the leading edge and the more traditional business. When we look at the leading edge business, including AI and high-end networking, business is continuing to boom. We cannot seem to get the capacity installed fast enough. For us, this is also not a single customer phenomenon either. And the interesting part is that it appears to be accelerating.

On the more traditional business side, there is some excitement regarding potentially shortened refresh cycles on a number of products due to new AI features. But by and large, our customers are still proceeding cautiously. For us, that means our customers generally are in a trial-and-error mode. They start with a conservative forecast.

If their products hit, they will quickly adjust their outlooks. There is an increasing optimistic sentiment in the market, and we are looking forward to when such optimism manifests into increasing forecasts.

If we look at our overall cost environment heading into the third quarter, we will need to continue to manage our costs of production with higher utility costs and ramping labor needs. On the operating expense front, we also see continued spending on R&D labor, tools and equipment related to leading-edge advanced packaging and testing.

And as I stated earlier, we will continue to bear higher operating expenses before the majority of revenues come into place towards the latter part of this year and next year. We are looking to hold our full year operating expense percentage increase to within 75 basis points of that of 2023.

With that, I would like to hand the floor over to Joseph Tung to speak to our corporate outlook.

Joseph?.

Yes. Thank you, Ken. Let me give you our guidance for the third quarter. In terms of ATM, in NT Dollar terms, our ATM third quarter 2024 revenue should grow by high single digits quarter-over-quarter. And our ATM third quarter gross margin should be between 23% to 23.5%.

In terms of EMS, in NT Dollar terms, our EMS third quarter 2024 revenue should grow mid- to high teens quarter-on-quarter. Our EMS third quarter 2024 operating margin should be slightly above fourth quarter 2023 level of 3.5%. In addition to the revenue and margin guidance, I would like to also provide an update on our CapEx.

In anticipation of the booming demand for leading-edge ATM capacity and further technology advancement, we believe we will need to step up our CapEx investment, starting from raising full year 2024 consolidated machinery CapEx to double from last year's USD 914 million levels.

And as these investments carry higher value contribution and are becoming more critical to the overall semiconductor manufacturing, we believe such investment will be both return and margin accretive. On geographical expansion.

We started our Malaysia expansion 2 years ago, and our Phase 1 has been completed, and we will start volume production in quarter 1 '25. Also, we're soon to close the acquisition of 2 Infineon operations, one in the Philippines and one in Korea, and expect revenue contributions of these 2 sites starting from third quarter this year.

We have also acquired land in Mexico and we've identified potential locations in Japan. The land will be developed and buildings be constructed in anticipation of future customers' demand, and all these efforts are made to shorten the lead time for capacity ramp-up. With that, we will open the floor for questions..

[Operator Instructions] Our first question is from Mr. Gokul Hariharan of JPMorgan..

My first question is about the guidance. So when you talk about ATM high single digit, could you also help us to understand the full year growth prospects for ATM? Because I think back in February or January, I think Tien had talked about 6% to 10% for industry and similar growth for ATM.

Any updates there given year-to-date, it looks like you'll be tracking at about flattish year-on-year on USD terms. And also gross margin. Previously, we were expecting 25% to 30% range in the second half of the year.

Any reason, Joseph, why it's a little bit lower? Are there any start-up costs or any other additional costs that have come in? Or is it just that utilization is tracking lower. And lastly, on CapEx, This double from FY '23, obviously, a bigger number than previous.

Any breakdown in terms of test versus advanced packaging versus traditional packaging in terms of where you are putting in incremental CapEx. That's my first question..

I guess the full year -- in terms of a full year perspective, I think the -- in the general market, the recovery seems to be a bit slower than we were expecting. So in the -- although in the second half, we'll start to see the -- general market start to bottom out, and we'll start to have some recovery -- more recovery there.

But the pace of it seems to be slower than it originally expected. But on the leading edge, we are still is booming, but we are aggressively trying to catch up with the capacity to meet the growing demand.

Well, all in all, put everything into consideration, I think for the full year, we're now looking at a more moderate type of growth in terms of our top line.

And in terms of margin, I think, again, because of the more muted recovery of the general market, I think the -- in third quarter, we will come in a little bit short, reaching the structural margin of 24%.

But on the full year basis, we still remain hopeful that going into second half, we will have a better pickup and we're still -- I think the -- coming back to the future margin is still in play at this point.

What's the third question?.

CapEx breakdown..

The CapEx breakdown. With the new CapEx that I just mentioned, I think the breakdown will be roughly 53% for assembly and 38% for test. We have another 1% for material and about 8% for EMS..

That's very clear. My second question, Joseph, is on the 2.5D advanced packaging where you're obviously spending a lot of CapEx and the prospects are good. I think last time, I think Tien had mentioned NTD 50 million additional revenues from this segment.

And I think recently, I think in your AGM, you had indicated there's probably some upside to that number.

Could you talk a little bit about, a, what is the scale of this kind of business? Is it -- do you think that this could be 10% of your ATM revenues next year? And secondly, could you talk a little bit about your partnership with the leading foundry given they also called out increased usage of OSAT partners to fulfill customer demand? How widespread is your kind of customer base? Is it mostly coming from this leading AI accelerator customer? Or is it kind of more broad-based you're having multiple projects from many different customers for 2.5D packaging?.

Well, I think in terms of leading edge, I think mostly is still coming from AI or high-performance computing. And I think it's a broader base. It's not just one customer phenomenon, as Ken mentioned earlier on. And we are having reactive engagements with both of our foundry partners as well as customers directly.

And I think the raising of our CapEx is towards building the needed capacity for this type of -- including packaging as well as tests..

Next question is from Ms. Sunny Lin of UBS..

Could you hear me?.

Yes..

So my first question is to follow up on the testing opportunities that you just mentioned regarding the advanced packaging business. And so could you share a bit more details regarding the customer engagement, and for some of the key projects, how you compete with the existing suppliers? ..

Existing suppliers?.

For testing, for these HPC projects..

I think the -- there's plenty to go around as we see the demand for leading edge, both for packaging and tests are booming. And we are currently still a little bit capacity constrained, and we are beefing up our CapEx in this area to build the capacity as well as to further our technology investment.

I think the ambition, demand is coming from different customers and also a part of it from the foundry as well. So we will entertain whatever business that is needed from our customers. And in terms of meeting competition, there's always competition, we will do whatever we can to further penetrate whatever business that's in front of us..

Maybe let me try to ask from a different angle. And so for some of the AI accelerated products, there are some pretty solid existing suppliers that have very close relationship with the customers. They are willing to customize the burn-in tools and also the Board's. And therefore, I just wonder what's your competitive strategies.

And if you manage to get some of these products, when should we expect the revenue to decide to come through?.

We are gearing up on our leading-edge testing capacity and capability, including burn-in. We are basically leveraging on our turnkey services, we do expect to have good progress starting from next year. ..

Our next question is from Mr. Charlie Chan..

So, maybe to begin with, I'm wondering, Joseph, your view about Kenneth and Tien talking about the Foundry 2.0. They include the began foundry as to their 10.

Do you think going forward, would be more competition from foundry? Or do you believe there should be more collaborations?.

That is really a testament that the back-end packaging and test is becoming a more and more critical part of the overall value chain. And we are excited about the opportunity in front of us. and we will be closely working with our partners both on the upstream and also downstream to make this thing happen.

So I think the overall relationship with both our customers as well as upstream foundries. We are working very closely with both of them, and we'll try to come up with the capacity and the technology needed to suit their needs..

So my follow-up question is also to clarify your potential expansion of the advanced packaging business. So a couple of speculation, maybe it's a good chance for you to clarify here. So first of all, my understanding is that you only do the Wafer-on-substrate WoS right, the OS. But the chart suggests that you want to expand to chip on wafers.

That's the first part. And secondly, there's a rumor talking about you probably will do a new fab very close to Taichung or Changhua area, and there's a kind of offloading TSMC's burden. And number three is really the capacity number. As far as I know, probably in terms of WoS capacity, you have 60,000 wafers per month.

But rumors talking about you're going to more than double the capacity for WoS.

So those kind of rumor or speculation, can you clarify a little bit?.

We're not going to clarify the speculation, by the way. But I will give you a qualitative description about the current engagement model. The technology takes years to develop.

So whether the OS or the CoW or the equivalent of OS or the equivalent of CoW has been developed together with the foundry partners as well as key customers for years, so in the current landscape because of the capacity constraints, there has been active engagement with our partner as well with our customer to see which route will give you the most efficient alternative to fulfill the customer demand in the shortest amount of time.

So this is kind of the partnership that we have developed with the leading foundry suppliers. We have also gained confidence from multiple customers over years of product development trial-and-error process, materials or different equipment set and the reliability. So in this round, as we can see, the OS ramp has been pretty successful.

And we will continue to measure and monitor the progress of ramp-up to make sure the yield, the quality and the cost model are all in check. We cannot give you the specific comment based on the number that you have outlined.

However, if you go back to the CapEx number we have provided, I think you can clearly understand that we're trying to ramp up rapidly on the OS. The CoW has been engaged with multiple customers as well as with our foundry partner.

So we have active conversation together with partner and customers, again, just to seek the best alternative way to fulfill the customer demand. I think the -- we're not in a position to comment detail of the engagement model as well as the quantity.

But over time, just to give you a rough number, at the beginning of the year, we estimated we will have about NTD 250 million incremental advanced technology revenue. I also commented we are slightly ahead of that curve. So right now, that status is we are ahead of the original number. The momentum most likely will accelerate into next year.

Also, as indicated by Joseph and Ken's comment, that we're adding the -- we're doubling the CapEx. In 2025, we will continue to monitor the situation, and based on the business landscape, that we will decide what will be the right way to do it.

Okay?.

Next question is from Mr. Rick Hsu of Daiwa Securities..

Hello.

Can you guys hear me?.

Yes, we can hear you. Yes..

Just my -- the first question is my usual housekeeping question about your utilization rate for wire bonding and the testing for Q2 and Q3.

And also a little follow-up, when Ken say something about your advanced utilization rates was already full, what's your quantitative definition of the full?.

In terms of utilization in quarter 2, we have both our packaging and testing running at slightly above 60%. And that ratio will go above 65% for both in quarter 3. When we talk about full, we mean everything is being used up. All capacity....

Okay..

Adding to what Tien just mentioned, I think when we talk about expanding our capacity, we are talking about expanding capacity for all kinds of demand coming from our customer, whether so CoW or OS or testing.

Also, not just on the machinery itself, we're also adding spaces, factory spaces, new factory buildings that are being built or being putting on the stream, online. We are also investing heavily, like we're beefing up our R&D as well, making a lot of investment in R&D and also in terms of hiring talents.

It's an all-around effort for us to meet the customer demand..

The second question is about your CapEx. We see a very strong commitment in building the advanced package and testing capacity, such as on substrate and also the core equivalent.

Would you worry about any potential risk of overflow business model, i.e., that when the foundries resolve their bottleneck issue, will they cancel the outsourcing? I mean, what is about the overflow risk?.

I think the business model, the -- of course, anything is possible. But the collaboration has been in place for many, many years. We're also very used to the up and down cycle. So in this round, really depends on the -- across the whole platform.

We're looking at multiple customer product development, that we look at their utilization and the volume road map. Then we look at our investment. Of course, we will examine the flexibility of the investment we put in place.

So the nature of the OSAT business is we will build a flexible and fungible capacity, and that can be easily moved around to accommodate different product mix as is volume. So right now, we're quite confident the capacity we put in place will be here and will be very useful for our customers and our partners for quite a few years, at least.

So the short answer is no, we don't worry about it..

Next question is from Mr. Brad Lin of BOA..

I have two questions. So the first one is on the overseas expansion. So does the firm have any new thoughts on the US expansion? I know that we comment about overseas, but seems to -- well, we did not mention much about the US side.

So on the new subsidy program by the US government and also the increasing political dynamics there, so if -- does ASE have any new thoughts on the US expansion? And also, we also see the expansion of the ISE labs there in California.

Does that qualify for apply to the new subsidy programs?.

Just to give you an answer, the -- well, let me comment on the ISE facility. The ISE facility only has one purpose, that is to support the innovation at a chip level or at a system level in Silicon Valley for either AI or the future edge device or any advanced technology, high performance computing or silicon photonics or power management.

So the lab is put in place to provide system-level testing, system-level development as well system-level reliability. The lab is in California, Silicon Valley.

The main mission is to bridge whatever technology are available domestically in the US and whatever technology that are needed that are available globally, right? So the CHIPS Act office just announced the R&D funding as well as the manufacturing funding. We will consider how to apply.

And for ISC and IC is similar with the Phase 3 or Phase 4, we will consider the CHIPS Act R&D as well as the manufacturing. The US high-volume manufacturing establishment are still being considered. We do not have a tangible plan yet, mainly because the customer request on the volume are still being investigated..

So can we say that, well, with recent developments, there is a higher possibility that we will do the so-called volume production side there..

We do have a Mexico facility, and that is also available for North America. Just to give you the clarification..

We have a question from Ms. Laura Chen of Citigroup..

Can you hear me?.

Yes..

I have questions about the gross margin trends. On one hand, we are seeing that we are accelerating our investment in the advanced packaging, but the broad based recovery on the traditional application seems to be -- remain muted.

So my question is that how would we expect this advanced packaging investment to bear fruit? If it's reached the stated 5G array and also the loading, how would you see that will be the normalized gross margin? That's my first question..

As I mentioned earlier on, the CapEx that we're making for the leading-edge packaging and test, those investments are of higher margin -- or these investments are of accretive margin as well as return.

So as we continue to grow that part of the business also on tests, I think it will help the overall margin for our ATM business as we -- as time goes on..

So can we expect that there will be over -- like if the overall ATM business reached, say, like 80% utilization rate, usually, we can expect like 25% gross margin. But with the cohorts, if that's reached a satisfied utilization rate and also the loading, that will be much higher than that what we expect before..

Such capacity is also higher than the corporate average margin..

And also, my next question is on the EMS business. You mentioned that the better than expected near term is due to the earlier ramp for the new products.

So do you expect the stability will continue into the peak seasons into Q4? Or can we expect that Q4 will be even stronger like the historical pattern?.

No, I think the first half versus second half pattern should be similar to a typical seasonality..

Next question is from Mr. Bruce Lu of Goldman Sachs..

I think I want to ask again for the advanced packaging, all right? So Joseph used to comment that the CapEx to revenue ratio. Can you comment about like what is the CapEx revenue ratio for the advanced packaging? You spend a lot of money this year.

How much revenue you can generate next year? In addition, TSMC commented that the advanced packaging gross margin is somehow approaching to the corporate average with 50-plus percent, right? So can we expect -- is there any reason that we can assume that your gross -- your profitability was somehow similar with TSMC for your advanced packaging?.

I don't think we should make that kind of the simple comparison. Because TSMC, the packaging they do as was the business model are slightly different for ASE. Joseph has already talked about it. The -- the OS that we're doing right now, the margin is higher than the ATM, the corporate average. That's where we're going to stand.

We're not going to comment comparing to the TSMC margin because the equipment, the process, all of the -- everything they use are slightly different than the ASE. Now in terms of the -- how do we bridge the dollar, I think normally, we're talking about the CapEx dollar, I think it's dollar for dollar..

So for advanced packaging, it's also dollar for dollar..

That's the current supplement.. We will actually need to -- once it gets to volume, then we will have better assessment of that. That seems to be still the case, yes..

The second thing is that I want to ask about seasonality for your EMS business. I think the third quarter guidance seems to be a little bit lighter than historical seasonality. So what was the reason for lower seasonality because -- we didn't hear any delay for the flagship model smartphone.

It's because of the smartphone or because of other applications?.

I think the people are expecting the new form [indiscernible]. Seems people are sending an uptick on the shipment. So this quarter -- and because of the earlier launch of the product, we're seeing a better-than-expected second quarter. And that same kind of momentum should go into quarter 3.

But as a whole, the other general markets, the recovery is still a bit slow. So I think from a full year perspective, I think EMS will have -- also have a muted year for the -- in terms of our top line..

[Operator Instructions] Next question is from Jason Tsang of CLSA..

Can you hear me?.

Yes..

My first question is regarding on the proportion of your advanced packaging. Can you give us how many percentage is that advanced packaging account for your total revenue? And you mentioned about NTD 250 million.

Is that your target in coming years? Or is expectations on the contribution from advanced packaging in this year?.

As we guided before, this year we're expecting to double the -- our revenue in terms of leading edge packaging tests. And we are -- and the incremental revenue should be over NTD 250 million. As Tien mentioned, we are currently slightly ahead of our schedule. And I think the momentum continues to be strong.

And going into next year, we're expecting another doubling of revenue in this segment..

My second question....

And the percentage of our leading-edge revenue is about 5% this year, going from 2.5% to over 5% this year..

5% of ATM business, right?.

Correct..

My second question is in terms of the industrial and auto business. Because practically, we merged the new fab from ID employers. And we know that currently, those segments remain weak.

So have we found any synergy or upside coming from those kind of new business? Or can we comment that whether we see any improvements or recovery on those segments?.

Well, I think in terms of auto, overall sentiment is still very soft. But we have been continuing our growth in the automotive business, largely because of the market share gain, which we are leveraging on our automation to continue to expand our market share. For this year.

I think in terms of ATM, the automotive part of the business will represent roughly 11% of the overall sales. And we'll see the momentum continue to grow...

We have a question from Mr. Charlie Chan of Morgan Stanley..

So actually, two follow-up questions from me. So first of all, I think Tien seems to suggest that the new CapEx for advanced packaging margin should be above the corporate margin.

But Joseph may clarify, is that both corporate average margin or above ATM margin?.

Above corporate ATM margin..

Corporate ATM margin. That's clear. Yes. And also doubling this revenue next year.

So it's doubling the 5% or doubling that USD 250 million incremental revenue?.

Doubling, I mean, next year's leading-edge revenue will be doubling this year's leading-edge revenue..

So if this year is around 5%, next year, it should be close to 10% or more than 10%..

Well, we are expecting other parts of the business to grow as well. So that remains to be seen, yes, whether it will be exactly 10% or a little over or a little under 10%..

Next question is from Mr. Bruce Lu of Goldman Sachs..

I want to know again for the advanced packaging in terms of investment direction. Because in advanced packaging for me, there are like 10 different technology from TSMC, ASA, ORR, SOI, CPX or whatever. I mean, it seems to me it's very complicated. And how do you decide which one is the direction to go for ASE to invest? That's one thing.

The second thing was TSMC is talking about ramping up the panel level packaging in 2 years, so which means it could be the case that they don't want to invest in the current cohorts. Because in 3 years, they can migrate to panel and they've got to leave that to the current OSAT partner. So that could be a short-term business potentially.

So how can we prevent this kind of read or what is the direction to go to?.

the common usage, the sheer usage and the flexibility to convert if needed. I don't think easy to say when technology can easily replace another technology. Each technology will cater for a different set of target customers. So when you talk about the cohorts, there will be different level cohorts already said that.

Even with the panel, the panel we're targeting at different line width, different lines of space. Each one is different. It's very complicated. But our job is to prepare the whole portfolio of toolbox, whatever the customer have a specific product mix they need to ramp. We are readily available to do that ramp for them.

And that's why we need to work very closely with our foundry partner as well customers because it's a very, very dynamic time. You can't really know a priori what the volume mix is going to be. But our job is to make sure we have all of the technology development, we have all the building ready.

And then when we need to, then our team will be very busy to convert one line to another or the other back to another just to make sure we have the operational efficiency..

Your panel size seems to be different with TSMC what's trying to do.

Is that going to be different? Is there going to be an issue?.

I'm not going to comment on TSMC's panel size. I think you have to ask TSMC. But ASE is pretty determined. We will do 600 x 600. Our customer seems to be comfortable with that dimension. The equipment will be different. The process will be different. But again, as I said, we cannot say one panel, there is no a universal, the definition of any process.

Well, ASE is working on the 600 x 600 as a certain feature size, as a certain cost model. And then we are working with our partner as well as our customer to make sure they're comfortable..

Can you disclose the predication for the panel’s customer? Is this is for power management or for AI or....

Actually, it's for all of the above..

I see. Okay. My next question is for gross margin in the second half. I think, Joseph, you mentioned that in the second half, ATM gross margin...

Our next question is from Mr. Gokul Hariharan..

I had one quick question. On your smartphone segment, which is the largest segment, I think, Joseph, you talked about some improvement in demand from a particular set of customers. Could you talk a little bit about what is the status of the smartphone segment, both for high-end SoC as well as for other accelerate -- auxillary products.

How is the inventory level? And are you seeing a more broad-based enthusiasm from customers to actually order more? Or things are still quite sluggish? Second part of that is for Tien. Tien, I think you talked about the toolbox of portfolio of packaging technologies that you have.

Smartphone has been an area where the packaging technology has largely been stagnant for quite some time, I guess.

Could you talk a little bit about are we seeing anything in the horizon where packaging technology could evolve towards something similar to the 2D to 2.5D fan-out for the smartphone segment in the next couple of years?.

The first question is regarding the cellphone. And this year, we're seeing an uptick on cellphone. If we look at the Q3 and Q4 forecast, the cellphone is ramping up in the -- just overall. So I think overall, this year, the cellphone over shipment, I mean, we're not in the position to come on that.

But based on the forecast we're looking at, it seems to be moderately positive comparing to, for example, automotive and industrial. Okay. That's the first comment. In terms of the toolbox, also the -- how could cellphone specifically, that segment, can utilize some of the new technology we put in place.

We believe on the technology road map, with all of the AI initiatives, over time, you have two types of devices. First, when do we expect the high-performance computing type of architecture to permeate into the cellphone. That will be the first question.

And we believe with all of the AI features coming in, if the market response is very good, I'm pretty sure the leading cellphone technology suppliers will start adding either more higher level of memory or higher-level, bigger chips into it.

If that is the case, then the current technology that's only be applicable to high-performance computing naturally will take some shape into the cellphone market. Now the second type of thing will be the edge device.

I think one of the earlier comment that we did not get a chance to answer is with all of the fabs, I think statistically, we have like 84 fabs coming up right now. They're pretty much in the traditional, not the leading edge.

So in the next few years, with all of this traditional fab, capacity coming online, how do we balance the semiconductor over demand versus supplier. I think the general industry view is with AI, at the top, over the next few years, there will be more edge devices such as autonomous driving, such as robots, such as drones, start coming out.

Now in any kind of edge device or edge device system, you will utilize a lot of actuator, MEMS, sensors, microcontrollers in an integrated and hydrogenous integration fashion.

So if you think about the potential robots, the potential autonomous driving, the potential drones, they will use a lot of conventional semiconductor devices as well as the AI engine or the AI brain. Now the semi talked about it. In 2030, '31, '32, the industry will hit about $1 trillion.

If we want to hit $1 trillion, naturally, we will use up all of the fab in the leading edge as well as the traditional package level.

And the OSAT or ASE, all of the automation that we put in place, all of the building blocks we put in place are targeted to capture not only the high-performance computing as well as the next generation AI brain to the edge device as well as to robotic fingers for all of the heterogeneous integration or any kind of existing package that our customers or system designer would like to do.

It's a very long answer to your question, but I hope that's comprehensive..

Yes, that's very clear.

So one follow-up on that, Tien, is if you had to hazard a guess, when do you think smartphone or cellphone SoC starts using chiplet? Is it in the next 2 years, next 3 years? Or is it much further than that in terms of technology readiness and market acceptance?.

I think the chiplet has already started. The chiplet really is a concept. In other words, when you have -- when you start integrating chip, it becomes a chiplet. So today, we're seeing the chiplet architecture in a supercomputing, high-performance computing.

But even when you clearly go to the electrical vehicle or as the current generation of cellphone, you can pretty much argue this is the beginning of the chiplet. So I think the chiplet concept, the platform are being validated by the market. Over time, I think the chiplet will become more pervasive.

Therefore, the hydrogenous integration, the power management, the embedded, all of this technology that we're developing today will become very useful. And that's where the panel, for example -- the panel, I did not get a chance to comment, we think about, at the power management level, there will be a lot of integrated devices, integrated package.

They're thinking about it. So the panel needs to be lower cost, flexible enough to accommodate the power management integrated package, but also needs to go high enough to accommodate the current form of high-performance computing. But any kind of package will have a different set of target customer.

If you're talking about 3D, you're talking about 2.5D or interposer, that feature size can continue to move up become smaller and smaller, to get the proportional to the lithography and also the density requirement. So our job is to make sure we have a whole spectrum of technology in our pocket.

Whenever the customer comes in, they need 10% of high end, 40% of mid-end and 50% of low end, we can easily provide them a comprehensive selection of tools in an automated manner..

[Operator Instructions] We have a question from Bruce Lu. Bruce, sorry about interrupting your question earlier..

Just one quick question. Joseph, you did mention that earlier, the second half gross margin for ATM will go back to 25%, 30% due to better product mix, better testing, more events. It seems to be slightly lower than guidance. Can we expect that to go back to 25%, 30% in margin in fourth quarter? What slows us down for the gross margin recovery? ..

I think the overall recovery was slower than we were expecting. And because of the volume shortage or the gap we are now -- the margin prospect for the second half seems to be a little bit lower than what we were originally expecting.

We do expect that going into fourth quarter, we should be able to go back into our structural margin range and when the overall utilization goes beyond 70%. And as I mentioned, it's more challenging now.

But I think for the whole year, we're still trying to struggle to see if we can still reach the lower edge of the structural margin, which is about 24%..

But not -- but the product mix improvement and more testing revenue is still intact in third quarter, is that right?.

I'm sorry, the what?.

The more testing revenue or the more advanced packaging contribution in third quarter is still intact? Just the traditional packaging that you guys are trying to get lower. That was the main reason for lower gross margin in third quarter.

Is that right understanding?.

That's correct. In terms of testing, we are progressing as planned. But the overall general market recovery seems to be slower. That's dragging our margin a bit, yes..

There is no question from the floor..

If there's no more questions, we will end the conference call today. Thank you very much for attending, and we will see you next quarter. Thank you..

Thank you..