日月光投資控股 (ASX) 2023 Q2 法說會逐字稿

Hello, I am Ken Hsiang, the Head of Investor Relations for ASE Technology Holdings. Welcome to our second quarter 2023 earnings release. Thank you for attending our conference call today. Please refer to our safe harbor notice on Page 2. All participants consent to having their voices and questions broadcast via participation in this event. If participants do not consent, please disconnect at this time.

I would like to remind everyone that the presentation that follows may contain forward-looking statements. These forward-looking statements are subject to a high degree of risk, and our actual results may differ materially.

For the purposes of this presentation, our dollar figures are generally stated in New Taiwan dollars, unless otherwise indicated. As a Taiwan-based company, our financials are presented in accordance with Taiwan IFRS. Results presented using Taiwan IFRS may differ materially from results using other accounting standards, including those presented by our subsidiary using Chinese GAAP.

I am joined today by Joseph Tung, our CFO. For today's session, I will be giving the prepared remarks. Joseph will then be available to take your questions during the Q&A session that follows. The following financial references we'll be referring to the first half of 2023 as compared to the first half 2022.

The first half of the year has been characterized not just by an unprecedented inventory correction, but also broad-ranging declines in electronics demand, as consumers catch up on travel and socialization. Issues relating to a slower-than-expected economy in China exacerbated already lackluster electronic spending. The excess inventory environment appears to be stretching out well into the back half of 2023.

In this climate, the holding company first half revenues declined 12% year-over-year, while our ATM and EMS first half revenues declined by 17% and 7% respectively.

For our ATM business, our automotive-related application bucked the trend and grew by 15%. Advanced packaging revenues declined slightly more than wirebond packaging revenues during the first half of the year, 20% and 17% respectively. Our advanced packaging products have more exposure to the more heavily impacted Communication segment.

Meanwhile, our testing business fared relatively better during the first half, declining 10% when compared with last year. This was driven by more stability of our China test business. Our EMS business declined 7% in the first half of 2023 when compared with the first half of 2022. This represents the overall weaker environment during the current year.

Holding Company first half operating income declined by 53%, ATM first half operating income declined by 57%. As can be seen here, our business is highly dependent upon the leverage generated when equipment utilizations are full and operating efficiency is more optimized.

For our business, it may appear things seem pretty bad with the overall economic environment and lengthening inventory digestion. But if we see this as the trough of all troughs, ASE appears to be doing okay. Of course, we would like a rapid return to health for the entire industry, but as a point of historical comparison, the last major downturns ASE saw were as follows.

In 2001, revenues were down 27% year-over-year with gross margin at 13%. In 2005, business was down 20%, with gross margin at 19%. In 2009, business was down 12%, with gross margin at 20%. Arguably, the current downturn is more severe in duration, and at least equal in terms of correction percentage. And yet our margin structure has held up and we are making more EPS in this downturn during the 2 trough quarters than compared to any of those full years. We believe this downturn is proving that our market position has significantly increased our resiliency and improved our structural business model.

For the second quarter, our ATM business experienced a relatively sluggish environment, but was somewhat better than original expectations due to higher-than-expected rush orders. Overall demand for services slightly improved off a first quarter trough levels.

Customer wafer inventories at our facilities appeared to be in the initial stages of decline. However, given the overall tepid market environment, customers are becoming even more conservative on their inventory levels.

For our ATM factories during the quarter, key equipment utilization rates were still low, staying near 60%. On a more positive note, we are seeing more incremental pickup related to R&D, new product introduction work. Our EMS business picked up slightly as anticipated. This is in line with our outlook and in line with our typical seasonality.

With that, please turn to Page 3 where you will find our second quarter consolidated results. For the second quarter, we recorded fully diluted EPS of TWD 1.76 and basic EPS of TWD 1.80. Consolidated net revenues increased 4% sequentially and declined 15% year-over-year.

We had a gross profit of TWD 21.7 billion, with a gross margin of 16%. Our gross margin improved by 1.2 percentage points sequentially, and 5.4 percentage points year-over-year. The sequential improvement of margin is principally due to higher loading in the current quarter. The annual decline in gross margin is principally the result of lower loading during the current downturn.

Our operating expenses increased by TWD 0.7 billion sequentially and declined by TWD 1.5 billion annually. The sequential increase in operating expenses are primarily due to higher R&D expenses at our ATM business and labor-related costs as customers are starting to ramp new product introductions.

The year-over-year decline was primarily attributable to lower bonus and profit-sharing expenses across the company. Our operating expense percentage increased 0.1 percentage points sequentially, and 0.4 percentage points year-over-year to 9%. The sequential operating expense percentage increase was primarily related to higher R&D costs relative to revenues generated. The annual operating expense increase is primarily due to lower overall loading relative to semi-fixed operating costs.

Operating profit was TWD 9.4 billion, up TWD 1.7 billion sequentially and down TWD 11.2 billion year-over-year. Operating margin was 6.9%, improving 1.0 percentage points sequentially and declining 5.9 percentage points year-over-year. During the quarter we had a net non-operating gain of TWD 0.7 billion. Our non-operating gain for the quarter primarily consists of net foreign exchange hedging activities, profits from associates, and other non-operating income offset by net interest expense of TWD 1.1 billion.

Tax expense for the quarter was TWD 1.9 billion. The effective tax rate for the quarter was 18.9%. Net income for the quarter was TWD 7.7 billion, representing an increase of TWD 1.9 billion sequentially, and a decline of TWD 8.3 billion year-over-year.

The NT dollar depreciated 0.6% against the U.S. dollar sequentially during the second quarter, and 4.5% annually. From a sequential perspective, we estimate the NT dollar depreciation had a 0.16 percentage point positive impact to the company's gross and operating margins.

From a year-over-year perspective, we estimate that the depreciating NT dollar had a 1.28 percentage point positive impact to gross and operating margins. As a rule of thumb, for every percent the NT dollar appreciates, we see a corresponding 0.29 percentage point impact to our holding company gross margin.

On the bottom of the page, we provide key P&L line items without the inclusion of PPA-related expenses. Consolidated gross profit excluding PPA expenses would be TWD 22.7 billion with a 16.6% gross margin. Operating profit would be TWD 10.6 billion with an operating margin of 7.8%. Net profit would be TWD 8.9 billion with a net margin of 6.5%. Basic EPS excluding PPA expenses would be TWD 2.07.

On Page 4 is a graphical presentation of our consolidated financial performance. You can see the impact of the current weak environment here. It does look like we are looking at the first quarter as the bottom though.

On Page 5 is our ATM P&L. It is worth noting here that the ATM revenue reported here contains revenue eliminated at the holding company level related to intercompany transactions between our ATM and EMS businesses. For the second quarter of 2023, revenues for our ATM business were TWD 76.1 billion, up TWD 2.8 billion from the previous quarter and down TWD 18.9 billion from the same period last year. This represents a 4% improvement sequentially rather than a flat quarter. While on a year-over-year basis, we declined 20%.

Gross profit for our ATM business was TWD 16.2 billion, up TWD 1.4 billion sequentially, and TWD 11.6 billion year-over-year. Gross profit margin for our ATM business was 21.2%, up 1.1 percentage points sequentially and down 8 percentage points year-over-year. The sequential margin improvement is the result of slightly improved loading offset in part by higher utility costs, while the annual margin decline is primarily the result of lower loading due to the current downturn.

During the second quarter, operating expenses were TWD 8.7 billion, up TWD 0.4 billion sequentially and down TWD 1.1 billion year-over-year. The sequential increase in operating expenses was primarily driven by higher R&D expenses related to higher labor and new product introduction costs. The annual operating expense decline was driven primarily by lower labor costs due to lower profit sharing and bonus accrual.

Our operating expense percentage for the quarter was 11.5%, up 0.1 percentage points sequentially, and 1.2 percentage points year-over-year. The sequential increase was due to increased R&D expenses, while the annual increase was due to lower loading relative to semi-fixed costs.

During the second quarter, operating profit was TWD 7.4 billion, representing an increase of TWD 1 billion quarter-over-quarter and a decline of TWD 10.6 billion year-over-year. Operating margin was 9.7%, improving 1 percentage point sequentially and declining 9.2 percentage points year over year.

For foreign exchange, we estimate that the NT to U.S. dollar exchange rate had a 0.28 percentage point impact on our ATM sequential margins and a 2.21 percentage point impact on a year-over-year basis. Without the impact of PPA-related depreciation and amortization, ATM gross profit margin would be 22.4% and operating profit margin would be 11.2%.

On Page 6, you'll find a graphical representation of our ATM P&L. As can be seen here, current year loading levels are still significantly lower than 2022.

On Page 7 is our ATM revenue by market segment. It's fairly similar as last quarter. However, over the course of the last 1.5 years, our Computing segment does appear to be gradually taking on larger segment share.

On Page 8, you will find our ATM revenue by service type. There isn't a significant change here.

On Page 9, you can see the second quarter results for EMS business and a graphical representation of its market segment allocation. As usual, the second quarter is the dull financial quarter during which not much changes from the first quarter. But in actuality, our EMS business is in the midst of preparing for its seasonal up cycle.

During the second quarter, EMS revenues were TWD 60.4 billion, improving TWD 2.7 billion or 5% sequentially, and declining TWD 5.8 billion or 9% year-over-year. Sequentially our EMS business's gross margin improved 1.4 percentage points, while our operating margin improved 1.2 percentage points. The margin improvements were driven primarily from favorable foreign exchange impacts to raw materials, and overall product mix.

Our EMS second quarter operating profit was TWD 2.1 billion, up TWD 0.8 billion sequentially and down TWD 0.6 billion annually. For our EMS market segment, our Consumer segment picked up as Industrial and Automotive segments declined. This was driven by slightly stronger demand in the current quarter related to our Consumer SiP product and temporarily weaker demand environment related to industrial products. Our Automotive business looks to be impacted by end market dynamics.

On Page 10, you will find key line items from our balance sheet. At the end of the quarter, we had cash, cash equivalents and current financial assets of TWD 66.4 billion. Our total interest-bearing debt was down TWD 3.2 billion to TWD 187.1 billion. Total unused credit lines amounted to TWD 384.6 billion.

Our EBITDA for the quarter was TWD 25.8 billion. Net debt-to-equity was down to 41% at the end of the quarter. We expect our debt position to increase during the third quarter as a result of incremental cash usage to pay our upcoming dividend.

On page 11, you will find our equipment capital expenditures. Machinery and equipment capital expenditures for the second quarter in U.S. dollars totaled $209 million, of which $107 million were used in packaging operations, $60 million in test operations, $33 million in EMS operations and $9 million interconnect material operations and others.

Current quarter EBITDA of USD 0.8 billion continues to outpace our equipment capital expenditures of $0.2 billion.

Before we get to our outlook, I would like to spend a bit of time on what we believe drives our business and touch upon the topic of the moment, AI. Our core business is to provide our customers unprecedented scale and repeatability of manufacturing. With such scale, we are able to achieve the lowest manufacturing cost without compromising quality. Interestingly, this improved scale of manufacturing also delivers the highest packaging yield.

We consider this to be our competitive advantage. I've included a rough and very simplified chart related to our business. The X-axis represents the number of package I/Os, and the Y-axis is a smooth exponential scaled estimate of volumes. We've placed some of our offered package types roughly along the X-axis. These package types actually overlap much more than what is shown here. But the basic message follows.

Our business is driven by the center area of the plot, where there are mass volumes. The more units we manufacture of a particular package type, the better able we are to wrap up economies of scale and manufacturing efficiency.

The right side of this curve identifies leading-edge advanced package types. As we've seen time and time again, as the industry adopts higher I/O counts, new package types reach a manufacturing critical mass and start climbing up the curve. This represents when lower-volume R&D lines become suitable to be scaled up for high volume mass production. We're starting to see this with some of our fan-out base packages.

For us, having substantial capabilities and leading-edge technologies is necessary as we want to be ahead of the curve, quite literally as in this reference to this graph. But the most technologically advanced packages neither drive significant volume nor significant profitability. For the back-end, package technology migration matters, but volume fundamentally determines manufacturing efficiency, and thus for us, profitability.

Now to the business of AI. Recently, we have been frequently asked to comment on our view of AI. Often, these discussions have centered on products that are very specific to the core enablement of AI training.

We understand the excitement and importance of such specific products which are enabling AI development. However, we believe that zeroing in on a couple of specific tip-of-the-iceberg products tend to be somewhat missing the bigger picture.

AI's impact on ASE will be much broader and multifaceted than focusing in on specific leading edge units. We believe AI is substantially a much larger phenomenon. We believe AI serves as a catalyst to the next super cycle for the semiconductor industry.

At a basic level, AI means more information will be collected, stored and analyzed everywhere and pertaining to anything and everything. From AI's edge data collection to the inference of that data, represents new volumes all along AI's processing chain. Not only do we see incremental product volumes ahead, future AI features will require an incremental step up in capabilities, increasing both die and package level requirements.

If we try to show this on our chart here, as incremental complexity ramps, higher I/O counts push the entire curve to the right. As you can see, this step-up will also drive adoption at the leading edge of our vertically integrated solutions. In this graph, the line pushes up for volumes and to the right for the technology step-up.

We also see further opportunities for system level complexities that may force further parallel design versus historical monolithic design tendencies. As overall system architecture must accommodate increasingly power hungry applications such as AI, we believe it becomes increasingly necessary to solve power issues near or at the package level. These are all frontiers on the heterogeneous integration roadmap.

AI only serves to add to this need and opportunity for us. The chart here extends once again for the potential proliferation of heterogeneous integration, all across our package types. We may see AI as a novelty now, but it will be ubiquitous soon enough.

As we come back to look towards the nearer term, we are seeing a ramp in manufacturing for the coming third quarter across our businesses. It's definitely not the pickup of a typical manufacturing year, but given the entire macro climate, including post-COVID spending patterns, lacklustre China demand and tightening inventory control by our customers, it's a decent start.

The current year does not look to be the V-shaped recovery that the industry was hoping for. It's looking more like a checkmark with a return to optimal manufacturing, unfortunately, stretching into future quarters.

From the cost perspective, the coming quarter's cost environment will be impacted by higher summer consumption and 3 full months of higher summer utility rates. Further, we have a smaller set of products ramping in the third quarter as -- versus normal. Given the selective product mix, we see a temporarily higher raw material content environment for the coming third quarter.

We would like to summarize our outlook for the third quarter as follows. For our ATM business in NT dollar terms, our ATM third quarter 2023 revenues should grow quarter-over-quarter by mid to high single-digits. Our ATM third quarter 2023 gross margin should improve 75 to 100 basis points versus the second quarter of 2023. For our EMS business in NT dollar terms, our EMS third quarter 2023 revenues should increase 20% quarter-over-quarter. Our EMS third quarter 2023 operating margin should be similar with the second quarter of 2023.

This concludes our prepared remarks. We would like to start the Q&A here. Thank you.

Now we would like to open the floor for questions.(Operator Instructions) We have a question from Mr. Randy Abrams of Credit Suisse.

I'll ask the first question on the sales guidance. For the IC ATM, it looks like it's tracking fairly close to TSMC, but leading the Tier 2 foundries. Could you discuss a bit more on the split by applications, where you're seeing the incremental growth momentum continuing? And if you see that continuing into fourth quarter?

And then for EMS, the 20% sequential looks like it's lagging traditional seasonality. Could you talk if it's delayed timing on product ramps, softness in certain applications, or just a timing issue that could help shift some of the business out to fourth quarter?

I think for both ATM as well as EMS, the delayed of the -- somewhat delayed new product launches, this has an impact on the second half performance -- revenue performance as the -- we are expecting sequential growth continuing for the second half. Also into fourth quarter, the inventory digestion as well as the weak end market consumption is still continuing. So we are now taking a more conservative outlook on the second half. And so the -- what we're showing here is really the forecast with some judgment with the focus that we got with our own -- some of our own judgment to come up with the numbers.

For the profitability, you discussed a few impacts on utilities and materials. Could you discuss a little more of the mix? What's driving that short-term material shift? And with the leverage where it's about -- it looks like about a quarter or two light of kind of a normal pickup with that type of volume. Are there headwinds from pricing environment? How you're seeing pricing or other factors?

I think pricing still remains to be resilient at this point. And going into third quarter, we do have a higher cost structure given the fact that the -- we're in the summer time, and we're subject to summer rates for energy and that put some pressure on our margin expansion. Typically, when we see our overall utilization goes up and revenue goes up, the margin should -- expansion should be a bit higher than what we are presenting here. But given the higher energy cost, I think that's putting some pressure on the margin expansion.

And also in terms of product mix, I think some of the higher material content products are being put out in this coming 2 quarters. That's also putting some pressure on the material cost of ours, and therefore, the margin expansion. But I think overall the -- in terms of material content or product mix effect, we'll start to improve in fourth quarter.

Next question is from Mr. Gokul Hariharan of JPMorgan.

My first question is on the state of the inventory cycle. Could you give us a little bit more color on what you're seeing overall on wafer bank inventory? How quickly are customers drawing that down? And also, could you give us some more color on the Q3 revenue guidance of sequential growth? Big segments are driving the growth of mid to high single-digit. Is there any divergence between what you're seeing on your communication platform versus Automotive, Industrial, and Computing?

I think the overall pick up in the third quarter, I think it's across the board throughout all applications. As we see new products coming out, we see Automotive continue to be the bright spot for us.

And the first part of the question?

Wafer bank. I think it's being gradually worked down, but I think the -- nowadays, I think -- all customers, given the soft market condition, I think people are getting more cautious. So the -- I think the inventory digestion will continue for may be the next 2 quarters or even further quarters into the future. But as a whole, I think going into 2024, I think things will start to improve quite a bit, and we are expecting a much more healthier growth here for us in 2024.

My second question is on the AI-related demand. Any estimate of exposure to AI for ASE as of now? I think TSMC talked about roughly 6% of revenue this year. And how are you planning for capacity on AI starting from some of your 2.5D and fan-out-related products to also -- like Ken mentioned, moving into some of the more mainstream higher volume kind of products as well? How are you thinking about -- like when do these products start to kind of kick in into your packaging and testing portfolio?

Well, I think if we are narrowly defining AI today, the associated revenue for us, it's roughly at the low single-digit of our ATM revenue at this point. But I think more importantly, it's really given time, AI will be adopted into all existing and new applications. Therefore, given time it would create tremendous volume and business opportunities for us, covering all kinds of different chips that will be coming on stream, as Ken has demonstrated in the chart that we've shown earlier on.

In terms of exact timing, I think that's -- it's -- right now, AI is still at its early stage. And I think it will take some time for us to -- for this segment to continue to grow, but not particularly just AI itself. But as AI gets emerged into all different applications, that's when we will start to see explosive -- above volume growth and that -- and driving the whole industry into the next super cycle.

Next question is from Mr. Brad Lin of BofA.

I have a question on advanced packaging. So basically, Ken just mentioned that critical mass is quite important for us to ramp up the advanced packaging capacity. So are we reaching a so-called sweet spot yet? Or when should we expect the company to -- or the industry to reach the sweet spot? And do we have any incremental CapEx plan on it? That's my first question.

Well, it will definitely go into that stage, and we are preparing ourselves by keeping us at the forefront of the technology development. We will be making the necessary investment as we see fit when the time comes. And exactly when it will start to happen, I think it will be a progressive -it will take steps and it will progress very nicely going forward. We do have the capacity in place, meeting different kinds of requirements, but what we're waiting for is for the volume to start the increase and to grow, and we will make our necessary investment accordingly.

So also, we have learned from UMC's call yesterday about the -- its collaboration with ASE and some IC design service firms on a so-called open ecosystem for advanced packaging. So would you please share the pros and cons associated with this open system that ASE and UMC are forming? And also -- would you please also provide some insights into the so-called optimal collaboration model versus the current dominant close system by TSMC?

Well, we're definitely one of - we're definitely the initiator of the collaboration, particularly on the interposer. And this is a major component of the overall chip and wafer type of process. And -- so we are collaborating with the foundries to complete the full process for us. And on top of that, we also have our own solution focus that is progressing well. We have very active engagement with a lot of -- quite a few of our customers, and we believe mass production will start soon going to the later part of the year or maybe early next year.

Next question is from Mr. Rick Hsu of Daiwa.

First question is just a very general housekeeping question. What's your utilization rate across the board that your packaging test in Q2 and Q3?

Q2, we are roughly at 60% across the board. And going to Q3, given the pick of our business, we believe the average utilization will be hovering around 65%.

Second question is just a quick follow-up on this open ecosystem. You guys -- cooperation with the foundries like UMC producing the interposer, and you guys are in charge of the back-end services like CoWoS, 2.5D and 3D. A further follow-up is do you guys have any -- do you have any exposure right now in the core GPU area for the CoWoS? Any participants in this area for capitalizing the AI server demand growth?

Well, right now, the AI-related exposure is still low. We have -- as I mentioned earlier on, a single -- low single-digit percentage of our overall ATM revenue. Not commenting on the specific products, but we are defining this as our leading edge packaging technology that that's AI related. I think mostly in the networking, in HPC type of applications. And I think the potential of this business is also pretty good. I think going next year we could probably see this part of the -- or this particular segment of business should easily double for -- in next year.

Next question is from Sunny Lin of UBS.

So my first question is about longer term. And so obviously, last few years, ASE has been outperforming the semi industry probably because of the share gain and customer supply diversification. And so I wonder, now looking ahead, how much of the growth upside you think ASE could continue to realize? Meanwhile, China OSAT companies are also expanding overseas. So would that affect ASEs opportunity to some extent?

Well, definitely, we're the clear leader in our space, and we will continue to gain shares going forward, leveraging on our scale and our technology competencies as well as our proximity to the foundries, particularly on these more advanced technology that's coming on stream.

Going forward, I think geographical expansion will continue. That's really the going trend as we see demand starting to polarize between China and outside China. On that, we do have very good footprints around the world. We are currently expanding in Malaysia, in Poland, in Korea, in Japan as we see fit or as required by our customer. When we see business opportunity that we -- that justifies our expansion, we will continue to do so. And I think…

Sorry.

Obviously, the Chinese will do the same, but then given the scale and also the leadership of ours in all aspects, I think, we will continue to outcompete our peers wherever they are.

I recall previously the company had a growth target to grow about 2x of the large SMEs. Is that still a value target that we could continue to anticipate?

Yes. Well, of course, 2023 is a very different year because we are all going through a very challenging year. But as I mentioned, I think, going into 2024, things will start to pick up and will go to a much more healthier condition. And our target is to continue to grow twice the logic semi-growth.

Next question is from Liyen Chen of Citigroup.

My first question is actually about our IC ATM growth margin outlook. We know that this year we still see the inventory correction continue. But previously, we are looking for like mid 20% to 30% gross margin for IC ATM. So I'm just wondering that what kind of utilization rate we can reach that level? And also considering that we are also doing some advanced packaging for the AI chip. So I'm just wondering that for that kind of business, what the -- will that be a margin dilution for the overall IC ITM business?

Right now, the -- what we call the leading edge technology or leading edge packaging, I think the overall profitability as well as return, is actually above corporate average. And in terms of the structural margin that we mentioned before, I think it takes us to go to a utilization rate of around 70%. Anything below that will be very difficult for us to reach that structural margin.

And also on the EMS business, I think just following -- Randy previously also asked about the seasonality into Q3 seems to be milder than historical. So just wondering that is that just due to the muted demand? Or we see any significant delay for the new product launch in second half?

No, I think it's just overall the muted demand that's causing the weaker seasonality -- seasonal pickup for our EMS business. But typically, if you look at a whole year from a quarter-to-quarter, typically, the third quarter will be the peak quarter. And at this time around, I think -- not trying to be the spokesman for our customer, I think the market knowledge about some of the products seems to be launched later than normal, does have some impact in terms of some of the order being pushed back to fourth quarter. So I think this year will be a little bit different from -- on a quarter-to-quarter performance perspective.

Our next question is from Zheng Lu of Goldman Sachs.

Let's start with some easy ones. So I think 3 months ago, you talked about the ATM business kind of declined by high single-digit to low teens. As we get closer to the end, can you give us some outlook for the full year for the ATM business?

Yes. I think it's going to be low teens to mid-teens.

So based on this kind of low-teens to mid-teens, we should see what kind of profitability?

I think it will be a substructure margin.

Then I still go back to the advanced packaging business. I do acknowledge that you guys talked about a larger addressable market for the AI potential business. But which part of the AI packaging business will have faster growth for ASE? I mean for the real advance one, for the CoWoS one or for the wafer or chip? Which one will have some higher profitability? And in terms of like CoWoS or related packaging business, what's your return threshold and margin threshold for you to continue to invest a lot more -- invest more for this business?

We're not still trying to figure out what kind of return that we can get out of the full process of CoWoS. I think that's depending on a lot of different factors, most importantly, whether we have a stable volume that we're going through. We will have -- how wide the adoption will be. And I mean the customer will be using that, so we can diversify the risk.

And there are a lot of factors that have been considered. I don't think that it's now at a more as mature stage as we expect. So I think the -- how much or how soon we will start investing into full process of CoWoS, it really depends on the situation that we're looking at. Right now, we do have some CoWoS capacity, but that helps maintain our technology prior acquisition. But for the -- I think the real bread and butter for us is really once the AI starts to proliferate into other applications going into automotive, communication, computing, all sorts of new applications. And that's where we're going to see the mainstream packages being started to be required for these ships. And that's where the volume is. And I think that will be the main investment area that we'll be focusing on.

Next question is from Szeho Ng of China Renaissance.

2 questions. First one regarding this year's CapEx budget, can you provide us an update? And in what areas we are spending CapEx budgetary money?

I think the CapEx for equipment, our guidance, it remains the same as 3 months ago. In terms of its breakdown, I think roughly 55%, 56% will be for assembly and the bulk of it is for always packaging and roughly 25% for test, and then 15% or so for EMS, and then another 3% for material.

And then going forward in the next couple of years, should we expect the CapEx intensity to rise because of our more involvement in the advanced packaging area?

Yes. Really -- again, it really depends on where the volume is and where we should be investing. But all in all, I think given the fact that this year, we have been fairly cautious and conservative about our CapEx, and we're expecting growth to start resume next year. I believe the capital investment or CapEx for next year will be somewhat higher.

Next question is from Mr. Gokul Hariharan of JPMorgan.

My question is that we are seeing a lot of capacity expansion on the front end happening in China, especially for holder technologies. What do you envisage for ASE? Do you think that you will also -- you will kind of miss out on that potential packaging demand given you have lesser exposure in China today? And is that potentially a competitive threat down the line, given that we are seeing some of that competitive threat emerge for the foundry side at this point in time? How do you see this evolve? Because we had a view maybe about 1 year back that the market is kind of becoming a little bit more divergent between China and non-China. You still think that's the case? Or you think some of the -- given that there is a lot of capacity getting built in China, even some of the non-China fabless companies will eventually go back, find their way back to the China capacity to leverage the lower pricing?

Well, I think the demand is diverged at this point. And -- but that also creates a lot of requirements in China for localization. And so we are seeing that. And we do have our -- aside from selling our 4 factories 2 years ago, we still remain to have a very sizable operation in Suzhou, which we can accommodate the -- whatever the opportunity there is. Plus the Suzhou factory that we have is of more advanced technology and more higher yield and efficiency. So I think it's a very, very competitive operation of ours in China to address the local demand.

As we are also seeing that a lot of investment being put in China is for the mainstream technology type of products. And so we do feel that there is a fairly good potential in that area as well. So I don't know if you all are aware that we are investing into the combination of our -- I should have put this -- We are putting back some of our investment into the 4 factories that we sold. That 4 factories is now combining with another entity today.

So we are taking a two-pronged approach. One is using our own facility, leveraging our own facility to capture some of the opportunity for the more advanced type of demand and also some minority interest in the 4 factories that we sold and continue to benefit or to share some of the prospects in the mainstream or the legacy kind of demand in China, which has a fairly good potential at this point.

And just a further question on the AI-related demand. For the cohost like technologies, what portion of the demand does ASE expect to kind of fulfill next year? Do you expect to -- I think TSMC talked about expansion of their own capacity and they probably account for the vast majority of the capacity right now. But what percentage of like -- roughly how much of that demand does ASE expect to satisfy once you start bringing some of your capacity online next year?

I think -- by and large, I think most of the demand will still have to be satisfied by the foundry itself because this type of technology or process is mostly wafer process, and I think they should be taking the leading role on this. And -- but as we've been saying all along, as time goes by, as the technology becomes more mature, the adoption becomes wider and the multi-customer using that and volume becomes more stable and much bigger in size, there will be a natural division of works between us and the foundry. And without specifying what kind of share we will be having, I think that is the going trend. It really depends on how much demand there is and what's the suitable working model between us and the foundry to share the work and to satisfy the demand.

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

Can I know what's the CapEx for the second half? And what's the CapEx allocation for 2.5D packaging in the second half.

I think for second half, our overall CapEx will be in the range of TWD 580 million to TWD 600 million.

For 2 quarters?

For 2 quarters, yes -- No, for the second half.

And location?

We are not commenting on the 2.5D CapEx at this point.

So what's the threshold for you to invest more in the 2.5D?

As I said, the business has to justify the investment. Economic...

What is the threshold?

What is the threshold? Well, in terms of margin, in terms of the lifespan, in terms of the return, it has to be above the corporate average.

But it looks like the revenue is going all in a meaningful way, believe that or not, but that's the current outlook for all of the project.

Yes.

It seems to me that you are not aggressive in terms of investing in these 2 systems.

Like I said, we are not focusing and we should not be focusing on the these so-called AI chips as narrowly defined, which maybe includes CPU, GPU, or AI accelerators. I think that's really just the catalyst of our business expansion going forward. So that's really not the part that we should be focusing our investment in. Once it gets to proliferate into other applications then volume be created for the -- our mainstream packaging and test, that's really when we start to invest in a major way. So I don't think it's wise to just focusing on the -- this CoWoS or 2.5D specifically because that's -- we will be missing out the whole -- the full picture here.

And one more question is what's the -- what do you think about the system level testing business? What's the revenue contribution from SLP for you? And is that a growing business? Is that a profitability business? Is that a high ROE business for you in the future?

Well, we do have system level test, and it's making a reasonable return. Just like every other business, as we see, there is real business potential. There is enough return for us to make, and then we'll make the necessary investment.

(Operator Instructions) There is no more question.

Okay. Thank you, everybody. And hopefully we will clear some of the concerns or questions you may or may not have. And we -- I think we had a decent second quarter and we'll have a decent second half to look after as well. So thank you very much. We will be seeing you next quarter.