All lines have been placed on mute to prevent background noise. After the presentation, there will be a question and answer session. Please follow the instructions given at that time if you wish to ask the questions. For your information, a webcast replay will be available within an hour after the conference is finished. Please visit the website www.chroma.com.tw/investor/index under the Investor Relations section. I would now like to introduce our CFO, Paul Ying. Paul Ying, please begin.
Thanks, Frank. Hi, everyone. This is Paul Ying, the CFO of Chroma. Welcome to the 2023 Q2 earnings conference. Today, it will be myself and Jennifer to give you a presentation regarding the Q2 earnings and the forecast for the H2. I think we have already put the presentation material on the website. You already get that. In the slide number five, you can see the condensed, the consolidated income statement for the Q2 . You can see from here, the net sales for the Q2 will be NT$ 4.41 billion.
Compared to the Q1 , NT$ 4.39 billion, it's a bit of a flat, which is only 1% growth. Compared to last year, NT$ 4.93 billion, it's 11% drop. Well, in there, you still can see that for the sales of the testing equipment, which is occupied like 95% of our total sales for the Q2 , is nearly NT$ 4.2 billion. Compared to the last quarter, it's a 4% growth, compared to last year, Q2 , it's a 10% drop. In there, you also can see that the gross margin for the Q2 reaches at somewhere like NT$ 2.75 billion.
Compared to the Q1 , it's a 3% growth. Compared to the last year, Q2 , it's a 5% drop, mainly due to the new material business was terminated at the Q1 of last year. I think in there, you will also will see the OPEX is pretty much consistent with the the top line and approximately 6% growth on a Q over Q base, and only one flat to the last year for the Q2 . For the operating income, reaches at the NT$1.2 billion compared to the last quarter, it's a 2% growth.
Compared to the last year, it's an 11% drop. For the non-operating part, the main difference for the Q2 compared to the last quarter, it's a 144% growth, but compared to last year, it's a 79% drop due to the disposal of the investment, approximately NT$ 500 million. That's a major difference from this quarter compared to last year. The net income we make at a little bit over NT$ 1 billion. Compared to the Q1 of this year, it's a 9% growth.
Compared to the last year, it's a 37% drop due to the differences of the last year, the disposal of investment, a little bit over NT$ 500 million. That's the main reason. For the earnings per share for the Q2 is approximately NT$ 2.42. Compared to the Q1 , it's a 9% growth, but compared to last year, it's a 39% drop, again, due to the disposal of the investment. For the H1 of 2023 compared to last year, I think we can, you can refer to the slide 6, which is the H1 of this year, which is at the NT$ 8.8 billion.
Compared to last year, NT$ 9.3 billion, it's a 6% drop. With the total amount drop, but you still can see that the consolidated sales of the testing equipment for this core business, it reaches at the NT$ 8.24 billion, occupy like 94% of our total amount. Compared to last year, H1 , it's a 2% growth. For the margin, again, the differences of the H1 of 2023 is NT$ 5.4 billion. Compared to the a little bit over NT$ 5 billion of last year, it's a 6% growth.
Again, for the OPEX, for the H1 of this year compared to last year, it's a little bit over, like 5% to 3%. The operating income, we're reaching NT$ 2.4 billion compared to NT$ 2.2 billion. It's a 9% growth. Again, for the non-operating items, the differences will be mainly from the disposal of the investment, makes a little bit over NT$ 500 million of last year. This year shows a bit of a drop, like 76%. Again, the net income for the H1 will be a little bit over NT$ 2 billion compared to last year, NT$ 2.54 billion.
It's a 21% drop. If we skip the, we take the disposal of the investment, well, barely, I think the, the, the achievement of the earnings is pretty much similar. For the H1 , the earnings per share for 2023 H1 will be NT$4.65 compared to last year, a little bit under NT$6. It's a 22% drop. This is the financial statement for the H1 and also for the Q2 . As to the balance sheet, you, you also can re- refer to the slide number 7. Here, you can see the, the, the level of the cash and the short-term investment a little bit drop.
Again, at the same time, the short term, the short-term debt is also being paid off. Here, again, inventory is still growing. Well, well, a lot of other reasons, but that also should prepare for the future shipment. For the inventory turnover, a little bit over our expectation, which is the 266 days compared to last year, a little bit under 150 days. I think it's beyond our expectations, we expect it will be dropped back to somewhere like, you know, two times every year. We are expecting it returns to the 180 days to 200 days, somewhere over there.
For the return on equity, it's a double digit, 19% for the, for this, this H1 of 2023. The return on assets is also 12%. As to the for free cash flow, I think it's a little bit low, I think due to the CapEx we're spending for the for our expansion for our second phase of our headquarters. I think this is the reason for that. This is all the financial numbers. I would, well, pass that to Jennifer to give you the the H1 , Q2 and H1 product mix and sales breakdown. Jennifer?
Okay. Good afternoon, everyone. I will give a summary for our H1 , the product mix, and also give some color regarding to the H2. As you may see from the product mix, the H1 compared to last year, I think the biggest drivers come from the test instrument and ATS. The biggest driver behind that is mainly come from the EV drive. H1 , we actually grew like 46% year-on-year for this test instrument, ATS. Last year, I think the biggest driver is actually come from semiconductor, especially for SLT, started to deliver from Q2 . H1 this year, actually, semiconductor and photonics was the biggest drop, around 56% compared to last year, first six months.
Turnkey solution, because based on project basis, it would be better to look at the whole years. I think the whole year is supposed to be, we're able to reach somewhere around last year or even better, and we can go down to more details later. Regarding to the overseas for from overseas operations, H1 , we generate about NT$ 1.9 billion, around NT$ 1.9 billion. We only booked around NT$ 100 million to around NT$ 150 million for EV battery cell. Most of the China EV battery cell orders will start to deliver in the H2. Looking forward for the H2 this year, apparently the H1 was the biggest drivers come from the EV industries. The H2, I think the situation will start to change.
Due to China's EV consolidation factors, we do see the slowing down from this EV drive contributes to this test instrument, the ATS. However, I think this factor is still able to reach about double-digit growth year on year comparisons. The driver for the H2 will be definitely come from the semiconductor and photonics factors, particularly come from first one, is AI-related chips and HPC market demand for SLT testers. Plus, we do receive some rush order for this optical fiber communications. We think these factors overall for this year, you could see quite a decent growth, much better compared to H1 , okay?
Because overall for this year, we still believe that the H2 will be better than H1 , but the whole year status is probably flat or slightly declined. Okay. Gross margin overall, I think we still just do our best to maintain the gross margin level around ±55% . I think it will be swing somewhere around this kind of a level. Okay. Now we open for Q&A.
Thank you. We'll now begin our question and answer section. If you have questions for any of today's speakers, please press star one on your telephone keypad, and you will enter the queue. After you are announced, please ask your questions. If you find that your question has been answered before it's your turn, please press star two to cancel the questions. Once again, if you wish to ask a question, please press star one on the telephone keypad. Our first question is come from Kevin Chang with CTN. Kevin, please go ahead.
Hi, thank you. Thank you, Jennifer, for taking my questions. Congratulations on the nice results. I have two questions here. First of all, a little bit on the margin side. I, I see that in the Q2 , gross margin, we have improved quite significantly quarter-on-quarter. I was just wondering, what's the main reason behind this strength? Was this the product mix or there's some other specific reasons? Because I recall earlier this year, we're expecting the full year gross margin to be around 55%. Just wondering, has something fundamentally has changed? Thanks.
Gross margin.
Gross margin, we still think that we still consider 65% will be somewhere that we are pretty much satisfied. Over that will be pretty much like a bonus. Again, in consolidated gross margin, we used to enjoy pretty much like 40-some% up to 50%. Well, mainly due to that, the dilution from the material business. Starting from last year, Q1 , I think we don't have this kind of a dilution anymore. We terminated the material business. Moving forward, I think, again, we still pretty much like focus on, you know, 55% as our expectations, and over that will be somewhere like, you know, overall satisfaction. Thank you.
Right. Thank you, Paul. I think next question will be coming back from the AI-related prospects. I, I think, I'm quite curious on, on what the management thinking about AI-related revenue, especially around the SLT growth in the H2 of this year. From my understanding that the H1 was probably turned out to be a little lower than previously expected. I was just wondering, what do we think about this growth in the H2 of this year and possibly next year as well? You know, the TSMC also mentioned that we're probably they're expanding their CoWoS by 2 times next year.
Is this something that's being more accelerated than previously expected or, or something that's still somewhat within our expectations? Just wondering how our management is thinking about this part of revenue growth going forward. Thanks.
Okay. Hi, Kevin, this is Jennifer. Yes, we do see, yeah, I would say quite significant pick up the order, especially from the H2. As you know, I think previously because some of the output is pretty much stuck at the foundry site, and I think customers is pretty much ready to ramp out, I mean, for their new product, which they planned for the early next year. What happened in the Q2 is, I think we, we ever mentioned in the beginning of the year, they already placed the order, even with the down paid, and they decide to maybe half the order, they decide to upgrade. Because it no longer meet their product roadmap for the coming, I would say, next one or two years.
You could say half the orders, which is the original place, has been pushed out to the H2 because they want to do further upgrades. That is, we need to charge for their further upgrades. They also placed another order for building out the new capacity for their new products. The current market that you have seen for those AI chips is not, I would say, currently our customers are planning to build out the capacity for their new products and not really intent on their existing products. Couple of the new customers, someone who planning to for the ASIC, I think they already start to buy testers....
The biggest, biggest capacity ramp supposed to be next, the planning is 2024 and 2025, since they already gave us their product roadmap and the capacity plans for next two years. If we lump sum everything together, as you know, just like we mentioned before, this year, as you can see, interest rate continue to increase, the CapEx become more expensive. In this state, people mostly spending their CapEx either on EV related or AI related, and these two, two industries, normally quite high. I mean, it's quite expensive. Yeah. Other sector, we haven't seen any sign of a recovery. I mean, especially for, if you just talk about additional CapEx or additional demand.
All right. Thank you, Jennifer. Just one quick follow-up. When you're talking about, half of the order, being upgraded, was this?
Yeah.
upgraded to for, the higher specs or higher throughput? If so, is that is that going to be how much up ASP upside would that be?
We actually, yeah, since our chairman mentioned about this, I just recap a bit. We actually, regarding to this further upgrade, we actually raised out around 70% of pricing, yeah, for the higher. It's not, cannot be a throughput because in current markets have two parties. One party mostly focused on the big dies designs, and the other party mostly focused on chiplet. Either side, they couldn't enjoy due to their design and the die size. It is very difficult to come up with, like, a huge throughput. Normally, even our highest throughput, the most higher levels of high class of the SLT, currently we're able to process the one, the throughput is 20 dies, ±. You need to use, like, four times, like, either 16 or 24. It is not really like.
This is why we find very tricky from some other analysts, how, how come they calculate such a big amount of AI chips? So far, we haven't seen this kind, that kind of throughput at the upstream.
All right. Thank you, Jennifer. It's very clear. I'll get back to you. Thank you.
Thank you. Our next question is come from Jerry Su with Credit Suisse. Jerry, please go ahead.
Thanks for the question. I just want to go back, you know, circle back, back to Jennifer's previous comment about semiconductor that gonna have to be better than H1 . How should we think about full year basis, how will semiconductor and photonic segment, you know, perform, compared with last year?
Well, we can't give a guide, numbers guidance, but we can give you some color. Well, to do better than last year probably will be very challenging. To do better, similar as last year, this is something we can work on. If you look at our semiconductor sales, we only, we only book NT$ 1 billion, and then you can think about how much we need to book in the H2, and that probably equivalent to how much order we have been received. As a usual, we normally comment the order we receive. We don't give a, like a, you know, some kind of a dream or covers, you know, for unreachable stuff. Yeah.
Okay. probably just, are we seeing like, you know, like, 10% different compared with last year, that's something we can, target for?
I don't comment on numbers or full count numbers.
Okay. All right, thank you. Then as, our full year basis, you mentioned that the revenue will be of around flattish to slight decline. I think I'm looking at those numbers, it, it, it, it seems like that the testing instrument, the ATS, this should still be one of the... Although the growth will be a little bit accelerating, but I think H2 should still be similar or compared to H1 . Should we think it that way?
No, I think H2, I mean, the sales for testing and measurement, I mean, the sales for test measurement in the H2 will be lower than H1 . The overall this year, we're still able to achieve double-digit growth, but not just.
Okay. Okay.
Double like the H1 .
Okay. Got it.
Our biggest concern-
Thank you.
Just like we guide in the beginning of the years, our biggest constraint comes from the China semiconductors, which is a legacy product. Just like we guide, in the H2, semiconductor sectors, biggest sales contributions, mainly number 1, come from SLT. Second, large order for the photonic sectors, optical fiber communication.
Okay. Okay, thank you. Those, that's all my question.
Yeah.
Thank you. Thank you. Our next question is come from Jeff Orla with Macquarie. Jeff, please go ahead.
Yeah, great. Thank you, Paul and Jennifer. First question, you mentioned rush orders from fiber optics. Is that U.S. or China?
Can I not comment on client faces? That is. Yeah, because actually Q2 , we already start to, actually, Q2 , biggest contributions come from optical fiber communications, I mean, for the semi side. We also received the rush order for Q3 . Yeah.
How about if you can't comment on customer, how about just hemisphere, Eastern Hemisphere or Western Hemisphere?
Yeah, sorry.
Okay, never mind. I'll go, I'll go on next. Long-term growth, you know, if you're looking at next year, 2025, do you still think-
Yeah.
ATS and EV downstream package-related testing can still see very strong growth? Do you think we've kind of peaked short term? Kind of the same question for semis. I mean, what do, what do you think, looking at the next couple of years, with- without talking about numbers, you know, what's going to be the fastest growth area, do you think? And where are we going to see maybe some negative growth?
Your first one is asking about ATS, right?
Yes. Because it's been so strong over the last, last, you know, I'd say, say four quarters at least, it's been very strong. You say, you know-
Yeah.
down half and half in the H2. Is, do you think we've peaked, or do you think it's just a little?
Yeah. This year mostly come from EV. I don't think H2 will have. We, we, we only could hope for next year, we'll have some recovery from consumer electronic goods. We didn't see any kind of CapEx recovery come from like a traditional sector, like consumer electronic goods, et cetera.
Mm-hmm.
Yes.
Okay.
for semi-
The second one.
Yeah, I think semi, because you need to take a long-term view. Yes, 2024, 2025, you know, our customers, they do have a schedule of capacity build plans, especially for those ASIC customers. We are currently bringing some other HPC makers, you know, we are almost to finalize couple new or couple new customers.
If you look at just this year for SLT, is your larger customer still a very large % of the SLT sales, or is it getting more diversified?
Start to diversify. I have to say, okay, we are not serving one customers, but just that customer has been worked with us for 10 years.
Mm-hmm.
We actually cover most of the ASIC and the HPC makers. More or less, depends on which product they default. Because some customers, like Qualcomm, Intel, they probably only focus on auto. They just-
Mm-hmm.
For today, they probably mostly focus on buying thermal control. They probably don't looking for some, like, a full testing function, like, our series of 3200, something like that. It really depends on the customers' data plans. Yeah, we actually cover most of the HPC makers.
Okay. All right, maybe one last question. I know, Paul, you, you're getting a lot of questions on gross profit margin, which, you know, from our perspective, is very hard to, to forecast, given, you know, you're over 60% in the H1 . You're guiding for 55%, but you're certainly tracking well above that number, so you'd have to have a very low number in the H2 to, you know, get near your 55. It sounds like you're being very conservative and, and still trying to understand better. You know, if you look at the parent company gross margin over the last five years, if you, just don't even include NMC, you know, you're, you're tracking well above that. I'm just trying to understand, is, is...
I mean, 60%, just a, a couple quarter anomaly, or is that possible to sustain or, you know, or at least well over 55? That sounds like a conservative number.
Well, agree and disagree. Basically, I think for the parent company, we always show you the expectation for, for, for the gross margin, somewhere like 55. For the consolidated, again, due to the product mix, we're still expecting that, you know, 55% as an expectation. Over that will be pretty much, you know, beyond the satisfactory. Again, it's it's on customer's call on, on the combination or the product mix. I'm not, you know, telling you that the for the first two quarters, it's somewhere over 60, and it definitely will be over 60 for the, for, for the next few quarters. I think we're still expecting that somewhere like 55, and that's somewhere that, yeah, we expect it.
Okay, great. Thank you.
Thank you.
Thank you. Our next question is come from Faisal Chang with Newton, and please go ahead.
Thank you very much for taking my question. I'm just wondering if you can please introduce a bit more about your competitive advantage in an SLT product. I'm just wondering if you can maybe share some data point on it as well. Like, say, for example, what was your share of SLT testing your biggest customer? Also, I know other, your bigger competitors like Advantest, and they are ramping up their SLT testing capability as well. Just wondering how, how confident are you in kind of taking majority of the growth going forward? Thank you.
We couldn't do this kind of comparison because both of us were Chroma to other peers. We actually launched a different stack, our strength is mostly focused on thermal control. If you look in the market today, well, if you look at SLT markets, you need to honestly, SLT mostly benefited by 2 things. One is advanced packaging, the other one thing is power density continue to increase. I think last year, HPC average power density already, I think average about 800 watts to 1,000 watts. According to, as I just mentioned, our new older customers are planning to escalate more than 1,000 watts. Based on these patterns, we're able to, capable to manage the power and also do the thermal control. Currently in the markets, we couldn't see any second party.
We couldn't do this kind of comparison because both. I think each of us actually focus on different section of testing. We only could comment only AI-related or HPC-related need to do this SLT process, because the SLT process is final, final testing, actually using advanced test testers for signal testing, and then after that, you do this kind of so-called physical testing. Regarding to thermal, actually, we re-registered several patents, and even next year, we're going to launch a new, new design, especially, you know, further improve our thermal control on a socket basis. This is we just, you could say, very exclusive, and then currently, customers already decide to adopt. We do have a patent couple years ago. It's very hard for us to say.
We, we won't position ourselves as sole vendor, but based on this kind of technology, this is like why most of the customers pick our solutions.
Okay, thank you.
Thank you. As a reminder, please press star one on your telephone keypad for ask questions. Just a reminder again, if you wish to ask a question, please press star one on your telephone keypad. Our next question is come from Vera Liu with HSBC, please go ahead.
Hey, Jennifer, I have two questions. My first one is just regarding your competitive edge in SLT. Can I clarify that your competitive edge is really regarding SLT? Because you customize on the SLT set, you are able to provide better test coverage compared to larger peers like Advantest. That's my first question. Thank you.
Okay. Our position is not for SLT as a whole market. I think our position is SLT for AI-related and HPC use. We don't cover SLT, like, for example, handler for front, for signal testing, like a different test tester, because even you do signal testing, you still need to have a handler. Can you say handler is not part of SLT? I think they're part of SLT, because SLT need to have a handler. We don't cover SLT for CPU. Memory, you don't need to have a SLT. The better described our position is, we only, we mostly cover the SLT for AI-related and HPC.
Got it. My second question is just regarding our battery cell formation project.
Yes.
Can I get some color just in terms of the delay, in terms of the Q2 ramp? What was the reason from your Chinese customers? Sorry, your European customers via China. Thank you.
Oh, this is based on the contract. They, they pay us a down payment, and the payment schedule is based on the contract. We couldn't convert those sales revenue, you know, which is, you know, against the contract terms. Mostly booked in the H2.
Thank you.
Thank you. Our next question is come from Aaron Wu with Point72. Aaron, please go ahead.
Hello. Thanks for taking my question. My first question is that for SLT, because this is the very final, final product testing, I just wonder if this kind of ASIC or GPU become more complex, do you think we will need more SLT going forward? I don't think, I was just wondering, can we improve the SLT throughput? If layer has more transistors, I don't think the ATE can cover all the testing. I just wonder what kind of a competition between the SLT and ATE going forward. Thank you so much.
I think the key point is not 100% come from, okay, the throughput is determined by, number one, the customer's die size, and second, the power density. Because the power density actually is quite a critical point that related to thermal control design. Thermal, you know, it's quite a physical features. You can't use software to simulate this kind of testing. Yeah.
Yeah, if this kind of GPU, design architecture is becoming more complex, it will require-
Mm-hmm.
some more high voltage and I would say, higher temperature for thermal testing. Can we increase the throughput for the clients, or the clients need to upgrade or buy more SLT for the testing? Thank you so much.
Okay. According to the current customers, their next, well, maybe not say in five years, but next, say, three years, for their roadmap, and according to their designs and the power density, I don't think they could increase the throughput. Because, as I say, one party is mostly focused on the big die models, and this kind of die size will continue to increase, according to current plan.
Got it. One more follow-up is that, in addition to ship the new equipment to the clients.
Mm.
The client needs to upgrade their current installed equipment for the, the next generation or the further, testing requirement going forward?
Sorry, I don't quite get this question.
In addition to ship the whole new equipment, testing.
Mm.
clients, can client just upgrade, or can client just upgrade, or they need to upgrade, their already installed equipment for the further, because, for the next generation, GPU testing?
Okay, I know your, I know your point. Okay. The answer is no, because that is, yeah, probably our customers have no plan to add the current model of yeah, you, the way market normally say A and H. Because the reason why they do, need to do these kind of upgrades, because they decide to move on to new products. If they want to do new products, then they need to go, go for our higher throughput, which is 3200, this model. Because the 20, I would say between 16 die to 24 die, this is the maximized throughput. Current backend process, they, most of their tester is using legacy models, and this is not, unable to upgrade and not meet the customers' of new products need.
Got it. Thank you so much. My next question-
Yeah.
is regarding the CoWoS, because I, I remember-
Mm-hmm.
You, you management team mentioned this kind of CoWoS installation system. Could you just give us some color, what kind of product you provide for this CoWoS technology? Thank you so much.
Yeah, our chairman particularly highlights. I think it's on the slides. He mentioned about on the slide 20. I think there's a serial system called 79A0. This is mostly to test TSV and RDL, this slide. I think he highlights this one. I think the other model, I think the other model is only is for this kind of stream crystal stuff.
Crystal.
Yeah. Mm. It's mostly for... Yeah, mostly for RDL process. Yeah. I think we start to ship, I think this quarter. Yeah.
Because TSMC mentioned that this kind of AI demand will like, have, like, 50% CAGR, and they need to increase a lot of capacity in terms of CoWoS. I just wondering, how the company calculate or estimate this kind of CoWoS demand or contribution to the firm? Thank you so much.
I probably couldn't comment how many testers that customers place to us. As I, as you can see, even they did half of the first order, upgrade is still not enough. They need to place another order. We can't interpretate the upstream, which is the foundry process, their capacity, but we can tell you that, most of the new capacity or the orders that customer place to us, is for their new product early next year. Those upgrade orders and the new orders have the, we have to deliver 100% by end of this year. Yeah, sure. It will be all contributing in the H2. I don't know how this equivalent to the foundry capacity. Yeah.
Got it. Thank you so much. I'll back to you. Thank you.
Yeah.
Thank you. As a reminder, please press star one on your telephone keypad, if you wish to ask a question. Once again, if you wish to ask a question, please press star one on your telephone keypad. There are currently no questions. Thank you.
Okay. Thank you, Frank. Well, thanks, everyone. I think this is the end of the first Q2 earnings conference. Moving on to the Q3 and the Q4 of the H2 of this year, we're making more efforts in trying to make our commitment all happen. Thank you so much. Until next time, bye-bye.
Thank you. Thank you for your participation in Chroma's conference call. There will be a web class replay within an hour. Please visit our website, www.chroma.com.tw.