[Foreign Language] Happy New Year, everyone. Welcome to TSMC's Fourth Quarter 2017 Earnings Conference and Conference Call. This is Elizabeth Sun, TSMC's Senior Director of Corporate Communications and your host for today. Today's event is webcast through TSMC's website at www.tsmc.com. [Operator Instructions] As this conference is being viewed by investors around the world, we will conduct this event in English only. The format for today's event will be as follows: first, TSMC's Senior Vice President and Chief Financial Officer, Ms. Lora Ho, will summarize our operations in the fourth quarter 2017, followed by our guidance in the first quarter of 2018. Afterwards, TSMC's 2 Co-CEOs, Dr. Mark Liu and Dr. C.C. Wei; CFO, Ms. Lora Ho; and TSMC's Chairman, Dr. Morris Chang, will jointly provide our key messages. After that, Chairman Chang will host the Q&A. For those participants in the call, if you 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 earnings conference presentation. As usual, I would like to remind everyone that today's discussions may contain forward-looking statements that are subject to significant risks and uncertainties, which could cause the actual results to differ materially from those contained in the forward-looking statements. Please refer to the safe harbor notice that appears on our press release. And now I would like to turn the podium to TSMC's CFO, Ms. Lora Ho, for the summary, operations and current quarter guidance.

Thank you, Elizabeth. First of all, Happy New Year. Good afternoon. Thank you for joining us today. My presentation will start with financial highlights for the fourth quarter and a recap of 2017 for the whole year. Then after that, I will provide a guidance for the first quarter. Fourth quarter revenue increased 10.1% sequentially to TWD 278 billion, mainly driven by major mobile product launches and continuing demand for cryptocurrency mining. Gross margin increased 0.1 percentage point sequentially to 50.0% as the benefit from higher capacity utilization and the inventory valuation adjustment helped offset the impact from continued 10-nanometer margin dilution and NT dollar appreciation. Total operating expenses increased by TWD 1.3 billion but, thanks to operating leverage, only represented 10.4% of revenue versus 10.9% in the prior quarter. And operating margin increased by 0.3 percentage point sequentially to reach 39.2%. Overall, our fourth quarter EPS reached TWD 3.83, and ROE was 26.9% for the quarter. Now let's take a look at wafer revenue contribution by application. During the fourth quarter, Communication and Computer increased 20% and 8% from the prior quarter, respectively; while Consumer and Industrial/Standard decreased by 38% and 4%, respectively. On a full year basis, Computer and Industrial/Standard increased 25% and 14% year-over-year, respectively; while Communication and Consumer decreased 2% and 10%, respectively. Now let's take a look at the revenue by technology. 10-nanometer process technology continue to ramp strongly, accounted for 25% of total wafer revenue in the fourth quarter. The combined 16/20 contribution was 20% of total wafer revenue. Advanced technologies, meaning 28-nanometer and below, accounted for 63% of total wafer revenue, up from 57% in the third quarter. On a full year basis, 10-nanometer contribution reached 10% of total wafer revenue in 2017. The combined 16 and 20 contribution was 25% of total…

Thank you, Lora. Good afternoon. Firstly, I would like to talk about our -- everyone's foundry strategy. Being everyone's foundry is the strategy TSMC takes by heart. Through our technology and services, we build an Open Innovation Platform, where all innovators in the semiconductor industry can come to realize their innovation and bring their products to life. As our customers continue to innovate, they bring new requirements to us, and we need to continuously develop new capabilities to answer them. In the meantime, they utilize those shared capabilities such as yield improvement, design, utilities, foundation IPs and our large-scale and flexible capacities. In this way, this open innovation ecosystem expands its scale and its value. We do not compete with our customers. We are everyone's foundry. Now on cryptocurrency demand. In the past, TSMC's open innovation ecosystem incubates numerous growth drivers for the semiconductor industry. In the '90s, it was the PC chipsets; then in the early '20s, the graphic processors; in the mid- to late '20s, it was chipset for cellular phone; recently, start 2010, it was for smartphones. Those waves of innovation continuously sprout in our ecosystem and drive the growth of TSMC. Lately, we observed the demand from cryptocurrency mining surged. Cryptocurrency mining requires massive, high-performance and low-power computing. TSMC's advanced technology and productization serves -- services suits it well. We have sized 2018 cryptocurrency mining demand carefully. Since it is still in its early stage of development, it is difficult for us to forecast its demand too far into the future with accuracy. However, as long as the cryptocurrency miners can derive positive returns, demand for TSMC wafers will continue. Furthermore, we are quite certain that deep learning and blockchain technologies, which are the core technology of cryptocurrency mining, will lead to new waves of semiconductor…

Thank you, Mark. Let me talk about the near-term demand and inventory. We concluded our fourth quarter with 10% Q-over-Q growth in U.S. dollar, which was slightly above the guidance we gave 3 months ago despite a continuing inventory reduction by our customer. Fourth quarter's strong result was mainly driven by a major smartphone release, which is based on our 10-nanometer process with InFO and the strong demand in cryptocurrency mining. Concluding 2017, semiconductor, excluding memory, growth was 9%, while foundry grew 8%. TSMC's revenue grew 9% in U.S. dollars and accounted for 56% of the total foundry market segment share. Moving into first quarter 2018, our revenue in U.S. dollar is likely to decline by about 8%, as Lora just have guided. This is mainly due to smartphone seasonality, offset partially by continued demand for cryptocurrency mining. Fabless DOI exiting 2017 was a few days above seasonal level. Moving into 2018, we forecast fabless DOI to trend up but will attract seasonal pattern. Now let me talk about N7 and N7+. So far, we have already taped out 10 customers' product in 7-nanometer. And we are currently qualifying this product in 2 different fabs in preparation for volume production starting in second quarter this year. During first quarter 2018, we expect to tape out another 10 products. In total, we expect to have taped out more than 50 customer products in 7-nanometer by the end of 2018. Major application covered by these tape-outs are in mobile, game, GPU, at PTA, networking and AI. In essence, almost every area that requires high-performance and lower-power consumption will benefit from TSMC's 7-nanometer solution. In terms of performance, power, area density and schedule, we believe our 7-nanometer solution is leading in the foundry industry. We will introduce our N7+ technology later this year.…

I'd like to make -- let me make a comment on CapEx and outlook for future capital intensity. Now let me start with CapEx. Our 2018 CapEx is expected to be between USD 10.5 billion to USD 11 billion, which is essentially flat with our 2017 CapEx. About 73% of the capital budget will be used for capacity buildup for the advanced technologies, mainly 7-nanometer, followed by 5-nanometer. Another 17% of the capital budget will be used for R&D, packing and mask. As I stated in our last investor conference 3 months ago, in order to support our 5% to 10% growth in the next few years, we expect our CapEx in the next few years to stay a few percentage point more than USD 10 billion. This implies that our CapEx in the next few years will remain at a similar level to what we have spent last year and what we expect to spend this year. Although our leading node capital costs continue to increase due to increasing process complexity, we are able to offset its impact to our CapEx by further optimizing our capital planning and productivity improvements. With the flattish CapEx going forward and the revenue CAGR of 5% to 10%, we expect capital intensity, which is the ratio between CapEx and revenue, to gradually decline from the current 30% to 35% level to 20% (sic) [ 25% ] to 30% level in the next few years. However, if any special revenue growth opportunities comes along, we will certainly respond to it with appropriate CapEx. Now our Chairman may comment now.

I will now make some comments on TSMC 2018 outlook. First, some comments on the semiconductor market. We estimate the supply chain inventory at the end of 2017 to be still a few days above the seasonal normal level. We estimate that the total semiconductor market will grow 6% to 8% in 2018, 6% to 8% this year. The semiconductor market, excluding memory, will grow by 5% to 7% this year, and total foundry revenue will grow by 9% to 10% this year. Now for TSMC, we expect 2018 TSMC revenue in U.S. dollars will grow by 10% to 15% over 2017, 10% to 15% in U.S. dollars over 2017. Our revenue seasonality will be similar to 2017, stronger in the second half than in the first half. Our revenue Y-o-Y, year-over-year growth rate now, however, will be stronger in the first half than in the second half. Basically, we expect the first half 2018 revenue growth rate in U.S. dollars will be slightly above 15% over the first half of 2017 and the second half 2018 growth rate to be slightly under 10% over the second half of 2017. All these are in U.S. dollars. So the slightly above 15% growth rate in the first half and the slightly under 10% growth rate in the second half make up the 10% to 15% total yield growth that I spoke about. Is that clear? Good. No worries [ around this ] -- everybody understands, okay. Our strong growth in 2018 is fueled by the growth of 3 of our growth platforms. Those 3 growth platforms that fuel our strong growth this year are the -- are high-performance computing, HPC, IoT and automotive. HPC will grow strongly due to continuing expansion of AI applications in all electronic devices and in continuing demand for cryptocurrency mining, GPU, et cetera. Mobile platform will be flat. And then, for the IoT and automotive platforms, there is considerable strength. You might describe this situation as the acceleration being enormous, but the velocity is still small. For the last 2 years, we've been planning 5% to 10% compounded annual growth rate in U.S. dollar revenue growth in the 2017 to 2021 period. We reaffirm that plan. In 2017, our revenue grew 9%. In 2018, we expect to grow our revenue 10% to 15%. 2015 will be an above-average year.

All right, this concludes our prepared statements. Before we begin the Q&A session, I would like to remind everybody to limit your questions to 2 at a time to allow all participants the opportunity to ask their questions. Questions will be taken both from the floor and from the call. Should you wish to raise your question in Chinese, I will translate it to English before our management answers your question. [Operator Instructions] Now let's begin the Q&A session. All right, first question will be from Goldman Sachs, Donald Lu.

[Foreign Language] My first question is to the Chairman, and very nice to see you here. What is your view on -- in cryptocurrency, which is very -- have a surge in growth in recent months? So that's my first question. The second question is on N7+. I think if I hear correctly, the performance of N7+ is similar to N7. Is that correct on speed, performance and power? Just want to clarify that.

Well, let me answer the first. I think you ought to come along here and repeat -- and I heard that he was asking about cryptocurrency.

Cryptocurrency demand. [ So message ] on cryptocurrency demand question.

The urge to mine cryptocurrency is very strong. The incentive, of course, depends on the price of cryptocurrency. And the price of cryptocurrency is very volatile. But the demand right now or for the last year has been very strong, and we expect it to continue to be strong. However, we have, in Mark's words, sized the demand carefully, we have sized demand carefully. So I believe that we will both satisfy our customers, and we'll not be too optimistic. So you asked the second question on Mark? Oh, C.C., okay.

Okay. Well, let me repeat your question, the N7+'s performance versus N7, right? Okay. N7+ actually is about 10% better performance than N7. And in terms of area, it would be also about 10% smaller. But if you put all together, like a critical layer, that if I use an EUV, you reduce that critical layer number, so we expect, per wafer, you are getting more -- get 10% more dies from N7 as compared with -- N7+ as compared with N7.

Can you explain the last point you have at the -- about the critical layer, you have less layer with N7+?

Yes, because we're using the EUV to replace some of the immersion layer. So I give you one example. One EUV there can replace a 3 piece [indiscernible] -- replace 3 layers of immersion. So in calculation of the defect density, you'll get more dies.

All right. The next question will be coming from Credit Suisse and Randy Abrams.

The first question, I wanted to ask, your growth, both for the industry and TSMC, is above trend. The prior few years were only low single digit for the industry. I'm curious how much you think is from the structural factors, meaning the new growth platforms, and how much you see is coming from cyclical factors we sometimes see in an upturn.

All right, Randy, your question is, with respect to our projected growth of this year, both the semiconductor and TSMC are above the industry average trend. And so the higher number, is it due to structural or due to cyclical factors?

By structural, you mean permanent?

Or I'd say a new trend -- based on the growth platforms, a new trend to higher growth.

Say it again?

By structural, a new trend to higher growth, say, from these drivers like IoT, high-performance computing.

Well, it has always been our intention to be fully competitive, therefore, the least we want to do is to maintain our market share. And now, as I said a little earlier, the total foundry -- all the foundries in the world, total foundry revenue will grow 9% to 10% this year. And I expect TSMC to grow revenue 10% to 15%. So that means perhaps a small gain in market share. Now what do we do to earn that? Now then I have to go back to the offset strengths we have, yes. We lead in technology, we lead in manufacturing capability, and I believe we lead in customers' trust.

Okay. The second question I wanted to ask was, this year, you already generated excess cash flow after paying the dividend. And you're talking about lower capital intensity the next few years. So if you could give a view now on outlook for dividend or broader payout, just with this cash generation, if you could start to accelerate what we've seen in the past couple years.

Lora will...

Randy, we had a very good free cash flow last year, and we also believe the free cash flow is going to be good in the future years, with what you just said. So we will stick on our plan to gradually increase our cash dividend every year. For next year, I think announcement will be made in February after the board meeting, yes.

All right. Next question will be coming from Deutsche Bank, Michael Chou.

Regarding the 2018 U.S. dollar sales guidance. So did you have any color for what area actually exceeded your expectations 6 months ago in terms of demand? Because in the past, you mentioned that long-term sales CAGR could be 5% to 10%, right, so is any area demand actually stronger than you expected before or really exceed your internal planning in terms of demand outlook for 2018?

All right. Question from Michael really is that since we have stated our CAGR to be 5% to 10%, and this year, the 10% to 15% is higher than the CAGR of 5% to 10%, so he wants to understand where the extra strength is coming from.

Where is what?

The extra strength. Which area we are doing stronger than...

Well, mainly, it's in the high-performance computing platform. And as I said a little earlier, that is due to the expanding usage of AI devices in all devices. And I also said it's the continuing strong demand of cryptocurrency mining.

For follow-up question, so for AI, if [ 8AI ] do cap -- put the [ category ] smartphone AI into the HPC or are you still putting the smartphone segment.

Since the AI in...

The AI-related application, if it is used in smartphone, do we categorize it in HPC or in smartphone? I believe we put it in smartphones.

Okay. My second question is regarding 7...

Okay, is that -- is it -- is that one -- that's what -- you are asking a second question now or what?

Just a follow-up -- a second question.

Are you asking a second question now?

Yes, second question.

This is your second question, actually, right? So anyway, let's answer that question. Let's answer that question. I think, well, either Mark or C.C., you can take it, yes. Okay, C.C., you want to take it?

All right. The AI in the smartphone will category into the mobile smartphone.

My second question is regarding your 7-nanometer sales portion in Q4 this year, in 2018, 7-nanometer sales portion in Q4 in the 2018, Q4 this year, and the whole year, do you have any forecast?

Well, I can say probably the whole year is roughly the 10% revenue. Okay.

All right, we have to move to the next analyst, which is Morgan Stanley's Charlie Chan.

Mr. Chairman, TSMC is really phenomenal. So my first question is kind of long. It's, again, on cryptocurrency. First of all, can you quantify the cryptocurrency contribution as you did the last quarter? I remember, it was like USD 350 million to USD 400 million now in 3Q. Can you give that data for first quarter, in 1Q? So given the strengths, can TSMC's 16-nanometer fulfill the demand? And will TSMC expand capacity for this cryptocurrency mining? This is the first question.

All right. Charlie's question is, can we disclose the proportion of revenue we get from cryptocurrency mining in the fourth quarter last year and the first quarter this year? And also, since this requires our 16-nanometer capacity, are we able to fulfill the demand, or do we plan to expand capacity?

Stop there.

Okay.

I heard the first part, but what's the second part?

Second part is, will we be adding capacity to satisfy the demand?

To satisfy crypto demand?

Right.

I don't think we want to disclose the revenue amount for crypto. Do we?

No.

No, we don't. Now as to the second question, no, because of the volatility -- as we -- as I said several times, because of volatility of the cryptocurrency demand -- the possible volatility, I should say, we do not -- we are not going to add capacity specifically for crypto. But we look at cryptocurrency demand as a part of our high-performance computing platform demand now. There are other parts of the high-performance computing that are also fairly volatile, though not as volatile as the -- maybe as the cryptocurrency. So we look at the whole thing as a whole, and we are definitely adding capacity for the high-performance computing platform.

So my second question is going to be very brief. It's about your antitrust implication, given TSMC's strong industry position. So how the company is going to deal with this antitrust issue, and will they impact your 2018 operation?

I think Charlie is referring to, recently, the EU Commission is investigating whether or not TSMC has violated any antitrust.

Well, we do not think that we have any problem. However, we'll cooperate with whatever government agency that investigates it. We’ll cooperate fully.

Next question will be coming from Daiwa's Rick Hsu.

Dr. Chang and all the senior management, my first question is, could you give us some -- maybe some ballpark number about the revenue breakdown of your 2018 revenue outlook in terms of the application HPC, IoT, automotive and mobile?

Well, those 4 are our growth platforms. We do have other platforms, which happened to be declining, actually. So giving you the percentage actually doesn't tell you anything, but can we -- Lora or C.C., do we -- yes, I mean, he wants to know the percentage of revenue in HPC, in mobile, in IoT and -- I think maybe you can give them some approximate numbers, yes.

All right. Approximately, actually, the mobile smartphone -- you want the half of what it is...

Mobile, mobile.

Mobile will be about 1/2; and then followed by the HPC, that would be about 25%; and another probably around 10% from automotive; and IoT, and then rest of that. Did that answer your question?

Yes, that's very clear. That's pretty good direction. Then the second question is about the inventory. I know Dr. Chang was talking about your end Q4 inventory was a few days above seasonal. And I'm not sure if you also commented about your inventory view for the first quarter this year is going to be coming down and tracking the seasonal average, or am I missing something?

We exited 4Q '17 with a few days above the seasonality. And then we expect this start to increase again because of our customers start to build for the New Year's. And so the inventory will increase a little bit, but it's still the seasonal subpattern, I'd say, is nothing surprised as compared with the previous years.

Next question will be coming from Credit Lyonnais, Sebastian Hou.

So my first question is on the -- what's your expectation for the growth from your Chinese customers? If I calculate correctly, it seems like your revenue from China has doubled for the -- in the past 3 years. And do you expect such doubling pattern to proceed in the next few years?

So Chinese customers have been growing quite fast. Do we continue to expect the percentage to our revenue will double in the next few years as it has done...

Do we continue to what?

Expect the customers' revenue to double in the next few years.

First, the Chinese customers have grown very fast. I guess -- you want to answer it? I was going to say I think the cryptocurrency thing, which is the volatile one, is the part of the Chinese customers. Yes, it has been growing fast. But you go ahead and answer the rest, okay?

Well, if you ask me whether China customer will grow, will double in the next few years, certainly. I think it has full potential to achieve that. The cryptocurrency is part of it, but even with -- after cryptocurrency cool, another application will continue to sprout in China I think.

You said cool. Don't say cool. It may not cool.

Just a follow up on that, how do you see -- besides cryptocurrency, how do you see the -- also there is a rise in numbers of the Unicorn. The startup in China. They want to design their ASIC or a lot of them are working on AI. And do you also expect that to be an important growth driver for your Chinese revenue in next few years for that doubling pattern?

So Sebastian is wondering if beyond the cryptocurrency mining, there are other IC designers in China that focuses on AI, whether or not these type of applications will drive the doubling growth of the customers.

Well, look. We are everyone's foundry, we are everyone's foundry. And being the technology capacity manufacturing leader, we are really in a very good position. So all right, so if they grow, great. We will grow with them. And we don't particularly care customers in which region grow. So we are trying to region blind, okay, as far as where the growth comes from. We're everyone's foundry, everyone in the world, we're everyone's foundry. And usually, the customer that has innovations -- has innovative designs that grow the fastest. And so I'm sure that the Chinese innovations will grow very fast. But then, the rest of world is not just staying put. They are growing too. I don't know, does that answer your question? Or...

Yes. Second question yes, the second question is on the -- I wonder what's the TSMC expectation and your estimate -- or evaluation on the raw wafer supply shortage. And remember, the same time last year, CFO gave an estimate about the [indiscernible] the price hike will impact the margin -- the profit. I wonder if you have similar numbers you can give this year.

Do you want to answer the question?

The raw wafer has -- sitting in short supply in 2017. We expect the short supply will continue through this year as well. Unfortunately, because we are big purchaser, so we were able to engage some of the vendor in the long-term contract. So we should be safe in terms of supply.

And how about the price increase impact on the profitability?

Yes, the price has indeed increased and has been affecting our margin as well. I think maybe 6 months ago, we have indicated the margin impact for 2017 will be around 0.2 percentage points something. With the continual hike in price and the impact in 2018 will be bigger. So we expect an impact will be maybe half point -- 0.5 to 1 percentage point impact of gross margin. Well, anyway, we'll try to offset that with some of the productivity improvement and other cost reductions.

Is it possible to pass the cost increase to your customer?

No, I don't think so.

Okay. I think it's about time that we go to the call first because there are quite a few analysts queuing on the line. So operator, could you please go to the first caller on the line.

Your first question comes from the line of Mehdi Hosseini from SIG.

My question has to do with the impact of EUV around your capital density. As we look into 2019 and 2020, is EUV actually happening with your capital intensity? Is it helping you with lower capital expenditure? And if not where should we expect to see that? And then I have a followup.

We spent several hundred million EUVs since last year. And going forward this year and next year, the EUV investment will continue to go up. However, the number has been included in our overall CapEx budget, which I was just mentioning, $10 billion to $11 billion. While it is true that the CapEx per 1,000 wafer investment will be higher if you move into the more leading-edge technology, particularly 5-nanometer, but also, the purpose of introduction of EUV to simplify the process. So with the process simplification, therefore, the overall patterning cost will not increase compared to the multi-level patterning or the immersion version. So that was the simple answer to your questions.

And moving on to the packaging, nonwafer revenues in '18, '19 and 2020, how should we think about the diversification, scaling of InFO then also CoWoS [indiscernible] the question is to do with the revenue contribution from [indiscernible] package to InFO and CoWoS in '18, '19 and 2020.

So Mehdi's question is regarding advanced packaging in terms of its revenue contribution to TSMC in 2018 all the way to 2020.

Say it again?

The advanced packaging such as InFO and CoWoS, the contribution to TSMC's revenue in percentage.

Lora, yes.

We talk about the whole assembly and packaging service in TSMC instead of just single aisle InFO or CoWoS, okay. The back end service accounts for about 7% of TSMC revenue...

Several, several...

7%.

7%, 7%.

Of TSMC revenue 2017. With the continued growth in -- particularly InFO and CoWoS going forward, we expect the percentage will go up slightly in out years.

Okay, we should go to the next caller on the line. Operator, please.

Our next question comes from the line of Roland Shu from Citigroup.

Mark, you mentioned the wafer increasing and also [indiscernible] economy. I have [indiscernible] prior year. So for this year [indiscernible] or the increase [indiscernible]?

Roland, I think you are asking if TSMC will continue to see the IDM outsourcing as a trend. And then...

Yes.

And then are you asking some proportion or percentage? No?

Yes, a proportion, yes, [indiscernible] revenue contribution might be, yes.

Revenue contribution or proportion.

Of IBM outsourcing?

Of IBM, yes.

Do we even have a number on that, Lora? The answer to the first question, whether the IBM outsourcing will continue. Yes, I think it will, but on the second part -- second part of the question is a number, [ that indicate ].

We don't have a number, okay.

Yes. Actually, frankly, I would say that almost without exception, every IBM has become fab light, almost without exception. There are exceptions. Yes. And all I'm saying is that those that are already fab light will become fab lighter and lighter.

Okay. Then [indiscernible] market share [indiscernible]. Mark, do you have more color on [indiscernible] credit margin impact the [indiscernible]?

Yes, Mark.

Yes. IDM has not been building new fabs for quite some time. But as technology is progressing, and when they reach to below 90-nanometer and they almost solely depend on the foundry support, and that's what we see the continued trend. And all those are specialty technologies.

But this is something new to TSMC? Or this actually is new to the IDM?

What was the question?

Yes, right. The technologies, are these new to TSMC or new to IDM?

Both, both for -- new to the IDM and also new for TSMC.

Okay. So there's, pretty much, some kind of application for an IOT?

Applications.

Well, mostly in the automotive and IoT, yes, in those 2 categories.

Operator, can we continue to the next caller, please?

Yes. Our next question comes from the line of Steven Pelayo from HSBC.

Just one quick follow-up on crypto. I know you don't want to disclose what it was on the fourth quarter. I just wondered if you could at least disclose what it was for full year 2017? And then maybe just directionally, is this increasing every quarter for you?

All right. Cryptocurrency proportion to TSMC's revenue. Can we disclose what's the impact? What's the proportion in 2017 and what's the direction quarter-over-quarter during this year, whether it is increasing every quarter or increasing in some quarter, decreasing in some quarter?

Well, C.C. or Mark, you want to answer that question?

Well, the cryptocurrency mining, the demand to TSMC. Actually last year, of course, we saw a strong increase in the second half of last year. But after that, I think it's -- keep a little bit flat for this year -- throughout this year.

Okay, fair enough. And then I guess I want to ask about -- a little bit about the 10-nanometer. If you look on the fourth quarter, it looked like if you exclude the 10-nanometer from the rest of the company, roughly 75% of revenue was down by 8% quarter-on-quarter. And think about -- the consumers were offered -- 20-nanometer about 15%. So I guess I'm trying to understand, is that normal seasonality? I really thought we were seeing a lot of strength from an overall centered cycle when we look beyond just the 10-nanometers?

So Steven's curiosity is about our fourth quarter. Since 10-nanometer accounted for 25% of our fourth quarter revenue. If we take that out, the non-10-nanometer business is a decline of 8% quarter-over-quarter. In a seasonally -- supposedly seasonally strong quarter. Is that normal?

Well, to answer that question, you actually -- if you take out the 10-nanometer, that representing a big portion of the smartphone business. So I don't know if you take out the smartphone, which is about a 50% of TSMC's revenue, cannot be, say others is -- continue strong first quarter or something like that. So it's not really appropriate to just take out the one node by talking the whole market segment. I'm talking about a smartphone HPC. In fact, really the HPC increase. But we enter into the first quarter, the smartphone seasonality dropped and HPC continue to be strong.

What do you think 10-nanometer will then be in the first quarter and not be able to [indiscernible] in the first quarter.

He's asking us to give a guidance on the first quarter this year, 10-nanometer contribution.

The first quarter 10-nanometer is decreasing because of, as I said, the seasonality of the smartphone business.

Okay, I think we have answered Steven's questions. And now we are coming back to the floor. First, we are going to ask JPMorgan's Gokul to ask his question.

I had a couple of questions on 7-nanometer. Could you talk a little bit about what is your expected market share if we think about N7 and N7+ over the next couple of years? The second part is, I think Mark mentioned in the last call, more than 50% of the tape-outs for 7 are HPC-related. Could you talk about what could be the size of HPC-related production revenues? Because not every tape-out is equal in volume, now that you're getting closer to production if you could give some color on that?

Could you repeat the question?

Right, Gokul's question is first asking us to quantify over our market share at the 7 and 7+ nanometer. That's his first question for the next few years, market share...

Market share on 7 and the 7+?

7 and 7+, right.

Well, we haven't even started producing 7 here. Is that a question or...

Yes, that's your question, right?

Yes.

We intend, I can only say -- we intend to have a very high market share. At least on 7, which we will start producing...

This year in the end of first half.

June, he means June, the end of first half, Okay, which we'll start producing in June. And we intend to have a very high market share. In fact, I would say 100% close. Close, anyway. Now the 7+ comes, even what about, a year later?

Yes.

A year later. So why don't you ask the question about 7+ a year from now, okay.

And Gokul's second part of the question is, since we have said that more than 50% of the tape-out is for HPC, so he wants to know whether or not it is also revenue percentage. Number of tape-outs HPC is more than 50%, whether or not the revenue contribution will also be more than 50%.

Yes.

Well, I would say initially, actually, smartphone business still occupy the majority of 7-nanometers of business followed by the HPC. But that would be a few quarter away.

So HPC will be more the 2019 kind of...

Yes.

All right. Then there will be a follow-up question from the floor coming from Credit Suisse and Randy Abrams.

Actually wanted to go back to mobile. I think in the remarks earlier, you mentioned it will be flat for the mobile platform this year. Could you talk -- is that your view on the market say for volume or content, maybe a change in that. Or is it a view temporarily there's some market share swings on that front.

You want to answer, C.C?

Yes, let me come in on the mobile -- the smartphone unit first. This year, we expect smartphone unit cost will be in the low single digit. However, the high-end smartphone was decreasing for this year. And middle end and low end will be increasing a few percentage points. So in total, it will be a low single digits growth. And for TSMC that we saw that we combined all the smartphone together, and wafer revenue would be flat as compared with last year.

And the second follow-up, it's more broadly on the margins. You mentioned the wafer price impact. If you could give a broader picture on the other swing factors, a comment on depreciation, the 7-nanometer ramp; and FX, a view on that. And the second part is, is there a new seasonality for margin? It seems like the last year or 2, you've had better margin in first half, but then you have the new technology ramp in the second half. So if you expect that same type of seasonality again?

There are several things that will affect our 2018 margin. Number one is wafer price, we were just talking about that. The other one is the ramping of 7-nanometer. We expect ramping profile for 7-nanometer will be very similar to 10-nanometer at the same time, and that we expect the margin dilution for second half of this year will be 2 to 3 percentage point, which is also similar to 10-nanometer. Other than that, the foreign exchange is -- it can move positively or negatively. It's totally out of our control. I don't know what to say about that. Other than that, we have been doing all our effort trying to plan the capacity better, reduce the costs, improve the productivity, and that would hopefully we can offset some of the potential negative factors. But having said so, as we have said, TSMC's financial goal for revenue on the next few years will be 5% to 10%, and the gross margin will be about 50%. But of course, every quarter can be different. For example, the first quarter, as I just guided, actually included a -- about 2 percentage point inventory variations, which is a positive to first quarter. As utilization move every quarter, that could be a factor, okay.

Depreciation...

Depreciation was $10.5 billion to $11 billion. We expect the depreciation grow year-over-year by mid-teens.

All right, then a follow-up question from Morgan Stanley's Charlie Chan.

So actually, 2 small follow-ups. So first of all on the AI semiconductors. So now those ASIC design customized chip are taking place and to replace the GPU, right? So if that trend continued to happen, do you think they will impact your GPU foundry business? That is the first question. And the second question is regarding the IBM outsourcing. I think the logic chip outsourcing I think is already very clear, right? But for those specialty semiconductor outsourcing like power management IC, does that need some technology or process transfer from your IBM customers to TSMC. How does that work?

First question is with respect to the growth of the ASIC customers, whether or not ASIC will gain more popularity over GPU. And if that is the case, what's going to impact TSMC because we have GPU business.

Why don't you answer that? Mark will answer the first question.

On the ASIC -- AI ASIC, right? I think it -- of course, more -- as more customers get into the AI ASIC, that part of the business will grow. However, it's also very clear that GPU, in some part of the market segment, for example, at least the data center, is very solid. So I see in the future different design, occupy different part of the market, and that is the situation.

And there is a second question.

Second question is the outsourcing from IDM beyond logic such as specialty technology, do we see -- how do we work with IDM's on those specialty technology when we phased in their technology...

I think you answered that question earlier, I think. While you -- while that was the point earlier or somebody...

Okay, both. I think we have a very deep specialty technology development. Some of the IDM just use our own technology, particularly moving into the finite nodes. But they are in older or -- older nodes definitely, as their demand increases, we are also doing this phasing, not transfer, phasing. That means using our technology base to stimulate their technology to do the foundry services for them.

Okay, follow-up question from the floor, and that will be coming from Deutsche Bank, Michael Chou.

First question is regarding your 16-nanometer [indiscernible]. Is that right, your 16-nanometer revenue could be up year-on-year in 2018 or do you have any color for that?

16-nanometer?

Yes.

It's a little bit increasing.

Okay, one follow-up question is, you mentioned cryptocurrency, right, so do you think in 2019, most of the cryptocurrency demand will shift to 7-nanometer or do you think that people will still try to use [indiscernible] node in the future?

Yes, a lot of applications from this cryptocurrencies mining; Bitcoin, Litecoin, they are all different kind of applications. So they're using a lot of technologies. But most of them are advanced technology from 7, 10, 16, 12, it's all...

Okay, second question is, do you think that your 28-nanometer sales will be up year-on-year this year or be flat?

28-nanometer?

Yes.

This year, it will be a little bit decreasing.

Okay, we still have quite a few analysts waiting on the queue on the call. So I think we really need to go back to the call. Operator, could you please get to the next caller on the line?

Next question comes from the line of Bill Lu from UBS.

I've had a chance to visit a few TSMC's customers recently especially in the areas of HPC and cryptocurrency. What I seem to be hearing is that foundry suppliers that should be at the leading edge is getting quite tight. Now your guidance for the full year is obviously very good, but that 10% to 15% is that constrained by supply or the net environment?

Bill, your background noise is actually quite strong, so let me see if I understand your question. You are asking whether or not 2018's 10% to 15% growth, are we constrained by capacity. Is that your question on the leading edge?

Yes, does that reflect the net environment or is that constrained by capacity?

What was the question again?

Whether or not our growth this year is constrained by capacity.

Is what?

Constrained, limited by capacity.

Huh?

[Foreign Language]

Oh the question is, do we have -- is our 10% to 15% growth being limited by our capacity. No, no, it's not. Is it?

A little bit.

A little bit? A little bit, yes. A little bit.

Yes.

But...

So the follow up is, is your situation where you may raise capacity, increase CapEx a little bit, when would that be?

Please repeat your question again.

I'm sorry about the background noise. I'm just wondering, given that it is somewhat capacity-constrained, what do you have to see to increase your capacity and CapEx?

Whether or not we will increase CapEx this year.

No, we have already given our guidance. Lora has already given her guidance on CapEx. And that is what we think we will spend. It's what -- you said 10?

10.5

$10 billion to $10.5 billion to $11 billion.

$10 billion to $10.5 billion to $11 billion, yes.

Bill, so we have answered your questions, right?

Yes. My second question is on 28-nanometers. Now it looks like it's all a bit weaker in the short term. TSMC I think has been saying pretty consistently that the 28 demand is going to be long-lasting. I'm wondering if you still feel that. Or any changes to the long-term outlook for 28.

So as we said that we expect 28-nanometer to decrease a little bit this year, but at the same time, we also said the demand for 28-nanometer is long-term, it's lasting. So is this decline a temporary situation?

Judging from the tape-out activities, yes, we expect this decreasing is just a short-term phenomena. In the long term, we hope that the business will grow as the tape-out has shown.

So judging by the tape-out activities, what do you think the 28 demand will inflect?

28 will be flat? You mean...

No, when is the inflection point for 28-nanometer demand going to start going up again?

When will it go up again, when will 28-nanometer business to go up again.

We certainly hope as soon as possible.

Look, there is the 22 you know, and 22 is made on the same capacity as 28, and the 22 is growing.

Yes, in the next few years, the 22 will play the more important role.

So the -- the answer to the question, when will 28 go up again, my answer is that when the 22 grows sufficiently, and we expect it will.

Okay, I think we have answered Bill's question. Let's move on to the next caller on the line. Operator, please.

Yes, next question comes from the line of Patrick Ho Liao from Macquarie.

I have only one question. Will 7-nanometer yield rates stop from a higher base since more than 90% of the equipment is compatible with 10-nanometer?

Did you hear the question? Would you please repeat?

Yes, 7-nanometer yield rate, Patrick is asking whether 7-nanometer yield rate will start from a higher base because there's 90% equipment incumbent with the 10-nanometer.

Our progress in the 7 nanometer's yield definitely is a little bit better than 10-nanometer at the same period of time.

Better. It's better.

Okay, now I know there are still quite a few hands on the floor. So we are coming back to the floor. That will be Goldman Sachs, Donald Lu.

Yes, my first question is on Moore's law. I think 5-nanometer is around the corner and in the back, so to speak. How about the 3-nanometer? And how about 2.5, if it stops at some point, what would TSMC do? You have such a technology lead scaling. How you build the next technology barrier? The second question is on industry consolidator. I think chairman has commented before, but just want to hear it again, I guess. If just assuming Broadcom and Qualcomm merge at some point, you're going to have a huge customer. How you deal with this kind of huge customer?

Will you repeat the whole thing because...

Right. First question is with respect to Moore's law. Donald is asking us...

Moore's law?

Moore's law, and that what is the status of 3-nanometer and what will happen to TSMC if Moore's law stops at some point in the future.

Let's answer that question first. Mark?

I think C.C. mentioned, 7-nanometer already in the fab. So the original 7-nanometer team, most of them is starting work on 3-nanometer already. So -- but the company has a pipeline of technology development. We also have a big team of pathfinding, pathfinding that is developed technology to see 3 and/or possibly beyond. And we also have a small research team that are exploring further outing of technologies. So our R&D is organizing in full spectrum, exploring to the future technologies also.

Can you comment on the progress at the 3? Is this feasible and...

Progress on the 3-nanometer path find. Well, go ahead. It's positive.

It's possible. I think from our program lead. So we have a monthly review and including Chairman also. So it is real.

Look, I review it every month. I review all the advanced technologies, 7 and 3. And EUV and the CoWoS, InFO, every month. And on -- specifically on the 3-nanometer. The program manager has become increasingly positive. It is still pathfinding. So when it first started up, what about more than a year ago, there was a question of whether it was even feasible. Now he has become increasingly positive that it is feasible. And in fact he is now working on concrete ways. Well, that's the pathfinding. I mean he is defining the path. That's a fair assessment. Yes, yes. And...

Second part of the question is industry consolidation. How would TSMC deal with big customer?

I just refuse to comment on that because I think you mentioned a pair. And both of them are very good customers of ours. So -- and we stay neutral. And I'm -- yes, okay.

Follow-up question. Yes, follow up question from Credit Lyonnais, Sebastian Hou.

My first follow-up is on the 8-inch revenue. It was growing year-over-year in 2017. But I remember a year ago -- initial last year was guiding -- the company was guiding my decline. So I was wondering what has changed. What's the change that lead to the upside and what's your outlook for 2018 on 8-inch?

Okay, Sebastian's question is a year ago, we were guiding 8-inch revenue to decline. But then in reality, that revenue actually went up. So what happened and what's the outlook for us this year?

Do you -- does Lora want to answer -- no, C.C., you can answer it. You have picked up the microphone already so...

I picked up the microphone to give it to Lora.

Okay, yes.

It is indeed that our 8-inch demand is very strong, and I think it's mainly from these IoT and automotive stuff, right. So in this year, we still believe the 8-inch revenue will continue to grow.

Thank you. Second part of my follow-up is the -- is a little bit different questions. Not on the silicon. But I wonder if TSMC's -- what is TSMC's strategy on compound semiconductors. It seems like there are more applications down the road like [ party buys ], [ LuxVue ] and RF, microLED, et cetera. So I was wondering what's the TSMC strategy on this because we haven't heard about this before?

We did not specify so many compounded semiconductors at development but we did say that gallium nitride is a one that we are developing. And for the power management IC, and for power devices also, that we are developing it for the -- working with the customer. And today, the business is still very small but we saw a high potential out of that.

So when do you expect this potential to be realized in the revenue?

It is starting from this year, but probably in the 2019, 2020 you will see kind of reasonable amount coming out. It's in the 6-inch wafer right now.

All right. Due to the consideration of time, I think we will stop taking analyst questions. But Chairman still have a few remarks to make.

I just want to say that even though I will continue to be the chairman until June 5 this year, June 5 this year, but this is the last time that I plan in this conference. And for the last 2 years now, I have appeared only once a year in January. In 2016, I appeared once in January. And 2017, again, I appeared only once in January. And so this year, this will be the last time. And I'll be -- I really have spent many years with some of you, many years, more than 20 years. Although I think most of you probably haven't attended this particular conference that long. But I've been here almost 30 years I think, yes. And I enjoyed it, and I think that we all -- at least I hope that -- I had a good time. I hope that you had a good time, too. And I will miss you, and thank you very, very much. Thank you.

All right, thank you, Chairman. And before we conclude today's conference, please be advised that the replay of the conference will be accessible within 3 hours from now. Transcript will become available 24 hours from now, both of which will be available through TSMC's website at www.tsmc.com. Thank you for joining us today. We hope you will join us again next quarter. Goodbye, and have a good day.