[Foreign Language] Welcome to TSMC’s First Quarter 2016 Earnings Conference and Conference Call. This is Elizabeth Sun, TSMC’s Director of Corporate Communications and your host for today. Today’s even is webcast live through TSMC’s website at www.tsmc.com. If you’re joining us via 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 first quarter of 2016, followed by our guidance for the second quarter. Afterwards, Ms. Ho and TSMC’s two co-CEOs, Dr. Mark Liu and Dr. C.C. Wei will jointly provide our key messages. Then we will open both the floor and the line for Q&A. 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 website at www.tsmc.com. Please also download the summary slides in relation to today’s earnings conference presentation. As usual, I would like to remind 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. And now I would like to turn the podium to TSMC’s CFO, Ms. Lora Ho, for the summary of operations and current quarter guidance.

Thank you, Elizabeth. Good afternoon, everyone. Thank you for joining us today. My presentation will start from financial highlights for the first quarter, followed by the guidance for the second quarter. First quarter revenue came out slightly better than we have expected. Although the February 6 earthquake caused some delay in wafer shipments in the first quarter, we saw business upside resulting from demand increases in mid and low-end smartphone segment and customers inventory restocking, therefore our first quarter revenue was essentially flat from the fourth quarter last year compared to about 2% decline we guided three months ago. Our first quarter gross margin was 44.9% and operating margin was 34.8%, both are 3.7 percentage points lower than fourth quarter last year due to lower utilizations and the negative impact from the earthquake, which I will provide more detail later. Overall, our first quarter EPS was $2.50. Now, let’s take a look at revenue contributed by applications. During the first quarter, communication and consumer increased 4% and 15% respectively, while computer and industrial/standard decreased 7% and 9% respectively. Now let’s take a look at revenue by technology. 16 and 20 nanometer contributed 23% of our total wafer revenue in the first quarter. In addition, 28-nanometer saw a nice rebound in demand and contributed 30% of our total wafer revenue. Together, these three advanced technologies accounted for 53% of total wafer revenue. Now moving in to the balance sheet. We ended the first quarter with cash and marketable securities of NT$648 billion, an increase of NT$62 billion. On the liability side, current liabilities increased NT$24 billion. The increase included the reclassification of NT$10 billion bonds payable from long-term to current. During the year of 2011 and 2013, TSMC issued corporate bonds totaling NT$211 billion to support a capital expenditure. These bonds…

Good afternoon, everyone. Let me start to give you a report on our new turn demand and inventory outlook. Our 1Q 2016 revenue is essentially flat from 4Q 2015, however exceeding our February 2017 revised guidance. This result was helped by an increase in smartphone demand for China and emerging markets pull in from second quarter and faster recovery from the earthquake, which offsets the weakness and seasonality in the high-end smartphone sector. We estimate our fabless income days of inventory to increase from 2 days below seasonal, exiting 4Q 2015, to about seasonal level ending 1Q 2016. Looking ahead for 2Q 2016, we forecast a growth of 6% to 7% quarter-to-quarter in NT dollars or 8% to 9% in U.S. dollars. Most of our fabless customers may see an above seasonal growth in 2Q 2016, driven by OEMs’ new product launches and inventory restocking in the supply chain. We estimate days of inventory will still be close to seasonal level exiting 2Q 2016. Due to the world macroeconomic uncertainties, we reduced our estimate of 2016 smartphone growth from 8% to 7%, PC from minus 3% to minus 6%, tablet from minus 7% to minus 9%, while maintaining digital consumer electronics growth rate at minus 5%. In spite of the reductions of these growth rates, there are still growth areas in smartphones, broadband network, wireless infrastructures and gaming. For smartphone growth, momentum comes from China’s 4G Plus deployment, through increases of operator subsidy and the continued 3G to 4G upgrade in emerging markets. These factors will also drive the associated infrastructure growth. Recently, we also see good demand from gaming GPU and game console processors for VR applications. For the whole year of 2016, we estimate the growth of world semiconductor to be about 1%. We maintain our estimate for…

Thank you Mark. Good afternoon, ladies and gentlemen. Today I will update you on the status of 16-nanometers, 28- nanometers, InFO and specialty technologies. I will also talk about their role as a growth driver. First one 16-nanometers, we continue to ramp 16-nanometer with yield and cycle time better than our target. Applications in mobile processors, cellular baseband, video game player, PC graphics will contribute significantly to our 16-nanometer shipment this year. Meanwhile we also begin shipment for other applications, such as the Ethernet switch, CPU, networking processor, programmable logic device and others. We expect 16-nanometer will contribute above 20% wafer revenue this year. As I reported here last quarter, we have completed the development of a low-power and low-cost version 16FFC. I can inform you today that 16FFC is now the most adopted solution by our 16-nanometer customers and we have entered mass production since first quarter this year. In addition to 16FFC, we also developed an ultra-low-power solution, with much lower operating voltage down to a level below 0.5 volts without too much sacrifice of the circuit speed. We believe this is a superior solution compared to other technology approaches available in the low-power area today and is ideally suited for the mobile and IoT market. Both mobile and IoT will be the main growth driver in the future. Now let me move on to 28-nanometer. Since our last conference, we have seen strong demand for our 28-nanometer, coming mainly from the mid and low-end smartphone related applications. As a result, the utilization rate of 28-nanometer has remained well above 90%. We think it will stay at a high level throughout this year. Key to this strong demand has been TSMC’s 28HPC and 28HPC+. In addition to the mid and low-end smartphones, applications related to networking processors and…

All right. This concludes our prepared statements. Before we begin the Q&A session, I’d 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 floor and from the call. Should you wish to raise your question in Chinese, I will translate it to English before management answers your question. [Operator Instructions]. Now let’s begin the Q&A session. First we will ask Credit Suisse, Randy Abrams.

Thank you. First question, just want to ask about the business outlook. For second quarter it looks like somewhat mild growth, so like mid to high single-digit growth, Just factoring in the earthquake, some shipments pushing out, smartphone lines chasing relatively lean inventory, could you talk about some headwinds you may be facing just to offset some of those positive drivers? And then if we look at the second half, what does that imply factoring in the 16-nanometer and InFO starting in second half? How do you feel about second half at this stage?

Well the second half, the headwind is not much different than the first half – the second-quarter headwind is not much different than the first quarter. But we do have – we do see the customers and even the end market demand is increasing. The inventory restocking is still cautious. And going forward, we’re still watching this month by month to see if demand will change or not. But this is currently our perspective.

Okay. And if I could ask, for the forward technology, you’re doing a much faster move to 7-nanometer, where it’s a new bump every one year, where so many other foundries, like Intel, is going to an optimized, taking longer. And I’m curious on two fronts. One if there’s any trade off or risk to your strategy moving faster. And the 7, your density shrink is about a 60%, so not at full shrink. If, as we go through the node, if competitors could leapfrog with a full 7 nanometer or you can upgrade the process as you move through. So maybe talk about that strategy for what you’re doing on 10 and 7.

Well I cannot tell you all our strategy. But we think we are developing a technology fit for our customers’ product announcement. So all this strategy and roadmap is collaborating with our customers at the shortest time to be able to reach the maximum technology development benefit for their products. So that has been the past two or three generations, but I don’t see no difference in going forward.

A quick follow-up. Can you put EUV, and you gave the update, but could you bring that in partway through 7 to enhance the process, say, partway through? Or is it more targeted now for 5 nanometer?

Yes. Our EUV, even though the recent progress is good, however, as you see, our 7-nanometer development is already on full steam and is getting to real enhancement mode now. And production is imminent. It’s beginning of 2018. So looking at this schedule, EUV is hard to reach for – be able to contribute. So for N7, it is purely a development vehicle for EUV. We do plan to use EUV on our N5. And that is our plan. And we will – we look at the schedule here that will fit properly EUV. Given the margin, it still will fit. It depends on the following progress for EUV.

Thank you.

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

Thank you. Regarding your comment on 10-nanometer, so you mentioned you have more than 20 customers for 10 nanometer. Am I right?

That’s it. That’s it.

Yes. So does that mean you actually see the stronger-than-expected 10-nanometer compared to six months ago based on your progress? Because in the past it seems that – I remember management mentioned 10-nanometer will be smaller than 16-nanometer. So based on latest progress, are you seeing the better demand outlook for 10-nanometer?

When I talked about 20 customers, it’s on 7-nanometer. Okay? So 7-nanometer indeed, we see stronger adoption in the past quarters. So that’s what we are – I was addressing.

Okay. Second question is for IoT. Is any way management can quantify the IoT sales portion or – because it seems that a lot of people are saying IoT will be an important growth driver in the long term but we still cannot see a lot of IoT products coming out, and mainly for infrastructure buildup. So is there any way you can try to quantify?

Weill I don’t think that we can really quantify what is the IoT market today. But a lot of products that, for example, connect the sensors that essentially that IoT’s most important product. One of the IoT’s most important is the sensors, including CMOS image sensor and the MEMS that you can see today. But you want to quantify the whole IoT market? No, we cannot.

One final question, is that will some of the IoT use advanced nodes or will it – they just use the 28 or 16 in the long term?

It should be because the IoT related to a lot of data transmission, data communication. So the computation, the high-speed computation is one of the essential parts of the IoT area.

Thank you.

Next question will be coming from JPMorgan’s Gokul Hariharan.

Thanks. My first question is on a little bit of granularity on the 2Q outlook. Could you talk a little bit about advanced mode versus older node given that there is a lot of restocking at 28 and below? Is it fair to say 16-nanometer is probably weaker than other nodes in second quarter?

All right. I think you are asking us about the restocking in the second quarter, whether it is on 16 nodes or –?

If it’s primarily on older nodes to date or 28 and above.

Whether the restocking is on the advanced nodes or across the board.

The restocking is mostly on 28, 40, 65.

Second, I think that you mentioned about cloud computing, high-performance compute as a potential market and you’re working with ARM on 7-nanometer high-performance compute product as well. I think here, a couple of years back, you mentioned that software ecosystem readiness has still been an issue for this market. Any thoughts on what you’re seeing from your customer side on this front? How far they have gone and how is this market likely do more given that – as you said, right now it’s not really interested in the meat of the market, which is servers; it’s primarily on the periphery.

Their momentum has collected quite a bit recently. As you hear from Red Hat, with the software developer and from the – some of the customer software alliances, as you know, has collected quite a bit. And there are services, cloud services for application software development on several sides. So the application software development also are progressing. So those part of the momentum we see increases. And for TSMC, of course, we will support our customers in that high-performance computing platform. For us mostly it’s the technology development, technology development also including the interface circuit development with our customers, also including the packaging solutions for that application. So I think that compared with a couple of years ago, indeed you are seeing a lot of reports that momentum is collecting quite a bit recently. So all the parts have received drivers for that into the high-performance computing market.

And just one follow-up. If you have to take a guess right now, is that going to be mostly machine fabless or system fabless, like you mentioned as the potential growth driver in the last couple of calls.

We see both. We see both. Including the service provider. They participate too.

All right. Next question will be coming from Citigroup’s Roland Shu.

Thanks. Good afternoon. First question is for the second quarter guidance now is TWD215b to TWD218b. So if we exclude the push-out, push-out should be – the shipment pushed out from first quarter, that should be about TWD7b. Right? So that actually is about TWD208b to TWD211b. So that actually is much smaller. Then if we take March, March 3, that’s TWD219b. So therefore the growth in the second quarter, for this growth, I think it’s this – what you expected early this year or are you going to see – because of some weakness on the shipment?

There’s a factor that affects our second quarter. One second. It’s exchange rate. We’re using TWD32.30, which is very close to today’s market. And in first quarter our exchange rate was TWD33.1. So there’s a 2.2% depreciate – appreciate. So that affects 2% of the revenue. Another reason is what Mark was mentioning, that we see the inventory restocking continues throughout the second quarter. So that’s another factor that we have. And a weaker-than-expected second quarter.

Okay. And also for the shipment delay, I think that according to your news release previously, you said about 100,000 of 12-inch wafer shipment had been delayed. And however, I think earlier you said you probably shipped about 90,000. So why is the difference between this 10,000, because they were cancelled by customer or we are going to further delay to 3Q?

No. Our people work really hard to recover the customer shipments. So actually we do better. I think we have announced 120,000 in our press release and it s around 90k. So it’s about 30% lower.

Okay. And I think my second question is again Gartner just reported for TSMC last year, the overall foundry market share was about 64.5%. So I think for this year, we are recovering 16-nanometer market share aggressively. And also – we are also going to capacity in 10-nanometer, with very high market share next year. And I believe for the 7-nanometer capacity in 2018, that that will be much ahead of the peers. So question is actually the three-year technology leadership, what do you think your growth – your market share will be?

Well, the data is ahead – still ahead of us. So I cannot claim any territory yet. So we try to improve our market share just according to what you mentioned, the factors.

Well is 50% of market share achievable?

Anything is achievable.

Of course.

At least 50%?

That’s not what I said.

Okay. Okay. Thank you. I think with this three-year technology leadership started, are you going to have an upside for your PBT target? I think now you said for five years the PBT target to grow by 10% in CAGR. So with this technology leadership, any upside on this PBT target?

Well the competition is still ahead of us. Right now it’s just not the time to talk about what the trophy is.

Thank you.

Next question comes from Daiwa’s Rick Hsu.

Yes. Hi. Good afternoon. Just one question from me. About your second-quarter guidance, I think it looks to me it’s a bit below seasonal in terms of sequential growth momentum. And you also mentioned that the MEMS is mainly from the 28-nanometer, 40- and 65-nanometer, not much for the 16-nanometer. And you also mentioned that you 16-nanometer for this year, the full-year revenue contribution will come to above 20%. Can I take that as a – your Q3 momentum is going to be really strong and likely above seasonal?

Yes.

Can you elaborate a little bit more?

Actually the high-volume ramp we expect in the second half of this year for the 16 FinFET. That’s because of leading-edge or the high-end smartphone. And other applications actually in the 16 FinFET continues on and we saw a lot of tape-out since last year. So the momentum accumulated and I believe in the second half of this year we will see higher growth.

Thank you so much.

We – I think we need to go to the line for some questions. So, operator, could you please have the first caller on the line? We’ll take the next question from the call.

Our next question comes from the line of Mr. Donald Lu of Goldman Sachs. Please ask your questions.

Good afternoon. I have two questions. One is on InFO. I think C.C. just commented that you package more chips and the cost will come down. And in what year, which year do you think the InFO cost will be similar or even lower than the conventional packaging? And also assuming it will happen with 10-nanometer, 7-nanometer then, will TSMC expect your mobile market share to increase significantly at 7-nanometer versus 10-nanometer or 16-nanometer global market share? So that’s question number one. Question number two is on the – in second quarter, TSMC’s 16-nanometer utilization is going to decline from Q1. Thank you.

So Donald’s first question is if we can package multiple chips in InFO, which is the chip partitioning and reduce the cost, he asked which year we will begin to see that the cost on InFO is lower than the cost from conventional packaging. And if we can apply this very useful technology on 10-nanometer and 7-nanometer, does that mean that our mobile market share will be higher at 7- and 10-nanometer compared to our mobile market share at 16-nanometer? That’s first question. Second question is will do have a utilization decline at 16-nanometer in the second quarter.

Okay. Let me answer the first question on the InFO. I mentioned that the InFO was of the chip partitioning and that will serve reducing the cost. Actually this kind of thing will incur the design architecture difference and changes. So we are working with the customer. I cannot give a very certain date and which year that it will be in the market and then compare with the conventional packaging method. But our analyses show that in any year when we introduce this kind of technology and this kind of new approaches, it certainly will be better in performance and in cost as compared with conventional packages. Whether it will help us to increase market segment share in mobile, certainly we hope so. But I cannot give you a number.

But if you could – sorry. Just particularly at 7-nanometer would you see new customers will adopt TSMC’s 7-nanometer together with InFO at this stage?

We are working with customers. That’s all I can answer you questions. Okay. So the second quarter our utilization rate for the 16-nanometer…

16-nanometer

Decreasing?

Decline.

No. That’s it.

Hi. I can add some comment on 16-nanometer. We are ramping at a very fast speed on 16-nanometer across the whole year. So the second-quarter utilization for 16-nanometer will be much higher than first quarter.

All right. So we will go to the next caller on the line. Operator, please help.

Your next question comes from the line of Mr. Brett Simpson of Arete. Please ask you questions.

Yes. Thanks very much. I just had a quick question on 28-nanometer. We saw a nice recovery there in Q1, but you have a lot of 28-nanometer customers migrating to 16-nanometer going into second half 2016. So how do you see 28-nanometer outlook as you start to see this big shift in mobile, low-end smartphones, gaming? What impact do you think that has in your 28-nanometer outlook and what it is might backfill that 28-nanometer node? Thank you.

So, Brett, your question is even if we are seeing very good recovery of 28-nanometer in the first quarter, but thinks a lot of the customers’ products will be migrating to 16-nanometer in the second half. You are asking us what kind of applications or business we can have to sell 28-nanometer in second half of the year. Is that your question?

That’s right. And whether you think 28-nanometer can maintain these revenue levels.

Okay. Let me answer the question. Actually, he asked whether this 28-nanometer, the strong demand can continue or not because of smartphones adopting the 16-nanometer technology quickly. Actually that in the smartphone we have high-end smartphone, mid and low-end smartphone. For the high-end smartphone moving to the 16-nanometer, it’s a natural choice and they are moving very fast. Yes. But then the mid and low-end smartphone, that demand increases, so a lot of them adopting TSMC’s 28-nanometer technology. That’s where the very strong demand comes from. And we see that strong demand throughout this year. In addition to that, I just mentioned that we also have the networking processors, consumer products. They are all adopting TSMC’s 28HPC/HPC Plus. Furthermore and in the future, we saw that IoT mobile applications for the low power consumption is very important. And TSMC offers a very competitive platform in this area. So we believe that 28-nanometer demand will continue to be strong. Thank you.

Thank you. And I just had an industry question as my second question. There’s a lot of talk about the rising cost of developing chips at the leading edge. How much do you think it costs your customers to develop chips at 10-nanometer and 7-nanometer compared with 28? And I think you mentioned 20 customers are – you’re engaged with 20 customers on 7-nanometer. Can you talk a bit about how big the customer engineering teams are that you’re engaging with at these nodes? Thank you.

Well I cannot comment on the exact number of the NRE engineering expense for the new end. Of course, each node, the development cost is higher. And – but on the other hand, the chip integration is also increasing. Where it used to be several chips, right now is being developed by one – in one chip. So secondly, actually, although we have – I’ve mentioned 15 tape-outs, typically the first tape-out incurs the highest cost and the following tape-out doesn’t – it can be synergistic. So indeed, customers see the 7-nanometer is a longstanding, long node. Therefore I think the customer is willing to put into their investment resources in it for the – for many years to come.

Thank you. And could I ask a third question?

All right.

So maybe just – I wanted to ask on graphics, because you talked a number of times in your prepared remarks about VR and graphics and game console. And there’s quite a big performance jump coming as some of these customers go to 16-nanometer later this year. What sort of growth are you seeing in gaming and graphics specifically at the moment this year? Thank you.

VR and AR, it still is an emerging application. So the growth rates come from very small number. So probably I cannot put into accurate number. Currently this application, this product is under the application software development, I think is being tried, several VRs and AR products. Many of the application software is a demo software. And both – all the users of the chips right now create their software development platforms. So this is at this stage. So in terms – we just see it starting shipment, but it’s yet to be a volume quantity this year for this application.

Okay. Thank you.

All right. We will continue to have the next caller from the line. Operator, could you please have the next caller on the line?

Your next question comes from the line of Mehdi Hosseini of SIG. Please ask you questions

Yes. Thanks for taking my question. Going back to your prepared commentary where you describe smartphone unit growth, which is going to be less than your prior expectation, which is also, in my opinion, is having an adverse impact on your assumptions for semiconductor revenue growth. How is it going to impact your revenue expectation growth for this year? Your – the range of 5% to 10% has remained unchanged. But the largest end market smartphone you have down-ticked. And then the semiconductor industry revenue is also going to be up by only 1%. So in that context if you could help us understand how this downtick is impacting the range. Do you see your revenue coming to the low end or the high end? Or are you still on track to hit the midpoint of the guidance given all the programs that are coming to the fruition?

All right. I think the question was that since we have revised down some of the growth rates, like smartphones and semiconductor revenue, etc., how would TSMC still keep our revenue guidance. So he was saying whether we are on the high end of the guidance range or low end of the guidance range, etc.

Well on semiconductor growth, we forecasted 1%. And part of it is memory market appears to be shrinking this year. Excluding memory, will be about 2%. But our system is very complicated. It’s very difficult for me to describe everything for you here. But the smartphone growth this year is mostly from the mid end and the low end in terms of unit growth, so is that one to reconcile for your numbers. Our number for this year, we maintain 5% to 10%. That is nor aggressive nor conservative. It is still the same number.

Sure. Let me rephrase the question. Historically you’ve had more content in high-end smartphone but the growth rate for low to mid range is much higher than the high-end smartphone growth rate. Would it be safe to assume that if the growth rate for the low to mid range is upsetting lower growth rate for high end, even though the high end gives you higher content?

It’s possible. We see the over-TWD500 phone is reducing, but the TWD400 phone is increasing quickly. And both high end and mid end silicon content are increasing with high single digit. But the unit of the mid-end phone increases and high-end unit number decreases this year.

Okay. Then I have a follow-up – I have a follow-up to my second question.

Yes?

So there’s been a lot of emphasis on this, on 7-nanometer and your progress. It’s great to see you have maintained your lead. But in that context should we assume that the 10-nanometer will be a shorter node? And if so, what happens to backfilling? You mentioned the 90% of the equipment for 10 and 90% of the 10-nanometer equipment could be used for 7-nanometer. Is that going to have an impact on your spending which ties into backfilling? Any comment or any color there would be great.

Okay. Mehdi’s question is that he’s happy to see that we have a leadership in 7 – at 7-nanometer. However, does that mean that we will have a short node for 10? And if it is a short node, what would happen to the common tools, this equipment being 90%-plus common? What’s the impact to our capital spending?

I think this 10-nanometer will ramp very fast, but it appears to be a shorter node than that of 7-nanometer. So the picture could be similar to our 20 and 16. And also we anticipated before enough that we try to maximize the common tools from 10-nanometer to 7-nanometer. Therefore this is 95% is in – is by design so that when the 10-nanometer, after three years, it reduces, the tool can be readily expanded into the capacity for 7-nanometer ramp.

The impact to spending.

The question is impact spending?

Impact to our capital spending.

That has been – long been planned into our capital expense. It’s an indicative number what we offer to you.

Okay. So now we come back to the floor. And the next one will be from UBS, Bill Lu.

Thank you very much. Going back to InFO. I’m trying to figure out when this might broaden out. And I’ve spoken to a couple of your customers and maybe some of the backend players. And what I’m hearing pretty consistently is cost and the fact that this is TSMC-proprietary. Other than that, people seem to love it. I know you’re working very hard on cost. In terms of this being TSMC-proprietary, I’m wondering what you are telling your customers to convince them. And I think it looks to me like it’s going to be TSMC with InFO, everybody else with some sort of other planar technology. Is that the way that you see it developing? And number two is would you consider possibly licensing InFO, because we’ve certainly seen success stories in terms of licensing in semiconductors. I think that would help out your customers. Do you just want to give your thoughts on that? Thanks.

Okay. We developed InFO technology that – with much lower cost than the cores that we introduced earlier. And InFO give advantage, as I said in the presentation. We have mentioned in the last quarter that we only focus – right now we only focus on the few high-volume customers because we start to ramp it up. And I also said in the presentation that we are working on the yield improvement and the cost reduction, thus meaning that today we still have some improvement ongoing. And once we complete it in our -volume production this year, we believe it will be very competitive. However, in every technology node we work with all strategic people and to better serve our customers. For now we also see that the cooperation between TSMC and backend people. But for this year, before we complete the volume production and ready for other consideration, this year we only focus on our own – bringing enough of this InFO volume production.

Yes. Thank you. That’s very clear. I’m guess I’m just wondering what happens beyond this year as you start thinking about more customers.

We will tell you that next time.

Okay. So my second question was what Mehdi just asked, but maybe I’ll just follow up. If it turns out that – 10-nanometer turns out to be a little bit smaller than what you expected previously but 7 turns out to be much bigger and you only spent TWD1b in the first quarter this year versus your guidance for the year, TWD9b to TWD10b, is it a possibility that maybe the CapEx becomes very backend-loaded or some of it gets pushed out into 2017?

It is true and we spent much less in the first half. And this year the CapEx will be very much backend-loaded, the reason being because we are preparing the capacity for 10- and 7-nanometer, though spending gets mostly spent in second half of the year.

All right. The next question will be coming from Morgan Stanley’s Charlie Chan.

Thanks for taking my question. So a very quick follow-on on Bill’s question on InFO. So what will be the margin impact to your corporate margin from the InFO business when you get to a mass production?

Since the InFO volume is still small for this year, and as C.C. was mentioning above $100m by the fourth quarter, compared to the total revenue it’s still small. So it will have very small dilution to this year’s corporate average margin.

Okay. So as yet the margin is slightly below corporate average, even it is small?

It is below.

Okay. Thanks. And my second question regarding the EUV adoption. So is it like it is already a done deal that you will adopt EUV at 5-nanometer, so the production timing is sometime like 2020? So from now to 2020, do you think the capital intensity will increase? I know the Company gave guidance before that the new node will be 3% – 35%. But in terms of the ratio, how do you think your growth, especially when you mass produce 7-nanometer? Thank you.

Well I’ll try to answer what I heard from you. Yes, we plan to use EUV on our 5-nanometer. And we have already three yielding tools on the floor and the fourth is coming. So all these are using actively for the development. So until 2020, when N5 is currently planned into production, yes, it will be adopted. But just to remind you that in – if you think about CapEx, don’t think about everything we’re going to change to EUV. There are 80 layers and only a 10th – there are about 50/60 if you discount multiple patterning. Only the 10 of them roughly will use EUV. So a lot of tools still be able to common tools. So did I answer your question?

Yes. So I just want to clarify because I thought the quadruple patterning or multi-patterning will cause a lot of CapEx. But also you mentioned that you will use a common tool strategy to reduce the CapEx. So just mainly in terms of CapEx, in terms of absolute versus 10b level or in terms of capital ratio versus 35%, in the coming four years do you think you will increase or decrease? So that was essentially my question.

Cannot be so specific as you said. But in general speaking, conversion of tool is cheaper than buying a new tool. So more conversion is beneficial to the CapEx.

Okay. Thanks.

All right. The next question will be a follow-up from Credit Suisse, Randy Abrams.

Yes. Thanks for fitting me back in. For the second half you mentioned a lot of the ramp will be 16-nanometer, which is still a relatively new node. In the second quarter you’re already at 49% to 51% gross margin. I’m curious, with further growth to come from that project, if 16 is approaching corporate average and we could get further leverage to the model in the second half from that.

For this year, 16-nanometer profit margin is improving as we have much higher volume. But it’s still below corporate average this year. We expect the 16-nanometer will get close to corporate average by second half of 2017.

Okay. I guess for second half 2015, net out the ramp of 2016 still below corporate, can we get some incremental leverage from that filling up utilization that we normally see? Like from the first half to the second half if there’s a potential for further margin expansion, would we normally get pick-up from utilization improving?

Our utilization will be better in the second half. That’s number one. And we have other technologies who have high utilization with very good profit margins. So that’s the next issue.

Okay. Okay. Thanks. The second question I want to ask, you’ve mentioned, Mark, in your prepared remarks about the server opportunity and development moving on software applications. If you could think about for your technology when you think we start to get the first volume work and, say, get to 1% or 2% of revenue, if that’s timed, say, around 7-nanometer or you think it’ll take more time, it might be further out?

We think it’s on 7-nanometer that we will see significant inroad.

Okay. Thank you.

All right. We need to go back to the line. Operator, could you please have the next caller on the line?

Our next question comes from the line of Mr. Steven Pelayo of HSBC. Please ask your questions.

Yes. Just first a follow-up there. On capital spending you mentioned it being backend-loaded this year. I’m curious, does that, from a run rate basis, really start to suggest that may happen in 2017 or do you think 35% of sales is a normalized CapEx level even though you’re getting high reuse rate at 16 today but maybe not so high at 10-nanometer before 7 kicks in? So CapEx to sales next year, I’m seeing the 35% range. What do you think?

All right. The question is this year’s CapEx is backend-loaded like 35 to 65. Would the same pattern, a similar pattern in 2017? And what will be our capital to – CapEx-to-sales ratio next year?

I probably cannot tell about the 2017, whether it’s going to be frontend-loaded or backend-loaded. But in terms of capital intensity, we have gone through this high-intensity period since last five years and it has come down to mid-30 range. And we believe this 30% to 25% range will be for the next few years.

Okay. Excellent. Thank you. In your 20-F filing this week we see that your largest customer fell from 21% of revenues to 16% of revenues. You obviously made it up with some other customers and with good growth last year. But I guess my question is expectations as you go through 2016, do you – do we have to worry about that kind of significant decline in your largest customers continuing?

All right. The question is we used to have a very big customer that accounted for 21% of our revenue last year and this year is probably in the mid teens. What will be in the future? What is the percentage drop? What’s the impact?

Yes. Just to clarify a little bit. I think it was in 2014 it was 21% of revenues. 2015 it was 16% of revenues. I’m curious if you’ve any concern over that trend continuing. Could they fall to single-digit percentage of revenues?

Well we will support our customers fully and let them grow.

All right. I think, Steven, I think we have answered your questions. Thank you. Operator, could you please have the next caller on the line?

Our next caller comes from the line of Mr. Donald Lu of Goldman Sachs. Pease ask your questions.

Hi. I have a question on 28. I think the conventional wisdom is that 28 will be a long and a very strong node. But I checked in Q4 last year, 28, it’s only maybe less than 5% higher than 65-nanometer at the same stage of the node. So my question is have you done any study to see how much bigger 28 will be for the next few years relative to 65? And the reason I ask this is a lot of your competitors are actually increasing CapEx this year for 28. So I’m afraid that we might have a price war.

Okay. Donald’s question is that 28-nanometer last quarter, 4Q 2015, which I think we said it was 25% of our revenue, and he said that this is only 5 percentage points higher than 65-nanometer at the same stage. So since many of our competitors are building up capacity for 28-nanometer, Donald is asking us how big can 28-nanometer really be and whether we will have a price war.

Well I cannot to speak specifically for the quantity, but I say that we saw the strong demand that due to TSMC’s technology very competitive. And for this year that the mid/low-end smartphone increased a lot. And then we also developed a very competitive low-power technology which we think is much better than our competitor. And so we continue to have confidence that we will continue to have strong demand and hold our market share.

But in the next two years, if you look out, compared to 65, do you think 25 – 28-nanometer will have 10% more revenue than 65 at the same stage after launch, or it’s 5%, 10% or 15%? Is there a number that we can use in our model, approximately?

Donald, this is Elizabeth. Let me answer your question. This percentage basically is not directly comparable because the base is different. 65-nanometer 30% back in, say, 2012 is a much smaller number than 28% – 28-nanometer at 30% in today’s number. In fact, if you look at same stage, say, 65-nanometer after five years and 28-nanometer after five years, 28-nanometer in terms of dollar is almost twice as big as 65 in terms of dollar

Actually, I’m sorry, I was – I’m comparing if I compared 65-nanometer revenue, absolute numbers of all the foundries I can find, versus the same foundries, total 28-nanometer revenue. So the total 28-nanometer revenue in Q4 last year was only 5% higher than 65-nanometer, total 65-nanometer revenue at the same stage after launch.

So you are ask – you are talking about market share percentage not revenue percentage?

Exactly. Not the market share but also the total absolute amount. Because there I think TSMC has much more information than us in terms of looking forward, in terms of tape-outs, et cetera. So using 28-nanometer is a long node or stronger node than 65 and how much, and for the whole market…

Well, let me answer the question again. It seems to me that you are not so confident on the 28-nanometers of demand, but it is strong. And this is a very sweet node, offered with low power but very competitive performance. And the cost structure is also good. So I want to say again that TSMC, the technology is very competitive. And we all hope our market share that we covered it some.

Let me add some comment to this one. Donald, I think you are trying to figure out how big is 28 versus the 65-nanometer. Is that what you’re asking for?

Exactly. That is for the whole market. Yes. That’s my question.

Okay. I think I don’t know about the 10, but for us, if I look at the number, 65-nanometer revenue versus 28-nanometer revenue, if you look at high – the peak revenue versus 65-nanometer, I would think that 28-nanometer revenue is much bigger than 65-nanometer at the same stage.

Okay. Yes, I will follow up with Elizabeth on the question. Yes, I think the 28 is ramped up very fast and dropped very quickly too for the whole industry. So the question is going forward it may – whether it will be bigger, how much bigger it will be versus 65. But I will follow up. Thank you.

Thank you. Let’s come back to the floor. Next question comes from Citigroup’s Roland Shu.

Thanks. Yes. Just a follow up on the InFO question. I think I feel just as a part of collaboration, with InFO, I would say likely candidate. But I would like to hear some other from you about the competition for the InFO or InFO like. Are you seeing – is there any close competitor who can do InFO capacity as what TSMC is doing in the near term? Thank you.

Well there is a lot of fan-out technology available today. TSMC developed the InFO because of – we developed technology that we have some certain advantages. Just like I mentioned, we have single thickness that’s suitable for your smartphone, because today’s smartphone needed to be thinner and thinner. And also that we found out our approaches can give benefit or advantage for the higher bandwidth that which is important for the smartphone for all the applications also. So we believe among all fan-out technologies, TSMC’s InFO is the best. And when we ramp it up, that will be the highest volume of fan-out technology in the industry. But our main progress is to hear about customers to gain – to have their product into the market, with very competitive performance and cost.

Thank you. So the conclusion is I think that you think probably it’s a very big chance for any competitor to do a similar job at TSMC to compete with you at the same customer or same application. Am I reading you right?

I don’t want to say that, but you might be right.

Thank you.

Okay. A follow-up question from Deutsche Bank, Michael Chou.

Hi, C.C. Is there any way you can mention what other modules be using InFO as well, after transceiver, this kind of product?

Yes. To answer your question, today we’re focused on a very limited customer with a high volume. But in the future, of course, it will be adapt by a lot of customers in a lot of different applications. That we expected.

Thank you.

Okay. I think in the interest of time – sorry, there’s a question. Sebastian Hou, in the back, from CLSA.

Hi. Thanks for taking my questions. So my first one is if you look – based on the inventory cycle, so we started to see some rush orders towards the end of last year. And first quarter we also have some business upside because of the inventory restock. And now the second quarter will continue to – you just comment that you continue to see some inventory restocking into second quarter. So that’s about –this is the fifth month or fourth month of inventory restocking. So where do you think that most of your customers are now in their inventory cycle? So apart from your largest customer, which are starting to bill soon, but apart from that, what about the rest of your customers?

Well what we see, the fourth quarter last year, you remember, is people drastically reduced their inventory. We only see the restocking almost after the Chinese New Year. So that’s where you count the cycles. And what – I don’t see very, very – I don’t see signs of overbuild at this point. People are still cautious. So that’s what we see, the cycle. How long it will last, I don’t know. Depends on – I just mentioned – I addressed on my prepared message, even to the end of Q2 appears the inventory was still above the seasonal level. It’s restocking, but at above the seasonal level of the end of Q2. So that will be four months into the building. But it’s still a cautious condition.

Thank you. My second question is on your revenue application side, because you break down revenue into four parts. And one of that is the industrial and Flash standards. And we’ve seen very strong growth for the past four years, a 27% CAGR, very similar growth to the communication category. So I wonder, can you provide some of your expectation for the CAGR for the next couple of years? And if you can talk about what the specific driver inside that industrial Flash standard and what kind of probability is above Company average or above the other application? Thank you.

Actually the industry – industrial standard, for example, including the MCU, we include in that category. But – and power management in that category. Both of those components still also used in smartphone. So when we talk about smartphone growth in the past few years, it includes some of the – what we categorize into the industrial and standard. That’s the reason you see the growth. Part of it is contributed from the smartphone.

So what’s the growth outlook for the next couple of years?

I cannot forecast into the exact number for you. But it will be a growth situation. It will be still growth pattern, I think.

Will it be continuing similar to the communication application growth or will it be higher or lower?

That’s hard to predict. I hope the communication grows faster, but that’s hard to predict.

Okay. Thank you.

Okay. So before we conclude today’s conference, please be advised that the replay of the conference will be accessible within three hours from now. Transcripts will become available 24 hours from now, both of which will be available through our website. Thank you for joining us today. We hope you will join us again next quarter. Goodbye and have a good day.