MUFG Securities. Wadaki-san, please go ahead.
Congratulations on a very strong performance. In March and April, I've spoken with investors about questions and discussions. I've spoken with more than 100 investors about your company. And it seems that the greatest interest from investors about your company has to do with high-end servers' yield issue, which is causing "special demand." Investors are afraid that this "special demand" may level off from July onward. If possible, I'd like to know if the yield-related special demand for high-end servers, would that come off? Are you concerned? Can you quantify, give any quantitative suggestion? Maybe you could give the split between the first half and second half revenue split. If revenue is not going to go down, maybe you can give some supporting evidence for how demand can be supported despite the concerns about this special demand. Thank you.
Wadaki-san, thank you for your questions. I believe that there were two parts to your question. One is special demand, whether your guidance includes the special demand. You would also, secondly, like to know the split between the first half and second half sales.
Yes. Thank you, Saragi-san. Thank you for the question. This is Doug. First of all, certainly there's always going to be certain devices that go through different yield profiles, and that will definitely affect the amount of test capacity needed. Generally speaking, there's no anomalies that we see currently right now that would affect the overall demand for test capacity. Most of the demand that we see is a continuous demand leading from the AI, HPC devices that are either currently at a peak ramp or transitioning to the next generation devices. That's both the traditional GPU players as well as the upcoming Custom ASIC and CSP players. Generally, we are not seeing anything that's a one-time yield effect. In fact, we're seeing more complexity leading to more test content, either in test times or additional insertions.
As far as the first half versus the second half guidance, usually we do not have as much visibility into the second half just because most of the visibility we have is six months out. We do see a stronger first half right now as we stand at this point in the year.
Thank you. It is not that you are seeing—you are going to see a little bit of a decline in the second half compared to the first half. Can you quantify that level? What kind of decline?
Nakahara will take that question. As Doug just explained, there might be, in comparison, first and half, the first half is going to be a little bit higher, and it will not be a huge gap. With regards to the gap, as Doug just explained, for the second half, we do not have full visibility. There might be some uncertainty. We cannot have a full read-through for the second half. When it comes to generation upgrades or device transitions, I am informed of strong demand. It's just that there's some uncertainty, which is why we expect a bit of a moderation in the second half, but we don't see a big gap.
Yeah. One more thing to add, Wadaki-san. A lot of it's visibility-driven, and there is a chance that our customers want to put test capacity in towards the latter part of our fiscal year in preparation for next generation device ramp. That is an area where things can fluctuate either from our fourth quarter into the first quarter of 2026. It just depends on the timing of the ramp.
Thank you very much. Thank you very much.
Next question, please. Goldman Sachs, Nakamura-san, please go ahead.
Thank you for my question. My question, I just wanted to double-click on your guidance for fiscal March 2026. Looks like even on a cost and currency basis, your revenue is only slightly up. In terms of your calendar 2025 time out look, you are still expecting a healthy growth. I just wanted to understand what kind of assumption you have made into your revenue guidance for March 2026. Is it mainly concerned on perhaps lower market share or lower test times? You mentioned increasingly invisible sort of environment for the macro environment. Are you being conservative, especially into the second half of the fiscal year? Thank you.
Yeah. Thank you, Nakamura-san. Yeah, I think the assumptions are that even at the midpoint, if you look at our SoC numbers and account for foreign exchange changes, we do intend to pick up some market share. On the Memory side, the expectation is to remain relatively flat. As to the point about conservatism, I think everybody right now, including us, are kind of watching the macro economy and trying to judge what effects it may have down the road. We are taking a cautious approach, particularly beyond what we can see in the first half.
Thank you. That's very clear. Just a quick follow-up. Is your fiscal March 2026 guidance based on the midpoint of the calendar 2025 tester time? Also, you mentioned the visibility is very low, but are you seeing any changes in the customer behavior after the announcement of the U.S. tariffs and the increasing macro uncertainty? Thank you.
Yeah, the guidance is essentially at that midpoint. I think the numbers should—you can kind of back calculate from our numbers in the slides and the midpoint. As to customer behavior right now, no, we're not really seeing any behavioral changes from our large customers with respect to tariff and trade concerns. We've listened closely to other earnings calls recently from our customers and peers. I think that's a message that's being consistently sent from those companies. We see the same thing. No real effect from tariffs as of yet. Especially directly, there's very, very small levels. The concern would be if there's any kind of macro level effects indirectly that would come through the global recession. Even there, the HPC, AI area seems to be not fully immune, but somewhat protected from some of these geopolitical systems.
Thank you, Nakamura-san. Next, CLSA Yoshida-san, please go ahead.
Thank you for taking my question. It sounded like you're communicating with your customers very closely. Could you please share your insights on the SoC and memory tester market size towards the calendar 2026, if you have, as well as supplementary information such as the test intensity outlook and opportunities in new test solutions such as the die- level testing? Thank you.
Thank you, Yoshida-san. I'll try. We don't guide on the 2026. The only thing I'll say is that we're optimistic about 2026 for a couple of reasons. One is there's going to be a next generation HPC devices from the traditional players. In addition to that, the cloud service providers and their Custom ASIC partners are releasing lots of devices during this year that we expect to ramp in 2026. I think that's going to be an additive piece to the overall TAM. The associated DRAM, particularly in the HBM, that corresponds to those devices should give some kind of an uplift. I can't give any numbers except to say that we're fairly optimistic about 2026 at this point.
Test intensity is a difficult one. I've been trying to get people not to talk about test intensity so much, more test dollars, just because the ASPs of the semiconductors can have sharp upticks and that can throw the test intensity percentages out of whack sometimes. From a test content perspective, we still see a really nice growth trajectory in that respect.
Additional product lines like our die- level prober, we think, are going to be also additive to our overall business, both from an overall standalone standpoint, but also, more importantly, from the strategic ability to address the test challenges that are going to be required for advanced 3D packaging. Those things go together along with our tester and our service environment and an important part of the company strategy.
Thank you. Just for the brief clarification, you said that 2026 is quite optimistic. Is that meaning the year can be another growth year as a total market and also for your company as well?
We hope so. For sure, we hope so. I should say that even on the 2025 midpoint, there was really a lot of internal discussion around this, which was we were without some of the macro effects going on right now in the world with tariffs, especially things with the test content increasing, we are trending us up towards the higher end of that TAM. Given the uncertainties that kind of offset that, we felt more comfortable keeping the midpoint. Honestly, we felt like 2025 is also a potential nice year, but we do not know what the future holds with respect to the macro environment right now.
Thank you. Thank you very much. That's very helpful.
Yoshida-san, thank you very much.
Next, Macquarie , Damian Thong. Damian, please go ahead.
Maybe follow up on your engagements with customers. You talk about that you benefit from launching new products for your customers. Can you maybe speak a little bit about the pipeline of, especially ASIC products? Can you characterize, for instance, how many more customers or how many more design starts would you be exposed to, ASIC designs you'd be exposed to, and how you see that trending in the next couple of years?
Yeah, thank you, Damian. Junko-san told me to speak more slowly. I'm sorry for everyone. I'll go a little slower. I can't talk about individual company devices or releases. The only thing I can say is there's a very robust list of devices that are being brought out from our hyperscale customers, either by themselves or with their Custom ASIC partners. I will also say that the amount of Custom ASIC providers has also expanded. There's new players in that market as well. As a company, we're very well positioned both at the hyperscale community as well as those existing Custom ASIC and new players in the Custom ASIC market.
Thank you. Would it be possible for you to maybe indicate your confidence as to whether you can increase your market share when those devices are launched? Obviously, your SoC tester market share is heavily anchored to, I think, a big GPU customer. With the ASIC side, will your market share tend to go higher?
Yeah, we expect our market share to go higher as those devices come out. We're very well positioned with our 93K platform at nearly all of those accounts.
Thank you very much.
Damian-san. [Foreign language]
Damian-san, thank you. Next, UBS. Yoshida-san, please go ahead.
Thank you very much for taking my question. My question is overlapping the previous question from Damian-san. Once again, the Custom ASIC. Could you elaborate the growth plate for this year? You have an assumption TAM for this year's SoC tester market. GPGPU or Custom ASIC, which one contributes more to the growth this year? If you can manage to describe like 2026, 2027, like a long-term trajectory, it's kind of tough to estimate, but in terms of SoC tester, which is going to be the higher driver for the SoC market? Thank you very much. This is the first question.
Okay, for 2025, I think that's an easy answer. That'll be more of the traditional market for us. As I said before, there's a lot of new devices in the Custom ASIC world that are being brought out that we expect to ramp into 2026. As for the split between the "traditional" and the Custom ASIC, we can't yet speculate. We do think the Custom ASIC area is going to be very large for us. I think there's a lot of discussion around whether for the inferencing, if the traditional players will be displaced by the specific workloads of the Custom ASIC. We're under the assumption that with the growth that both are going to coexist. We're very fortunate that we are well positioned both in that traditional and that newcomer space.
I apologize. I can't speculate a split between traditional GPU and Custom ASIC at this point. Maybe next year, I could give a better idea.
At least Custom ASIC demand is going to grow this year, at least?
Yes, a little bit this year, but the larger ramps are most likely going to happen in 2026. I can point you to just the companies in that Custom ASIC who have cited the serviceable market for them. You can kind of back- calculate the level of test requirements that they'll need. It is fairly large when you run that calculation.
Thank you very much. The second question is really the GPU margin in the March quarter was 60%. It is very high. Is it sustainable if your demand for SoC tester is remaining at the level of like JPY 148 billion-JPY 150 billion each level, or is there a special item that was included in the March quarter? Thank you.
It's a blend. I mean, our SoC testers always have a higher gross margin. So that was a higher mix in the fourth quarter. We also had upgrades that we were able to install. And upgrades are always a higher profit than the full systems. And so there was some uplift from that. Sustaining 60% probably is going to be difficult. It'll fluctuate quarter by quarter based on the product mix. I don't want to call that the high because we were always working on increasing our margins. But we had some kind of alignment of things that happened in the fourth quarter that really helped us on the gross profit side.
Thank you very much.
Yoshida-san, thank you. Next, BofA, Hirakawa-san, please go ahead.
One question for kind of near-term visibility. You mentioned that you have a good visibility in the first half, March 2026, and the second half is not that much. In the first half, do you see any skew in demand in SoC and memory testers either in the June quarter or September quarter? Or are you expecting the similar quarters in June quarter and September quarter? That is my question.
Yeah, I'll try to answer, and then maybe I'll ask Mak to help me. Pretty much the demand is consistent. It's a matter of planning out our production cycle with the long-term forecasts that our customers give us. We work closely with them to make sure we're managing our production capacity with their demand. That leads to less lumpiness because we're working together to keep that as a consistent volume. I think that's kind of at a high level. I don't know, Mak, maybe you had some.
As Doug just explained, in the first quarter and the second quarter, we expect almost a plus-ish trend, firstly because in the first quarter, we already have a good level of bookings and a good pipeline. Given those statuses, we can see that it will be even or flat. As for the second half, we do not have full clarity yet.
In FY 2024, we saw a sequential increase quarter by quarter in the fiscal 2024. By Q4, we had good capacity in place. We did not struggle too much with shipment and delivery. In the first and second quarter, we expect a flattish trend. Thank you very much.
Hirakawa-san, thank you. Next, Jefferies. Nakano Mio-san, please go ahead.
This is Nakano Mio from Jefferies. Can you hear me?
Yes.
My question is about PS5000. In the equity market, investors are talking about replacement demand for PS5000. The investors think that that has made good contributions to your earnings. Maybe investors are concerned that it may peak out from now onwards. As for incremental, given the incremental complexity increase, is that going to be a constant trend? Once PS5000 diffuses among customers, could it peak out? Is that a fair assumption? That is what many investors believe.
Nakano Mio-san, let me just clarify your question. With regards to PS5000, in 2024, we launched PS5000. You assume that PS5000 has already populated the market. Therefore, for the going forward, there will not be incremental deliveries or demands of PS5000. Are you asking us that we expect a peak out as well?
Even if PS5000 peaks out, will there be an increase in other modules? How should I think about the overall demand?
In the stock market, the market people understand that PS5000 is a standard module for the EXA Scale V93000. If the PS5000 is already installed fully, then whether this is peak out for the demand for PS5000, even the next generation customer's device is lower.
Just first to clarify, the PS5000 is our standard digital module that ships with all of our EXA Scale platforms. Any new mid-level or high-level digital application requires us to ship PS5000s. We always design our digital instruments and power supply instruments so that they have some longevity.
The PS5000 certainly has a long life because of the data rates and the vector memory within that instrument. First of all, there is no need for there to be a technical inflection point in the near term that would create a new digital instrument. Most of our sales are forecasted into the future to be provided by the entire systems, which would be digital instruments, power supplies, sometimes mixed several instruments or RF instruments. We see it as a benefit to have a common uniform digital instrument with a long lifetime because it lends itself to many different applications. Hopefully, that helps. Maybe Mak, try.
Let me add a comment or two. In 2024, the pace of ramp-up was really fast when it comes to launching EXA Scale and PS5000. Our customers back then, like foundries and OSATs, were suffering from low utilization rate. There were many new device launches, and we booked sales. Now at customer sites, our system's utilization rate is almost full. As new devices launch at customer sites, which is what we expect, we are not really concerned about slowdown of PS5000. Rather, we expect that higher level of utilization is going to encourage them to place new orders.
Do you think that in 2026, this trend will still continue? There will not be much change?
When it comes to logic devices, if I look at new development and how complexity will continue, I hope that will be the case. The question mark for us is external factors like economic impacts from macro. When it comes to technology factors, we'll have new modules as well. We think this trend can be intact.
Thank you.
Nakano Mio-san, thank you very much. We are almost close to the end of the hour. We'll take one final question. Tokai Tokyo Intelligence Lab, Kamisaki-san, please go ahead.
I'm Kamisaki from Tokai Tokyo Intelligence Laboratory. My question has to do with market share. In 2024, your market share was 59% and followed by 56% market share in the following year for SoC testers. You mentioned the reason is emergence of local suppliers. How is your market share in the high-end segment? For memory tester segment, will there be emergence of local suppliers as well?
Question. First one is, is there any other risk in high-performance computing area in SoC or competitive environment? The second question is in memory tester market segment competition, are there any risk on the local supplier?
First, with regards to logic, as Doug mentioned, it was down by 3 percentage points compared to 2023. For us, the core market is high-end market. For that segment of the market, we believe we have gained market share. As you may be aware, in China, Made in China ATE is emerging. The local Chinese suppliers are now targeting the low-end segment. Chinese customers are now tending to buy local Chinese testers. I believe that there is some business that is going to the local players, but that is not a core market for us. Again, our target is the segment of the market that benefits from rising complexity. With regards to memory testers, Korea and China, there are local players, maybe just single-digit market share each.
Similarly, we do not think that a large part of market share is going to them because customers are demanding high-end testers because of the market shifting to a more high-end segment overall.
Thank you. That was clear. This is it from me.
Kamisaki-san, thank you very much. We have other questions in the queue, but we would like to conclude the session. Thank you very much for joining Advantest FY 2024 financial briefing despite your busy schedule. Thank you very much.