[Foreign Language] Happy New Year to everyone. Welcome to TSMC's Fourth Quarter 2018 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 live through TSMC's Web Site at www.tsmc.com. If you are joining us through the conference call, your dial-in lines are in listen-only mode. As this conference is being viewed by investors around the world, we will conduct this event in English only. The format for today's event will be as follows: first, TSMC's Senior Vice President and CFO, Ms. Lora Ho, will summarize our operations in the fourth quarter 2018, followed by the guidance for the first quarter of 2019. Afterwards, Ms. Ho and TSMC's CEO, Dr. C.C. Wei, will jointly provide company's key messages. Then TSMC's Chairman, Dr. Mark Liu will host the Q&A session where all three executives will entertain your questions. For those participants on the call, if you do not yet have a copy of the press release, you may download it from TSMC's Web Site 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 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 microphone to TSMC's CFO, Ms. Lora Ho, for the summary of operations and current quarter guidance.

Thank you, Elizabeth. First of all, happy new year, everyone. Thank you for joining us today. My presentation will start with financial highlights as usual for the fourth quarter and a recap of 2018 for the whole year. Then after that, I will provide a guidance for the first quarter. Fourth quarter revenue increased 11.3% sequentially to NT$290 billion as our business benefited from the strong demand for our 7-nanometer technology covering both mobile and high performance computing applications. Gross margin increased 30 basis points sequentially to 47.7% reflecting the absence of the virus incident that occurred in the third quarter, an improvement in back-end profitability, higher capacity utilization and the more favorable foreign exchange rate these factors helped offset the unfavorable technology mix. Total operating expenses increased by NT$2.7 billion, thanks to the operating leverage, only represented 10.6% of the revenue versus 10.8% in the prior quarter. And operating margin increased by 40 basis points sequentially to 37%. Overall, our fourth quarter EPS reached $3.86 and ROE was 24.6%. Now let's take a look at revenue by technology. 7-nanometer process technology continued to ramp strongly and accounted for 23% of wafer revenue in the fourth quarter. 10-nanometer was 6% and the combined 16/20-nanometer contribution accounted for 21%. Advanced technologies, 28-nanometer and below, accounted for 67% of wafer revenue up from 61% in the third quarter. On a full year basis, 7-nanometer contribution reached 9% of wafer revenue in 2018. 10-nanometer was 11% and the combined 16/20-nanometer contributions was 23% of wafer revenue. Advanced technologies accounted for 63% of total wafer revenue up from 58% in 2017. Now let's take a look at revenue contribution by application. During the fourth quarter, communication increased 27%, while computer, consumer and industrial/standard decreased 2%, 35% and 3%, respectively. On a full year basis,…

Thank you, Lora. Good afternoon, ladies and gentlemen. Let me start with our 2019 full year outlook. We forecast a slowdown in global GDP growth from 3.2% in 2018 to 2.6% in 2019 due to the weakening macroeconomic conditions leading to a low growth for the semiconductor industry. For the full year of 2019, we forecast the overall semiconductor market excluding memory will grow 1% while foundry growth will be flat. For TSMC, we estimate our business will grow only slightly in 2019 given the slowing economic environment. Our 2019 business will be supported by the continuing demand for our 7-nanometer where we see strong interests from our customer in high-performance computing, mobile and automotive. Even with a slow year like 2019, we firmly believe AI and 5G are the megatrends that will drive the future semiconductor growth. And we reaffirm our long-term growth projection of 5% to 10% CAGR. Now I will talk about our 5-nanometer status. Our N5 technology development is well on track, with customer tape-out schedule for first half 2019 and volume production ramp in first half 2020. We are already in preparation for N5's ramp. All applications that are using 7-nanometer today will adopt 5-nanometer. In addition, we are expecting the customer product portfolio at N5 and see expanding addressable market opportunities. We expect more applications in HPC to adopt N5. Thus we are confident that N5 will also be a large and long-lasting node for TSMC. Now I will talk about the ramp-up of N7 and N7+. Our 7-nanometer has been a very steep and successful ramp in second half 2018 and we expect continued ramp through 2019. Our customer portfolio is growing stronger while more applications such as HPC and automotive are coming to N7 as well. Customer tape-outs activities at N7 continue…

All right. This concludes our prepared statement before we begin the Q&A session. I would like to remind everybody to limit your questions to two at a time to allow all participants the opportunity to ask questions. Questions will be taken both from the floor and from the call. Should you wish to raise your questions in Chinese, I will translate it to English before our management answers your question. For those of you on the call, if you would like to ask a question please press the '*' then '1' on your telephone keypad. Now questions will be taken in the order in which they were received even at any time you would like to remove yourself from the queue please press the '#' or the hash key. The first question will be coming from Credit Suisse, Randy Abrams.

Yes. Thank you. Thanks for the details. The first question I had was on the guidance both for the sales and the CapEx. You're lowering sales outlook to just slight growth, but only making a small change to CapEx. Could you talk about the assumptions you're making for further market share gains or demand recovery? And also the assumptions for 5-nanometer and EUV to keep the CapEx intact.

Good afternoon. I want to say Happy New Year again. Yes, welcome to the investor conference. Your first question I would like C.C. to answer.

You're asking about -- we've only grow slightly but CapEx did not decrease dramatically. Good question. We now become more conservative on CapEx it's because of we forecast a slower -- a lower 5-nanometer demand at the initial ramp. However, we still have to prepare enough capacity to support our customers. And we believe that today, with more than few hundred million U.S. dollar is to reflect the reality of expecting the high-end smartphone a little bit slower growth. But this year, this year is because of 7-nanometers of capacity has been built and the demand -- actually, we forecast the number of the units of the smartphone, especially high-end smartphone to be negative growth. Although, we are still want to say that. For TSMC, the year-on-year from 2019 compared to 2018, the smartphone -- the mobile business, we still grow slightly on this year regardless if the unit is dropping, okay. And as a result, as I just said, we expect the foundry business will be flat, but TSMC still grows slightly.

Yes. One follow-up to that. And then, I'll ask a second question, for that capital spending change you mentioned on EUV, could you talk now about kind of magnitude and steepness like what type of ramp we should expect next year for that node? And on the CapEx, are you making any consideration with the slower environment to reuse more equipment? So that's a follow-up to the first question.

Yes. Yes, we do expect that we -- in the 5-nanometer we'll use some of the 7-nanometer's equipment. To answer your question more correctly actually why it only drop a little bit. A few hundred million U.S. dollar because of EUV tool actually we already buy a lot. All right. And we expect that will continue to grow also. So the ramp-up of the 5-nanometer now definitely was slower than we expected. But we still think it's a long-lasting node because of all our customers who are using 7-nanometer today are engaged with TSMC today in their product design. So and in addition to that, we also increased our product portfolio, our customers' product portfolio. And that will add something to HPC application. So look, it's not a very promising in this year but we still have the confidence that 5% to 10% CAGR will be sustained.

A change on the second question, a follow-up 1 more. For the 5-nanometer, if you could just talk a little bit more what's driving some of the change in consideration if it's anything about the EUV readiness versus the cost structure, mature 7 versus 5. Maybe if you could go into if now you expect a bigger 7 initially, what the factor for that is?

Let's talk about EUV's readiness. We are progressing really well. So that's why I said the 7+, which is using the EUV will start production in second quarter this year. And our 5-nanometer's technology development, a lot are based on the EUV's progress. And 5-nanometer's technology development is on track. So that's one thing. Why we lower down CapEx is because we expect the high-end smartphones growth will not be as strong as we used to project before. We lower it down a little bit but still, we still think the high-end smartphone will grow because of 5G and AI's application.

I guess the question was more EUV for it to be slower than original maybe what drove the change to the EUV or to the 5-nanometer?

I did not say that EUV will be slower.

5-nanometer.

5-nanometer no. It continues…

The CapEx this year, majority of it has been on 5, not even 3-nanometer. EUV's productivity continued to improve. And that modulate the CapEx we need to invest. So from the number we are very confident that our 5-nanometer will ramp according to our previous plan. However, we do put a more conservative, tighten up our CapEx under those confidence to improve the productivity and make sure the lead time is tightened up enough to minimize the CapEx.

Okay. The lower 5-nanometer initial ramp, could you talk about that like the rationale for that lower initial 5-nanometer ramp?

Okay. Well, I will just say that we expect a higher growth, but right now, we are more conservative on the high-end smartphones, of course.

Thank you.

Next question will be coming from Citigroup's Roland Shu.

Thank you. Happy New Year. First question is, C.C., you said for the mobile business this year, you still expect some growth, but under your overall year, on the whole year revenue growth just slightly. So how about the growth of the other platform products, IoT, HPC and automotive?

Okay, Roland, you asked a good question. Smartphone grows slightly. IoT grows double-digit. Automotive will be flat. HPC, if we're excluding the cryptocurrency mining, HPC also grows slightly. But cryptocurrency is a big drop from 2018 to 2019. So if we put the cryptocurrency together in the HPC, it's a big drop. It's almost a double-digit.

Okay. Thank you. May we have more color for the breakdown for these four platforms, the revenue breakdown in last year in 2018?

This one maybe, Lora, you can directly answer on the number.

For 2018, smartphone roughly accounts for 45% of our wafer revenue. HPC, about 32%, IoT, single-digit 6%, automotive 5%, digital electronics 6% and there's other 5% here.

Okay. Thank you. And may I know how big of the contribution for cryptocurrency last year?

A lot.

That we don't want to specify too much on the segment, particularly it belongs to one of the big customers.

Okay. Thank you. And the other question is Lora said both 7-nanometer utilization will be dropped a lot in first half. But the whole year, the 7-nanometer revenue contribution commented by C.C. will be about 25%. So that means that above 25%. That means the second half the 7-nanometer contribution will be very strong. So may we have more color, what kind of application is driving such a strong 7-nanometer demand?

Okay. A lot of it because of seasonality of the high-end smartphone. So in the second half, we expect to ramp up the high-end smartphone again. The first half is a little bit kind of cyclical in the high-end smartphone.

To be honest, this high-end smartphone drop in the first quarter came a little bit sudden. So the inventory in the supply chain is quite a lot, so that may push the first half drop. But the second half, we expect the new product launch will carry on another wave of ramp.

So Roland, I'll give you more color. How about that? The second half, we already said that we have more customers product portfolio. So that will expand into the high performance computing. Some of the product will enter into production slightly from fourth quarter and extended to the next year.

Okay. Thank you. For the 7-nanometer follow-up, I think 7+ will be also included in the whole 7-nanometer. And N7+ total revenue will still be somewhere around the NT$1 billion.

That's correct.

Okay. Thank you.

7+ is not a major node, but rather it is a technology good for second wave and third wave customers. So it is come up slower when the first wave products or first wave customers ready to convert to another improved version. So it doesn't conflict with our 5. 5 is another major node.

Next question will be coming from Goldman Sachs' Bruce Lu.

I think I will be back. So I want to ask about smartphone semiconductor addressable market. TSMC management used to provide semiconductor for high-end smartphone, medium smartphone and low-end smartphone in the past. Can we ask what kind of semiconductor growth either come in 1 or 2 years with different segments smartphone? And also, what kind of semiconductor growth for smartphone moving from 4G to 5G, especially for the addressable market for TSMC?

You want to specify what is semiconductors' content into the high-end smartphone, actually, you want to narrow down to a specific number just like before, I don't think we could do it today because of there's a lot of application, new application that add into, that's one, such as AI and something. We expect in the future AR/VR will be inside. The second thing is 5G itself will increase a lot in frequency band and so that will add the silicon content inside. I'll give you a kind of feeling that the smartphone unit will drop a few percentage, but TSMC will still grow the smartphone business. So that means there's a lot of increase in that revenue, it's because of silicon content in terms of the smartphone business.

But moving to 5G, as you mentioned that there are a lot of RF content those kind of semiconductor increase, do we see that having any impact to our business?

You have a good question. Yes, moving into 5G, the end of RF become more complicated. And so you have larger area in the RF chip and so we expect silicon content per se that will increase. And so the content in the smartphone will increase.

But can we somehow quantify the number?

Not very.

Let me put some color. I mean, this is based on of course forecast number may not be accurate, but we have a picture about the silicon content supporting C.C. Wei's description. On a high-end smartphone, we see the silicon content increase actually it's increased every year, very fast. On the mid-end, it will increase slightly. On the low-end, it's less predictable, but we take the model currently decrease slightly. So that's the picture. And most of our business in the smartphone belongs to the high-end. So we are going to enjoy the silicon content with the equipment of our technology -- leading-edge technology.

Okay. My second question will be that management just mentioned that 5% revenue growth CAGR will maintain. I think management first mentioned that in 2017. And with like slow growth in 2019, if you do that after math, you now expect that more than double-digit growth for the coming two years?

You can calculate it.

Do I read it correctly?

I want to reaffirm that the 5% to 10% that CAGR will sustain, we have confidence on that and you calculate it.

So the 5-year is 2017 to 2021, right?

Yes.

Thank you.

All right. Next question will be coming from CL Securities, Sebastian Hou.

Thank you. My first question is I would like to get TSMC's brains on the overall semi-cycle inventory. Remember, three months ago, the company talked about that you are not too worried about the inventory situation compared to the 4Q '17. But today, you just talk about that there's a lot of excessive inventory out there. So I was just wondering what has happened in the past three months, is it more due to demand slowdown a lot or it's still more due to there are some hidden inventory now that emerged?

I will say it's more due to the sudden drop in the demand rather than there's some hidden inventory that we cannot see.

Okay. Can you elaborate more on the -- which end application has seen the most significant decline in demand?

High-end smartphone is one thing, and then, others seeing the drop, actually you can imagine that cryptocurrency's mining that dropped quite a lot. And then, related to that might be some of the high-performance computing that you can see from other applications that related to the cryptocurrency mining.

Okay. Got it. Thank you. And in terms of the inventory digestion, I think Lora talked about that. The inventory will go back to a more seasonal average level by mid of this year. So usually, in the down cycle, we usually see the supply chain will tend to over-cut inventory to below normal. So in mid this year that means that we are now ready or is that below normal level can before or after that?

We expect it to be a reasonable level that it can extend the business. And we expect that this kind of a demand will recover gradually from first quarter, second quarter, and then, moving to the second half of this year. So that's why we expect in the middle of this year all the inventory will go back to the reasonable or seasonal level whether it will be much lower or something like that, we did not expect that yet.

Okay. Just one last question from me is, if you compare this down cycle to the past, I know that TSMC has not offered utilization rate details for many years already. But just to give us a rough idea. On the current utilization rate you are seeing now in first half of this year compared to the UTR we've seen in 2015, 2012 or say post a financial crisis, is it likely to be the lowest point that we have seen since then?

As Lora just presented, the utilization impact this quarter is really mostly come from 7-nanometer. Although other nodes utilization is lower, but if you compare with the last downturn, if you take 2009, for example, it's not that low. So 7-nanometer really is the major underutilized. We think it's temporary.

Thank you. Is it fair for us to assume that the UTR is not as slow as in the 2009 yet, but there is already lower than what we have seen in 2011, '12 that cycle in 2015 cycle?

You are right. You are right.

Okay. Thank you.

Next question will be coming from UBS, Bill Lu.

Hi. Thank you very much and Happy New Year. First question is, we've seen a pretty sharp drop-off in demand. And Dr. Wei, I think made a comment that in the future, we will work with customers to plan for capacity differently. Wondering if you could talk a little bit more about that as far as what can be done. Is it better planning; is it diversification; is it the different payment terms? Can we just talk a little bit about what are some of the things?

Well, in fact, this is kind of the way we deal with our customer. So there will be a better planning process. TSMC as well as TSMC's customer, I believe that we are learning a lot during this cycle and then, we will be more prudent or have found ways at TSMC and customer can work together for the better planning of the capacity in the future without -- to summarizing in one wording, we will be more conservative. But again, we will not lose our support to our customer.

Wider product portfolio is another factor you mentioned.

Great. Thank you. Second question is for Lora. Can you talk about the depreciation for this year? And I'm asking that because I feel like TSMC over the last year, year-and-a-half has brought quite a few EUV tools. And I believe you don't depreciate until these tools are in production, right? And so if you look at the CapEx trend versus the depreciation trend, is there a different linearity because all of a sudden the tools you bought, I don't know, 10, 15 tools need to be depreciated.

You're right. We started depreciating the tool when it gets into the productions. The depreciation pattern versus the CapEx changes is not linear, also because whether the CapEx is front-end loaded, back-end loaded also will impact the whole year depreciation. But I can tell you based on the 10 billion to 11 billion CapEx this year, we expect the depreciation will increase by mid-single-digit, which is versus last year was double-digit. So you see, it's all about 10 billion to 11 billion, but depreciation change can be quite different.

Sorry. I think one for the math a little bit. But, what is that assuming for, I guess, the 7+? Because I just want to figure out the next couple of years, as EUV ramps up, are we going to see a pickup or how does that work?

I don't recall there is a sudden increase. It's a more modest increase year-over-year.

Thank you.

All right. I think it's about time that we should go to the line for the questions from analysts waiting in the line. So operator, could you please have the first caller on the line?

Yes. The first question is from the line of Brett Simpson from Arete Research. Please go ahead.

Yes. Thanks very much. I did have a question on inventory levels. Can you help me frame where the industry inventory levels are at the moment, maybe the inventory days versus a quarter ago and what would you say is normal inventory levels in inventory days? Can you look exactly how elevated the inventory levels are at the moment?

Brett, I think you probably talk to too close to the microphone, so I'm not hearing you quite well, but I think you are asking the inventory level. The excess inventory that we see it now versus we saw three months ago. And you like to have a little bit color on that difference. Is that your question?

Yes. Thank you. I think and also just to understand what normal inventory levels are in your view.

Three months ago, we estimate the inventory exiting 2018 is several days above seasonal. Now we look at it more like a 10 days above seasonal.

Okay. Thanks very much. And you said Smartphone seasonality was also going to be weak event in Q1. What are you suggesting the communication division revenue in Q1 versus Q4 revenue?

Well, I think the main reason is because of high-end smartphone seasonality. So that's a -- and typically, Q1 is a low season for the high-end smartphone. But this kind of high inventory is because of a sudden drop from the 4Q last year and extended to the first quarter this year. And that's why the inventory suddenly increase so much. Did that answer your question?

Thank you. Can you perhaps quantify how big a drop in Q1 your Communication revenue will be because of the Smartphone seasonality?

Okay. Lora, would you answer this?

So you're asking the by segment how do we see the changes in first quarter. Communication in the first quarter will drop the most followed by the computer. Consumer actually will grow slightly and Industrial/Standard will grow very significant. So I think the decline in three sectors are all double-digit decline in the first quarter.

Okay. Thank you. And one last question on 7-nanometer, what percent of sales in Q1, do you expect to come from 7-nanometer and you are suggesting a very strong 7-nanometer ramp in second half when the industry gets back to normal inventory levels. Is there a risk that demand for 7-nanometer exceeds supply? And to what extent can you get customers to pull in orders 7-nanometer and start ramping earlier in Q2 of this year?

We're trying to understand your question. So you're asking us about 7-nanometer since we said that there's going to be a strong cutback on 7-nanometer. Are you asking what percentage of 7-nanometer is part of our Q1 revenue, and then, what will be the level in the second half for 7-nanometer?

Let me answer. Let me try to clarify the ramping of 7-nanometer this year. We already ramped the 7-nanometer last year and that's our first generation 7-nanometer. Coming to the first quarter, the 7-nanometer composition is about 21% of our corporate revenue already. Now this year, we are preparing to ramp the second generation 7-nanometer. We don't have a specific name, different customer have different flavors. But overall, it's their second generation products to be launched this year. So or the 7-nanometer number for the second generation will be drastically increased during this ramp in the second half of this year.

I'm sorry, just to clarify. Can you indicate what 7-nanometer as a percent of sales might be in Q1 and does this mark the trough for 7-nanometer in 2019?

Question is what's the percentage of 7-nanometer in our revenue in Q1 and whether or not that percentage is the trough, is the lowest for this year.

Lora, can you answer that?

Okay. Mark just mentioned the 7-nanometer accounts for 21% of our first quarter revenue. We see that percentage of total revenue will continue to grow. So I will say that percentage-wise first quarter will be the trough.

Okay. Thank you very much.

So operator let's go to the next caller on the line. Thank you.

Next question is from the line of Mehdi Hosseini from SIG.

Yes. Thank you for taking my question. I have one follow-up regarding your high-performance compute expectation for 2019. I believe cryptocurrency accounted for only a few of percentage of overall revenues in the first half of '18. So why is there still an overhang on your HPC revenue mix, I believe you said including crypto, it will be -- HPC will be down double digits. And I'm just trying to better understand how it has trended from '18 to '19? And I have a follow-up.

So Mehdi, I think you are asking us about HPC this year, if crypto accounted for a few points of our business in 2018 and how much crypto will be accounting for our business in 2019.

Yes.

Okay. This year, we don't forecast -- we become conservative in forecasting this volatile business. So the cryptocurrency for mining this year is much, much less than last year. And to what percentage, I don't think I can release it right now.

Sure.

This is a conservative estimation of cryptocurrency.

If you had a new HPC customer program, how come those ramps are not enabling you to ramp HPC in 2019 rather than just slightly up?

So Mehdi's comments is that if we have this expanding customer portfolio in HPC, why is it that HPC, excluding crypto mining can only grow slightly this year?

All right, good question. That's because of our customer we expand in our customer portfolio and product portfolio. But their ramp will start from probably in the second half this year with the small-volume and then going to the mass production next year. That's why this year, we saw just slightly increase on the HPC business excluding the cryptocurrency.

Okay. Very clear. Thank you. And my second follow-up has to do with the 5-nanometer ramp. And I'm trying to better understand your visibility and tape-out engagement. Should we expect a 5-nanometer ramp in 2020 to look more like a 7 or is that going to be a steeper? Any color would be appreciated.

Well, let me comment on that. 5-nanometer are ramping in 2020. I would expect that product portfolio is expanding more as compared with the 7-nanometer in 2018. How much of a steeper of that one, it will be similar or probably we're a little bit conservative. But today, we saw the better product portfolio, better customer portfolio, but steeper ramp probably will be similar.

Sure. Should we assume that 5 would become bigger than 7 in 2021? Is that when the true benefits of 5 and the customer diversification that will materialize?

From today's situation and customers we engage with, yes, 2021, 5-nanometer will be bigger than the 7-nanometer's same period of time.

Okay. Thank you very much.

All right. Let's come back to the floor. Next question will be coming from Morgan Stanley's Charlie Chan.

Thank you. So my first question is about smartphone semi-demand for this year because the management attribute the weakness in 1Q to a seasonality. But do you see any structural issues, for example, the lengthier replacement cycle [indiscernible] still have that issue of the smartphone semi-demand? And also I can because high-end smartphone, even it's a higher content it should be contributing more for your revenue. But you're assuming this segment will decline and units will decline, right? So why can you get a number that a full year mobile revenue can still grow slightly?

Good question. Actually, high-end smartphone this year at least for the first half still have some kind of inventory issue. But in the second half, we expect that new model coming out. It's ramping up starting from the second quarter and all the way to the end of this year with a new model coming out. However, the total units as we expected or we forecasted that what we drop slightly. The content will increase more than the unit drop in the percentage-wise. So TSMC still grow slightly on the mobile business. That's why. And your question is --

So I just want to get a confidence why you think second half that recovery could happen, right because it's possible that a new product may disappoint again. And also, yes, if I can get more color, do you think it's kind of a specific brand issue or is that across different brands for the high-end smartphone demand issue?

All right. Second half, that's a good question also. Second half, why we still expect a recovery, because we have a better customer portfolio. That's all I can say.

Okay, fair enough. So market share in there. And also I want to follow-up the previous question regarding HPC, right? So because compared to your previous target to grow 20% year-on-year every year for HPC, I feel like except for crypto there's also some demand issue, right, for example, I'm not sure what you will see the demand in AI cloud and those telecommunication, right, because that will be some demand issue to get only kind of slightly growth for HPC versus your targets of 20%. Can you comment on the demand?

Okay. Let me comment on that. We used to be very optimistic. We are still very optimistic because of the AI and 5G. But AI and 5G; AI is picking up, 5G will start from this second half. Start to install some of the base station, the infrastructure will start to build out. So we did not expect the 5G smartphone will be a big number this year. It will be a big number next year. And all the smartphone we expect that the AI will be included inside with some kind of a content that we did not get the full picture yet but we know is increasing. And you are talking about HPC. We used to say the 20% CAGR. Did we say that?

That is last year and the year before.

No. The last year compared with a year before because there is one uncertain -- not one uncertain, one surprise that's called cryptocurrency mining. It suddenly increased dramatically. And TSMC because of the technology offering, so we have to say that we are the -- we get most of the benefit of the cryptocurrency's mining business. Now, we saw volatility drop. However, without the cryptocurrency, we still see strong momentum on the high-performance computing; one, because we have very competitive, or actually I want to say the industry leading technology which fits the requirement of HPC business; second, again, I would have to say we have a stronger customer portfolio.

Let's look beyond 2019, right. So, for two or three very high growth profile segment, HPC, IoT and auto, which I was very surprised that your company is only flat year-on-year. So for coming in maybe two or three years, what do you think should be the right CAGR for HPC, IoT and auto?

I think the strongest growth is high-performance computing. The second strong growth in terms of the dollar increase is still smartphone, yes. So that's a dollar increase.

But auto includes those kind of [EUV] [ph] -- I'm not sure how you classify the autonomous driving, right? But auto I think it used to be kind of a segment the company expects strong growth. Why this year we don't --

Yes. Auto if you look at the past two years, it's a very strong growth, 27%. But as you know, since last year, suddenly the automotive markets almost stopped growing. Many of our customers say the same thing. Very consistently. And some of them has attributed to the steel and aluminum tariffs. And that is, I think is the structure of the automotive. In terms of the long-term, we still see the innovation of automotive will come out of this.

So yes, I guess one question from a lot of investors. It's about dividend payout expectation right, because comment was steady growth every year. But you have had some over the past two years EPS growth of around 3% per year and this year, I'm not sure, right. It will be flat, right? So does the company think this year you want to increase the dividend dollars again?

Despite the short-term market weakness, if we look at free cash flow that we can generate remain very strong. So we plan to further increase dividend in 2019. We will get the Board approval in February, so will make announcement after that.

All right. Next question will be coming from JPMorgan's Gokul.

Thanks for taking my question and Happy New Year. My first question, could you talk a little bit about any recalibration in the China expansion given what we have seen with cryptocurrency and some the China demand as well? Could you talk about the plans on China expansion? The last time, you mentioned maybe go up to 20k in phase 1 and then potentially expand beyond that. Was there any timeline difference on the Nanjing fab?

The plan did not change, because this is so short-term different kind of market softening. No, it did not change our strategy. The plan continues.

Is there any timing differential in terms of when you get into phase 2 given --

It will depend on the business situation, definitely. But 20,000 wafer per month is a plan, continue to be executed.

Okay. Next question I had was on N5. Just building on some of the questions as well. It is for many customers it's going to be the first node with significant EUV insertion. So could you help us understand what is the feedback you're getting from customers. Are customers planning to stay on at N7 longer even N7 has been successfully proven node and as you mentioned a lot more customers are coming through there, or do we expect to see the same number of tape-outs like 50 tape-outs that you had by the end of 2017. Are we going to be at similar kind of levels as we get into end of this year? Can you give us some idea about how that progress from our breadth of customer base is progressing in N5?

Okay. On the N5 progress, I just reported that our progress is on track and we work with the customers. And you say that whether our customers have worry or you have concern on EUV's readiness or the stability and those kind of things, no, they are happy with us. We worked with supplier, we worked with the customer on both side and our progress is so far, so good. Now whether that they want to move into N5 or stay in the 7, all I say is all the customer who are working with us on N7 are engaging with us at N5. The product, when they want to introduce into the market that is our customer's judgment because N5 did offer very good performance per se as compared with N7. So some of the high-speed computing, some of the high-end smartphone, they still need to go into N5. So you would expect those customer will adopt the N5 immediately that when it is available, okay. So number of the tape-outs, I cannot tell you right now, but I already told you that all the customers are engaging with us.

Okay. Thank you.

Next question will be coming from Credit Suisse with Randy Abrams.

Hi. Thank you. I wanted to follow-up on your guidance for 2019. Last year, you gave a guidance at the beginning of the year for first half year-over-year growth and also the second half year-over-year. So if you could give a view or maybe split how you see in first half versus second half?

Lora, can you give the number?

Second half definitely is better than first half, okay?

That's good, much better.

First quarter of '19 is quite weak, and we think it's going to take another quarter to recover. So from what I have seen now for this year, if you compare year-over-year, our first half may not be higher, maybe worse than last year. But second half definitely will be better.

Okay. A follow-up on the application and inventory issue, could you talk how broad? Because you did flag smartphone as inventory, but could you flag how broad you think that inventory issue is? Will cross applications in the guidance by product had industrial up significantly, consumer up slightly. If you could talk a bit about that factoring in the inventory comments, what was driving those increases?

You mean supply chains inventory?

Yes, because you mentioned it's a broad inventory or you have an inventory issue. How do see it across the other applications and what's driving the segments.

Exiting 2018, I think the inventory increase is broad range, because of the macroeconomic uncertainty in this trade tension. Pretty much put people really on hold very careful, so there's used to be across the board.

Okay. Then I guess, the rationale then for industrial up significantly and consumer up slightly.

Well, I think that the industrial --

Randy, your question is industrial segment is down in first quarter.

No, I think you said industrial was up significantly and consumer up slightly --

No, no, I said industrial down significantly. Only consumer up slightly. The other sector all down.

Okay. Thank you. And if I could squeeze 1 more in the backend, it's still another significant CapEx. Could you talk maybe a snapshot what you expect for the backend business this year and what you're seeing for InFO and CoWoS expansion this year?

The backend business is this year it will grow double-digit still a very good business because of we offer the solution to the customers so that they can get the higher performance and better cost structure that they can find in other piece?

The back-end last year is about 2.5 billion. And going forward, currently, we see the double-digit growth at least year-over-year. To clarify, backend business still is in initial stage because as you remember initially, our backend was adopted by the smartphone. But now more and more, we see the high-performance computing customer in the 5-nanometer. Almost all of them wants to adopt what we call advanced packaging. So we see the advanced packaging business is coming to support our high performance computing products across the leading edge customers. Just to make the record, from now on, we call -- we don't call backend. We call that advanced packaging because we realize that packaging is very different than the OSAT business we used to know.

Follow-up question from CL Securities, Sebastian Hou.

Thank you. First, follow-up is can you provide us some update on the tape-out numbers that you have received on 5 nanometers now or your expectation by the end of this year?

Well, I just say that I cannot give you the numbers today. But I repeat it again, all the 7-nanometer customers are working with us right now on their designed.

Okay. And the 7-nanometer tape-out number still on track to exceed 100 by the end of this year?

It's still.

Or it could be earlier?

Well, it will depend on how quickly that our customer can adopt our N7, N7+. But still, the numbers still meet our expectations on the HPC probably even more higher, yes.

And then, given the current macro uncertainty, have you noticed any of your customers which are during the tape-out on 7 right now, become more hesitant or cautious in terms of the production schedule?

No. They are being very aggressive because of 7-nanometer has very good performance. So they can sell them to gain the market -- to their product going to the market. So it's not slowing down. It's actually accelerating a little bit.

Okay. So I think last year and this year, I think probably only not a big portion of these 7-nanometer tape-out with customers are reaching production. Is that a fair assumption?

Yes.

Just one of the big customers account for big wafer volume but the tape-out number is small. Which means do you expect more of this smaller volume type customers, but to get account of the big numbers of the tape-out will reach production in 2020.

Well, I just mentioned yes, some of the second wave or third wave of product design get into the 7, 7+ will start to ramp probably second half or the fourth quarter of this year. Mass production will be expected in 2020.

Okay. So, it sounds like you don't notice or feel any bubbling in terms of the tape-out booming.

No, we did not see that.

Okay. It's great. My next question is -- sorry, I won't have another question. Thank you.

Okay. The right, follow-up questions from Citi's Roland Shu.

Thank you. Last year, the raw wafer negatively impacted gross margin. How about this year? Are we going to still see the same change, or we happen to see will be reversed maybe a benefit for this raw wafer price on margin? Thank you.

We have signed a long-term contract with our suppliers in 2017 and '18. So lock in the long-term supply and also lock in the price. So we do not see any further price deterioration for us, yes.

Okay. Yes. But, given now that demand is declining and are we going to renegotiate these longer contracts with the wafer suppliers?

Yes.

So this is for 2019 and 2020 or going forward?

Going forward. Going forward, we hope we achieve some cost saving.

Okay. Thank you. Second question is by when you will see your CPU revenue to reach 1% of your total revenue and also by when you will see the CPU foundry also coming from multiple customers? Thank you.

That's some very specific 1% but just say that we started working with the CPU customers and starting from 7, 7+ and all the way to 5, okay, and when time's up I'll report it here, say what is percentage that we got from the CPU, okay.

I think you can calculate from our customers' financial.

Okay. Just to clarify, C.C., you said that you start to work with a CPU customer or customers.

Customers.

Okay. Thank you.

All right. Next question will be coming from UBS, Bill Lu.

Hi. Thank you. A couple of questions on the trade tension in tariffs. With tariffs becoming, I guess more prevalent, does that change your outlook on where to place the fabs? And secondly, you've got more competition at the trailing edge if you look at the Chinese fabs. Are you seeing an opportunity to for it to take share with the trade tension?

Okay. You talked about fab location, right? When we build fab, nowadays we only build the leading edge fabs. That is we're doing 5-nanometer fabs, preparation and we're building 3-nanometer. And the decision was made to build in Tainan, okay? And of course, with across-the-board support from the local and administration here. Upon the trade tension, I think it appears that we will, at this point, it seems that there are much -- it is good to build a fab in Taiwan. But actually, we have almost none request from our customers to change the plan we have. But, however, we do constantly assessing and deliberation about what's the pros and cons of that plan and whether there are other options. But so far, we haven't changed our plan. The prime reason is this; when we ramp the leading edge fabs, as you know those are really for the high-performance computing or smartphone launch, the ramp is very tight. Time-to-market is critical. And to have a leading edge fab ramp not only this fab has to be closely coupled with an R&D fab, which is in Hsinchu, the two teams also work as one team, also this fab has to collaborate with other fabs in TSMC. As our founder mentioned, there are thousands of engineers, transport from fab to fab to cope with this sudden ramp of those resources needed. So those are the background of those leading edge fabs. When we open book with our customers, this is the best sure way to ensure their product announcement and product launch and the request conversation will take them back. But we're constantly watching this on bigger geopolitical change. Of course, at this present time, we do not have plan to change that.

The second part of the question is are you seeing an opportunity to take share, for example, a few of our customers don't want to use Chinese fabs?

The question is say that again?

Is there an opportunity at the trailing edge 28-nanometer, 8-inch to take market share?

From Chinese Fab? Well, most of the U.S. customer -- currently, the Chinese Fab is still not used by the local customers, when you put it this way. Intention of the other U.S. customer, we didn't encounter those comments.

All right. Follow-up questions coming from Morgan Stanley's Charlie Chan.

Yes. Thanks. So I want to follow up a little bit about the trade tension issue because another topic is Huawei's 5G and information security issue. So how does the company kind of evaluate this kind of business risk, I mean exposure to Huawei's supply chain and have you got any concern from governments or U.S. governments on this topic.

No, and we are everybody's foundry. And we did not see any kind of instruction or information from the government so business as usual.

Okay, great. And also, another question is regarding your specialty semiconductor strategy. As you mentioned, you want to build up a new 8-inch in Tainan. I would guess is for lots of niche demand. But how do you deal with this kind of outsourcing or partnership with your subsidiary Vanguard going forward in the 8-inch business?

Well, if you look at the market, 8 inch as a business that is actually some time the capacity is in shortage. But regardless of Vanguard or not, TSMC did not increase the capacity. We increased the specialties capability and also our service to the customer. So we built the new building to give more room to put the specialty tool inside to support our customer.

Okay, okay. And that's I guess, this year could be tough, right, could be the first downturn that 2 gentlemen become the new management, right? So is there any initiative that you want to take the company, get through this potential downturn any new initiatives in the company?

Why do you say this is a new management? We have been here for more than 20 years at least. The new measure, I mean, that's -- we follow a very good guidance from the previous chairman. And so far, so good. The strategy continues, the management style continues. So new management, hard to say.

Sorry for the phrasing, sorry.

Let me comment. This is -- we, over the years, I mean, right now, probably is the -- in terms of technology portfolio TSMC has built, both leading edge and specialty, I think this is the strongest position at present time, do you agree? And also about the customer engagement or customer portfolio, this time, it's about the widest customer portfolio we ever have. And C.C. is talking about getting into high-performance computing even the CPUs, including data center CPU, accelerators and the client CPUs. It's probably the widest addressable market we ever have. So we're going to continue these three thrusts. We're going to invest further on the R&D and we're going to continue to engage with our customers' design, closely building the team. Actually, we sent people to our customers to build the design of 5-nanometer some of the 3-nanometer discussion. And also, we just want the high performance computing can adopt our technology development features more tightly. So we think this is a -- although we're facing some headwinds on the macro economy and just another uncertainty is not in our control. It's the trade dispute tension, we just hope the 2 country can come to a win-win or at least not lose-lose solution. I think we're on the road to climb our next peak, yes.

With TSMC's business under the steady stable hands and the expanding addressable market. We will conclude today's conference. So please be advised that the replay of the conference will be accessible within 4 hours from now. The transcript will become available 24 hours from now both of which will be available through our Web Site 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.

Thank you. Thank you, everybody.