Welcome to TSMC's first quarter 2014 earnings conference and conference call. This is Elizabeth Sun, TSMC's Director of Corporate Communications and your host for today. Today's event is webcast live via tsmc.com. If you're 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's Senior Vice President and CFO, Ms. Lora Ho, will summarize our operations in the first quarter 2014, followed by our guidance for the current quarter. Afterwards, TSMC's two Co-CEOs, Dr. Mark Liu and Dr. C.C. Wei, will jointly provide a couple of key messages. We will open both the floor and the line for the Q&A.
For those participants on the call, if you do not yet have a copy of the press release, you may download it now from 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. I would like to turn the podium to TSMC CFO, Ms. Lora Ho.
Thank you, Elizabeth. Good afternoon, everyone. Thank you for joining us today. I will start my presentation with financial highlights for the first quarter, followed by the guidance of the second quarter. You may have noticed from the information we provided to you today, starting from this year, we have changed the wafer unit to 12-inch equivalent. A 12-inch accounts for the majority of our production capacity. The first quarter came out better than expected. Our revenue, Gross Margin, and operating margin were all exceeded the revised guidance. In the first quarter, the demand for TSMC's wafer was much stronger than we had initially predicted in mid-January. The strength came from customers' better fourth quarter results, which led to a more positive outlook for the whole year, prompted to the active replenishment of inventory from a very low base in the fourth quarter last year.
We were able to capture the bigger market share of customers' upside demand thanks to the better performance and higher yield and reliability for our advanced technologies. First quarter revenue increased 1.7% on sequential basis and 11.6% on year-over-year basis to reach TWD 148 billion. Gross Margin was 47.5%, up 3 percentage points from the fourth quarter last year, mainly due to higher capacity utilization, a favorable foreign exchange rate, offset by unfavorable inventory valuation adjustment. Operating margin was 35.4%, up 2.6 percentage points from the fourth quarter. Non-op item was a small gain of TWD 0.78 billion. Overall, the first quarter EPS was TWD 1.85, and the ROE was 21.9% in the first quarter. Let's take a look at revenue by application. If you recall, the fourth quarter inventory correction was mostly serious in the communication-related applications.
When customer demand came back in the first quarter, communication shows the strongest increase. Compared to the fourth quarter last year, communications increased 8%, computer increased 2%, and industrial-related revenue increased 2%, while consumer declined 14% during the first quarter. By technology, 28-nanometer revenue continued to grow and account for 34% of our total wafer revenue in the first quarter. 40-nanometer has a nice rebound, now represents 21% of our total wafer revenue. The two advanced technologies, 28-nanometer plus 40-nanometer, represented 55% of our first quarter total wafer revenue, increased from 51% a quarter ago. Now let's move on to the balance sheet. Cash and marketable securities decreased TWD 10 billion sequentially to TWD 235 billion at the end of first quarter. Current liabilities decreased by TWD 15 billion, mainly because of the decrease in accounts payable to equipment suppliers.
Meanwhile, we borrowed TWD 9 billion in short-term loans for hedging purpose. Accounts receivable turnover days decreased three days to 45 days. Days of inventory increased to 52 days, reflecting the strong demand in the second quarter and the starting of 20-nanometer ramp. Now let me make a few comments on cash flow and CapEx. During the first quarter, we generated TWD 95 billion cash from operations, invested TWD 115 billion in capital expenditure, and borrowed TWD 9 billion short-term loans. At the end of the first quarter, our cash balance decreased TWD 11 billion to TWD 232 billion. Free cash flow for the first quarter was an outflow of TWD 20 billion due to higher capital expenditure in the first quarter. Our CapEx and capacity. First quarter CapEx was US$3.8 billion. We expect 2014 CapEx will be about US$10 billion, with front-end loaded pattern.
About 70% of the budget will be spent in the first half to support the quick ramping of 20-nanometer. On a full year basis, we plan to increase our capacity by 10% from 2013. So the total annual capacity will reach 8 million 12-inch equivalent wafers. So I have finished my report on the financial part. Now let me turn into the second quarter outlook. Based on our current business outlook and the forecast exchange rate of 30.10, we expect our second quarter revenue to be between TWD 180 billion and TWD 183 billion. This would translate into around 22% quarter-over-quarter increase. On the margin side, we expect the second quarter Gross Margin to be between 47.5% and 49.5%, and operating margin to be between 36.5% and 38.5%.
At the January investor conference, I talked about the tax rate for the year will be about 13%, and the second quarter will carry more burden. Let me give you an update this time. According to the accounting principle, when shareholder approves the earnings distribution in June, we need to accrue the 10% retained earning tax for the undistributed earnings. Therefore, our second quarter tax rate will go up to 21% and then go back to the normal level of around 11% in the third and fourth quarter. Full year tax rate, however, will still be 13%. This concludes my remarks. Let me turn the podium to our co-CEO, Mark and C.C., for their comments.
Good afternoon, ladies and gentlemen. Let me cover the first three items on the agenda. Let me begin by saying our 2014 market outlook has improved since our mid-January investor conference. The first quarter is typically a slow season for our customers and for TSMC. Since mid-January, we started to see strong orders across all segments. We now have an improved demand outlook from the following three perspectives. First, the demand in smartphone appears healthier than we expected last quarter. The acceleration of LTE infrastructure buildup, LTE smartphone proliferation, and the increased silicon content of smartphone improve our demand outlook. Silicon content increases come with multi-core 64-bit application processor, multi-mode baseband, multi-band RF transceiver, image sensors, MEMS, near-field communication, and fingerprint. These are all included in the smartphone. The second, TSMC's 28 nanometer technology performance and quality suffice to fill more of those customer demand.
Fabless DOI was substantially below seasonal when we exit 4Q last year. We still expect fabless DOI will be below seasonal in 1Q14. For second quarter '14, we expect our demand will continue to be strong and above seasonal in all major applications. I have covered the market supply chain and demand of second half and full year for 2014. Since first quarter fabless DOI is below seasonal, the second quarter demand on us is unseasonably strong. We expect fabless DOI will return to seasonal level at mid-2014. Our second half demand would be more normal. TSMC will gain 28 nanometer market share with 28 nanometer High-K metal gate transition in the second half of 2014. TSMC also expect to gain overall foundry market share in second half 2014, when we ramp up 20 SOC in the second half 2014.
Since our first quarter and second quarter will establish a higher base, the quarter-to-quarter growth of third quarter and fourth quarter will both be positive but will be more moderate than our second quarter. For the full year of 2014, our outlook improves from last quarter as follows. Semiconductor revenue growth from 5%-7%. Fabless revenue growth from 8%-9%. Foundry revenue growth from 10%-14%. For our growth for full year of 2014, it will be higher than the forecasted foundry growth by several percentage points. I cover the updates on 16 FinFET, 16 FinFET Plus, and our 10 FinFET. First, we have two general offers for customers, 16 FinFET and 16 FinFET Plus. 16 FinFET Plus offers 15% speed improvement, the same total power compared to 16 FinFET.
More importantly, 16 FinFET Plus offers 30% total power reduction at the same speed compared to 16 FinFET. Our 16 FinFET Plus matches the highest performance among all available 16 nanometer and 14 nanometer technologies in the market today. Compared to our own 20 SOC, 16 FinFET Plus offers 40% speed improvement. The design rules of 16 FinFET and 16 FinFET Plus are the same. IPs are compatible. We will receive our first customer product tape-out this month. About 15 products planned for 2014. Another about 45 in 2015. Volume production is planned in 2015. The 95% tool of 16 and 20 are common. We will ramp them in the same GIGAFABs in TSMC. 16 FinFET yield learning curve is very steep today and has already caught up with 20 SOC. This is a unique advantage in TSMC 16 nanometer.
10 FinFET, 10 FinFET offers TSMC's third generation FinFET transistor designed to meet the power and the performance requirement of mobile computing devices. 10 FinFET will offer greater than 25% speed improvement, the same total power compared to 16 FinFET Plus. More importantly, 10 FinFET offers greater than 45% total power reduction at the same speed compared to 16 FinFET Plus. 10 FinFET will offer 2.2x of density improvement over its previous generation, 16 FinFET Plus. Currently, 10 FinFET development progress is well on track, but risk production will be in 4Q 2015. Above are the key messages on the three items. Thank you.
Thank you, Mark. Following Mark, there will be two remarks made by C.C. Wei.
Thank you, Elizabeth. Good afternoon, ladies and gentlemen. While you mark the exciting forecast on our second quarter and the second half of year 2014, I would like to take this opportunity to share with you two topics with you. The 20 SOC ramp and TSMC's advanced assembly solution to our customer. First, I will brief you on the status of 20 SOC ramp. Let me recap what we have said in last meeting here. We started 20 SOC production in January this year, and by first quarter of this year, the 20 SOC will account for 20% of that quarterly revenue, wafer revenue. For the whole year of 2014, we expect 20 SOC will be about 10% of our total wafer revenue of the year of 2014, of course. All these expectations remain the same today.
There are some major achievement I would like to share with you. First, on the ramping speed. 20 SOC by far is the fastest ramping in TSMC's history. Of course, this fast ramp is to meet customers' strong demand. I believe with the production of 20 SOC in TSMC, we present one of the largest mobilization in semiconductor history. Let me share with some number so you can have a snapshot on this ramp. In about one years of time, we have built a manufacturing team of 4,600 engineer and 2,000 operators in two fabs, Fab 14 in Tainan and Fab 12 in Hsinchu. More impressively, in the same time period, across to 1,000 engineer has been relocated among TSMC's fabs in Hsinchu, Taichung, and Tainan. All these are prepared for the 20 SOC ramp-up. This magnitude of mobilization, I believe, is not an easy job.
We move people around that show our strength in manufacturing, this high mobilization is not moving the tool or just a handful around. We're talking about we're moving the engineer and operator among TSMC fab. Meanwhile, we have installed more than 1,500 major tools for this 20 SOC ramp. Of course, the faster ramp has come with a very good device reliability and a very good wafer defect density. Without those, the fast ramp would make no sense. Now, how important of this 20 SOC ramp? Well, we knew that 20 nanometer provided the engine of TSMC's profitable growth in the years of 2012 and 2013. Similarly, we expect 20 SOC will provide the engine of TSMC's profitable growth in year 2014 and 2015. Now let me switch gear to advanced assembly technology.
The purpose for us to develop the advanced assembly technology is to provide our customer as a better performance at the lower power consumption, while at a lower cost as compared to the previous assembly solution. For example, we have developed CoWoS. CoWoS has been developed to connect two dies or more dies together to have a very high performance and a very low power consumption. Today, CoWoS is in a small volume production already. However, the cost structure of CoWoS has made CoWoS only suitable for some very high-performance application and the products. To address the cost structure issue, and for those very large volume mobile devices, we have developed a derivative technology called InFO, that stands for Integrated Fan-out. InFO will have significantly lower cost compared to CoWoS.
At the same time, InFO also can have the same capability to connect multiple dies together just as a CoWoS did. Currently, we are working with major customers on the InFO to incorporate this structure into their future product. We have delivered many functional dies to our customers already, and the process optimization are ongoing. In fact, we are very excited about TSMC's advanced assembly technology development, as we are building an innovative solution for our customer's product, which requires high performance, low power consumption, and has a very reasonable cost structure. Now let me sum up of today's key message, which is presented by Mark and myself. First, we have revised up 2014's forecast for the semiconductor industry and for foundry segment, as the outlook has become very positive.
More importantly, TSMC will be able to strongly outperform our foundry peers, and we will continue to increase our market segment share as we have done in the past four years. The demand from 28 nanometer and 20 nanometer will provide a foundation for our profitable growth and the stronger outperformance of this year. The 20 SOCs ramp is a new record. That is one of the largest mobilizations in the semiconductor industry. Our 16 FinFET, especially 16 FinFET Plus, are highly competitive, and we have a very similar good defect density performance already, which, of course, just like Mark said, is 95% similar to 20 SOC, which we already have a very good defect density performance. Our 10-nanometer technology development are on track, and we are working on a cost-effective, advanced assembly solution, InFO, for our large volume mobile devices.
We believe all these activities will pave the way for our continued profitable growth for the next few years. This ends our prepared remarks. Thank you very much.
Thank you, C.C. This concludes our prepared statements. Before we begin the Q&A session, I would like to remind everybody to limit your questions to no more than two at a time so that we can allow all participants an opportunity to ask questions. Questions will be taken both from the floor and from the line. Should you wish to raise your question 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. Let's begin the Q&A session.
First question, we will invite Goldman Sachs, Donald Lu.
Mark. smartphone semiconductor content? iPhone Galaxy high-end customer? semiconductor content last year and this year, or high-end, low-end? TSMC last year, this year, and next year, average semiconductor content question? 16-nanometer FinFET on track? ASML mobile FinFET delay? TSMC mobile SOC which company delay?
All right, well, since Donald prefers to use Chinese, then it will be my job to translate to English first. First question Donald has is with respect to smartphone semiconductor content. He likes to know what is the semiconductor content for the smartphone on average in 2013 and 2014, and what's the semiconductor content to TSMC per smartphone in 2013 and 2014, and If we could do that, we also want us to break it down into low-end and high-end smartphones. That's the first question. The second question is with respect to 16 nanometer FinFET progress. According to Donald, ASML said they saw mobile FinFET delay, and he was wondering whether the delay is caused by TSMC or another company.
Do I use?
English, please.
Okay. I probably cannot give you as detailed numbers as you wish, but let me give you some numbers from my perspective, from my granularities. For the smartphone total in 2013, TSMC's average wafer value per unit is TWD 7 last year. It will increase to about TWD 8. This increase, of course, include the silicon content increase as well as the market share increase. At this point, I cannot distinguish those two. For this increase, significantly stands out is in the high-end smartphone. High-end smartphone, last year, we are about TWD 10.8 per unit on average. This year, we expect to be TWD 13.9, so that's a major increase. For the mid-end and the low-end, we see similar. Last year was about TWD 6, and this year will also be TWD 6. Low-end is TWD 3.6, and this year will be TWD 3.6 also.
We do see the low-end performance spec continually increasing. Mid-end and low-end smartphone specs continually increasing. We see a lot of features, regardless of the mid-end or low-end, and we're still increasing almost like the high-end of last year. These are the perspective we can see now. Secondly, about the ASML's message. I really don't know what he means. Let me give you some comment on that. If you talk about the FinFET technology difficulties, I must tell you that our 16 FinFET technology development is well on track, and our yield improvement is well on track, and we are working with customer closely, and we expect to ramp up 2015. I think one unique feature of our 16 FinFET is our 16 FinFET has the same design rule, back-end design rule of 20.
We can leverage all the yield learning, all the massive work C.C. had talked about in this year into next year's 16 FinFET. Secondly, if you observe the mobile device industry, in the past six months, we do see the 32-bit conversion to 64-bit in the processing after the Apple's announcement. That change, that transition include indeed bring a lot of attention of the product development back to 28-nanometer and 20-nanometer product design. Indeed, we see increased demand on the 28 and 20 this year as well as next year. That's a second message. The last message I like to comment with that is, if you combine in 20 and 16, our Chairman has mentioned to you that combined 20 and 16 total revenue in the first eight quarters or first two years will be even bigger than the 28-nanometer revenue in total.
That's how I give you this above. Thank you.
All right. Next question will be coming from Credit Suisse, Randy Abrams.
If I could ask a follow-up on the 16. Could you talk about the timing for the 16 regular version versus the Plus version, if there's a difference in timing and also the customer adoption? Could you also talk about your expectation at this early stage of market share for the 16 node?
Randy, your question is with respect to the timing, the availability of the 16 FinFET versus the 16 FinFET Plus.
Yes.
The way the customer adopts, whether they are more adopting 16 FinFET or the 16 FinFET Plus.
16 FinFET Plus will be qualified in September. Remember, we and our customer work on 16 FinFET design one and a half years before. All the customer already design. The design is on 16 FinFET. For those customers, when the product tape-out, for example, we have a first product tape-out this month, it will run on 16 FinFET processes. For those customers taped out in the second half, I would say mostly will be run on the 16 FinFET Plus. I would think majority of our product customer will run on 16 FinFET Plus. Looking into the volume and coming up for the next year, I would say that most of the product will be run on 16 FinFET Plus.
Okay. Thank you.
The second question, if we look ahead to second half, you talked about normal profile for growth. If you could give a characterization what that normal would be for third quarter, fourth quarter, how you think about normal seasonal. Then in that context for Gross Margin, given 20 nanometers ramping and depreciations going up, if potential with normal Gross Margin can move up.
Talking about our second half growth will be normal. As I just give you the guidance, our second quarter will grow 22%. We believe second half will be normal, means less than 22%. I cannot give you a specific number, sorry. You also talk about the Gross Margin. I think last quarter, a lot of analysts asking how can we maintain even better structural profitability, given 35% year-over-year depreciation increase. Let me elaborate. I will talk about the 20 nanometer impact. The reason we can improve structural profitability are the following. Number one, you see the 35% increase in depreciation. That's on dollar-to-dollar. We also increased 10% capacity. If you divide that by unit basis, actually depreciation or whole basis go up by 22%. That's number one. Number two, depreciation accounts for about half of our manufacturing cost. We have another half.
That's basically variable cost, indirect material, and other fixed cost. We work extremely hard to drive those costs down. With very good progress. With that, along with the better blended ASP, thanks to the technology migration and our higher yield and better performance, we are able to raise the overall corporate SGM level. With the ramping of 20 nanometer, which just started in the second quarter, we will have very, very small volume shipment, and we have much more volume in third and fourth quarter. Any new node, starting with low Gross Margin, we expect there will be some dilutions to corporate level Gross Margin starting from second quarter. The magnitude of that will impact about 1% in second quarter. It will be slightly bigger than 1%, will be a very low single-digit impact on our second half.
for the whole year, we still expect to see a slightly higher SGM compared to last year.
Next one goes to Bank of America, Merrill Lynch, Dan Heyler.
Thanks. Good afternoon. Thanks for all the clarifications on 20 and 16. It's very helpful. I wanted to follow up on Donald's questions on the mobile numbers that you put out, which were helpful. That's a very significant increase in high-end content per phone for TSMC in terms of market share. Congratulations there. I wanted to talk more about your mid-end number. I'm a little bit surprised to see that you talked about the mid-end phones being basically flat, but you also commented that the specs are increasing for mid to low-end phones. Would that suggest that there's significant pricing pressure and you're not benefiting from that content increase? Thanks.
This is the number for the full year. The current we can see is from 5.9 to six. I really cannot distinguish the silicon content versus the market share. I think we will hold our market share fiercely as we had before. The price, as you know, is always there.
Okay, yeah, I think that mid-end market is where there's significant units, in terms of globally, it represents the biggest part of the units. I am amazed at how much that content is increasing in the mid-end phone. I was a little surprised as to why you're not seeing TWD value in the mid-end market go up more, or is that maybe happening in the second half of the year in 2014 or 2015? How should we think about that?
I think it has to do with 28nm SoC also. Which is mostly second half.
Okay.
Dan, do you want to reduce your microphone?
Sure. While they were chatting, I thought C.C. was going to comment there. No, I will not give up my microphone. I have one more question. I wanted to ask a bit about if you could elaborate on what your feeling of the progress on EUV is, if you think it's coming along slowly, or are we still treading water? Your latest best estimate on when there would be an insertion into your progress. Again, timing. Thanks.
I think we work very closely with ASML. With their comment about EUV yesterday, I think that we hold the same perspective. Today, of course, the EUV is not up to the production spec. Actually, the most recent breakthrough was a 30-watt, now they have a higher 80-watt machine, we're still working towards that goal. Also the same as Peter mentioned yesterday, that the EUV will not be inserted in 10-nanometer at the start because that's already past the window. However, our EUV team is still continuously working on EUV, hopefully to insert a few layers after the 10-nanometer process starts to qualify as a follow-up process simplification. That will meant to be second half next year. Thank you.
Thank you. Next one goes to Deutsche Bank, Michael Zhou.
C.C. Wei, could you give us more color on the advanced packaging you just mentioned? What is the difference between this one and CoWoS? Thank you.
The difference between the InFO and the CoWoS is actually the geometry to connect multi-dies together. In the CoWoS, actually, we are using very small geometry, like the 65-nanometer of geometry to connect the multi-die together. In InFO, we're using the larger geometry, which is still technical, confidential information. The cost is much, much lower.
Could we expect that will be adopted by mobile customers?
I would like to say that we are working with major customers.
Thank you. The second question is regarding Q2 outlook by segment. Outlook by segment.
Second quarter, all segments will grow more than seasonal, especially strong in communication, consumer, and industrial-related applications. Computer will grow less than the other three segments.
Thank you.
Next question goes to J.P. Morgan, Gokul.
Thanks for taking my question. My question is about the second wave of demand, both for 20-nanometer, I think some of it is already coming through, and how do you think about second wave of demand for the future process, like 20 or 16? Is it going to be much smaller, and how does that have an impact in terms of your thinking on investment as well as future returns? Thanks.
I think, Gokul, your question is with respect to the backfill of the 20-nanometer and 16-nanometer once the first-wave customer migrate to the next node, whether or not we will have sufficient demand as big as what we had in the past, and if not, whether this will change our investment profile and change our return on invested capital.
Okay. Let me answer this question. The 20-nanometer SoC is not a transition node. Okay. We have worked with our customer already, come out with very powerful products and in the ramping. These customer, some of them indeed will quickly transition to 16 FinFET Plus. However, still, some other product will stay on 20 SoC for quite a long time. Not every product requires the highest speed, for example. The key is we manage this transition by the tool commonality between 20 and 16 are 95%. When the customer moves from 20 SoC to 16 FinFET, we only need to increase much smaller, very marginal amount of the capital to suffice that demand. Of course, the ASP and the product will be more competitive for our customer. We consider 28nm SoC CapEx-wise, are very similar. We always put it together as one node.
It does provide our customer, their product grade or product spec be continually improved year after year.
Okay. Just one more follow-up on that. I think you have indicated that 2015 CapEx will start to come off a little bit. Should we expect that to continue going into 20-to-16 and 10 nanometer, or is it intermediate stop, then we need to ramp up as we ramp up maybe EUV or triple pattern and those kind of stuff?
We have not decided CapEx for 2015, but from what we are seeing right now, we expect the CapEx for next year will be similar to this year level. However, because our revenue will continue to grow for this year and also for next year. The capital intensity will go down this year and next year as well. Next boost of CapEx will be 10 nanometer. I think that will be coming on maybe the timeframe of 2016 and 2017.
Next question comes from HSBC, Steven Pelayo.
Hi. We focus so much on 28, 20, and 16 nanometer, I guess one thing I noticed here in the first quarter is your 40 nanometer actually grew about 25% quarter-on-quarter. That was a bit of a surprise. What's the driver of that mid node? The second question would be the More than Moore strategy. Are you out of capacity on 200 millimeter? Are you able to transition some of those products to 300 millimeter? What are we thinking for the higher nodes that are still half the business?
I'll answer the 40 nanometer demand. I think our 40 and 40 nanometer market share holds up this year much better. The product mostly is associated with the connectivity. The connectivity integration become very big. We hold a very large capacity for our customers, the demand is very strong this year. The second question is more than more?
On the even higher level nodes. Obviously things like CMOS image sensors, fingerprint sensors, they suck up a lot of capacity. Historically, these are being built on 200 millimeters, I'm curious, are you out of more mature node capacity, what are you doing about that?
That's a good question. Actually, we did not increase huge amount of capacity for those specialties. We did modify and buying some bottleneck tools to increase a little bit and also to convert the logic capacity into the more than more specialties capacities. We did all the time. We continue to see the strong demand on those specialties, such as like you mentioned, CMOS image sensor, fingerprint, especially, and embedded flash for automotive, for those kind of products.
If I can just sneak one last question in. We're so focused on an intensifying competitive landscape potentially at 16, 14 nanometer, it seems to me that at 28 and maybe now 45 as well, you guys are just extending or lengthening your period of dominance. This fear of this talk of potential second sourcing, it almost seems like competitive landscape's almost got less intense. Would you agree with that?
I think it will be more difficult today, even more difficult in the future. Take 28-nanometer, for example. This is already the fourth year we ramp 28, the third year we ramp 28 High-K metal gate. First of all, the complexity is hard. Secondly, when we ramp 20, our performance did not stand still. We continue to improve the product grade in the 28-nanometer. There are several ways of product grade improvement. That's how we try to defend our market share.
Next question goes to Citigroup, Roland Shu.
Good afternoon. I think my first question is to C.C. C.C. are talking about how the engineer mobilization is the key for the fast ramp on the 20-nanometer. I think that is the reason we highlight that TSMC has enough talent pool for that. I think this is for sure. I think going forward, I think last couple weeks, TSMC would like to hire about 5,000 talent. Also at the same time, Hon Hai and the other tier 1 company in Taiwan also would like to hire thousands of the talent. Are you worried about the shortage of the talent pool in Taiwan?
I think TSMC today is a preferred employer in Taiwan, I don't use the word worry.
Okay. Yeah, very nice to hear you are so confident about that. I think another side question is, how are you going to motivate current employee going forward or the new employees, and to motivate them to work more hard, more smart, more innovative, and do more contribution to TSMC?
How to motivate the existing employees. Actually, TSMC continue to expand, and our young engineer or the employees all have a very bright future, because there are a lot of new openings in the higher position. Company's performance is getting very good. That's enough to motivate our existing employees. We never have that problem.
Okay. Sounds like TSMC has very good career path to all of your employees. Okay. My second question is to Lora. I look at your margin boom, from the continuous work cost reduction is one of the reason. I also look at your other manufacturing cost in first quarter have been declined about 7% quarter-on-quarter, even though you have the same amount of wafer shipment. My question is, how much room for you to continue cut cost and to squeeze their Gross Margin?
I also ask myself the same question, how much room? Every time I was surprised how much the engineer in the factory can do. I'm confident we will continue to drive that in old technology, current technology, and leading-edge technology.
Okay. Thank you. The follow-up is, TSMC is loading at very high utilization, what's your expectation? If TSMC is loading at 100% utilization, what is the margin expectation now for TSMC at the current SG&A? Thank you.
You're trying to understand whether we can achieve 50% again, if the capacity utilization gets to 100%, right? I can say, we have the capability to achieve 50% if the utilization reach 100%.
Okay, I think this is about time that we should open the Q&A to those people who are on the line. Operator, could you proceed with the first caller on the line?
Yes, ma'am. First question from the line of Mehdi Hosseini from SIG. Please ask your question.
Yes, thanks for taking my question. Going back to the commentary on 16 FinFET and 16 FinFET Plus, it seems to me that 16 FinFET Plus will have a larger mix of revenue, and 16 FinFET revenue contribution will be limited. Could you please clarify on that? Once we think about the actual contribution, is this going to be more of a Q4 2015, or would it be more meaningful in early 2016?
All right. Mehdi, I think his question is with respect to 16 FinFET versus 16 FinFET Plus, whether or not we will have a bigger business volume from 16 FinFET Plus, and if so, whether or not the volume will be coming from Q4 2015, or we have to wait until 20-to-16.
From 2015, first of all, the volume of 16 FinFET Plus will happen in 2015. Let me drive that. Even for some of the customer, initially their product sits on 16 FinFET. They also would like to migrate their second if market opens the opportunity to upgrade their product. That would be the majority. I really mean massive majority would be 16 FinFET Plus. Yeah.
Okay. One follow-up question for Lora. You talked about the HCNA trend into the second half. What should we think about the R&D? Should we also assume that the R&D percentage of revenue will go higher in the second half and into 20-to-16?
I think our current R&D to revenue is ranging from 7%-8%. If I look at the second half, I think that's still in that range, from 7%-8% range. Mm-hmm.
Okay. Thank you.
I think we can still take another call from the line. Operator, could you please proceed?
Yes, ma'am. We have the next question from the line of Brett Simpson from Arete Research. Please ask your question.
Thanks very much.
Just have a quick question. Can you give us a sense, within the 28 nanometer nodes, how this splits between polysilicon and High-K, and how do you think this might trend through this year?
Brett's question is what is really the mix between polysilicon, that is our 28 LP versus our High-K metal gate, and what is going to be the trend with respect to that kind of mix throughout this year.
Allow me to answer that. Our 28 nanometer High-K metal gate has three options: 28 HP, 28 HPM, and 28 HPC. This year, these 28 High-K metal gate technology will be about 85% of the overall 20 nanometer in terms of the wafer. Yeah.
Great. That's very helpful. Another follow-up question on InFO. Can you maybe talk a little bit about attach rates for InFO over the next couple of years in smartphones? Is this something you expect all major smartphone chip makers to adopt, or is it something that you think is more targeted at the high end? Maybe as a quick follow-up on that, what is the real benefit that InFO is bringing to chip makers? I don't know if there's a performance or power saving you can share with us. Also, how should we think about the margin structure for this within TSMC's business? Thank you.
Okay. You asked whether we are working with the customer and when you will have a product out. We are working with the major customers. That's all I can say, and the outlook is very good. Performance is very good. The cost is low. I believe that we will see the product out, but that will be our customer's schedule. Technology is close to being in production, ready probably early next year.
Okay. If we look out maybe a couple of years from now, just to get a sense, is this something you think will be a very high attach rate? When you sell SoCs at the leading edge, do you think the percentage of chips you sell at that leading edge will be attached with InFO? I'm just trying to get a sense for the penetration that you think InFO will have within the bigger smartphone chip makers.
We certainly believe that the attach rate, you are using that working attach rate. We certainly think it will be very popular among the mobile product. Because it provided a very low-cost solution and continue keep the high performance and the lower power consumption. Whether it's going to be. What is the percentage of attach rate, I cannot answer that question right now.
Okay. Thank you very much.
All right. Let me go back to the floor now. The next one will be Morgan Stanley's Bill Lu.
Hi, good afternoon. I'll start with a housekeeping question for Lora. A quarter ago, management said that the company was pre-building, forgot the word was exactly, but building some of the demand in 2Q and in 1Q, and that impacted margins. If you look at your 2Q guide, if that didn't happen, can you tell me what the margins would've been? I think that would've been 1Q margin a little bit lower, 2Q margin a little bit higher, right?
2Q will be a very strong quarter, and our capacity are essentially full. There's no need to build inventory and no room to do building.
Yeah. Sorry, I wasn't clear. What you did in 1Q impacted 1Q margins and 2Q margins, correct?
Yeah. Bill's point is that since we did it in the first quarter, which probably represents some sort of a pull-in from the second quarter, so had we not done that, our second quarter utilization probably would be even higher, and then second quarter's margins would be higher. Right?
I have to talk from first quarter, okay? I said first quarter, we did do some inventory, the impact to first quarter, about one percentage point margin. For second quarter, had we continued doing that, well, you have to look at which node you're building. For example, for certain capacity, it's already full. We can only run that much, right? I don't know how to assume. We will be exceeding 110% utilization. That's essentially not possible. Am I answering your question?
Not exactly. I guess I can take it offline. That's okay.
Okay. Maybe I should say that the inventory build for the first quarter has no impact on second quarter. It will impact first quarter because we utilize first quarter of the capacity ahead of time.
Doesn't mean that second quarter margins could have been a little bit higher.
Second quarter order already coming in. It's a regular order.
Second question is on this inventory cycle that we've seen. If we look back into last year, 1Q, 2Q were also very good, things dropped off in 3Q and 4Q. Can you talk about what you're seeing now that is different from a year ago, in terms of potentially a broader base recovery, maybe more end market, more customers? What are you seeing that gives you confidence that that won't happen again this year? I guess if you look at this more broadly, TSMC used to have a very diversified customer base. In the not-too-distant future, I could see your top two customers being 25%, 30% of total revenues, both in the mobile segment, right? How do you manage that concentration going forward?
Bill, you actually have two questions packaged into one.
Okay. I think, as you know, typically in the past few years, we see the end of the year, people can try to control inventories. Toward the end of the year, it's about back to the seasonal level. However, last year, we look at our fabless DOI, in the industry, they drastically control the inventory, reduce the inventory. Even at the end of Q3, we see them get to seasonal. Going to Q4, everybody find that their inventory is below seasonal. What our data shows below seasonal by six days, which didn't happen before. The reason they did this is because of their perspective for the outlook when they were in the third quarter last year. Okay. For this quarter, we see the inventory not even back to seasonal, from our data, is minus five days below seasonal.
They don't have time to replenish inventory given the better outlook. What happened to refill the inventory will happen in the second quarter. Now, whether this year will be as bad as last year, I think I'm typically more optimistic because even down to the second quarter, it's only to the seasonal level. Okay. Even we assume it doesn't overshoot, it shouldn't be very much below seasonal. This is a difference we see this year. Okay.
Customer concentration risk.
Customer concentration. We just work very closely with our customers. Okay? Remember, if you are IDM, if you have big market share, you have to face the same consumer product demand decontrol. This problem is more an industry problem than our problem. But we deal with this problem by working very closely with our two biggest customer. Today we are already together planning our 2015 supply and demand very closely. We recognize these challenges. Our customer also recognize the same challenges. The only solution is work very closely together. Okay.
All right. Next question goes to Barclays, Andrew Lu.
How does one forgot my name?
I forgot your company.
We still survive.
The first one is the CapEx number for next year, about TWD 10 billion. For this year, we are going to spend a lot of money on 20-nanometer capacity, where there's a very likely one or two customer is going to migrate direct a large customer to 16 FinFET Plus or something. Earlier you say the capacity is 95% convertible. I believe the spending is very low. Assuming we have a 40,000, 50,000 capacity by the end of this year, and next year, 30,000 convert to 16 FinFET instead of adding a new capacity. Supposing next year, CapEx should be sharply decline compared to this year. I don't know what to mix on this one. Thank you.
You said next year, the capacity will suddenly decline?
Yes, because your 20 versus 16 is 95%.
Okay
convertible.
What we see is, this year we ran 20-nanometer very deeply.
Yes.
Toward the end of the year, we see next year, the combined 20 and 16 capacity will continue increase. Next year. Some of the 16 FinFET, we will convert 20-to-16. We will also add new 16 FinFET capacity next year. That's what the most of CapEx this year deals with. The total 2015, 20 and 16 capacity will continually increase next year.
total combined 20 plus 16, the amount, the capacity you are going to add, quite similar for each year. Can we say that? For example, just using examples, 40K capacity by the end of this year, next year, and next year, about 80K.
No. Allow me say the total capacity increase next year, I think will be more than the conversion of 20-to-16.
Fine.
Andrew, I think you can go home and work out your numbers. Yes.
The reason I ask this question because I assume one customer takes out 50% of your capacity on 20. This customer is not going to use 20 next year. It's going to migrate 16 to another. You have a total capacity next year. You don't need to add that much on 20.
I think you're.
On 16, unless you add additional 20-nanometer capacity.
I think the key is this customer will still continue to use 20 SOC next year.
That's right. Andrew, well.
That's good enough.
That's right.
That's good enough. The second question I have is, when you calculate US dollars per smartphone, do you calculate based on the shares you have, or you use the total value divided by the global smartphone shipment?
Yes. We use the total TSMC with revenue divided by global smartphone shipment.
That's why the high-end side, if you gain shares, then your ASP change a lot.
Yes. That includes the second content increase and also the share increase. I cannot distinguish which portion is how much at this point.
Based on our calculation, it should be much higher than TWD 8.
Maybe your number is correct and you will be very happy. Yes.
Average, total average?
Yes.
Okay. I'll check with my staff. Thank you.
Okay. Next question will come from UBS, William Dong.
Okay. Thank you. My quick question is, I think in terms of competitive landscape, there's always been a lot of talk about design portability. Obviously, I think your competitor will try to come after your clients next year very aggressively. I wanted to check, what is your view on design portability? Is it a realistic threat as sort of with the move from 20-to-16, we're using the same design rule. Does that really mean that the competitors can actually try to get some of these customers?
William's question is with respect to design portability in the sense that if 20 nanometer and 16 nanometer shares the same design rule, will we open the window for our competition to learn something about our 20 nanometer design rule and then compete with us on 16 nanometer?
Let me answer the question. First, 20 SoC and 16 FinFET are totally different device structure. Even the back-end design rule are similar. You cannot port from one to the other easily. No, it won't be easy. What's your second question?
I guess from 20-to-16, it's difficult. In terms of perhaps you having 16, your competitors having 16 nanometer as well, obviously, I think there's been some talk in the market about them having some ability to port designs. Do you think that's a realistic threat, or is that really just a lot of wishful thinking?
Well, Mark already answered this kind of question, saying that porting from one foundry to the other foundry are getting harder and harder. The real reason is all the device characteristics right now is related to the strains, strain or those kind of things which you cannot just copy. You cannot de-layer, you cannot do the reverse engineering to look at what the device structure is and then get the same kind of IV characteristic. It's actually very hard.
Okay. Let me answer that. Well, the porting, it happens. I just want to bring out 2 points. We just have to live with it, deal with it. Let me point out 2 points. One is porting costs more and more nowadays for this generation. I think it takes a lot of R&D resources to do that. Secondly, we just have to do a better 16 technology. There are reasons to do that because first of all, we ramped our 20 SoC massively this year, a lot of learnings, a lot of process window control, a lot of design collaboration with our customer, we built ahead of our competitor
We believe our 16 FinFET, right on top of our 20 SoC, given their same design rule, will be more mature at the time compared with our competitor.
All right. There will be follow-up questions from the floor, that will be Bank of America Merrill Lynch's Dan Heyler.
Just wanted to come back a little bit to the 16 discussion. Again and again, this industry encounters some big node changes, 28 was fraught with problems that TSMC didn't anticipate, your customers did not anticipate when 28 was first rolled out. How should we think about 16? This is really uncharted territory. It is in a totally new device structure. Should we be a little more conservative on the possible unknowns coming in, that this could easily be pushed out, say, a few quarters? How much are you really setting expectations here very high that you can have 16 in the market? Should we be thinking of this as a pretty uncertain node, given how early it is? Thank you.
I think this is your comment. We'll accept that. I think, at this point, we develop our technology. We can only make sure our technology do not stand in the way. When customer has need for product meet, we'll be there and mature and support their business with enough capacity. Okay. As I mentioned earlier, industry market does change. For example, from 32-bit to 64-bit, that does sway some of the schedule of 16 FinFET. Today, I really talk about our readiness, okay, as to how much real business, and we'll let you know when we do the next year's financial forecast.
Just to be clear, a follow-up. Just to be clear, we should start to see volumes in first half 2015, where we'd start to see a few percentage of revenue based on your current expectations? A few percentage contribution by second quarter next year, is that fair? Lora's checking.
You mean the 16?
Yeah.
Okay.
We will see some revenue in 2015.
Good. Okay. Setting the expectations low. That's good. I want to follow up on this InFO. This is quite interesting. Could you maybe elaborate a bit more on what exactly are you going to be attaching? Which devices were you talking about in terms of-- With CoWoS, it was pretty much the FPGA companies were there and others, some baseband. But what devices are you attaching on the initial generation between the different chips? Second part of that question would be, how many customers do you expect to manage to have in this area? Because you start handling lots of devices and lots of customers, it gets a little complicated. You start to look more like an OSAT. I wonder if this is going to be a pretty small group of high-volume products.
Finally, as you attach, are you actually doing the chip attach, or will you be doing only the wafer level activity, and will you be working with the OSATs to do the actual chip attach? Thank you.
I think to answer your question, the InFO, actually, we are right now working on application processor together with the memory dies. If that's good enough for you. I cannot say anything more than that. We're working with mobile product customers, and we expect a very high volume, but we announce work with many customers as current takers. We are working on the wafer level, plus stacking die on top of it. We are able to do the complete line ourselves.
Thank you.
All right. We have a hand raised on the back of the room. I think this is from the media. Could you please say your name and the company you work for?
Hello. I'm Mark from UDN TV. Just want to congratulate TSMC on another great quarter. Just out of curiosity, I'm just wondering how is the management team right now? Because obviously Morris Chang is not here, and I'm just wondering if that is a sign that he's very confident that both CEO right now here. How do you divide work right now between Mark Liu and C.C. Wei? Just out of curiosity. Is TSMC shifting into a new generation management team?
Well, I'm delighted to work with C.C. Wei. Of course, Chairman. We three contact constantly to devise new initiative for the companies all the time. The rest, I think, is for you to judge.
How open is Morris Chang to both of your ideas to the company? I mean, does he just let you guys do your expectation, or does he still make the final decision?
Of course. Everybody contribute ideas, of course. Decision usually is made together. More than often, we find wisdom talking, discussing with Morris and our Chairman, and that we learn, and that we want to grow and to up to his expectation, I guess. Yeah.
Thank you.
All right. I think, in the interest of time, we will conclude our conference here today. Please be advised that the replay of the conference will be accessible within three hours from now. Transcript will become available 24 hours from now, both of which will be available through TSMC's website at www.tsmc.com.