6857.T FY2026 Q2 Earnings Call Transcript
Date: 2026-10-28
Source: Financial Modeling Prep

Speaker 0: We'd like to have the first question from Mr. Nakamura of Goldman Sachs Japan.
Shuhei Nakamura: Thank you very much for taking my question, and very strong results, so hats off to the management team. My question is related to SoC testers. You raised our SoC tester revenue outlook quite substantially compared to your July forecast. Just wanted to understand what's driving this upward revision for SoC. Is it mainly come from GPU, ASIC, or CPU? And also, you mentioned that you are expanding your production capacity to 10,000 units at an accelerated pace compared to your original plan. Just wanted to understand, when exactly do you expect to reach this 10,000 units of SoC tester capacity?
Doug Lefever: Thank you, Nakamura-san. Thanks for those questions. Oh, sorry. First--
Speaker 3: [Non-English content]
Doug Lefever: Okay, now I'll go. Sorry for jumping. To answer the first question, Nakamura-san, really across the entire device segment within the compute market, we're seeing elevated demand. For GPU and then for custom ASIC, definitely, those are still running very hot. But then probably the biggest factor is the CPU business. I think the industry did not foresee the amount of CPU demand that would go along with agentic AI. And so that is really a third layer now on top of GPU and custom ASIC. On the production capacity side, we originally said by the end of fiscal 2028 for our 10,000 systems. Not really prepared to give an exact date for the 10,000 systems, but it has to be accelerated pretty significantly given the demand, and we have high confidence that we'll be able to do that. I think we'll be able to pinpoint that time frame probably in the next quarter or so.
Speaker 3: [Non-English content]
Shuhei Nakamura: Thank you very much. Just for a quick follow-up, you mentioned CPU as a third pillar of strong server demand. Is it mainly driven by the CPU volume that we're seeing, or is it longer test time? I think you used to mention that CPU test times are not as long as for GPU or ASICs, but just wanted to see if you're seeing longer test times for CPU as well. Thank you.
Speaker 3: [Non-English content]
Sanjeev Mohan: Nakamura-san, this is Sanjeev Mohan. You're correct. The test times for CPU are not as large as GPU or custom ASICs in some cases, but they're also not small either. They're actually pretty significant. But as Doug mentioned, the volumes of the server devices is quite large, driven by agentic AI. Also for inferencing, the ratio of CPUs to GPUs have gone up as well. Combining all those factors, the CPU is providing quite a boost to the ATE results.
Speaker 3: [Non-English content]
Shuhei Nakamura: [Non-English content] Thank you very much.
Mr: [Non-English content]
Speaker 0: Thank you very much, Mr. Nakamura, for your question.
Mr: [Non-English content]
Speaker 0: Next, we would like to ask Mr. Shimamoto of Okasan Securities to ask a question.
Takashi Shimamoto: [Non-English content]
Speaker 0: Thank you very much indeed. My question is concerning your prospect for the tester market TAM for year 2027. You mentioned that the upcoming 2027 environment is very favorable, but what are the drivers to actually lead this kind of growth further? And is it possible for you to actually share with us some quantitative growth rate if you know some and share with us?
Doug Lefever: Thank you, Shimamoto-san. We won't quantify completely until January in our third quarter call as we normally do. But I do want to give you a little bit of information for your question. The last couple of years, the ATE TAM CAGR has been about 50% or a little above 50%. We don't see that slowing down. And it's going to be led by the same factors — GPU, custom ASIC, and CPU. We see the similar drivers as this year, just with continuing growth. We also can look at some of the statements from our customers about their expectation for year-over-year, both with the fabless customers we have and also the foundry partners who are looking at the CAGRs of mid to high 50s all the way through 2029. I think we feel quite confident that the TAM will be in line with the previous years and potentially even larger.
Speaker 3: [Non-English content]
Takashi Shimamoto: [Non-English content]
Speaker 0: Thank you very much. Over to you.
Mr: [Non-English content]
Speaker 0: Shimamoto-san, thank you very much indeed.
Mr: [Non-English content]
Speaker 0: Next, we would like to ask Mr. Wadaki from SBI Securities Company to ask a question.
Tetsuya Wadaki: [Non-English content]
Mr: [Non-English content]
Tetsuya Wadaki: [Non-English content]
Speaker 0: Thank you very much. I would like to ask you about the growth, the possible growth of tester market. You mentioned that the kind of prospect is visible for next two years or so. But my hunch tells me that the market is going to continue to grow for another four or five years. There's some reason to my hunch. That is the CPO development. Two days ago, there were articles about the CPO, and CPO requires quite long testing times as equivalent to that of a chiplet. Otherwise, there's the risk of the product may be discarded. The CPO development is one reason. The other reason is that Terafab, the test market. Because of those two reasons, I would think that the tester market will grow for another four or five years. What is your perspectives?
Doug Lefever: Thank you, Wadaki-san. On the question of CPO, yeah, you're right. The test times are quite lengthy in order to guarantee good yields for the optical engines. That does give quite a bit of a tailwind for that part of the business. We've said in the past that we see a double or tripling every year for the next couple of years, getting to somewhere in the billion-dollar ATE market size, which is great. It's great growth, and we're excited about that, and we're participating in all the insertions. But still, compared to the overall SoC market, that's above $11 billion and growing also totally at a 50%+ CAGR. It's still going to be relatively small compared to the rest of the market. But I believe it'll be something that will be a bigger part of our business in the foreseeable future. I happen to agree with your views also on the four to five years. We're not making any kind of disclosures about our view. We will be releasing our Mid-Term Plan 4 in the May timeframe, and that will give some visibility out into 2028 and through 2029. But I actually share some of your conviction about the upside. Maybe I'll pause there and let the translator--
Speaker 3: [Non-English content]
Doug Lefever: Okay. Onto the question about Terafab. Obviously, we can't make any detailed comments about customers, and they fall into that category. But the one thing I will say is that it's very real, and I think it will lead to meaningful business in the future. I think there's a genuine need for additional capacity in the world, both SoC and memory. We're excited about the potential prospects of that business for Advantest.
Speaker 3: [Non-English content]
Tetsuya Wadaki: [Non-English content]
Speaker 0: Thank you very much.
Mr: [Non-English content]
Speaker 0: Next, we'd like to ask Mr. Yoshida from CLSA Securities Japan. Please have your question.
Yu Yoshida: Thank you for taking my question. I would like to ask about your market share outlook. Three months ago, you explained that your SoC tester market share will be going up to 70% or more in future. Do you have a sense that you can reach the level of this market share in the next year?
Speaker 3: [Non-English content]
Doug Lefever: Okay. Yeah, thank you, Yoshida-san. I mentioned last quarter that we do see a path above 70%. I didn't give a timeframe, but I do stand by that remark. I think that we're seeing continued growth, not just in our current customers' expansion, but we're also picking up some market share now. We went from 56% SoC in 2024 to 66% in 2025. Based on some of the recent results, we have growing confidence that we will be able to meet that 70% threshold and maybe even beyond in the near future.
Speaker 3: [Non-English content]
Yu Yoshida: Thank you. Please let me clarify. I think this year market share can be more like a flattish but in a gap, so like towards next year. Is that correct understanding? And also, if that is correct, what would be the driver for next year? Would there be a favorable customer mix or application mix? Or are you regaining the market share from your competitor in some specific customers? Thank you.
Speaker 3: [Non-English content]
Doug Lefever: I think we do expect to gain share in 2026, so we don't see that as being a flat year-over-year trend. And then, I go as far as to say that in 2027, we'll continue that share gain. It's both that the customers where we have incumbency are continuing to expand at a high rate, and we're also in a position now to pick up meaningful market share in accounts where we weren't necessarily the incumbent. The industry now, particularly for high performance compute, is largely considering the 93,000 as an industry standard. We think that those share gains are going to continue to be meaningful year-over-year.
Speaker 3: [Non-English content]
Yu Yoshida: Thank you so much.
Speaker 3: [Non-English content]
Mr: [Non-English content]
Speaker 0: Let me add my remark as well. The market share of the 70% is not what exactly declaring it. Whenever we make that the official statement that it will be taking place in the end of fiscal 2026 and the beginning of the fiscal 2027, in other words, April.
Mr: [Non-English content]
Speaker 0: Mr. Yoshida, thank you very much indeed for your question.
Mr: [Non-English content]
Speaker 0: I'd like to ask Ms. Shikanai of JPMorgan Securities Japan to ask your question.
Mio Shikanai: Thank you for taking the question. I'm Mio Shikanai from JPMorgan. First of all, congratulations on another strong quarter. I wanted to ask you a question about your earnings upgrade. While you maintain your CY 2026 SoC tester market outlook unchanged, you've raised your full year guidance. Should we assume that this revision is primarily driven by stronger fourth quarter expectation rather than the third quarter? And what has driven the increase in your own sales outlook despite unchanged market forecast? And if you could add a little bit about the pricing. If I remember correctly, last quarter you mentioned about the DRAM pass through to the customers. If there's any updates there, it will be great to hear. Thank you.
Speaker 3: [Non-English content]
Sanjeev Mohan: Yeah, this is Sanjeev. Thank you, Shikanai-san. For Q3, we already showed higher revenue than Q2. Sorry, Q2 than Q1. And as Doug mentioned, we are continuing to expand capacity, and that's work in progress. We expect to ship more systems in Q3, and then we expect to ship even higher systems than in Q4. We definitely see that the trend will continue, in terms of increasing our revenue for the second half for SoC systems. For the memory systems, we also believe that Q4 for us will be quite strong. And all of that, as mentioned, is a result of expansion in capacity and ability to ship more systems as we move forward in time. As far as the pricing is concerned, we're maintaining pretty healthy margins already, and we're able to absorb the cost of increase in components and other things that we need. We don't believe there will be a material impact. And we will maintain competitive pricing as we move forward.
Speaker 3: [Non-English content]
Mio Shikanai: Thank you very much. Just to clarify, so you revised up both third quarter and fourth quarter, and the main reason behind that was because your capacity expansion is going faster than you have initially expected for this fiscal year.
Doug Lefever: Yeah, we can say yes to that follow-up.
Speaker 3: [Non-English content]
Doug Lefever: So you know the answer. Yeah, the assumption is true. Really, we're very proud that from the first half to the second half top line, when you adjust for the currency exchange, basically 28% growth, half-over-half, which is pretty significant and speaks to the ability for us to expand capacity and meet those demand opportunities.
Speaker 3: [Non-English content]
Mio Shikanai: Very clear. Thank you very much.
Speaker 3: [Non-English content]
Mr: [Non-English content]
Speaker 0: Ms. Shikanai, thank you very much.
Mr: [Non-English content]
Speaker 0: Mr. Alex Chang from BNP Paribas Securities, please have your question.
Alex Chang: Thanks for taking my question. Congrats for the great results. I have two quick follow-ups. The first one is that in terms of the gross margin, it seems like this quarter gross margin is a new high. Remember management comment that the memory price inflation would have some meaningful impact on the margin, but it seems like the impact is quite limited. Can we confirm that from this level, the memory price inflation impact has been completely absorbed by the company, and company didn't raise the price or pass this cost inflation to customers, and the company is comfortable with the current memory price, or if there is any potential impact if the further increase of the memory price? A second follow-up is that, could you give me some color on your market share difference between x86-based CPU versus Arm-based CPU? If Arm-based CPU share is going to increase in future, would that mean any negative impact to you or positive impact? Thank you.
Speaker 3: [Non-English content]
Doug Lefever: Okay. Thank you, Alex. This is Doug. Let's start with the margin question. Definitely, we felt the pain of DDR pricing. We consume a lot of DDR in both our SoC and memory testers. A good part of the first half or second half operating margin decrease has to do with that step up in pricing. In the first half, we enjoyed the old DDR pricing with our inventory, but now as we look into the second half and into the future, we'll have to be contending with these higher spot prices just like everybody is. We are definitely pursuing ways to counteract those margin pressures. We constantly have the ability with new configurations and licensing and new instrumentation to reprice our products. That's a big part of our strategy without just doing a direct price increase, which we don't like to do. We will continue to have those kind of headwinds with DRAM prices and also other component inflation. But still, we're operating at such a high level.
Speaker 3: [Non-English content]
Doug Lefever: Second question is about the CPU landscape. I answered this last quarter, and I'll say the same thing again, which is we're very well positioned across the entire CPU market on both the x86 accounts as well as the Arm-based companies. We feel very good about our position with the 93,000 across the board. For us, who wins and who gains market share between x86 or between the players and on the Arm side, we don't really care so much because we have the same tester positioned at all of those accounts. That's one of the reasons you're seeing this second half top line growth is because we're so well positioned in the CPU market.
Speaker 3: [Non-English content]
Alex Chang: Can I have a quick follow-up question?
Speaker 3: [Non-English content]
Alex Chang: Yeah, sure. Comparing your revenue growth versus your peers, it seems like for this quarter and last quarter, you outperformed a lot, and you mentioned that the CPU was the main driver or main contributor. Can we confirm that your market share in CPU tester could be even higher than your market share in the AI accelerator tester? Is that the key reason?
Speaker 3: [Non-English content]
Sanjeev Mohan: It's a good question. I think we don't like to break it down like that as you're asking. I think what's important to note is that in the entire high-performance compute market, we have a very high market share. Whether it's GPU, custom ASICs, CPU, we probably have a majority of the market share in all of these spaces. In addition to that, there's networking and a few other things. Silicon Photonics was brought up earlier, where our market share is also very strong and growing. So it's a combination of all of the above. Situation really has not changed in many years. We continue to be a dominant player for the accelerators. We definitely are the dominant player in the custom ASICs, and that goes for CPU as well.
Speaker 3: [Non-English content]
Mr: [Non-English content]
Speaker 0: Thank you very much, Mr. Chang.
Mr: [Non-English content]
Speaker 0: Over to you, Mr. Nakanomyo of Jefferies Securities, please have your question.
Masahiro Nakanomyo: [Non-English content]
Mr: [Non-English content]
Speaker 0: Yes, we can hear you all right.
Masahiro Nakanomyo: [Non-English content]
Speaker 0: Thank you very much. Now, I don't know whether it is appropriate to ask you this question. However, being an analyst, very active in the stock market, we actually look at the various data, especially we resort to the Taiwanese SPE import data. In the past, there was some kind of very good correlation between those data and SoC testers. But this second quarter, the situation is slightly different and there are significant gap. There's no correlation. And I wonder why. Is this something to do with SoC logistic issues? Or maybe that's the way that statistics data are taken, or maybe that the equation or formulas have been changed. Please explain why there's this gap in the second quarter.
Doug Lefever: Thank you, Nakanomyo-san. Unfortunately, we just can't comment on that. I've seen the reports where people are trying to track Malaysia exports, Taiwan imports, and I think there's a lot of noise in those models. I would really encourage you to maybe look at other KPIs rather than that. I think you can try to get a head start maybe on what will be our numbers, but I just don't know that that's the best way to do it because that's a very imperfect data, is what I would say. I can't explain why you've seen a difference in the trend, because we don't look at those numbers ourselves.
Speaker 3: [Non-English content]
Masahiro Nakanomyo: [Non-English content]
Speaker 0: Thank you very much.
Mr: [Non-English content]
Speaker 0: Thank you very much, Masahiro Nakanomyo
Mr: [Non-English content]
Speaker 0: We would like to ask Mr. Kinoshita of UBS Securities to ask you a question.
Atsuhiro Kinoshita: [Non-English content]
Speaker 0: Thank you very much. Once again, congratulations for a wonderful, superb result and performance in last quarter. My question is about the operating income margin and your improvement. The second quarter, you have registered a very good margin and margin improvement. And you mentioned just now about the annual margin will also be expected to be higher. As much as possible, could you please explain to us the contents or the reason why that margin is being improved?
Doug Lefever: Thank you, Mr. Kinoshita. I think I talked a little bit before about the decrease from first half to second half. A big part of that was the inflation on the components that we put into our systems. Another thing that in the prepared remarks, is that we did have a reversal of inventory that happened in the first half as kind of a one time event that did lead to some higher than normal first half margins. Those won't be carrying into the future. So, we're kind of settling in at this low 50s. And as I also mentioned, we have the ability as devices become more complex, the configurations of the testers continue to expand. And those are margin expansion opportunities for us, both on the SoC side and the memory side. And so we've been able to use that fact to offset a lot of the inflation that we're seeing with the components. Hope that answers.
Speaker 3: [Non-English content]
Atsuhiro Kinoshita: [Non-English content]
Speaker 0: Thank you very much indeed. I have another question, if I may. People say that the memory is in big shortage nowadays. And your company actually mentions that the SoC tester and CPU could be the very promising drivers to really boost the performance in the future. But considering the generative AI system you have to have many components, including memory, CPU, and custom ASIC, and all those components need to be synchronized. Otherwise, if there is a shortage of memory, and if memory is not incorporated into the system, you are not able to ship as a product. And when you consider the future performance, I wonder whether the shortage of memory has some impact. So could you please comment on that?
Doug Lefever: It's true. Memory is scarce in the world right now, and there's a lot of factories being built right now. But largely in the last couple of years, the wafer starts have been moved from older legacy DRAM into high-end HBM and server DRAM. And that's caused big problems for industrial and other automotive applications. So we're not immune to that at all. What I will say, though, is we're part of a community that includes our customers, and they understand that for them to get testers, they have to provide us the memory. We went through this in the supply chain shortages in 2021 and 2022. And the industry really rallied around each other to make sure that the components were going to be provided for the equipment suppliers, because there was an understanding that if they didn't do that would lead to greater shortages across the board. I think it's tight, no doubt. But we also have very, very good suppliers that are helping us.
Speaker 3: [Non-English content]
Mr: [Non-English content]
Speaker 0: Mr. Kinoshita, thank you very much indeed for your question.
Mr: [Non-English content]
Speaker 0: As it is almost time to close this briefing, we'd like to take our last question.
Mr: [Non-English content]
Speaker 0: Mr. Hanaya of SMBC Nikko Securities, please have your last question.
Takeru Hanaya: [Non-English content]
Speaker 0: Thank you very much. I'm honored to actually give you the last question. Now, my question is about the production capacity ramp-up. You mentioned that you are about to really achieve the goals of the 10,000 unit production. For timing, it's not disclosed yet, though. But of course, you made effort to really increase production capability because there are some needs that you have to fulfill. And do you have some aspiration that you're going forward from 10,000 to 15,000 and beyond? Or is this just too early to actually look for that kind of aspirational future? And what is your idea of incorporating, perhaps in the upcoming MTP, we will actually deal with that. Could you please share with us your thoughts about the further production capacity expansion?
Doug Lefever: Thank you, Mr. Hanaya-san. We have already some aspirations. From 5 to 10 was a big step, and we're accelerating that as fast as we can. Yes, we'll give some more details about the timing of that, perhaps at the next call. I'll say I believe the next step will be another doubling from 10,000 to 20,000. That's something that we're actively talking about inside the company based upon the customer demand. Obviously, we're not prepared to talk about the timeframe for that right now, but I think certainly by the time we release our midterm plan four, we may be prepared to talk about it then.
Speaker 3: [Non-English content]
Takeru Hanaya: [Non-English content]
Mr: [Non-English content]
Speaker 0: Thank you very much, Mr. Hanaya, for your question.
Mr: [Non-English content]
Speaker 0: Although we have received additional questions, we would now like to conclude the session.
Mr: [Non-English content]
Speaker 0: Thank you very much for taking the time to join our financial briefing despite your busy schedule today.