total capacity increased by 17% year-over-year to 13.2 million 8-inch equivalent wafers, while 12-inch wafer capacity increased by 29%. We expect our total capacity to increase by 5% sequentially to 3.6 million 8-inch equivalent wafers in the first quarter of 2012, mainly due to an increase in 28-nanometer capacity. Next, let me give you a recap of our performance in 2011. 2011 revenue was NT$427 billion. Gross margin was 45.4%, operating margin 33.1%, and the EPS was NT$5.18. 2011 was a challenging year for the semiconductor industry.

Despite a difficult macro environment, TSMC still achieved top-line growth. In U.S. dollar terms, our revenue increased 9% year-over-year to $14.5 billion in 2011, continually outperforming the foundry industry. Regarding profitability, despite rising depreciation, R&D expense, and unfavorable exchange rates, we still maintained gross margin at above 45% and achieved our long-term financial target of an above 20% ROE. Supporting bystander financial performance, we are committed to invest in R&D for future growth and strive for long-term shareholder value. Now, let's turn to the outlook for the first quarter 2012. Based on the current business expectation and the forecast exchange rate of 30.25, we expect our consolidated revenue in the first quarter to come in between NT$103 billion and NT$105 billion. In terms of margin, we expect our first quarter gross margin to be between 42.5% and 44.5%, operating margin to be between 28.5% and 30.5%.

Our first quarter revenue outlook is similar to revenue in the previous quarter, but the profit margin is slightly lower. The main reason is that our 28-nanometer wafer sales contribution will increase rapidly to around 5% in the first quarter from 2% in the fourth quarter 2011. As margins of any process technology are typically lower than corporate average at the initial stage of volume production, this increase in 28-nanometer contribution will have some temporary negative impact on our margins. This concludes my remarks today. Now, I would like to turn the call over to Dr. Morris Chang, our Chairman and CEO.

Speaker 2

Hi, everyone. First, I'd like to look at 2011 in retrospect. In 2011, the world semiconductor growth was close to zero. The world foundry industry growth was 4%. TSMC growth was 9%. All these figures were in U.S. dollars. TSMC growth was mainly driven by the strength of 40 and 45-nanometer nodes and the emergence of the 28-nanometer node. Revenues from our advanced technology, that is 65-nanometer and below, grew from 46% of total wafer revenue in 2010 to 55% in 2011. Now, a few words on supply chain inventory. The supply chain inventory has been working down throughout the second half of 2011. We expect that the supply chain's DOI in aggregate is about seven days below seasonal at the end of 4Q11 and will continue to be below seasonal throughout 1Q12. Next, I'd like to say a few words about this year, 2012's semiconductor and foundry outlook.

The outlook for 2012's economic condition is for world GDP to grow at 2.4%. That's our estimate. It's higher than some and lower than others, though. We expect the world semiconductor market to grow at about 2%. In the semiconductor market, mobile products will enjoy particularly strong growth. TSMC is well-positioned in the mobile product market with the right technology, and we have the manufacturing capacity. We should perform better than the overall semiconductor and foundry industry this year. I have asked Dr. C.C. Wei, Senior Vice President of Business Development, to give a special presentation on TSMC's positioning in the mobile products today. He will speak after me. TSMC's Q1 revenue is about flat from last year's Q4 but stronger than seasonal. According to our current booking forecast, our second quarter revenue will be stronger than the first quarter.

On our capital expenditures, in order to leverage our leadership position in 28-nanometer and to expand our development work in 20-nanometer and beyond, we expect to spend about $6 billion in capital expenditure. About 80% of the spending is for tools, buildings, and facilities for 28-nanometer and 20-nanometer nodes. 12% of the total spending is for R&D. I will say a few words on the status of our 28-nanometer ramp. Our 28-nanometer entered volume production last year and contributed to 2% of 4Q11's wafer revenue. Defect density and yield performance is ahead of schedule and is better than 40/45-nanometer at the corresponding stage of the ramp-up. We expect the 28-nanometer ramp this year to be fast, and we expect 28-nanometer will contribute more than 10% of total wafer revenue this year.

Our tapeouts on the 28-nanometer, we have so far completed 36 individual tapeouts and have scheduled another 132 individual product tapeouts in 2012. While three versions of the 28-nanometer technology, the LP, the HP, and the HPL, have entered volume production, the fourth version, the HPM, has entered risk production this quarter and is expected to begin volume production in the second half of this year. Now, I'll give you a report on some technology development. First, on 20-nanometers and beyond, our 20-nanometer development is on track to start risk production at the end of this year. We combine our high-performance planar transistors with system integration such as CoWoS and our ecosystem partners.

We think that with the combination of those three, the high-performance planar transistor, the high-density system integration such as CoWoS, that stands for Chip on Wafer on Substrate, by the way, and our ecosystem partners such as ARM and other EDAs, TSMC's 20-nanometer technology will be most competitive in the marketplace. For FinFET, TSMC will offer second-generation FinFET transistors at 14-nanometer nodes with risk productions in 2014. The feasibility has been proven. We're in the process of integrating it into the technology platform. To solidify and enhance our R&D efforts, we have invested a few hundred million dollars of additional equipment and put in hundreds of engineers to start up an R&D process center in order to deliver the 20-nanometer and 40-nanometer prototypes with quick cycle times. I want to say a few words on high voltage.

The fast growth of mobile computing devices has fueled the growth of demand for high-voltage HV technology. The HV technology is needed for power management IC and for display drivers. The mobile computing devices require a smaller footprint, high efficiency with predictable yield and large capacity to support. TSMC's strength of integrating logic and HV fits well in the strength. For power management, I see TSMC's 0.25 micron BCD is in volume production, and customers are designing to 0.18 micron BCD while 0.13 micron BCD is under development. For the display driver, the requirements include high-resolution thinner frame to narrow the edge and faster data rate to refresh the screen. TSMC's 110-nanometer HV and 80-nanometer HV fit well in the strength with the smallest design rules and the largest capacity for the market. We plan to offer 55-nanometer HV next year to further extend our leadership.

Next, chip on wafer on substrate, CoWoS. TSMC offers a robust CoWoS solution to enable future advanced products featuring low power and consumption. The solution integrates advanced silicon technology and interposer technology. We expect CoWoS to start at 28-nanometer and become significant at 20-nanometer and below. We've been collaborating with customers to develop CoWoS products for qualification in 2012 and initial production in 2013. That would be the 28-nanometer version. The CoWoS ecosystem includes memory manufacturers and OSEP partners. Initial applications will be in the areas of high-end GPU, FPGA, and networking. A few words on the status of our solar and solid-state lighting business. This year, we have seen very quick price drops in both solar and solid-state lighting. In this environment, we are proceeding cautiously and hopefully prudently.

We are spending somewhat longer time in getting the technology to an even higher performance than we originally planned before we switch to production. We are still focusing most of our efforts on development, and that will last well into 2012. Our development will last well into 2012. Although we do expect modest revenue in 2012 for both solar and solid-state lighting. Our goal in solar is to produce high-quality solar modules at competitive costs. We have started to move in tools into our new solar fab in Taichung. Recently, we were able to demonstrate a circuit efficiency of 17.12% on a 30 by 30 centimeter cells. To the best of our knowledge, the highest efficiency that has been achieved anywhere is 17.2% for a cell of similar size. As I said, we achieved 17.12%, very close to the highest efficiency that has been achieved.

In solid-state lighting, our goal is to produce emitters at a stable high-quality level. We are making good progress in developing our technology and converting it from R&D to pilot production. We're now approaching the 120 lumen per watt level, also bringing this new activity in the top bracket of technical performance. Now, I would like to turn the talk over to C.C. Wei. Dr. C.C. Wei, TSMC's Senior Vice President for Business Development, and he will talk about mobile computer products. C.C.

Speaker 8

Thank you, sir. Good day, everybody. Let me start with the two most important products in the mobile computing business. That is the smartphone and the tablets. In the year 2003 to the year 2011, smartphones grew from 10 million units to 456 million units. The year 2011 is also the first year that smartphone outnumbered PCs. Looking forward for the next five years, the growth will likely continue with 21% CAGR and reaches about 1 billion units probably in the year 2015, but no later than 2016. In addition to smartphones, the tablets grew from 18 million units in the year 2010 to 60 million units in that year. We also expect that a greater than 30% CAGR for the next five years for the tablets. Both smartphones and tablets are expected to have very strong growth in the future. Equally important is the mobile data traffic.

According to Cisco's estimate, the number of mobile data traffic by smartphones and tablets almost doubled every year from about 91 petabytes per month in the year 2009 to 546 petabytes per month last year. One increase to more than 1,163 petabytes per month this year and more than 2,000 petabytes per month next year. The dramatic increase in data traffic almost shows two things. First, the number of smartphones and tablets will have a very strong growth in the future. Second, higher transmitting speed and larger bandwidth will become necessary. As a result, better semiconductor technology is required. That will lead to my next topic, why TSMC has very strong positioning in the mobile computing market. As I mentioned, in the mobile computing world, device speed is very important, but there's another factor to be considered. That is the power consumption.

Low power is essential because you are determining the smartphone and tablet's use time. In summary, all in all, the speed, the low power that turns out to be the technology semiconductor technology is very, very important. In the last three years, we have increasingly invested in R&D to enhance our progress in the technology. Through working with our customers, TSMC can provide to our customers the leading-edge technology optimized for speed and low power consumption. In addition to processing technology, we also benefit from the ARM architecture, which is famous for the low power operation and has been used by almost all TSMC customers in their products. In fact, it is no surprise that TSMC customers are gaining their market share in mobile computing, and we are growing with them.

Further, needless to say that TSMC and ARM are both cooperating together to catch the great opportunity in this mobile computing era. Next, I want to talk about why TSMC's good positioning helps win the foundry market segment share. We have seen a faster technology migration in the past three years. For example, at the 65-nanometer node, we saw about 35 products at the initial volume ramp, and the number growing to 43 products at the 40/45-nanometer nodes and about 80 products at the 28-nanometer node. Not only has the number of products increased, we also saw the same trend in the volume ramp compared with the same period of time. The faster technology migration actually is mainly driven by the performance need in these three years: higher speeds, low power, everything.

In order to keep up this faster technology migration, that is, shorten the time to market, TSMC customers have found it is beneficial to have earlier and deeper cooperation in technology definition. As a result, we believe this faster technology migration will help TSMC to gain foundry market segment share. Indeed, we have observed foundry market shares gained about 2 points from the year 2008 to 2011 in spite of a new entrance in the same period of time. Thank you. That's.

Speaker 0

This concludes our prepared statements. Operators, please open the floor to questions.

Speaker 9

Thank you. At this time, we will open the floor for questions. If you would like to ask questions, please press star followed by one on your touchtone for now. Questions will be taken in the order in which they were received. If at any time you would like to remove yourself from the questioning queue, please press the pound or hash key. Please limit your questions to two at a time with one follow-up to allow all participants an opportunity to ask questions to the management members. The first question comes from the line of Dan Hale from BofA Securities. Please ask a question.

Speaker 10

Good morning, and thanks for taking my call. I wanted to ask you about the 20-nanometer ramp, which appears to be probably ahead of your previous comments, but it would be above 10% in the second half of this year. You're now indicating 10%, higher than 10% for the full year. What's the status of the high-K metal gate portion of that? Will that start to kick in soon, or would you expect that to be more meaningful in the second half of the year?

Speaker 2

It's going to be more than 10% for the year then. The revenue contribution of the 28-nanometer node is going to be more than 10% for the whole year this year. Roughly, in the first quarter, it will be 4%, 5%, 5%. As you know, we said that in the fourth quarter, it was 2%. This quarter will be 5%, and it will gradually build up so that for the whole year, it will be 10% or a little more than 10%. High-K metal gate, when does it kick in? It has already started to kick in. It's a little bit behind the oxy nitride, but it will build up quite fast. At the end of the year, we expect about half oxy nitride and half high-K metal gate.

Speaker 10

Agreed. I was just trying to get a sense of the dynamics there because how important that high-K metal gate is to achieve a high second-half growth. Is that, are you assuming that that will occur and that that's one of the major drivers to the second-half portion of 2028, if my thinking is correct there?

Speaker 2

Yeah, that's right. We don't really have any doubts that high-K metal gate is going to work. In fact, it's already started.

Speaker 10

Okay. Thank you. On the second question, maybe for C.C. and for Dr. Chang, the collaboration with ARM and TSMC is well known and has been going on for some time. I wanted to get a sense of your thought process in this ecosystem moving into the PC space. Is that something that we would, that ecosystem would be positioned to in the 20-nanometer node with your three initiatives taking place combined? Does that position?

Speaker 2

The collaboration with ARM will continue to the 20-nanometer and beyond in the PC space also. Yeah, I very much expect that.

Speaker 10

I'm just wondering whether you think the performance level would position TSMC and ARM for a PC presence at the 20-nanometer node, or could it come earlier, say, in the 28-nanometer node?

Speaker 2

C.C., you want to answer that?

Speaker 8

Okay, Dan. I cannot say that the high-speed CPU are entering that area, but you know there is some CPU combined with graphics, and we are gaining the market share from there.

Speaker 10

In other words.

Speaker 9

The low-end market?

Speaker 2

Combined the CPU and the graphics, you understand?

Speaker 8

Yes. Actually, I would not say the low-end, but let's say that in the mobile product.

Speaker 10

Okay. Thank you. I'm looking to hear more color on that going forward. That sounds like an interesting initiative. The final very quick question for Lora. If I could get some color on the different fab capacity movements in the first quarter, I don't know if you can pull up that slide. It seems as though some of the 8-inch fab capacity, you've got some fabs declining, fab 2 and 3 declining, fabs 5 and 6 are actually increasing. The only reason why I'm bringing it up is they've been pretty stable over the last couple of quarters, and the first quarter seems to have some capacity movements on the 8-inch. As you look on the 12-inch front, you know the two big increases that you've already talked about in the first quarter at fab 12 and fab 14, I presume, are 20 nanometer.

Finally, the third part of the question is, how much growth this year will you have in 40 nanometer? Because I do understand that you're upgrading some of your 65 nanometer to 40 nanometer. How much growth in 40 will there be as a result of those upgrades? Thank you.

Speaker 7

Okay. You were asking about the fab capacity increase in the first quarter. You just mentioned there are some changes in 8-inch. As I mentioned in my earlier remarks, there were some 8-inch coming down from the annual maintenance. The capacity for those fabs will be slightly smaller than the fourth quarter. The two big 12-inch fabs, Fab 14 and 12, are all undertaking the expansion for 20-nanometer as the majority. Fab 14 has some small increase from 40-nanometer. That's the increase part. What's your second part of the question?

Speaker 10

Will 65-nanometer capacity decline this year, and how much will 40-nanometer increase?

Speaker 7

Oh, I see. 40-nanometer will increase slightly. If you talk about year over year, 40-nanometer increased about 20%. 65, we are migrating 65 to 28. Overall, 65-nanometer capacity will go down.

Speaker 10

Excellent. Thank you.

Speaker 9

Thank you for your question. The next question comes from the line of Mehdi Hosseini from RBC Capital Markets. Please ask a question.

Speaker 10

Thank you. Thank you very much. If I can just follow up on the previous question, did you say that you are migrating 65 into 28-nanometer or 65 into 40-nanometer?

Speaker 7

65 into 28-nanometer.

Speaker 10

Okay. Another question on 20-nanometer. You said you would start risk production by the end of this year. When do you expect the wafer shipment for avenues?

Speaker 2

The 20-nanometer, we will start.

Speaker 7

2014.

Speaker 2

No.

Speaker 10

2013?

Speaker 2

Second half of 2013, yes. Not this year. 2013, second half.

Speaker 10

Okay. What was, so one last question?

Speaker 2

Yeah.

Speaker 10

One last question. How much capital are you allocating for 20-nanometer this year?

Speaker 2

20 nanometer, we're not spending very much tools money, very much equipment money, capital equipment money. However, we are building facilities, building plants in both Taichung and in Sengzhou. We are not spending very much equipment money, production equipment money. We are spending some R&D equipment money on the 20 nanometer.

Speaker 10

Can you give us an initial estimate if you have in terms of how much the cost will increase for 20-nanometer from 28-nanometer equipment cost?

Speaker 2

It's about a ratio of 1.45, I think. Yeah, 1.45 per 1,000 wafers per month capacity. If it costs $1 capital on the 28-nanometer node, it will cost $1.45 on 20. Have I explained myself?

Speaker 10

Okay. Thank you very much. Yes, thank you. Thank you.

Speaker 9

Thank you for your question. The next question comes from the line of Mehdi Hosseini from Susquehanna Financial Group. Hannah, please ask a question.

Speaker 4

Yes. Thanks for taking my question. Going back to the margin commentary on 28-nanometer, when this year would you expect the 28-nanometer gross margin to come up in line with the corporate average? Would there be, beyond that, would margin actually for 28-nanometer increase above corporate average? I have a follow-up.

Speaker 2

It will come very close to corporate average in the fourth quarter this year. I really hope and expect that it will be above corporate average next year.

Speaker 4

Okay. Regarding the CapEx, it is my understanding, based on what I am noticing at equipment vendors, that maybe the majority of the CapEx has already been communicated, POs, CLAs. Would that be fair to say that the CapEx in 2012 is going to be front-loaded, with spending in the first half much higher than the second half?

Speaker 2

No. What are you wanting to answer that question?

Speaker 7

It's only slightly more than 60% in the first half, not too much.

Speaker 4

I'm sorry, more than 60%?

Speaker 2

A little more than 50%. The number I remember is 52% or 53% in the first half and 47% to 48% in the second half.

Speaker 4

If the overall macro picture were to improve, what are the prospects or the probability of increasing CapEx, like let's say by summertime? Is that even under consideration?

Speaker 2

Yes. Yeah. There is some possibility. If you recall, last year in January, a year ago, we guided our corporate spending, I mean, sorry, capital spending last year at $7.8 billion. It ended up $7.3 billion. The year before, that's two years ago, January, we guided our capital spending in 2010 at.

Speaker 7

4.8.

Speaker 2

4.8%. It ended up.

Speaker 7

5.9.

Speaker 2

In 2010, our capital expenditure exceeded our guidance early in the year. In 2011, our actual capital expenditure was less than what we guided. Both years, you know what happened in both years. In 2010, the climate was better than we expected. In 2011, it was the worst that we expected. This year, depending on the macro situation, it could be up or down. That's right.

Speaker 4

Got it. Thank you.

Speaker 9

Thank you for your question. The next question comes from the line of Gokul Hariharan from JPMorgan Chase & Co. Please ask a question.

Speaker 6

Hi. Good evening. Just out of curiosity, with regard to your R&D budget this year, what % will be allocated for the advanced assembly?

Speaker 2

I'm sorry. What's that?

Speaker 7

What percentage of R&D? You mean capital expenditure?

Speaker 6

For this year.

Speaker 2

R&D expense, right?

Speaker 6

R&D operating expense or R&D capital expense?

Speaker 8

Operating expense.

Speaker 7

Oh, operating expense.

Speaker 2

As a % of what?

Speaker 6

The R&D expense allocated for backend.

Speaker 7

Backend. Backend. Backend R&D.

Speaker 2

Backend R&D. I don't have that number, and it's not a major part of our total R&D.

Speaker 6

All right. The other question I have is on the tax rate for marketing purpose. Because the last couple of years, the percentage was pretty low. I just want to know any number you have for marketing purpose.

Speaker 7

You're asking about tax rate?

Speaker 6

Yeah.

Speaker 7

Last year, we were close to 7%. 7%. This year will be 8%.

Speaker 6

Do you mean for this year?

Speaker 7

This year will be around 8%, including the tax cuts for tax credits.

Speaker 6

Okay. All right. Got you. Okay. Thank you very much.

Speaker 9

Thank you for your question. The next question comes from the line of Michael Chou from Deutsche Bank. Please ask a question.

Speaker 6

Oh, good evening. Just a follow-up question for 20-nanometer high-chemical gate. Chairman, do you expect an accelerated adoption of 20-nanometer high-chemical gate in mobile devices in 2013 or second half this year, given that it seems that there's a limited adoption.

Speaker 1

Michael, could you please repeat? Do we expect accelerated, what?

Speaker 9

Adoption of.

Speaker 1

Oh, accelerated adoption of 28-nanometer node high-k metal gate in mobile devices in 2013, given there seems a limited adoption in 2012.

Speaker 2

I don't think the adoption in 2012 is limited. It is, frankly, limited by our capacity. It's limited by our capacity and our ability to ramp up the production. To answer your question, yes, I do expect it will accelerate in 2013, but it will accelerate in 2012 also.

Speaker 1

Thank you. That's a no for your question.

Speaker 9

Thank you for your question. Next, we have a follow-up question from the line of Dan Hale from BofA Securities. Please ask a question.

Speaker 10

Thanks for the follow-up. Dr. Chang, today you had commented on the solar portion of your business, perhaps contributing somewhat to revenue towards the latter part of the year. I was looking through my own notes to see if there are any comments on the LED bit. Do you anticipate some contribution to revenue this year as well as the solar, or is that?

Speaker 2

Yeah, I said both. It will be very modest. It's not nothing, it's modest.

Speaker 10

Okay. Excellent. Another follow-up was in relation to the backend strategy. I know that it's still a work in progress, and you're consulting with your partners and customers on this. I wanted just to clarify, the ecosystem itself, you're working on the CoWoS with your partners, and the substrate attachment part would still be done by your backend partners. Are you 100% sure of that? That's the strategy now, that the backend partners, the OSAT, the ASCs, and the Spil of the world will still be doing the substrate attachment. I presume then also the encapsulation and the packaging and test as well, which is.

Speaker 2

Oh, that Dr. C.C. Wei answered it.

Speaker 8

The substrate will cooperate with OSAT partners. Whether we put the die on substrate, we'll put it by ourselves or we'll put it by the partners. At the beginning, it will be done by TSMC. Finally, I think that we will cooperate with each other.

Speaker 2

Did you get that? He said that.

Speaker 10

Yes, I did.

Speaker 2

Initially, we'll probably do it ourselves, putting the chip on the chip.

Speaker 8

On the interposer and then on the substrate.

Speaker 2

We will always put the chip on the interposer ourselves.

Speaker 8

That's right, yeah.

Speaker 2

That's our business model of CoWoS. We will not use OSAT to put the chip on the interposer. As far as putting the interposer on the substrate is concerned, initially, we plan to do that too. Eventually, I think that's what C.C. said. It will be done by OSAT. Is that right?

Speaker 8

Yes.

Speaker 10

Can you explain why the change? Why would you do it first and then move it to a substrate?

Speaker 2

On the wafer on substrate, why do we do it first?

Speaker 10

Yeah, why?

Speaker 2

You're talking about the wafer on substrate now, right? That's right?

Speaker 10

Correct, yes. Moving that to OSAT.

Speaker 2

Yes.

Speaker 8

Okay. Let me answer the question again. At the beginning, we want to do it by ourselves because we have a concern about the stress, about the device matching, and all kinds of things together. Once we are very confident on doing that, because we have to buy the substrate—I'm sorry, we have to buy the substrate from the OSAT people. We want to do it by ourselves at the beginning due to the concern of mechanical stress, thermal stress, and everything reliability-related. Once we are very familiar and we have confidence, we will work with the OSAT transporter technology. We'll cooperate with them so that in the mass production, they can also handle it.

Speaker 10

That's great. I'm trying to conceptualize this technology as how we should think about it. Is this something that we should think about as keeping Moore's law going in another way? Given that a lot of chips are increasing their memory size, many SoCs, the memory bed is getting larger and die sizes are getting quite large. Is perhaps the longer-term outcome that, you know, moving the memory off chip on an interposer, that this is a meaningful cost reduction for your customers, and therefore, we're creating more value? I think some investors are confused whether this is you moving to the backend and taking away a business that, you know, from a backend process, or is this really essentially a frontend value creation exercise? If you could just conceptualize this for the investors, that would be great. Thanks.

Speaker 4

Do I have to answer?

Speaker 2

Yeah. Okay, Dan. Actually, you're talking about quite a long sentence, but let me answer in a very short one. We put the chip and put the memory on interposer so that we can first move the Moore's law ahead. Secondly, it's for the performance because when you put two together, you can save the power consumption. You eliminate a lot of I/Os, all right? The speed is also being reserved. Because you're limiting the I/Os, you save a lot of power consumption. That's the idea. The cost is lower.

Speaker 10

This is a way for chip companies to effectively lower their die size, but maintain the same performance level. Is that right?

Speaker 8

Yes.

Speaker 10

Thank you. Is it important for you to have a memory partner here? There's been also some arguments that those that have vertical integration and have the memory technology, is that a differentiator for someone that knows memory, or is that not the case?

Speaker 8

Yes, we have partners in the memory field.

Speaker 2

Is it important for us? I think so. Yes. This is Morris. Your question, is it important for us to have a memory partner? Yes, I think so, because we need to put memories on the interposer also.

Speaker 10

Does your competitor have an advantage? I see Samsung with those memory partners.

Speaker 2

I don't know. I don't know about that because we're working with almost every memory manufacturer except what you call our competitor. Every one of them is quite eager to work with us because they know that we are actually their defense against their competitor too.

Speaker 10

Thank you.

Speaker 9

Thank you for the question. The next question comes from the line of Mehdi Hosseini from Susquehanna Financial Group. Hannah, please ask a question.

Speaker 4

Yes. Two follow-up questions. First, on the depreciation, can you remind me, please, the year-over-year growth from 2011 into 2012?

Speaker 7

Okay. Maybe $6 billion CapEx will increase our depreciation based on current estimation, about 20% year over year.

Speaker 4

20%. Okay. One other thing I noticed, the mix of your revenue by Fabless versus IDMs. Back in 2007, IDM accounted for a third of your revenue. Now it has gone down to about 15, 16%. Is that a reflection of the secular change, more of an ARM ecosystem, or does that mean that when IDMs see full utilization rate, they're going to come back to you and it would have additional revenue opportunity for you?

Speaker 2

Let me try this answer. This is Morris. The line between IDM and Fabless is a very, very much blurred one now. This has happened just in the last four or five years. At leading-edge technology, in fact, it started with the 40-nanometer node. There are only, oh, I would say, two IDMs. That's true at the 28-nanometer node also. All the rest at 28, yeah, 28 and the 40, everyone except those two is a Fabless.

Speaker 4

Okay.

Speaker 2

Yes.

Speaker 4

Now, just as a follow-up, when we compare the 28-nanometer transition to prior transitions, let's say 65 to 40, and with IDMs not having any 28-nanometer capacity, does that mean that the pricing power for TSMC is better compared to the previous time? It is 28 is more capital-intensive. That's why your margins are below corporate. Once you get a scale, does that change your pricing power compared to previous transitions?

Speaker 2

The fact that the 28-nanometer node margin right now is below corporate average is a very natural one, I think. It happened every generation, and I expect that it will happen in the future, in future generations, 20-nanometer and 40-nanometer and so on. This is the well-known learning curve effect. When you start making something in production, it's always a pretty high cost. The scale is low, and the yield is generally still on the very sharp upslope. You can't possibly, if you try to charge a price that compensates you for the very inefficient manufacturing at the early stage of a technology, then you will find that nobody's willing to pay it. The fact that the margin is low this year to us, it's something that we accept in every new generation.

As I said earlier, I expect that the margin will close, will become very close to the corporate average by the fourth quarter of this year. I expect it will actually surpass the corporate average next year.

Speaker 4

Sure.

Speaker 2

You also asked a question about pricing.

Speaker 4

Pricing power.

Speaker 2

Yeah. I don't like to use that term because I don't like it. All I would say is that we try to make the pricing.

Speaker 4

Let me simplify for you. Your only other competitor this year is still on a gate-first technology. Even beyond this year, let's say at 20.

Speaker 2

They are our only competitor, I guess so. Yeah. I'm sorry. Yeah. Yeah. Okay. Yeah. Yeah. Yeah.

Speaker 4

Right. The one or two competitors are still at a gate-first technology. As you migrate to 20-nanometer, they still have some catching up to do. I'm just trying to stay rational. You're still our only viable foundry solution, but also, you have to be mindful of your customers' own pricing power. I'm just trying to reconcile if you would use your competitive advantage to ask for higher prices or if overall margins are going to remain close to what we have seen in the prior cycles.

Speaker 2

Our customers are our partners, and we are going to, we're not going to do anything that would destroy that.

Speaker 4

Okay, got it.

Speaker 0

All right. Operator, in the interest of time, we will only allow two more callers' questions. Please go ahead.

Speaker 9

Okay. Next question comes from the line of Steven Tulayu from HSBC. Please ask a question.

Speaker 5

Yes. Great. I want to ask a little bit about the 40-nanometer and 45-nanometer node. Right now, is your visibility suggesting that that continues to grow, or are those people migrating now to 28-nanometer? I guess I'm just trying to understand on a dollar basis. As you look in the first quarter and second quarter, are you pretty confident that 40-nanometer and 45-nanometer grow?

Speaker 2

Yeah. 40-nanometer and 45-nanometer is continuing to grow. The volume this year will be higher than last year.

Speaker 5

Earlier today, you had hinted that once you get up the yield curve, I think you said it takes about three quarters or so, it surpasses the corporate average margin. Is it safe for me to assume that your 40 and 45-nanometer today is surpassing your corporate average margins already?

Speaker 2

Yes. 40-nanometer and 45-nanometer is higher than corporate average.

Speaker 5

Okay. My last question is, you know, a lot of questions are being asked about structural profitability with your depreciation growing, increasing your revenue growth for a couple of years here or so. We keep talking about capital intensity being 1.4 times greater at each node. I know we don't want to call it pricing power, but I guess I want to make sure your customers, your partners really do understand the value you're delivering. Are the ASCs that you're seeing, you know, at this particular stage, whatever it is, 10,000, 20,000 wafers a month or something, the 4 and 20-nanometer are also at similar multiples, where it is 1.4 times greater? I'm just trying to understand the pricing multiples.

Speaker 2

You're saying that you hope our customers understand the value we are offering to them. Is that what you're saying?

Speaker 5

I just want to make sure you're able to price for the value that you're bringing, given that you have to pay so much more for the CapEx.

Speaker 2

We spare no effort in convincing our customer, persuading our customer that we are really giving them big value, big value, even bigger value than he is willing to pay for.

Speaker 5

Is there any way you can quantify kind of in color on what's a 28-nanometer ASC or the multiple of your 40-nanometer, 65-nanometer? If you're blowing it out, just give me a general idea to help me understand, given you have to pay so much more for the CapEx.

Speaker 2

I can't comment. I don't want to comment on 20-nanometer pricing. No. No. Okay. Thank you.

Speaker 5

Take care.

Speaker 9

Thank you. The last question comes from the line of Brad Lin from Arete Research Services LLP. Please ask a question.

Speaker 10

Yeah. Thanks very much. I have a question for Dr. Chang on Samsung. Dr. Chang, Samsung's logic business has grown something like 70% last year, and they're talking about a big CapEx year again in 2012. How do you view their overall manufacturing capability? Given they're the only ARM chip maker with their own leading-edge fabs, how do you think about Samsung? Do you think they are a long-term risk for the foundry sector?

Speaker 2

I think they are a formidable competitor, and I do expect that they will grow their, maybe not their foundry, but they will certainly grow their, what do they call it, system LSI. Yeah. Remember now, it's system LSI has a major role in supplying Samsung itself.

Speaker 9

With logic products, and of course, Samsung's use, their own use of their logic ICs, you know, in smartphones and tablets and even in consumer electronics. The other consumer electronics has been growing, and now foundry, the way we do it, is only a part of the system LSI business. I expect their system LSI business to grow very fast. The numbers you cited are really applied to their system LSI business, and foundry is only a part of it. They are a formidable competitor. They are a formidable competitor in the foundry field. Yeah.

Speaker 0

Great. Thanks. Maura, just a follow-up. I think there's an early question on depreciation. I guess the last few quarters, 300-millimeter shipped wafers have been sequentially pretty flat, and now you're talking about it growing again in the first quarter on a sequential basis. Now that that's kicking up again, can you talk a bit about depreciation over the next few quarters? How that might trend?

Speaker 3

First quarter depreciation will go up, roughly 5%. After the second quarter, it will go up more rapidly because we spend CapEx, slightly more than half of the first half of the year.

Speaker 0

Okay, that's great. Thanks very much.

Speaker 2

All right. This concludes our Q&A session. Thank you for joining us today. We hope you will join us again next quarter. Goodbye.

Speaker 0

Thank you. Before we conclude TSMC's 4Q11 results webcast conference call today, please be advised that the replay of the conference call will only be accessible through TSMC's website at www.tsmc.com. Thank you all, and you may all disconnect.