[Foreign language] 很 方 便 。 各 位 投 资 界 的 朋 友 跟 媒 体 的 朋 友 ， 大 家 午 安 。 我 是 台 积 电 企 业 讯 息 处 的 孙 又 文 ， 欢 迎 您 今 天 来 参 加 我 们 2018 年 第 三 季 的 法 人 说 明 会 。 由 于 这 个 法 说 会 是 向 全 球 投 资 人 同 时 连 线 转 播 ， 所 以 我 们 全 程 会 使 用 英 文 ， 请 您 见 谅 。 Welcome to TSMC 3rd Quarter 2018 Earnings Conference and Conference Call. This is Elizabeth Sun, TSMC Senior Director of Corporate Communications and your host for today. Today's event is webcast live through www.tsmc.com. If you are joining us through the conference call, your dialing lines are in listen only mode.
As this conference is being viewed by investors around the world, we will conduct the event in English only. The format for today's event will be as follows: first, TSMC Senior Vice President and CFO, Ms. Lora Ho will summarize our operations in the third quarter 2018, followed by guidance for the fourth quarter. Afterwards, Ms. Ho and TSMC CEO Dr. C.C. Wei will jointly provide company's key messages. We will open both the floor and the line for the Q&A. For those participants on call, if you are not yet have a copy of the press release, you may download it from www.tsmc.com. Please also download the summary slides in relation to today's earnings conference presentation.
As usual, I would like to remind everybody that today's discussions may contain forward-looking statements that are subject to significant risks and uncertainties, which could cause actual results to differ materially from those contained in the forward-looking statements. Please refer to the safe harbor notice that appears on our press release. Now I would like to turn the microphone to TSMC CFO, Ms. Lora Ho, for the summary of operations and current quarter guidance.
Good afternoon, everybody. Thank you for joining us today. My presentation will start with financial highlights for the third quarter, and followed by the guidance for the fourth quarter. Third quarter revenue in US dollar reached $8.49 billion, which increased 8.1% sequentially, reflecting customers' new product launches using our 7-nanometer and a more favorable foreign exchange rate. Third quarter revenue came in stronger than the revised guidance that considered the impact from the August 3rd virus incident. As we were able to make up most of the delayed shipments. In TWD, revenue increased 11.6% sequentially. Compare with second quarter, gross margin decreased 0.4 percentage point to 47.4%. As the unfavorable technology mix was partially offset by the profitability improvement in the back-end business, while the more favorable exchange rate offset the virus incident impact. Total operating expenses increased by TWD 1.7 billion.
The increase was mainly for 5-nanometer and 7-nanometer+ development. Thanks to operating efficiency, total operating expenses represented 10.8% of revenue versus 11.3% in the second quarter. Our operating margin improved 0.4 percentage point sequentially to 36.6%. On tax expenses, after a step-up in tax rate to 17.5% in the second quarter due to the accrual of retained earning tax, our effective tax rate fell back to 10% in the third quarter. We expect the full year tax rate to be about 12%. Overall, our second quarter EPS was TWD 3.44, and ROE was 23.2%. Now let's take a look at wafer revenue contribution by application. During the third quarter, communication and industrial standard increased 24% and 6% respectively, while computer and consumer decreased by 35% and 1% respectively. Now let's take a look at revenue by technology.
7-nanometer process technology contributed 11% of total wafer revenue in the third quarter. 10-nanometer accounted for 6%, while the combined revenue from the 16-nanometer and 20-nanometer accounted for 25%. Advanced technologies defined as 28-nanometer and more advanced technologies accounted for 61% of the total wafer revenue. Moving on to the balance sheet, we ended the third quarter with cash and marketable securities of TWD 604 billion, a decrease of TWD 145 billion from last quarter. Mainly as we paid out TWD 207 billion of cash dividend, while we borrowed TWD 43 billion short-term loans for hedging purpose. Correspondingly, current liabilities decreased by TWD 141 billion. On financial ratios, accounts receivable turnover days remain at 38 days. Days of inventory decreased one day to 73 days due to stronger wafer shipment during the quarter. Now let me make a few comments on cash flow and CapEx.
During the third quarter, we generated about TWD 94 billion cash from operations and spent TWD 70 billion in capital expenditures. As a result, we generated free cash flow of TWD 24 billion. After we pay our cash dividend and borrow short-term loans, cash balance decreased by about TWD 143 billion to TWD 489 billion at the end of the quarter. In US dollar terms, the capital expenditure spent in the first three quarters of 2018 totaled $6.7 billion. Now let me provide you the fourth quarter guidance. Based on the current business outlook, we expect fourth quarter revenue to be between $9.35 billion-$9.45 billion, which is a 10.7% sequentially increase at the midpoint of the guidance. Based on exchange rate assumptions of $1 to TWD 30.80, our fourth quarter gross margin is expected to be between 47%-49%.
Our fourth quarter operating margin is expected to be between 36%-38%. This concludes my financial presentation. Let me follow by making a few comments about the near-term demand, inventory, capacity, and CapEx. Now on the near-term demand and inventory, we conclude our third quarter with revenue of TWD 260.3 billion, or $8.49 billion, which is above our revised guidance. According to the revised guidance, the computer virus incident on August 3rd was estimated to have impacted our third quarter revenue by about 2% and gross margin by about 1 percentage point, respectively. However, we were able to make up about 75% of the affected shipments in the third quarter. So overall, our third quarter result was mainly driven by strong demand from product launches using our 7-nanometer technology.
Moving into fourth quarter, despite the current market uncertainties, our business will benefit from a continued steep ramp of 7-nanometer for several high-end smartphones, as well as the demand for 16/12-nanometer for the launches of new generation GPU and AI. However, this growth will be partially offset by continued weakness in cryptocurrency mining demand and inventory management by our customers. Fabless DOI exiting third quarter 2018 was several days above seasonal level, slightly higher than what we expected three months ago. We forecast Fabless DOI to continue to stay at this above seasonal level exiting 2018. Let me make some comment about capacity in CapEx. At TSMC, we build our capacity according to customers' demand. We are continuing to increase 7-nanometer capacity to meet the strong customer demand. We reiterate our 2018 CapEx to be between $10 billion and $10.5 billion.
As I have talked about before, although our leading edges capital costs continue to increase due to increasing process complexity, we are able to offset its impact to our CapEx by productivity improvements and further optimization of our capacity planning. Going forward, we expect annual CapEx needed to support our 5%-10% revenue CAGR in US dollar terms. In the next few years, we will continue to range between $10 billion and $12 billion. I conclude my remarks. Let me turn the microphone to C.C.
Thank you, Lora. Good afternoon, ladies and gentlemen. Let me start with 2018's year's outlook. For the full year of 2018, we forecast the overall semiconductor market, excluding memory, will grow between 5%-7%, while foundry is expected to grow between 6%-7%. For TSMC, as Lora has just indicated, that our second half of 2018 business will be strongly supported by the 7-nanometer ramp-up, which is mainly driven by a few new smartphone launches. However, our business is also negatively impacted by further weakening of cryptocurrency mining demand. As a result, we estimate our 2018 growth rate will be about 6.5% in US dollar term, which is close to the country industry's growth, but slightly below our 7%-9% guidance given in the last conference. Let me update you about the August 3rd virus incident.
On August 3rd, TSMC experienced a computer virus outbreak which affected a number of computer systems and fab tools. The infection was due to misoperation and insufficient firewall controls. We have since corrected this problem to ensure such virus incidents will not happen again in the future. Our remedial actions including the following, implementing an automatic system to guarantee foolproof execution so that such misoperation will not happen again. Enhanced firewall control for fab isolation and network control to each individual computer. More enhancements are ongoing to further improve to immunity against future infections. TSMC sets top priority for such security enhancement. Let me talk about N7 and N7+ and the EUV progress. TSMC's N7 technology is now available for customers to unleash their innovations. This is the first time in the semiconductor industry, a most advanced logic technology is available for all product innovations at the same time.
We continue to work with many customers on N7/N7+ product design, and expect to see more than 100 customer product tape-outs by end of 2019. We expect 7-nanometer to be a long node and will attract multiple waves of customer adoptions. N7+ is in risk production now. Since N7+ has 15%-20% better density and more than 10% lower power consumption, we are working with many customers for their second wave product design on N7+. Although the number of tape-outs today accounts for a small portion of the total 7-nanometer tape-outs, we expect the activity to pick up at a rapid pace in 2020 and beyond. Because N7+ is using a few layers of EUV photolithography to have better cycle time and pattern control, we have made a steady progress on EUV technology development towards high volume production.
Tool availability, EUV power, productivity, defect reduction, mask improvement, material and process optimization are all on schedule. A few customers has already made plans to adopt our N7+ in their 2019 products. Let me move to our N5 status. Our N5 technology development is on schedule. We have completed the design solution development and are ready for customers design start. The N5 risk production schedule in first half 2019 stays the same. Compared to N7, TSMC's N5 deliver 1.8x-1.86x logic area reduction and close to 15%-18% speed gain on Arm A72 core. We expect to receive first customer product tape-outs in spring of 2019, followed by production ramp in first half 2020. Now let me talk about the N28, N22 and mature node strategy.
Due to faster than expected technology migration from 28-nanometer to 16-nanometer and below, 28-nanometer of overcapacity becomes an industry-wide phenomenon and is expected to last for a few years. TSMC's mature node strategy is to work closely with our customers to develop specialty technology solutions to meet customers' requirement. For example, we continue to develop 22-nanometer for scaling benefit and better performance. We are also developing 22-nanometer for CMOS image sensor, MRAM and RAM. For all mature nodes, TSMC will continue to develop a variety of special technologies such as power management IC, embedded flash, image sensor, MEMS to maintain our good capacity loading rate and to increase our technology value to customers. Now let me talk about advanced packaging update. TSMC has been developing advanced wafer-level packaging technologies to integrate advanced SoCs, memories, integrate the passive device to enhance system performance.
We believe our advanced packaging solutions will contribute to our business growth. We are now expanding the application of both CoWoS and InFO, especially for high performance computing. Most of the CoWoS products require integration of SoC with high-bandwidth memory, HBM. In 3D stack, we are making good progress in qualifying multiple HBM sources through close collaboration with customers and the DRAM suppliers. We are also working with a few leading customer on SoIC, which stands for System-on-Integrated-Chips, where multiple heterogeneous chiplets will be integrated with close proximity to deliver better performance. We target to start production in 2021 timeframe. That's all my report, thank you for your attention.
Thank you. This concludes our prepared statements. 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 an opportunity to ask their 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 star, then one on your telephone keypad now. Questions will be taken in the order in which they were received. If at any time you would like to remove yourself from the questioning queue, please press the pound or the hash key. We can start. Deutsche Bank, Michael Chou first.
Thanks. Two question. One is, what is the outlook for seven nanometer sales portion in 2019? Can you give some color at this moment based on your current visibility or some tape-out numbers?
Michael, your question is, what is the outlook of seven nanometer?
In 2019.
In 2019?
Yeah. In terms of sales portion or any indication.
I would like to answer the question, say, it's a very strong demand in all the platforms, in all the market sector, which include in the mobile phone, high-performance computing. That's included graphic, a lot of AI accelerator, FPGA, you name it. A lot of customer working with us, as I just stated, and we expect 100 or more than 100 tape-outs, product tape-outs in year 2019. Did I answer your question?
Is that fair to say the 7-nanometer sales portion will be more than 20% of total sales for the whole year next year?
Next year?
Let me answer that. You have seen our report. The third quarter, 7-nanometer accounts for 11%, and fourth quarter will be more than 20%. For whole year 2018, 7-nanometer will contribute close to 10% of total TSMC revenue. Go beyond 7, 2018, we will have very, very strong ramp in 2019 as well. We expect the revenue contribution will be much higher than 20%.
Second question. C.C., you mentioned, 28-nanometer oversupply in the industry could be a normal situation over the next few years. Will management consider converting some 28-nanometer to advanced node? I know you mentioned some specialty, especially PMIC 28, CIS in 28, so which should be a good volume product. Would you consider to convert some capacity to advanced node?
Good question, Michael. Actually, we are moving from 28-nanometer to 22-nanometer, which is a smaller geometry and more advanced. I think your question is that, are we consider moving into 16, convert some of them? We keep the flexibility to moving the wafer around so they can support each other. We already did that. There are a lot of commonality in the tools. Some of that capacity or part of the capacity can support other technology node.
It means that if, for instance, if your 16/12-nanometer demand is beyond your estimate, you can use some 28-nanometer, do some 16 node product in the future?
We have some tools that have a commodity.
Oh.
Say, for example, in implanter.
Yes.
Right? Clean up and a lot of things. Let me stress again, I say the overcapacity for few years. However, we are developing 22-nanometer. We are developing a lot of specialties. For TSMC, this only probably a couple of years, we are going to fully utilize 28-nanometer again.
Is that fair to say your 28, 22 UTR over the next few years, you are confident you can maintain maybe above 85%, above 90%?
I'm not going to answer this question, after a couple of years, you will be the high utilization rate again.
Sure. Thank you.
Next question will be coming from Credit Suisse, Randy Abrams.
Okay. Yes, thank you. I wanted to ask a question about sales and also gross margins. The sales in fourth quarter is actually fairly good, factoring in the business environment. Could you talk maybe the forward view? You mentioned also fabless days of inventory may exit the year a few days high. With the implications for first half, especially relative to the last two years, we saw declines in first and second quarter. If you have a view, kind of the rate or magnitude of decline relative to what you think normal seasonal is for first and second quarter.
Are we going to forecast the next year's first quarter and second quarter?
Initial view based on inventory and demand.
We have very strong demand in 7-nanometer. Let me just continue to say that. For the last quarter of this year, you're talking about the profit margin, right?
Yeah. The second question on margin I wanted to ask, it's about a point or two below the traditional, say, 49%-50% target.
Yeah.
Lora cited product mix. Next year, 7-nanometer will be more mature. Do you think you have the catalyst to get back to that 49%, 50% or other factors like 28-nanometer oversupply?
Well, let me answer your First quarter's profit margin is a bit lower than we used to have. That's between 49% and 50%. That's because of a product mix Lora just mentioned. Product mix, one of the major reason is the 7-nanometer ramp up much faster than we thought. The demand is very strong. That lower down our margin, a couple point. I'll let the CFO to explain the number.
C.C. Wei mentioned about a 7-nanometer being particularly strong, compared with our other leading-edge technology in the past. It does dilute our corporate margin level starting from the third quarter this year. Going to the fourth quarter, as I mentioned, the revenue contribution will be 20%, more than 20% of our fourth quarter revenue. The margin for 7-nanometer at the beginning stage is still below corporate average. The dilution in the fourth quarter will be more severe than the third quarter. Our estimation is between 2-3 percentage point for fourth quarter gross margin.
Yeah.
Going into 2019, as the RRAM will continue to be very strong, we believe the margin dilution will be eased off when we reach to the second half of 2019.
Okay. Maybe I'll just follow up on that first one about sales. You mentioned seven is very strong, I guess, do you have a view the rest of the business may be doing some further inventory adjustments in first quarter?
Laura?
There are several days above seasonal this quarter. Going into 2019, I think inventory will be gradually digest when we move into the second half of 2019.
Okay. The second question I wanted to ask was about the 7 plus versus 5 nanometer. You mentioned 2020 would see the very strong ramp up of tape out and activity and volume on 7 plus. Is it your view I think last conference Mark said 5 was a little bit more conservative at this stage. How is your view now for interest, activity, and expectation for a steep ramp-up of 5 into 2020?
We still expect a very fast ramp on 5. The reason is simple: because of a lot of product developing in the AI area, you need the speed, you need the low power, and you also need the small footprint. From today, we can see when we work with our customer, the ramp will be steep again.
Okay. Do you think mobile will also contribute in that steep ramp?
Still come from the mobile phone first, and followed by other applications.
Okay, great. Thank you.
Next question will be coming from Citigroup's Roland Xu.
Good afternoon. I would like to follow up for this 4Q gross margin guidance. If I compare with 4Q last year, actually, we have a similar percentage contribution from this leading edge. Last year, 4Q, 10-nanometer was about 25% of the total revenue. In 4Q this year, actually, we have a favorable exchange rate compared to 4Q last year. Also, we have this improved backend profitability. How come we still have a lower gross margin compared to 4Q last year?
There is about two percentage point difference, this 4Q versus last 4Q.
Yes.
There's one reason is product mix issue. I'm particularly talking about 28-nanometer. You know, 28-nanometer is very profitable, and the contribution to corporate revenue has declining from last year to this year. These are two reasons. Product mix, because 28 declining, and also because the utilization of 28 is not full. Yeah. That's one reason that there's a new phenomenon.
Okay. How about the standard gross margin? Does it change?
Structural profitability is still intact. No change.
Okay. Also, last quarter you talked about for longer term, our financial objective for gross margin is 50%. Is this a 50% a fixed number, or this 50% is average for the upcoming few years?
50% is our goal. We aim to achieve that. If we can have higher utilization, I am quite confident we can achieve that.
Okay. This year, I think that probably, we're not able to achieve this 50%.
No
because of the utilization.
Not this year, and also because their 7-nanometer ramp is more severe than previous node. First year dilution is slightly bigger.
Okay. Thank you. My second question is, AMD starts its 7-nanometer CPU from server, and also Arm recently also talk, they have this new brand server product. Is server CPU going forward to be a very significant part of our business going forward? Thank you.
It will be an important part. Significant? Maybe.
Your 4 growth platform, smartphone, HPC, IoT, and automotive. This server CPU will be in HPC side.
That's right.
Okay. Thank you. Can you just reiterate the growth breakdown for this fourth platform next year? Thank you.
Okay. Let me give you some color on it. In the next few years, if we look ahead, actually, the smartphone going to be in our daily life even more and more. We have four growth engine. One is a mobile phone, actually, this high-end smartphone. The second one is high performance computing, automotive, IoT. The mobile phone, probably for TSMC will have in five years a taker, if I look at it right from today, it will be mid-single-digits of growth. All other three platforms will have a very comfortably double-digits of growth in the five-year time frame. Did that answer your question?
Yes. Thank you.
You're welcome.
Next question will be coming from UBS, Bill Lu.
Hi there. Thank you very much. First question is on China. A lot of concerns about the trade tension, about the economy slowing down, et cetera. Can you talk a little bit about what feedback you might be getting from your Chinese customers, both short-term and long-term? Short-term, meaning whether you're seeing any increasing levels of conservatism on ordering, on demand. Longer term, is there any change in the strategies of these Chinese customers in terms of perhaps less reliance on the U.S., et cetera? Thank you.
To answer your question, for the short-term, we did not see any impact, if there's any at all. Our China customer, no, they did not change their behavior. We continue to work with the customer to produce their product. For the long term, due to the trade tension, your question is a dual question.
Yes.
Let me answer the question by another way. TSMC has been proud to be everybody's foundry. If there's a trade tension, if there is it, and if it continues, I think the impact to TSMC will be less or minimized, because we still need the semiconductor device, and TSMC is everybody's foundry, right? Whether produce here, produce there, is all TSMC's customer.
Great. Thank you. I know 2018 is not over yet, but if you think about the next couple of years, I know TSMC has talked about a long-term growth rate of 5%-10%. Now, I feel like more recently, you've talked a lot more about the progress on 7 nanometers. We all know about Intel struggles with their process technology, and it's public information. They've announced it, right? And then you've got some good design wins. Can you talk about your long-term outlook at 2019, given these drivers? You just said three out of the four new drivers will be above 10%, are we looking at something more towards the high end of that, or how do you think about that?
We continue to say 5%-10% growth rate. Probably, following your question, I would like to say, probably tends to be the higher side of that 5%-10%. That answer your question?
Yes. Thank you very much.
Next question will be coming from Daiwa's Rick Hsu.
All right. Thank you. My first question is, I know Cissy was talking about your more than 100 tape out for N7 and N7 Plus. Can you give us a little bit more idea about the breakdown? I would like to know how many tape out for N Plus, and also what kind of application for that.
Okay. N Plus is a product tape out. I am talking about product tape outs. It is still a small percentage of the total N7 nanometer node, as I just mentioned. After 2019, I expect that 2020, when the current customer, they start to design on their second-wave products, it will be more of them on N7+. I don't have a specific number for you on 2020.
Okay. Thank you. My second question is, you also mentioned that your N7 will likely be another long node. Presumably, probably something like N28, in terms of success. My question is, in terms of capacity build, I know you don't talk about the capacity number, but in terms of capacity build, would you benchmark your N28 into more capacity number?
We build our capacity according to the customer demand. I can assure you that. Compared with the 28-nanometer-
Right
I cannot make any comment right now.
Okay. Well, thank you so much.
Okay. Next question will be coming from Morgan Stanley's Charlie Chan.
Thank you. My first question is regarding the semiconductor inventory. In which segment do you see more inventory? Because we are hearing lots of noise about automotive, industrial segment weakness, and some cloud CapEx is cut, right? Can you give us some color about where do you see the inventory, and also your utilization rate? Thanks.
Well, I don't think we want to specify which product sector that have high inventory. Let me assure you that this year, at the end of this year, the inventory level actually as compared with the last year's same time, is much smaller. Even we have some kind of inventory adjustment, I don't think the impact to the next year's quarter or next year's first half will be as severe as we saw this year. All right. That's all I can say, because it's very dynamic. Lora, you want to add something?
I can add some color to your questions. When you look at the fourth quarter, we have just guided the 10.7% growth quarter-over-quarter. Looking at the segment, fourth quarter communication will grow very strong. Computer is small growth, lower than the corporate growth, so that area is relatively weak. Consumers is weaker. It's negative growth, industrial is a small decline. Maybe this can give you some color about which segment has more inventory.
Thanks. Maybe this is also related to Bill's previous question, right? When you analyze this inventory, customers' demand, do you think there is any impact from any macro factor including the China-U.S. trade tension, or it's just a normal semiconductor cycle?
I would say it's just a normal inventory cycle.
Okay, thanks. Next question is also related to 7-nanometer Plus, right? You mentioned only a few tape outs, sometimes one tape out can be very significant to revenue, right? Can you give us some comments regarding the 7-nanometer Plus revenue contribution next year?
No, I don't want to specifically say which one.
All right. Thank you.
All right.
I think, at this point, we would like to turn the questioning to the callers on the line. Operator, could you please put the first caller on the line first?
The first question is on the line. Your line has been opened.
Mr. Hosseini, your line is open.
Thanks for taking my question. Yes. Thanks for taking my question. I want to go back to your view on 2018. An interesting year because we have a migration from 4G to 5G telecommunications, which could adversely impact the demand. You also have AI and the IoT ramping. How should we think about the mix changing and would new applications and continued share gain enable you to actually offset any adverse impact as we migrate 4 to 5G?
Mehdi, in fact, we can only hear your voice in a discontinued manner. Could you please repeat your question again? Maybe keep your voice a little bit away from the microphone.
Okay. Sorry about that. I'm just trying to better understand how you're thinking about 2019. It seems like there are many new products that are ramping, given the number of tape outs at 7-nanometer. We also know that in the smartphone market, we're in transition from 4G to 5G. In that context, should we expect new products, especially for AI and IoT, be strong enough to offset any adverse impact from migration from 4G to 5G?
You want to repeat the question?
All right, let me try. I think Mehdi's question is that he is very positive about the outlook for us in 2019 because we have so many new product launches from our customers using 7-nanometer, and smartphone is entering into a 5G era. He thinks that we may get a lot of new products in AI and IoT, which would be able to offset the migration from 4G to 5G.
From 4G to 5G.
Yes.
Well, that's the question, huh? Looks like the question says, indicate that 4G to 5G, they will have a dip. Actually, it's not. 5G in 2019, we expect start to grow, but not a significant number. Most of the phone, the smartphone, I mean, it will be in 4G still in 2019. However, you mentioned about the AI and other products segment. Yes, that will start to grow in 2019. We expect a lot of new product other than the smartphone will contribute that 2019's 7-nanometer business.
Could new product actually be big enough to help you with a 5%-10% revenue growth? I'm asking you this because this year we've had adverse impact from customer inventory and cryptocurrency that impacted your overall growth target for 2018. In 2019, would new products offset any weakness that is associated with the mature products?
Well, in 2019, as I just mentioned, the inventory correction will be much less effect as compared with the 2018, beginning of 2018. Definitely our growth rate in the 5%-10% will be much likely and in the high side. That's what I say.
That's right.
Mehdi accept our very positive outlook for 2019. Good. Operator, could you please go to the next caller on the line? Thank you.
Thank you. The next question is from the line of Gokul Hariharan from J.P. Morgan. Your line is now open.
Yeah, hi. Thanks for taking my question. My first question is on the China demand and capacity build-out. Could you talk a little bit about the progress of the build-out for N16 in Nanjing? What is the schedule for phase 2? Has there any change in plan post the weakness that we have seen in cryptocurrency demand from some of the Chinese customers? That's my first question.
Can you repeat it?
Gokul's first question is he asked us to update our Nanjing fab's progress, and he asked if we have any changes in plans to this ramp-up in Nanjing fab because of the weakness in cryptocurrency mining.
The Nanjing Chairman want to answer that?
We had a 10,000 wafer per month install capacity currently, we're planning to increase the capacity to 20,000 wafer per month next year. We are developing customer as well as some derivative technology that are related to the 16-nanometer. We are progress as we plan.
Okay. Thank you. My second question is on N7 and N7+. TSMC appears to be in the best position to comment about the benefits of EUV, given that you have N7 and N7+, one in production, one in this production. Could you talk a little bit about the kind of improvements that we're seeing, either from an in-grade perspective or reduction in mask layers, et cetera, and the impact in terms of overall cycle time for the fab, comparing N7 and N7+?
Okay. Actually, the question is about the EUV and how much of the benefit we can get from the EUV, right? Usually, if we are not using the EUV, sometimes for the very critical dimension on the N7 you had to, or N7+, you had to using the four layer of photolithography to have a, to pattern one of the critical dimension. Now using the EUV, you just use one layer so that you reduce the cycle time by four times of photolithography, four times of etch. Now you become one lithography, one etch. In total, how many layers we reduce? That is depends on the customer's requirement. Usually, I just give you a hint already, right? Four layer can be become one, and we are replacing some of the three layers to become one, and we have a few layers of that.
That give you a hint. Cycle time reduction, definitely, because of you do four times and do one times, that's a big advantage. Productivity-wise, today EUV is progress very well, up to our expectation. In fact, TSMC has turned on the 250-watt power, and we believe we are the only one company continuous around the 250-watt EUV power so far today. Okay.
Okay. That's very helpful. Thanks.
We can come back to the floor. Next question will be coming from CLSA Securities, Sebastian Ho.
Yeah. Thank you. My first question is on the trailing edge probability and outlook. Couple of your peers, in the past two years have decided to stop going advanced. Now their strategy is to focus on the node they already have. China is also ramping up some trailing edge technology capacity. It seems like more focus on there. Understand that TSMC actually generate lots of profit, cash flow from trailing technology. How do you see this dynamic going forward? Is there a potential for this to negative impact your trailing edge probability? This first question.
Okay. Actually, I don't want to comment on my competitor's strategy. Let me, again, stress our mature nodes strategy. We continue to develop some of the specialty technology to meet the customer's requirement, right? I just stated in that. There's a lot of specialty technology we are doing. I give you some example already, power management IC, CMOS image sensor, MRAM, everything. That will help us to compete with our competitor. Actually, this kind of specialty technology, particularly, we have to work with the customer. That's why I say working with the customer to meet their requirement. That will in turn to give TSMC's business. That's a way that we migrate the technology, pure logic technology to the more advanced node. For the existing capacity, we develop into the specialty technology.
Our strategy is still meet customer's requirement, but we don't increase the existing logic capacity.
Right. I thought TSMC has been doing this for a long time.
Yes.
You'll keep doing this. The real change is not what you're doing, the real change is what your competitors, the dynamic, the more competition threat from.
We already have a lot of competitions. Before they announce that, we still think they are the competitor. No doubt about it. Not because of they drop out some of the technology node they don't want to do. We are not going to change our strategy because of that.
Fair. Thank you. Second question is, still on the node profitability. I remember, the rule of thumb for TSMC node profitability is the N minus two node usually have the very good or probably the highest probability for TSMC. Now we're already ramping on 7. The N minus two for you is 16. Understand that the 28 used to be very profitable, as Lora just said. I just wondering, if based on this, assuming, 16 and 20. We compare 16 and 28, assuming they are both depreciated at the same utilization rate. Will 16 be more profitable than 28 or similar?
This is very specific. You ask the profitability. Lora, are we going to answer that one?
I probably cannot quantify that, we don't know whether you said N minus two is the most profitable node is correct. Eventually, I think as we continue to improve productivity and we keep very competitive, especially technology for the mature node, our margin will be all very good.
Okay. Let me ask from another perspective, is that some of this capacity on 16 or 20 was installed in 2014, 2015. Which means that they will become fully depreciated very soon. Which means that these will become a margin profitability tailwind for TSMC starting from next year, partially offset some of the 28 headwinds. Is that the right way to look at it?
We certainly expect that.
All right.
You are talking about the fully depreciated tool to generate more profit. Yeah, you can calculate. We have a five years of depreciation period.
Right. Which means that by the same time next year, regardless of the utilization rate or whatever pricing, 16 will likely to be more powerful than 28.
Actually, TSMC's 28-nanometer is very profitable, and that is What should I say? You say 16 is better than 28, and I hope so, but I cannot tell you the true number.
Okay. Thank you.
I think our profitability is not depending on which node is better than which other node. We want all the nodes to be very profitable.
Thank you.
Okay, next question will be coming from Morgan Stanley's Charlie Chan.
Thanks for taking my follow-up questions. After the recent industry consolidation, does that change anything, like your bargaining power and also your thinking about the cash return? What does that change you're thinking about, for example, the payout ratio going forward, the industry consolidation?
Industry consolidation?
Yeah. I mean in GlobalFoundries exceed 7-nanometer, and Intel seems to be struggling with their leading edge. I would assume competition in the leading edge is getting less, right? In that case, maybe your bargaining power can increase, and also you have more cash may free up to return to the shareholder.
I don't want to say that because of my competitor, well, in fact, Intel is my customer, so we respect their performance, of course. TSMC's strategy actually is working with the customer. Whether that I have a competition or I don't have a competition, that's independent of that. We think we offer a very good technology value to our customer. Customer are happy, we are working together and growing the business together. I'm not going to tell you that's my pricing strategy. That's not in discussion.
Okay. Yeah, your current payout ratio is around 60%, right? In what kind of a circumstance in the long term would you consider to revise up your payout ratio to maybe 70% or even above?
We have been paying around 70% in the past few years. I have said many times, with 5%-10% growth, we are confident we can continue to generate increasing free cash flow, that will be the source for the increasing of cash dividend.
Yeah. We still with that kind of commitment.
Thank you. The next follow-up is regarding that EUV question, right? I know it could be too direct to ask about revenue contribution from 7-nanometer plus. From another angle, what keep your customer hesitating to adopt the 7-nanometer plus, right? Because you mentioned a few tape outs. Is there any concern about yield rate, sorry, the throughput of EUV or capacity of EUV? What keep a customer hesitating about EUV adoption next year?
Now, let me give you some explanation because of TSMC's 7-nanometer is very successful, customer using the design on 7-nanometer first. 7 plus certainly have some advantages, right? If we're using the EUV, you have a shorter cycle time, I just say that it's a kind of a better pattern control that means your critical dimension has been tightly controlled. That help the performance also. We expect the customer moving their second wave product. A lot of them will go to 7 plus. Have they any concern about the EUV's maturity? That actually is what you are asking.
Everything sounds very decent, right? What's the issue? Price or any issue?
What's the issue? Okay. EUV progress steadily to our expectation. As compared with the DUV photolithography, the maturity is not comparable. Not yet. EUV is still climbing up, and we hope that the productivity can be better and better. Today, not yet. You just look at how many DUV we have, and how many EUV we have. You know what I'm talking about. I mean that in the industry. Just say in the industry. Okay.
Lastly, if I may. I think AI semi has been a very topical growth driver. I think recently there's getting more clear that some ASIC claims their performers can be 8 times better than GPU. Those are general purpose. With now lots of new ASIC coming to light, do you think that it's going to impact your overall AI semiconductor revenue? Put it in another perspective, do you expect any re-acceleration or slowdown of your AI semiconductor business in the coming year?
I think the AI application will be everywhere, actually. From the edge server or to the end device, that's just like a smartphone in everybody. This kind of a development is to our advantage, because TSMC certainly have a technology leadership. All the AI will be effective. You need a very advanced technology for the highest performance computing. I don't see the effect that you're talking about, this application is better than that, so that will affect the course or something. No, it will be continuous to grow. I expect it grow much faster than I predicted here.
Okay. Thank you.
Next will be coming from Deutsche Bank as a follow-up question.
Thanks for taking my follow-up questions. Regarding the 7-nanometer tape out next year. C.C., can you give some color regarding the portion of 7-nano versus 7-nanometer plus EUV? Will customer shift to 7-nanometer plus EUV aggressively in the second half of next year, in 2020?
Your question is it 7 plus?
You mentioned more than 100 tape out numbers for next year, by the end of next year or so.
Right.
Will the majority of that be 7-nanometer or 7-nanometer+ EUV?
Next year is still 7-nanometer.
Okay.
Of 7-nanometer+, we're working with the customer, some of the customers moving a little bit faster, some of them. For those customers are already in the 7-nanometer, they are planning on the second wave product in the 7-nanometer+.
Okay. next year, I can say most of them are still in seven. activity will start to rapidly grow in second half, in 2020, that will be even grow faster.
Okay. For the whole year tape-out number. Can we say the 7-nanometer plus EUV tape-out number increase quarter-by-quarter through end of next year? Can we say that?
Will increase quarter-by-quarter, yes, you can say that. It is still a smaller portion of the total 7-nanometer a node.
Okay. Based on the 7-nanometer plus EUV tape-out by customer planning. Can you expect the 2020 we will see very big volume 7-nanometer plus EUV mass production and the revenue contribution? Because you mentioned increase quarter-by-quarter, right?
Yes.
We just try to find out if you will become very big revenue contributor. Yeah.
If I can forecast well, I mean that one tape out, especially for some of the applications, takes about one year to full ramp up.
Seven plus will be increasing, the revenue, I would expect 2021 start to see the big number.
Okay.
Not 2019, not the first half of 2020.
Okay. How about second half next year? You think that will be?
You continue to nail down the number?
Sorry.
Okay.
Yes.
2019, less than TWD 1 billion.
Oh, less than TWD 1 billion. Thank you so much. Thank you. That's good.
All right. Hopefully next question will be better, and it will be coming from UBS, Bill Lu.
Hi. Thanks again for taking my question. It's a bit more of a long-term question. Hoping to ask you about the R&D, and maybe this is both for Dr. Wei as well as Lora. If you look at the Moore's Law, certainly not getting easier. If you look at the number of players that can share the R&D burden, I think it's getting less. Number 3, TSMC is clearly ahead of the competitors now in terms of process development. As you look out the next two, three years, do you have to think about R&D increasing at a faster rate? Thank you.
We do think the smaller geometry takes a lot of effort to develop the technology. I would like to using TSMC founders wording to answer your question. Long as some people can develop it, TSMC will be there. That's our commitment to the customer. All right? In addition to smaller geometry, actually, TSMC also developed the wafer-level advanced packaging to help the customer improve their systems of performance. We think we have these two weapon in hands, and so we are in a very good position to compete in that field.
I guess I'm not doubting TSMC's abilities to develop it, I think TSMC is taking on more of the burden now. You also have to add the cost for advanced packaging R&D as well, right? I'm just.
Yes
asking about the cost.
The cost. Actually, I would like to say the value of our technology. All right? That's what we discuss with our customer. The cost, we continue to working on it, and we hope TSMC has a capability and ability to roll down the cost, as we already proving in the previous years.
I think Bill's question is whether or not R&D as a % of our revenue will increase in the future because of our taking up a bigger burden.
We might, but I cannot tell you the exact number.
Great.
That's a cost. Yeah. Thank you.
All right, follow-up question from Citigroup's Roland Shu.
Just a follow-up question for the profitability, but I'm talking about the back-end side. C.C., since last quarter, you kept talking about you have this improving profitability on back-end. Also in last quarter, you said the back-end actually was able to partly offset the lower gross margin from this poor product mix. Consider your back-end is just a relative small part of your total revenue. It has to be with a very good gross margin in order to positively offset the unfavorable product mix. How does your back-end gross margin compare to your corporate average right now? Thank you.
Back-end margin is indeed improving. We don't compare back-end margin with corporate average because the nature of back-end is there's a very high asset turnover and lower margin compared with wafer. The overall investment return is pretty good. If you look at our ROIC, it's very good. We don't compare back-end margin with corporate average. If our back-end revenue continue to increase, it may dilute a little bit of corporate margin, but the overall investment is pretty good. Okay. We have seen our advanced packaging. We continue to invest in capital, and we see more customer adoptions. We are increasing the size and the business volume of the back-end. Of course, we continue to improve the productivity. Back-end will be accretive to TSMC going forward.
Okay. Can we have the rough number of how much the revenue is coming from back-end so far?
It's around TWD 2.5 billion this year.
TWD 2.5 billion. Does it include wafer bumping, or this is purely from InFO and CoWoS?
Including all the back-end, that including the wafer bumping, testing, InFO, CoWoS, everything.
Okay.
That's roughly a little bit more than 7% of this year's revenue.
Okay. Lora, you said if the back-end revenue exceed a certain portion of the total revenue, probably the corporate average gross margin will be diluted. What is the threshold for the percentage of the back-end? Thank you.
The back-end is only 7%. It's still low compared to our very big wafer size. I think the back-end problem is cyclical because it was very concentrated on several segment. Cyclicity is more of a issue. In terms of financial return, we have that in our model, so we're not worried about that. Of course, we continue to improve the profitability.
Okay, thank you.
A follow-up question from Morgan Stanley's Charlie Chan.
Thanks for accommodating another follow-up. I think because the stock market weakness, so I need to do a bit microscope on your comments on near term. C.C. Wei, you previously mentioned that 2019 inventory correction in one Q wouldn't be as severe as the beginning of this year. I don't know why now you have this kind of a visibility. Is that because of customers' order behavior? What have you seen for the coming one Q that can make your inventory correction less severe than the beginning of this year?
You ask where I get those information to get the conclusion that next year's one Q inventory correction will be less
Severe than this year.
Less severe, huh? That's interesting. We work with customers, right? Lora just mentioned about those market segment, communication, consumer, and computer. We look at that, we work with the customer, and we understand their business. Not 100%, but we understand quite well, we can do this kind of a forecast.
Okay.
Okay. Thanks. Yeah. Is that fair to link the industrial segment decline to their edge utilization rate? Can I assume that edge utilization is not as full as before? Is that a fair assumption?
You can repeat what I said to answer that question, all right? You are asking very specific.
Okay
utilization and the customer demand
Right.
Actually, let me say that we are very optimistic.
We work with the customer. As I said, we work with the customer, we understand their business. Going forward, we know that they are doing the inventory adjustment, but they also discuss about what they want in the 2019, which I'm not going to give you some forecast today. Not yet.
Okay. Thanks. Sorry for those very detailed questions. Thank you.
Questions will be coming from Credit Lyonnais, CLSA Securities, Sebastian Ho.
Thank you. Follow up on the advanced packaging. I have the impression that, of this year, more than 50 7-nanometer tape out, a huge part of that or a large portion of the customers will use our advanced packaging solution. Moving to next year, our tape out number will be 100 or 100 more on 7. It's still that kind of the percentage, was still the same, which means that the advanced packaging customers on 7-nanometer will also double.
Let me answer the question that advanced packaging, that include CoWoS and InFO. The CoWoS is for very high-speed performance, such as like a graphic chip, such as some of the networking processor. Those are using CoWoS. For the InFO, that a lot of them are using for the mobile phone business. That I can answer you, but I cannot give you, say, which one.
No, I'm not asking which one. I'm just asking this, still of the next year-
Yes
100 tape out, a large portion of this using our-
Oh, okay.
Pack-
Sorry that I misunderstood your question.
No worries.
The next year, the InFO for the mobile phone application, some of them were moved into the high-performance computing also, because that offer a little better cost structure. The CoWoS continue to be the high-end, high-performance computing product. We also introduce SoIC, as I just mentioned, that even much better system performance. We continue to progress our technology, does that answer your question?
Sort of. Well, anyway.
Not 100%
My question is very simple, actually. Simply just want to know that your earlier comments on the large portion of the 7-nanometer customer tape out will use our advanced packaging. Now our tape out numbers, based on your comments, it seems like double from this year to next year. This large portion still apply on the doubling 7-nanometer tape out next year. Is that right?
I see. Develop one new back-end process, packaging process. It actually will last for many years. Next year's 7-nanometer or 7+ application for those products are still using the InFO and also using the CoWoS. Yes. Whether it's double or not, I'm not ready to answer that question yet.
Okay. Thank you. That's clear. Second follow-up is, I remember last quarter when, C.C., you talk about the 28-nanometer outlook, you mentioned the 2019 will come down and expect the recovery, nice recovery in 2020 because of the 22-nanometer adoption. Today, you just mentioned that you expect the overcapacity will be the norm for several years. What I interpret is that it sounds to me that over the past three months, you've become more cautious, conservative on 28 compared to three months ago. Am I interpreting that right?
You may be right. Because of recently, a lot of capacity has been built on 28-nanometer as compared with three months, six months ago. Right. That makes the overcapacity a little bit more severe than previously predicted. However, I would like to stress again, I mean, that TSMC develop shrink of the geometry for customer to gain the benefit of performance and better density. We also develop specialties. That's why I say that we have confidence a couple of years later that our 28-nanometer will enter into a new era with a high utilization rate again.
Do you view those excessive 28-nanometer capacity or the more than expected newly add 28-nanometer capacity, is it a real threat to you or they're a non-event? I mean, their effective capacity or their-
We always compete in the market, whether they build capacity. Of course, let me answer the question directly. Overcapacity is not good for any player in this field. Definitely. Okay.
Okay. Thank you.
All right. I think given the consideration of time, we'll just have Randy Abrams to have the last question.
Yeah, just one quick follow-up to that last line of discussion. Just in the last few months on the 28, have you changed your view on the 16 migration? One factor, some foundries, if they stop at 28, can address the 16. Are you taking a more aggressive approach? As part of your view on 28, are you getting more positive on 16, and you're trying to encourage more customers to migrate to that node where there's fewer players?
Your question is, are we going to repeat this kind of a 28 migration-
Well, just-
On 16?
In the last few months, are you also seeing, one, more customers demand to actually migrate to 16? Is it also more your strategy with some competitors not moving on, you actually have a bit less competition on 16 and below?
Yes, we have less competition in 16 and below. Yes. So long as you have one competitor, that's good enough, right? I mean, you have to work really hard to compete.
Okay, I guess on the customer side, are you starting to see more As part of your 28 coming down, you're actually seeing more customers with an urgency or a desire to actually do that shrink. You're seeing any change in that cadence of customers moving from 28 to 16?
Can you be more specific on your question?
You do feel like there's more capacity on 28.
Yes.
Flip side, are you seeing some of the demand you previously expected on 28 now shifting to 16 and 12?
Oh. Oh, okay. A lot of my competitor can do 28 nanometer, but not a lot of my competitor can do 16 or 14 nanometer, at least for today. So far, we did not see the same phenomena as a 28 nanometer. Does that answer your question?
Okay. Yeah, just more about demand. Do you see more customers now trying to make that shift to 16?
Oh, okay.
On the customer side.
It will continue doing this way, actually, that customer migrate faster than we expected into 16, right? You expect that in the 16 or 14 will migrate faster than we expected.
Yeah
into 10 or 7?
No, the question is now 28, you're more worried about oversupply. Is part of that from you're seeing more customers migrate to 16?
Yes.
Okay.
The answer is yes.
Okay. Thank you.
They migrate into 10 and seven also. They're faster than we expected. Yes.
Yes. Thank you.
Okay. This will conclude our conference today. Please be advised that the replay of the conference will be accessible within four hours from now. Transcripts will be available 24 hours from now, both of which will be available through our 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.