Welcome everyone to UMC's 2017 Second Quarter Earnings Conference call. [Operator Instructions] After the presentation, there will be a question-and-answer session. Please follow the instructions given at that time if you would like to ask a question. And for your information, this conference call is now being broadcasted live over the Internet. Webcast replay will be available within an hour after the conference is finished. Please visit our website www.umc.com under the Investor Relations, Investors Events section. And now I would like to introduce Mr. Michael Lin, Head of Investor Relations at UMC. Mr. Lin, you may begin.

Thank you and welcome to the UMC Conference Call for the second quarter for 2017. I'm joined here by Mr. Jason Wang, the Co-President of UMC and Mr.Chitung Liu, the CFO of UMC. In a moment, we will hear our CFO present the second quarter financial results followed by our Co-President's key message to address UMC's forecast and the third quarter 2017 guidance. Once our President and the CFO complete their remarks, there will be a Q&A session. UMC's quarterly financial reports are available at our website, www.umc.com, under the Investors Financial Section. During this conference, we will make forward-looking statements based on management's current expectation and beliefs. These forward-looking statements are subject to a number of risks and uncertainties that could cause actual results to differ materially, including the risk that may be beyond company's control. For these risks, please refer to UMC's filing with the SEC in the U.S. and the ROC securities authorities. Now I would like to introduce UMC's CFO, Mr. Chitung Liu, to discuss our second quarter 2017 results.

Thank you, Michael. I would like to go through the 2Q '17 Investor Conference presentation material, which can be downloaded from our website. Starting on Page 3. Second quarter of 2017 consolidated revenue was NT$ 37.54 billion with our gross margin at around 18%. The net income attributable to the stockholders of the parent was NT$ 2.1 billion and earnings per ordinary shares was NT$ 0.17 a share. And the utilization rate in quarter [one] stay at flat rate around 96% compared to the first quarter of 2017. On next page you can see our comprehensive income statement. Revenue was pretty much flat around NT$ 37.5 billion. This is based upon around 3% increase in shipment and flat ASP as the NT dollar appreciated around 2.7%, which offset the wafer volume shipment increase. Gross margin because of the NT dollar appreciation dropped around 1.9% to NT$ 6.7 billion. Operating expenses is under control and also helped by a one time around NT$ 400 million expense delay so it come down to about 14.2% or NT$ 5.6 billion. And operating income as a result has come up 21.7% quarter-over-quarter to NT$ 1.66 billion in quarter two of 2017. And EPS for this quarter is around NT$ 0.17 compared to NT$ 0.19 in Q1 2017. For the first six months of the year, revenue grow about 5% compared to the same period of last year, which is again because of the 6% NT dollar appreciation, which offset more than 13% wafer volume growth on a year-over-year basis. Gross profit improved to NT$ 14.1 billion compared to NT$ 13.3 billion in the same period last year. And EPS for the first six months of this year is NT$ 0.36 compared to NT$ 0.23 of the first six months of 2016 . Balance sheet…

Thank you, Chitung. Good evening, everyone. Here I will like to update the second quarter operating result of UMC. In the second quarter of 2017, UMC's foundry revenue was NT$ 37.45 billion with a utilization rate of 96%. It brings our wafer shipments to about 1.74 million of 8-inch equivalents. The gross margin was 18.1%. We continue seeing robust demands on computing and communication segments on both 8-inch and 12-inch mature technologies. Looking into the third quarter, demand for our mature technology remains strong. However, due to the softness out of the 28 nanometers, we expect sequentially flat quarter. Our 28-nanometer High-K-Metal Gate business has become more vulnerable due to a high dependency on a few customers. As a result, we are seeing a possibility of 28-nanometer High-K-Metal Gate business decline in the second quarter -- the second half of 2017. Recently UMC's board of directors have appointed Mr. Chien and I as the Co-Presidents. President Chien will take the lead in fab operation and technology development while I will be responsible for business management and corporate strategy. To enhance our financial performance, both of us will implement measures to strengthen our foundry competitiveness in core manufacturing and to drive operational efficiencies. The goal of course is to unlock UMC's value and lead to a positive cash flow. That was my second quarter results. Now, let me go over the third quarter 2017 guidance. Our wafer shipments would remain flat. ASP in U.S. dollar will remain flat. Gross profit margin will be in the mid-teens percentage range. And capacity utilization rate will be in the low 90% range. For foundry CapEx of 2017, as Chitung mentioned it, it will be revised to USD 1.7 billion. That concludes my comments. Thank you all for your attention. Now we are ready for questions.

[:

My first question maybe for Mr. Wang. For the roles you're taking as Co-Presidents, could you talk a little bit more about areas you see the strength in the foundry competitiveness and if you plan to make any changes in how aggressive you are investing either in the specialty technology or the advance leading-edge technology?

As I stated earlier to enhance our financial performance, well, first, we will implement the measures to strengthen our foundry competitiveness including [indiscernible]. So as a manufacturing company, I think that's very important. Secondly, operational efficiency will be the focus. And as far as the market landscapes, we believe with UMC's extensive technology offering capacity across the 12-inch and 8-inch, we can actually address many market opportunity in the future. But the first thing is try to improve our efficiency.

If I could ask the CapEx change that you made. Could you discuss the implication, how much that changes, how much, say, wafer capacity that you plan to add? I think you've outlined plans for Xiamen so if any change to how fast you ramp that up? And maybe if you could give a sense where your utilization is now on the advanced nodes, whether you feel that lower CapEx bias may continue into next year or we should think about closer to USD 2 billion level or being back in that range?

Well, that's actually a few questions. So the first is actually, you're talking about the trend -- trending down our CapEx to USD 1.7 billion. The way we're trending down is actually we try to optimize our utilization across the board from both 8 and 12-inches. So, it's truly just more aligning to our current demand outlooks. And the number is trending down from about USD 2 billion to about USD 1.7 billion. So, actually it's across the board, not something specific we're not too certain of.

Could you talk maybe your initial bias on CapEx I guess given that 28 outlook is softer, if your initial view is you may stay on a more conservative pace for CapEx or we should think of this level as a good level as you look ahead?

Maybe I can jump in for that. Again we really want to support the CapEx so you can see internally generate cash. So, our self-sufficient type of CapEx for us very important is return driven type of CapEx will be the key element in determining our future CapEx.

Okay. And if I could ask just one last bit. 28 High-K/Metal Gate the vulnerability, can you discuss the factors because you've had some good leading customers in the past even if it's a small base just what's causing that vulnerability? If it's -- is it safe foundry competition like whether technology or pricing and then maybe implication how we should think about 28 overall like as a percent of revenue in second half?

I stated earlier in my statement that one of the reason was our 28-nanometer High-K/Metal Gate is highly dependent on few customers today and those very high volume customers have migrated to advanced nodes. That's one. At the same time, we also experience a rapid evolving change of the High-K direct technology as you may know. And so therefore, it will take some time to catch the next wave of customer. Our plan is actually releasing the 28 High-K/Metal Gate via HPC plus version. By doing that, we expect this 28-nanometer node will improve in the next three to four quarters.

Okay. And with customers migrating to I guess advanced nodes like 16 and 14, how's your progress or plans or maybe even need to step up the pace on 14 if it's going to catch some first or second wave moving in that direction?

For 14 particularly, I mean our R&D and production team have already demonstrated our capability and we find our technology roadmap working now. And we will continue with that for the 14 and -- but still going back to 28 -- I still believe 28 there's very long like nodes. There are many application product will come into 28. So, we still remain positive about 28 outlook in the same time that we will continue our development with 14-inch.

And the next one is from Bill Lu from UBS.

First of all, Jason, congrats on the new appointment. Looking forward to working with you. So starting with 28-nanometers, it sounds like 2Q actually was maybe a little bit better than expected three months ago and yet there is uncertainty into the second half of the year. Is it that poly-SiON is actually little bit better or what exactly is happening there?

For the mix of the parties that are on the High-K, definitely we see actually much stronger poly-SiON momentum. At this point, the poly-SiON is actually better than what we anticipated and so the softness really happening with High-K area.

Is the strength in the poly-SiON business sustainable?

Yes. I mean we believe that and I think first of all, it remains healthy and we see many, many application actually coming into poly-SiON so especially in the consumer area. So, I mean, I think our poly-SiON will remain healthy, which is sustainable, yes.

Okay. I think the concern though is that for poly-SiON, you do have a competitor in China, so what do you think about the competitive dynamics?

I mean very good question. At first when we look at the 28-nanormeters, High-K is actually more challenging technology to develop and so we believe there will be many poly-SiON players in the market. So, we strategically positioned that High-K is more of a focus and that poly-SiON is more of a companion solution. So -- and I mean the intention is still want to capture our opportunity in the High-K area. And so as I mentioned earlier, I think that will give us probably 3 or 4 quarter time I think that the High-K will come back and so we have some confidence on that. Now coming back to the poly-SiON, it is lower base but the result of today, the momentum is better than what we anticipated.

And then following up on Randy's question on CapEx, so if we -- as we look out into next year, are you going to expand 28 in Taiwan anymore?

Actually it's subject to the outlook of the demand. I mean basically what's driving our CapEx. We're driving our CapEx; one, based on the demand outlook. Second is financially justifiable. So we trade out those balance and at the same time we look at our market position. So we will be cautious about our CapEx, first, we already revised our current CapEx number and we'll continue taking the same approach to our 2018's as well.

Do you have an early guess as to approximately what the CapEx number might be in 2018?

I think at this point -- I think it will be too early to conclude that, but by end of the year, we'll probably have a much -- a better visibility there.

And the next one is from Michael Chou from Deutsche Bank.

First question is regarding the poly-SiON demand, seems that your competitor is still, I mean [indiscernible], so what's your expectation for 28-nanometer gross margin in the second half of this year? Do you think that there could be some downside to 28-nanometer poly-SiON [indiscernible] 2018 evolving even softer? So, that's my first question. My second question is regarding first Q2 and what's your view for Q3? Thank you.

I will answer the second question first. Our weighted average NT dollar entry in quarter 2 was 30.3 versus 31.1 in Q1 so appreciates around more than 2.5% and net income about 1.5 percentage points of our gross margin. For the third quarter outlook, we currently only expect to see a flattish NT dollar for our third quarter budget.

So going back to the first question, I mean if I take the question right, you are asking about the gross margin of the 28 poly-SiON?

Yes. Given the conditions for poly-SiON, so what's your view for your total 28-nanometer gross margin, I think that High-K/Metal Gate UTI should drop in the second half of this year. So what is your view for gross margin trend?

Let me kind of answer that question. First of all, it's actually two questions, one is about the fairness of the market landscape of the 28 poly-SiON, and the other is the gross margin so I won't comment about the gross margin for the particular node, but let's talk about the market landscape of 28 poly-SiON. I think there are more players in the market for the 28 poly-SiON, but I think UMC is more in a unique position. First of all, we're offering both poly-SiON and High-K solution in the marketplace. There will be limited players who will do that. Secondly, we have both location a 12A in Tainan, a 12X in China for the 28-nanometer production. That enables UMC to further expand our support to many different customer world wide, in particular in China where we're ready to tap into China market and the third is actually the manufacturing performance because in 28-nanometer not only the technology offering, also you require the higher better cycle time and given our position, I actually feel comfortable about our 28 outlook. Definitely we'll see some turbulence in the next 3 to 4 quarters, but I remain positive about that.

Then how about the gross margin [indiscernible] is trying to get market share for shipment strategy for example. Definitely your rates much [indiscernible] but the pricing sense quite of course, so what if you still have a decent revenue momentum, which downside with 2018 28-nanometer poly-SiON [indiscernible]. So, does that means our margins in 2018 in light of the [indiscernible] continue?

Michael, about 28-nanometer and also along this UMCs splendid average selling price, we're actually give our guidance about flat ASP outlook in quarter 3 and 28-nanometer poly-SiON, also in terms of loading will remain for almost like the second half of 2017. So it's hard to believe there would be any significant impact from competition in terms of 28-nanometer poly-SiON gross margin. However, for High-K/Metal Gate , I think it is not so much as the competition today -- it's really due to our loadings. So, we believe over the next 3 to 4 quarters, we expect to see good improvement for our Tainan High-K/Metal Gate capacity utilization rate, which intentionally will help our gross margin recovery for High-K/Metal Gate.

Okay. Just a follow-up question. What is High-K/Metal Gate sales portion in Q2 I mean in total 28-nanometer?

So more than 50% is poly-SiON.

And the next question is from Sebastian Hou from CLSA.

So first one is I'd like to focus on CapEx first. So Chitung, you already mentioned that your CapEx will be a more return driven going forward to generate cash. So does that mean that your CapEx will be pretty much benchmarked to your depreciation will be or be even below to your depreciation going forward?

I think that that guideline and our depreciation right now is more than NT$ 1.6 billion, NT$ 1.7 billion per annum and we also have a new fab up and running in Xiamen, China. So I would say yes, that's pretty much the basic guideline, but whenever there will be opportunity in the market, UMC definitely are in a good position to grab the opportunity.

And can you give -- update us on the depreciation guidance for 2017 full year?

2017, in terms of the long-term number, we are talking about 0 to 5% increase year-over-year compared to 2016. However, the composition is rather opposite and for the Taiwan and Singapore part we will see the number at peak in mid of last year and gradually coming down. The new fab in Xiamen actually showing a strong growth in terms of depreciation. So it depends on which quarter and which driving factor is stronger. So from a quarterly point of view of 2017, we see a flattish quarter-over-quarter depreciation. Therefore, the year-over-year we are talking about 0 to 5% increase.

Okay. And if we build on the basic guideline of your CapEx will benchmark to your depreciation, which is below this year CapEx, can we assume that your full company depreciation should peak in 2017 and likely decline in the next few years?

I should rephrase that the guideline should be UMC's depreciation plus our earnings so it will be on cash earning basis, so first of all. And secondly, it's probably unlikely to hit in 2017. I would say that hit may happen in 2019.

So, which means that depreciation may likely stay either flat or a bit flat in the next 2 years or if not increase?

That is 0 to 5% type of growth.

Okay. Can you -- can we go back to your OpEx that you mentioned about in the second quarter, there was NT$ 400 million that went off. Can you elaborate that? You said is a delay expense and will we see -- what was that? And can you comment on your guidance for the OpEx to sales ratio in second half this year?

It's a one-off adjustment due to the delay for the marketing expense and also some reversal of R&D wafers becoming shippable wafers. Of course it will be our key focus to control the OpEx to a certain level, acceptable level.

Okay. So supposing we should see your OpEx to sales ratio back to the normal level in second half this year so Q2 might be extraordinarily lower.

Yes. Again the whole year number may be back to the normal, but it's now going to control the OpEx and stringent measurement.

Okay. And the second part question is on 28-nanometers. You mentioned about you're working on 28 HPC plus which we're contributing 3 to 4 quarters from now and what's the next derivatives? Because the reason I ask is that your competitor in Taiwan seems to be able to launch a new derivative on 20-nanometers almost every year and the reason I suspect do you think after you launch HPC plus, they have the 22 nanometers and maybe after we have 22 nanometers, who knows what they will have next? So my question is that when do you see able to narrow the gap of your peers in Taiwan on 28 derivative?

This is Jason. We really see this is -- the derivative was actually slowing down because after the HPC plus, there will be a 22 [URP] and after that, we believe this is for the High-K area. So we do have a plan to introduce our 22-nanometer as well and it will be available around near 2018. So when I say give me another 3 to 4 quarters, we already have that considered as well.

Okay. And last is on your advanced technology development, what's your strategy on 40-nanometers and below? Do you have any plan go down further to 10 or 7?

Well, I mentioned about 14 nanometer our R&D and production team already demonstrated earlier. So, I'll kind of address the question about the 10 and 7 because they are different. For the both 10 and 7, this point is I'll believe that there will be very limited number of application for UMC to address. So instead we will review our resources and reprioritize them and we'll focus on the addressable market that we believe we can enhance our trends.

Okay. It sounds like that you probably will focus more on the existing platform you have and maybe not just the modeling age technologies if I get it right. And with that kind of assumption, what's the revenue outlook CAGR you have right from this year to 2020 if there is not much migration further compared to overall foundry and semiconductor industry growth?

If you look at the market price today and taking all of the advanced nodes, there still are plenty of market opportunities and for our market shares, we believe there's still room to grow. And given our intensive technology offering and the capacity in both 8 and 12-inch, we believe we're really putting ourselves at the right position to capture those opportunities. So I can't give you a specific CAGR number right now, but we think that there will be significant opportunities there.

The next question is from Gokul Hariharan from JPMorgan.

Congratulations first of all, Jason, on your promotion. Just wanted to ask a couple of things. Just following on from Sebastian's question. Could you talk a little bit more in detail about what is your realistic growth expectations? You mentioned that you want to take share in foundry as well. Is it fair to assume that you expect that each foundry industry kind of growth over the next few years and how does that also gel with the kind of CapEx plans that you have for some of the business sectors because historically leading edge is the one which has grown faster than the overall foundry industry so if you take a little bit of setback on the income side, could you explain how the industry kind of growth can be achieved? And I have a follow-up to that.

Let me see if I captured the question correctly. Now if I look at the up market opportunities going forward, if I take year 2020 data coming back, the foundry market will probably result in $65 billion range and all the $65 billion, there will probably be about $20 billion to $24 billion for the intention of being below 14, okay. That will be 10, 7. So I think there will be plenty of advanced technology opportunity out there, but however, many of those applications may not be addressable by UMC. Now if I coming back to that figures, so that still represents about $40 billion to $45 billion market opportunity at $40 billion, $42 billion. Now given that UMC today is close to $5 billion of annual revenues and assume that we grab reasonable market shares, that will still give us opportunity to grow in those areas now and those areas will be driven by mainly in the IoT area, IIoT area, automotive area which many applications does not require themselves. So coming back to my point of with our technology offering based on capacity growth and we believe that will give us opportunity out there. Now we currently had a utilization of [90%] and so which means that we will continue spending CapEx, spending our capacity to address those markets and -- but maybe just in a different technology area and so that with our financial strengths, that will help us to fuel that growth. So based on our R&D, based on our manufacturing efficiency as well as financially, I think that will give us the chance to do that. And so I hope that will answer your -- I captured your question that way.

Just wanted to add on to that. If we think about it from a node-wise perspective, as we look out into 2020 and the growth in foundry and UMC's growth, is 28 nanometer still going to be the primary driver of growth? I think 28 nanometer, you reached close to about $1 billion I guess in revenues this year, slightly lower than that. Is that still going to be the primary growth area as we look out in 2020?

Absolutely. I mean the 28 nanometer will remain very important for our business and we believe the 28 nanometer will be a long life node and there will be multiple ways for the customer and application coming into 28 nanometer and that's also the reason that we are including the 28 capacity in both 12A and 12X. And so, the answer is a simple yes.

Okay. Also just wanted to understand you did allude to profitability improvement thus far and we're now running at mid-teens to low-teens kind of gross margins and Chitung mentioned that depreciation roll-offs is only going to start from really 2019 at the earliest. In the next couple of years, are you anticipating meaningful gross margin improvement because we're close to full utilization and we're running at 17%, 18% gross margin. I think in previous cycles, I think we have reached 22%, 23% gross margins when we were running at close to full utilization. So what is that incremental improvement, where -- which part will that come from, is it this 28 pricing or is it just 28-nanometer product mix or is it just industry competition inching off over the next couple of years for the gross margin improvement to happen?

Well, I mean it's our goal to improve our gross margin. I think if the depreciation doesn't come down this year, a couple of different areas that we will improve. Operating expense, OpEx area is currently under review and we hope to drive that. And secondly is with higher loading on the 28-nanometer once the 28-nanometer orders return and with better mix, higher ASPs; that will help our gross margin as well. So, I think there are a few areas for us to improve even before the depreciation started to decline.

Okay. Just a last question in terms of the 28-nanometer competition landscape. Typically your larger competitor in Taiwan and a big supporter of the pedal on older notes after a couple of waves seems like they seem to be still very insistent on keeping market share as in the announcement, they will keep getting ready for some new market share. Just wanted to take your perspective on what you think competitive landscape especially from your larger competitor appears to be on 28-nanometer compared with say what you're seeing in 40-nanometer or 65-nanometer at a similar stage?

I think the competition -- the market competition is going to continue. I mean we have been doing this business for 30-some years and we both existing for a long time and so the goal -- the way to winning this business or protect your shares is really of many different area. One is manufacturing excellence, you've got to improve your operations. You pass on your benefit to your customer, you stay competitive, technology readiness is also important and the customer relationship. We have years of relationship with some of our customers that we can continue serving them so I mean not only the one particular node, it's actually on the broader technology offering so the relationship is actually not just one particular node, actually dependent on multiple nodes. So I think with our current customer profile and technology offering and with our market position, I think we -- I would have to say we feel fairly competitive in this marketplace.

Okay. If I just ask a very short question, do you have any timeline in terms of when 28-nanometer can approach corporate average gross margin or maybe even corporate average 12-inch gross margins?

For corporate it has to largely depend on depreciation expenses as the other node has already pretty much complete the depreciation schedule. So again I think if we talk about company gross margin, it will have to be around after 2019 given before that these have plenty of areas we can improve the gross margin for current 28-nanometer gross margin.

And the next one is from Roland Shu from Citi.

First question for the CapEx spending this year. I think previously you guided 91% of the $2 billion CapEx spending will be spending on 12-inch and this time you revised down to 89% of the $1.7 billion will be spent on 12-inch. So the difference exactly is $300 million so that means that we can assume your CapEx cuts mainly from 12-inch and for this year your 12-inch expansion is mainly on 28-nanometer. So can we assume are you slowing down the 28-nanometer expansion plan this year?

Again for the CapEx reduction is actually across the board. It actually happened both -- for both 12 and 8-inch not just one particular node.

Yes, but for the 8-inch or for the 12-inch just according to my calculation, actually I believe there's a $300 million cut mainly from 12-inch. So that is I think it probably has nothing to do with 8-inch, right?

Again it's across the board, but [indiscernible] is really the auto making is pending compared to the rest of the inch. So this up and down sectors, but generally speaking it's across the board cuts.

Okay. So how about the 28-nanometer capacity in Xiamen by end of this year? How big it will be?

By end of the year, the number will be about 5k.

Okay. And then the total compared with the Taiwan, total will be 39.5k or 40k?

Around 40,000, yes.

40,000, okay. So, it's still the same. Okay. Second question is now you have a pretty high utilization on 8-inch and also on 12-inch except for the 28-nanometer High-K/Metal Gate and going forward, you said probably it will take 3 to 4 quarters to recover the High-K/Metal Gate utilization. So can we assume in the next 3 to 4 quarters at best the quarterly revenue probably will be just flattish as 3Q?

I mean we are only providing the outlook for the next quarter and so I think we're not going to comment anything going beyond the next quarter except for revenue numbers.

Okay. And next question, in your annual report actually you said that you are going to start a 10-nanometer [sharp hole] by this year and if this plan will stay on track, you are going to revise this 10-nanometer [indiscernible] plan?

First of all, we believe that 10-nanometer is more of a soft node. And so originally alpha engineering on the 10-nanometer as more the engineering research and exploration focus and so particularly for the dry -- the [indiscernible] progress, I have to go check but it's not going our main R&D focus.

Okay. So that means probably this CapEx probably is also something related to this 10-nanometer R&D as well?

Again it is across the board, I mean there's many different areas. It's not just one particular node.

Okay. My last question is that for your depreciation schedule, I think also just see in your annual report now you're taking 3 to 11 years for the depreciation schedule. So question is on your -- what kind of a tool you typically use for depreciation and what kind of tool you take for 11 years for depreciation?

The majority still remain unchanged 5 plus 1 or essentially six years. The 3 to 11 is really following the new accounting IFRS that certain different unique facility related equipment will have different life cycles -- lifetime in terms of depreciation. For the bulk of the manufacturing equipment, it's still five plus one.

Okay. So can we assume there is no big number change for the total depreciation numbers because of this depreciation schedule change?

No change at all almost.

Next question is from Donald Lu from Goldman Sachs.

My first question is the offload of 28-nanometer, I think we have been talking about this is 28-nanometer will have a long life. But if you look at in your 28-nanometer revenue since 2006, it has not really increased. So, I mean what would-- 2 questions. One is what would be the confident that you think more demand will come to 28? Secondly is that we noted that you're expanding more 28 in Xiamen fab capacity wise. I mean we will see more customers want to use your 28 capacity in Xiamen versus in Taiwan and if so, what's the reason for that? Well, that's my first question.

Well, I mean again as I stated earlier, I mean we believed 28-nanometer will be a long life node. There are many applications coming in and application areas like communications area, there will be a Wi-Fi and consumer area, there could be a digital TV set-top box and even sound IoT product will come into the 28. So, we remain positive about 28 outlook and we think that this will be a very healthy node. So, I mean that's actually what we believe is 28-nanometer outlook. And for the question about why we continue expanding and extend the 28-nanometer capacity in Xiamen; not only for China market, I think that China has a high growing semiconductor opportunity there and that's why we want to be there. But more importantly we expect 28-nanometer loading will be improved in the next 3 to 4 quarters. And so, we want to continue expanding our capacity, but what we're going to do is we want to make sure that business needs and [indiscernible] is there. So, based on that we will continue monitoring our CapEx decisions.

Do you see more demand for more capacity 28-nanometer capacity in China versus Taiwan, like for example, certain customers say needs to produce it in China versus Taiwan or-- ?

There is a lot to talk about that, and we -- in our numbers, we haven't really seem that yet. And China is now representing high percentage of 28 demand at this time. But there are many engagement that we have, customer engagement we have with China customer, that's because we have capability in China, yes.

I have another question, it's more DRAM R&D efforts with China DRAM fab. Can you give us an update in terms of what kind of -- the progress of that effort?

It's on schedule, we are scheduled to deliver two generation of technologies and we're working on the first generation right now, and our licensee, our partner company, which we do not invest is building their shell as we speak and so everything is on schedule for the time being and we haven't really going to the second generation of technology yet so its focusing on the first generation.

What is the first generation signing up?

The first generation is so called 3X technology.

3X DRAM.

Yes.

And this is planned to be -- to enter mass production, this 3X, sometime?

I think it depends on how quickly we can deliver the first one as a core and then more competitiveness technology, the second generation the 2X we will follow suit quickly.

Is it supposed to be in mass production next year -- end of next year?

The fab is currently under construction and normally take 18 to maybe 20 months so I would say yes, maybe end of next year the fab will be completed and by the time our first generation technology will have to be ready.

And congratulations to Jason and SC.

And the next question is from Charlie Chan from Morgan Stanley.

So, first of all just a follow-up on your CapEx outlook because you seem to maintain your 28-nanometer capacity, right? So, can you please further give us -- what were this capacity so for slowdown, for example, do you slowdown for 40-nanometer capacity expansion?

Really I have to say it's across the board and there's also timing related adjustment so the end of the year capacity number didn't really change. But the schedule in terms of capacity becoming available has already adjusted.

And I guess the next question is to Jason. So, you mentioned that there are some markets that is UMC can address, I have a guess it could be high performers competing through that because it needs a lot of leading edge capacity. But at the same time, I think that you honestly don't have great relationship with those fixed income companies like Broadcomm, like [indiscernible]. So, I do believe if you can further invest, you can capture with the future demand, right? So, am I getting anything wrong in a way why you don't think you can address those future demand? Why you don't think you can address?

Well, I mean, yes, I mean we have a long time relationship with many customers and those customers representing all different sort of application and products. Getting our product offering, we -- in the past we have continued developing on to the 14, right? I mean, until 14 and coming down from the largest process to the high-voltage, embedded flash, and we have a very expensive product offering. And that address all different sort of applications. And so we're going to continue doing that and we will continue advance our technology in those area to make it competitive and also in other emerging applications includes the ultra low power or emerging memory. So, those are the area we'll continue to invest as well. So, our technology development advancement is not going to stop and R&D will continue and the only area we believe is more difficult for us to address is really in an area like 10-millimeters or 7-millimeters area.

So, can I get your guidance regarding your premium capacity plan for this year and next year?

Our capacity for this year is close to 40k per month, which is 5k in Xiamen and 35k in Tainan. And for 2018, we will continue to extend our 28-nanometer capacity in Xiamen, but that depends on the market demand and also customer engagement in China.

And how about 40-nanometer?

Which one?

14-nanometer capacity. Yes.

Well, the 14 as of today in our 2017 CapEx has a little small portion of that, very small. And so, that will be -- I think in a range of about 2k.

And next year? I mean, you want to be fill it to sort of scale some normal scale like 10,000 capacity?

Based on our current customer engagement, it also -- we will actually adjust and the goal is simple, I mean there has to be a reason, we want to make sure that the customer engagement, [indiscernible] and before we come into any CapEx in 14.

And my last question is very brief. So can you say again that why third quarter gross margin would decline or provide sort of a conservative guidance about [19%] versus 2Q it was 18%?

Third quarter, we have lower loading mainly due to the High-K/Metal Gate decline. There was some elasticity which were about 1% of our gross margin rate. So I would say this two are the key effect.

And the next question is from [indiscernible]

First, [indiscernible] for 8-inch and 12-inch in 2Q and coming 3Q respectively?

Well, Q2 2017, the 8-inch is at full capacity we're running for 8-inch and for 12-inch is roughly 80% and the overall company is 96%. And for the Q3 '17, the 8-inch will remain full and the 12-inch will probably still be around 80%, but overall will probably be a low 90%.

Some questions about -- because you said you are going to expect 28-nanometer High-K/Metal Gate orders coming back in 3 to 4 quarter time, right, and can I clarify how we'll be driving those orders coming back? It's because of the HPC part and the 22-nanometer is coming back from existing customers or new customers? Can I have more information on that?

Well, there are many customers not just existing, there are many customers. And the engagement includes the 28 HPC following with the 28 HPC plus. And so at this point, the 22 ULP is still at a much earlier stage than the HPC plus. So as we said here, when we mention about 3 to 4 quarters, we mainly talk about the combination of HPC and HPC plus.

When you have this expectation, are you assuming limited expansion of 28-nanometer capacity in Xiamen over the next 1 year, which means you are assuming the attention rate of 20-nanometer to be full in 3 to 4 quarters' time based on current capacity. Is that correct?

I think we would expect the 28-nanometer node will improve and come back in 3 to 4 quarters and whether will be full or not, I mean I'm not making that projection right now. As far as the CapEx goes for Xiamen 12X, well, we will have to remain checking our engagement situation and once the engagement's solid and confirmed, then we will confirm our CapEx in Xiamen.

Lastly, about Xiamen fab because you already said the expansion depends on your engagement with customers and the [coming] from customers. However, Xiamen is a new fab. By end of this year, if I recall correctly, you are going to have 11k per month capacity including 20 and 40 nanometer. If I my sense is right, it's going to be around probably 30k or more to have the economical scale for a new fab -- new like the one in Xiamen so which means it's a matter of time for you to rent the capacity to your 30k per month, is that correct?

Yes. I mean we hope that we can run the 30k as early as we can to reach that economic scale. But we still have to follow the progress of the customer engagement.

Is there any timeline for this 30k; by 2 years time, 5 years time or is there any timeline for this to be reached?

It really depends on the return and the operating result and also it's not 30k. The fab 1 in Xiamen is designed for the house of maybe 25k, 26k as a maximum.

Because I just wondered you have a very good intention to keep your utilization rate tight by managing the capacity in a more different way and process in the past. But Xiamen is a joint venture so you are not the only one running the -- you are the major one, you are controlling the fab, but it's a joint venture. So I was wondering whether the other side, Xiamen government, who pays for this infrastructure will require you to grow the capacity even if before you have a customer.

UMC is in full control of this operation. We have the majority of the board and we consolidate this operation into our financial report. So, it will be a market driven return driven type of investment scale.

The next question is from Steven Pelayo from HSBC.

I'm little confused I guess on 28-nanometer. I know you only give one quarter guidance, but you're talking very -- multiple times about 3 to 4 quarters before it's going to recover as well. So I don't know, it just sounds like this isn't just a customer concentration or an in-demand issue. This sounds more significant, more market share shifts going on, and yet you're expressing confidence that it's going to come back in 3 to 4 quarters. So, I guess I'm just a little confused on what's really going on in 28-nanometer and [indiscernible] vulnerabilities and customer concentration. Why is it kind of being written off for the next 3 to 4 quarters, that seems quite a long time?

It takes time to recover because the 28-nanometer High-K/Metal Gate has continued coming out with a so-called derivative solution from the very early HPM to HPC and to HPC plus. So once you have continued evolving those derivative technologies, it's on the second wave or third wave customer, they want to get the latest technology offering. So when our HPC plus readiness comes behind that, then we have a difficulty in those areas. So, it's really true. When early customer adopting HPM and so we have a high -- and when those customers are migrating to the next generation, the second wave and third wave, they want to get on to the latest offering and then we have to [evolve] those. So it's really a combination of two. One is a high dependency of a few customers, early customer moving -- the early customer move on to the next generation while the next generation of customer waiting for the derivative technology to be ready. So, that is causing the time differences.

And how does that translate into your third and fourth quarter 28-nanometer revenue overall? So I guess I'm curious it sounds like you have kind of two companies right now above 28-nanometer and 28-nanometer. So for that portion of 28-nanometer, is that declining in both 3Q and 4Q and is the above 28-nanometer just completely offsetting it and that's why the total's flattish? What's the magnitude of the 28-nanometer weakness in the second half of the year?

I mean you're correct. I mean it is offset by the mature versus the 28 softness. So -- I mean so that's why it's a flattish outlook. I mean so you're right about that. And the 28-nanometer is actually much better compared to High-K, probably actually cross over the High-K ratio right now. So, overall the utilization right now is about [98%]. So, that's actually the idea of the softness behind High-K.

And then in terms of quantifying, is 28-nanometer declining double-digit quarter-on-quarter in the third quarter?

It's hard to tell. We don't have the number to comment.

Okay. And then, Chitung, one more for you just on the OpEx number. I guess you said back to normal delayed expenses and so I guess I'm a little confused on when we look specifically into the third quarter, does that NT$ 400 million delay all fall into the third quarter, does it get spread out? Can you be more specific on your OpEx guidance for the third quarter?

I think we're more back to the Q1 in range in terms of percentage of revenue. Again for the mid to longer time, it's the new President's goal to keep that under control and continue to review that with a better picture by the end of this year.

Ladies and gentlemen, we're running out of time, so we're taking the last question, and the last question is from Bill Lu from UBS.

Thanks for taking the follow-up question. So management team had talked about unlocking value in the company, I'm wondering if you could talk a little bit more about what that means, and what matrix are you looking at to measure that?

Well, I mean, I think the best way to describe that is our goal to design a business model, a UMC business model around our competitive advantage. And so, even what [where we do it] and try to be more selective in spending our resources and for revenue growth, and so that we'd be able to generate sustainable profit, now of course end of the day we've been to try to maximize the return for our shareholders. So that is truly our plan of just including a business model that you can do that. The other measurement that we look at is we look at the ROI and ROE, and we are not satisfied with the current level. And so, the overall objective we try to strive for of much better ROE and ROI.

Is there a target for what you think ROE and ROI can get through in the next several years?

Well, I mean it's not something I would like to elaborate right now. But for ourselves, I think I have a target, but I think it's not ready to elaborate that yet.

Yes. I mean, is it fair to assume that you have to get above obviously number one, your cost of equity and number two, you look at your peers and you look at ROE that way, are those fair assumptions?

Yes, that will be -- that will be a good assumption, yes.

And just a quick second question, you said that 14-nanometer is a [2K] this year in terms of capacity. How much CapEx was spent on 14 this year?

A very small fraction of about NT$ 1.7 billion.

We thank you for all your questions. That concludes today's Q&A session. I'll turn things over to UMC, Head of IR for closing remarks.

Thank you for coming to join us today. We appreciate your questions, as always if you have any additional follow-up questions, please feel free to contact UMC at ir@umc.com. Have a good day.

Thank you, ladies and gentlemen that concludes our conference for 2017 second quarter. Thank you for your participation in UMC's conference. There will be a webcast replay within an hour, please visit www.umc.com under the Investors Events section. You may now disconnect. Goodbye.