We will now begin the Q&A session.
[Non-English content]
All right. We'd like to have the first question from CLSA Securities Japan, Yoshida- san, please.
Yeah, this is Yoshida from CLSA. Congratulations on the strong earnings results. Um, as for the guidance, I understand there are some uncertainties, but concerning your strong progress in SoC tester sales in March quarter, I thought that your tester market outlook could be revised up furthermore. Uh, so could you please give us more details of the background of this new SoC tester market forecast, such as your, uh, market share assumption? And if there are any changes in the inquiries recently which are not yet reflected into this market outlook, such as the demands from GPUs and custom ASICs, could you also, uh, give us some colors of them? Thank you.
[Non-English content]
Okay. Yeah. Thank you for the question, Yoshida- san. You know, I think we are projecting pretty decent growth year-over-year from 2025 to 2026. You know, as shown in our, in our TAM numbers and we continue to progress with market share gains. In some cases, there's limitations not from the tester side, but just availability of capacity in the foundry and packaging area. Those are all baked in into our TAM estimates. A lot of it is kind of external factors. That's, you know, in the future, as we continue to expand in the industry, that's why we are continuing to expand our production capacity to meet the capacity increases in the market. I hope that helps.
[Non-English content]
So is there any changes in, uh, in your, uh, AI-related customers, uh, production focus in, uh, in the near term, uh, um, that, uh, hasn't been reflected into this market forecast?
[Non-English content]
Um, this is Sanjeev Mohan. Um, we don't necessarily see the changes, um, but what we see is the, uh, firming in the forecast. Um, uh, obviously in January when we did this, we were, um, estimating the, uh-- and using our judgment based on what we were seeing. But now, um, we-- its, um, uh, visibility has gone up, um, obviously, and we think it's, uh, firmer than before.
[Non-English content]
Thank you very much.
[Non-English content]
Thank you, Yoshida- san.
[Non-English content]
Next, Mr. Yoshioka from Nomura Securities. Go ahead, please.
[Non-English content]
[Non-English content]
[Non-English content]
I'm sorry in Japanese. I have a question regarding your silicon photonics business. In your presentation in highlight, you said you want the very large order for the high volume production tester area for silicon photonics. When it comes to the silicon photonics for the future, what sort of term would you expect in this area and what sort of business opportunities can you see in this area? I think at present, also, I want to know the current status. What is your current market share and what sort of, can I say, level you are expecting? In this area, your competitors are also very strong. I would like you to let us know how you can have the catch up plan or try to, can I say, innovate the market share. I'd like to understand the potential of the silicon photonics market.
Yeah. Yeah. Okay. Thank you. Yoshioka-san. Well, first of all, we don't have to do any catching up because we feel that we're leading in this category. However, the market size is still difficult to determine. It's a very new business area. There's many uncertainties when it comes to different test insertions. The test times are still being understood along with the yields. So this year is really the first year where there's going to be some volume production that gets put into place. And then in terms of trajectory, we see that next year should more than double. And then after that double again at least. So we're very bullish on the market size, but it still is very difficult to determine exactly how large it's going to be.
I think, you know, we're going to do some work throughout the year, and I'm hoping that we can come back later this year and give a more accurate report on TAM size. It's just a little bit too early to tell.
[Non-English content]
I just want to add one last comment, which is part of our leadership position is the result of us picking the right partners to collaborate with. You know, these test cell solutions for silicon photonics are quite sophisticated, and so we've been very deliberate in who we've selected as our partners, and that's been a real strength for our solutions.
[Non-English content]
[Non-English content]
Thank you very much. That was very clear. Thank you very much for a very detailed explanation.
[Non-English content]
Thank you very much, Mr. Yoshioka.
[Non-English content]
Now we like to take the next question from Goldman Sachs Japan, Nakamura.
Congratulations on your strong results once again. I wanted to ask you about the capacity outlook for your SoC testers. I think previously you mentioned that your plan was to expand it to 5,000 units by the end of fiscal March 27. Now you're saying that you're accelerating your capacity expansion to 10,000 units. So wanted to see when exactly are you planning to expand your capacity to that 10,000 unit mark? And also, you have maintained roughly maintain your guidance for calendar 26 SoC tester. However, you're expanding your capacity plans. So wanted to see what's the reason behind this.
[Non-English content]
Let me just be clear about the numbers. We had announced that we would have at least 5,000 systems per year by the end of this fiscal year, and we're tracking above that plan. We also had said that several years later, we would be at 7,500 systems per year, and we've been aggressive in pulling that in. Today, I wanted to be clear that now we're moving to even a third step, which will be 10,000 systems per year that we are targeting to be ready, hopefully the end of 2028, but more likely into 2029. That's the first part of your question. The second part is, you know, we have to plan this capacity more than, you know, one year out.
We're looking out into time into 2027 and 2028 and even beyond. You know, we're very bullish upon the market continuing its high growth rate. In order to put the right supply chain and production capacity in place, we have to make those kind of decisions, you know, a couple years in advance. That's what's driving these continuous step functions upward. Hope that helps.
[Non-English content]
Thank you. Just as a follow-up, what was the number for your SoC historic capacity at the end of fiscal March 2026? Is it fair to say that by March 2027, that's going to be something like in the neighborhood of like 5,500 or maybe 6,000 units or something like that?
[Non-English content]
Yeah, I don't believe, Nakamura-san, that we disclosed what we were going to build in production capacity for this past year. You know, it can be implied with some buffer what we can do. I think we're more interested in looking, you know, out into the future for the 5,000, 7,500 and 10,000 steps that we're pursuing now.
[Non-English content]
Thank you very much.
[Non-English content]
[Non-English content]
Thank you, Nakamura-san.
[Non-English content]
Next, Mr. Shikanai of JPMorgan Securities Japan. Go ahead, please.
[Non-English content] All right. I had a question on the replacement versus the new demand. Given the SoC testers can, theoretically speaking, that, could be used across generations, how should we think about the trajectory, the new demand versus new system demand versus, replacement demand or, upgrade demand?
[Non-English content]
Yeah, I think we-- this is Sanjeev again. I-- we expect the, uh, new system demand to continue, um, because, uh, right now we are still in this growth phase. And, um, we expect that this will continue for at least, um, next two to three years. Uh, beyond that, um, we believe that, uh, as these devices get more and more complex, um, they will require upgrades. Um, so at that point, um, there may be an upgrade cycle for the systems, but we still expect, um, the overall, um, use of AI and the tokens to continue to grow, um, for several years. So yes, um, for right now, uh, majority of our business is new systems. At some point in the future, you know, two to three years down the line, uh, there could be more upgrade business.
Um, but at this point it's, um, it's mostly, uh, new systems. Um, as far as being concerned, um, you know, will we ever plateau and because these systems can be used, uh, f or a long time, we really, you know, have not seen that. There's always new catalysts for the industry and new segments that emerge, and that is not our concern at this point.
[Non-English content]
Yeah, I would just add one other simple comment, which is, um, unit volume is gonna continue to grow, you know, as wafer starts and things like CoWoS capacity, new memory factories are brought online, unit volume will continue to grow. Then there's another component which is test content continues to grow. So because of those two things, there's always, uh, a need for additional capacity. Then within that capacity, certainly there's upgrade, uh, opportunities for things like, uh, you know, power supplies and so forth. But, uh, overall, the trajectory of, uh, the unit volume multiplied by test content gives, uh, two factors, uh, that drive, um, new capacity to be put in place for test.
[Non-English content]
Understood. Thank you very much.
[Non-English content]
[Non-English content]
Thank you very much, Ms. Shikanai.
[Non-English content]
Yes, we'd like to take the next question from BNP Paribas Securities Asia Limited. Alex Chang, sir, please.
Yes, thank you for letting me ask this question. Actually, I realized that company did a very strong full Q results because SoC tester, you know, it seems like gross was higher than the memory tester. Is this reason for your better margin, the full Q? Another following up question is, given your guidance for next year, you assume like largely flattish operating margin. According to your projection for the TAM, SoC tester will still be growing faster than memory tester. Is this a little conservative for the margin expansion next year? Thank you.
[Non-English content]
Okay, thanks, Alex. Good question. I think, you know, and for Q4, we had a really perfect blend of product mix along with licensing that really drove the super high gross margin. You know, we obviously SoC is one of those factors. Within the memory portfolio, we also have different memory systems and the ones that we're shipping in that quarter were on the higher end of the margin profile for us. Plus there was a good licensing revenue. Everything kind of came together in Q4, and that's really not indicative of what we'll see consistently going into next year necessarily.
We're also, you know, in the market, we're contending with, you know, some inflation and, you know, you know, we use a lot of memory in our systems as well. We're making sure that we don't overstate some assumptions because our bill of materials is being affected by some of the macro economic things along with just supply constraints in the memory area as well.
[Non-English content]
Thank you. Can I ask the second question?
Sure. Yep.
My second question is about the competitive landscape. I think your peers earlier indicated they target around 15%, 20% market share in the GPU market in the midterm. I was wondering now for your customers, you know, purchasing supply supplier strategy and for to pick for particular GPU model, will your customer introduce dual supplier or just, you know, for the, you know, just a dedicated supplier for a particular GPU model? And if your, you know, your peer, you know, will be it turns more aggressive on the pricing. Will you feel pressure on this and will this impact your margin? Thank you.
[Non-English content]
Yeah. Alex. I'm aware of, you know, these kind of discussions. You know, unfortunately we don't comment on customer specific activities. You know, all I'll say is that we still feel very comfortable with our market position at GPU customers and other accelerator customers. You know, we just have our heads down. We do our work and let the numbers speak for themselves, and I'll just leave it at that.
[Non-English content]
Yeah, I think I would also like to add that the pricing you mentioned, you know, is very important is one variable. You know, we always have to be very aware of that and be very competitive, which we, you know, obviously have already been doing. You know, as far as the dual vendor strategies are concerned. Yes, of course, as the business has gotten larger, you know, customers are concerned about the assurance of supply. Therefore, you know, we always want to stay ahead of that. There are also other reasons, technical capabilities that are very important in selecting a platform, which we obviously have a strategy to stay ahead of competition. Finally, you know, the dual vendor strategy can also benefit us in certain areas. We should always factor that as well.
[Non-English content]
Thank you Alex.
Thank you very much. Again congrats for the strong results.
[Non-English content]
Thank you, Alex. [Non-English content]
Next question is from Mr. David Dai from Bernstein. Go ahead please.
Thank you very much. This is David Dai from Bernstein. Thank you for taking my question and congratulations for a strong set of results. I have two questions. The first one is regarding the CPU testing opportunities, because for the AI data center so far, investors and the industry has been focused on the GPUs accelerators. Recently, with the rise of agentic AI, I think companies like Intel and Arm have both made a big cause that CPUs are going to become a bigger contributor incrementally pairing with like we're seeing more CPUs to pair with the GPUs and accelerators. Wondering what kind of impact is that going to have on the Advantest's tester market and whether that has been factored into your market outlook this year as well as the future, you know, capacity expansion to 10,000 units?
[Non-English content]
Yeah. Good. Thank you, David. That's a really good question. Yeah, you're right. Yes, there's, right now, you know, because of the move to heavy inferencing, there's going to be more CPU usage. There's, you know, through heavy math and pre post-processing, cost, power benefits. Those are all things that are driving things through the data centers, you know, more agentic tools, you know, calling on software that are running off of CPU. Yeah, CPUs, LPUs, TPUs, those are all part of, you know, a huge migration right now. You know, we're really well positioned for that part of the market. We have traditional customers, long time customers been where we've been in place for decades.
We have been very intentional for some of the fabless accounts that are of note, where we have a big ecosystem of partners that work on the fabless side. With the V93000 being really a de facto standard in the foundry and OSAT space, we work really hard with those partners to maximize the value of our systems, you know, in the OSAT and outsource space. Finally, you know, these devices are still very complex. Again, leveraging our partners to develop the full test solutions is critical for this area too. I think, you know, people have talked about orders of magnitude or several orders of magnitude higher compute for inferencing versus training. We have factored a lot of that into our 2026 forecast.
But we see that going throughout, you know, 2027 and beyond. And that's again why, uh, we were so aggressive with our production capacity expansion. Um, so hopefully that helps.
[Non-English content]
Thank you. That was very helpful. My second question regarding the China demand. As everybody can see, China needs to build or is building its own ecosystem of the accelerators as well as the HBM memory capacity. As they do, they are obviously going to need more tested capacity as well. Just want to confirm that Advantest we can expect Advantest to have a similarly strong position in China as well. If you have any visibility on timeline when that might take off, that would be extremely helpful. Thank you.
[Non-English content]
Yeah, I think for China, it's, you know, it's an important region for us. It always has been. And, um, you know, we continue to participate in that high end SoC, uh, market. Um, China is, you know, it's a complicated, uh, situation. There's a lot of domestic, uh, competition that is growing up in China. There's obviously a geopolitical, uh, component, um, with the decoupling and regulatory environments. Um, and so, you know, we are navigating that, um, but we still feel as though, uh, we have a leadership position in our, uh, technical capabilities that, uh, are needed, uh, for the high end SoC market. And so, uh, our, you know, we intend to continue to compete, uh, as much as possible in China. And, uh, you're right.
Yeah, there's a lot of activity there in, uh, in new device applications and, uh, pretty exciting things happening there.
[Non-English content]
Thank you very much and congrats again.
[Non-English content]
[Non-English content]
Thank you very much, Mr. David.
[Non-English content]
Thank you very much. There are still more some questions, but it's time for us to close today's session. Thank you very much.
[Non-English content]
Thank you very much for taking time to join our financial briefing despite your very schedule today.