Alchip Technologies Ltd

TWSE:3661

Q1-2025 Earnings Call

[Daniel Wang, executive]

[Operator Instructions] And this video and audio content of the meeting, we'll update to MOPS. It takes a while for upload. So probably after 2 to 3 hours of the meeting, you can tune the video and audio content from MOPS.

Okay. I guess, everyone is pretty much understand what we are in the background of Alchip, so we will skip these slides. And these slides, I would like to invite our CEO, Johnny Shen for briefing the picture and the company's direction and the outlook.

[Johnny Shen, executive]

Okay. Good afternoon, ladies and gentlemen. I'm Johnny Shen, President and CEO of Alchip Technology. Thanks for joining the investor conference meeting. We appreciate the opportunity to share our Q1 results and provide the guidance for our future business outlook.

For Q1, our Q1 financial results are in line with our plan. Revenue reached TWD 319 million, representing approximately 4% year-over-year decline. Net income came in at TWD 44 million with EPS TWD 18.1. So the detailed breakdown and analysis will be provided by Daniel in a later section.

There are a few key highlights for this quarter. First one is our gross margin improvement. Yes, we have achieved notable percentage increase -- percentage point increase in gross margin compared to last year. This was mainly driven by favorable [ mix ] and improved mass production margins. Second is our N3 project milestone. We successfully complete and tape out our most important N3 project. We expect wafer to come out during this quarter, with prototyping and evaluation beginning in Q3. Number three is our ADAS production progress. Prototyping and evaluation for our ADAS production is nearly complete.

Our end customer is satisfied with the result, has decided to place additional wafer orders beyond their original plan. As for the future business outlook, as I mentioned in the previous earnings call, our mass production line is currently in transition phase. The current generation is almost end of life, while the next-generation product scheduled ramping up starting from 2026.

Unfortunately, our highly expected IDM customer has revised down their production plan multiple times. As a result, there's a significant revenue decline compared to the number last year. On more positive note, we continue to see strong demand for HPC and AI-related design opportunity in North America region, utilizing the most leading-edge technology like N5, N3 or even N2.

We have already secured several designs and remain confident in our ability to grow our revenue -- our revenue this year. In addition to HPC business, mass production for new gaming device and smart speaker application has started ramping up. Overall, our annual revenue forecast remains consistent with the guidance we provide in the previous earnings call.

Yes. As for the geopolitical risk management, we have a significantly -- we have a significant progress in diversifying our business beyond China. In Q1, only about 5% of the revenue are originated from China. And we are also facing limited potential tariff impact as primary -- majority of our chips are shipped to the local CM partner rather than directly to U.S.

In terms of workforce, we have launched a very aggressive hiring plan to strengthen our engineering support beyond China, including Japan, Taiwan, Malaysia and Vietnam. Our Japan office can accommodate up to 200 employees now, and our Malaysia and Vietnam office has over 50 engineers combined.

Based on current growing plan, our Southeast Asia workforce were up to 120 people by the end of this year.

In conclusion, our revenue forecast remain unchanged. We anticipate flat revenue for the next quarter. However, our profit margin expect to further improve, driven by increased customers recognition, stronger contribution from NRE and pricing support from key suppliers. Despite the revenue adjustment, our earning number forecast remained consistent with last year.

Look ahead, we remain highly optimistic for AI market from 2025 to 2027, like the industry leader, we are confident in our ability to outperform the market CAGR in HPC applications.

[Daniel Wang, executive]

Okay. My turn, it is Daniel. So the number is really straightforward. For the revenue part, it is already a public information. The first quarter revenue is USD 318.7 million, representing 21.1% quarter-on-quarter decline and 4.4% year-on-year decline. However, because of the improvement of gross margin, our operating income in the first quarter this year record $45.4 million a week compared to a year ago, it's 17.9% year-on-year growth.

By comparing with last quarter, fourth quarter '24, the operating income is 16.3% quarter-on-quarter decline. So for the first quarter, the net income is USD 44.4 million, representing EPS of TWD 18.1. And for the revenue breakdown for the first quarter, by application, you may see that 95% of our revenue came from HPC-related applications.

So for the other three categories combined, we don't consider they are significant. They are consumer sector, networking sector and the niche markets.

[Technical Difficulty]

So I will repeat the breakdown again. We are facing some technical problem right now. And for the quarterly breakdown by applications, again, HPC is the application contribute as the biggest revenue of our company. 95% of our revenue contributed by HPC sector in the first quarter. So for the process node breakdown, again, we are very proud of our advanced of our process node technology contribution.

Okay. 41% of our total revenue comes from 5-nanometer or more advanced process node. And 55% of our revenue comes from 7-nanometer process node in this quarter, which means combined 7-nanometer and 5-nanometer and more advanced process node contribute us more than 95% of our total revenue in the first quarter this year.

And for the geographic breakdown, like our CEO mentioned, for the first quarter this year, revenue from North America accounted for 93% of our total revenue, while the Asia Pacific, which includes Taiwan and China contributed about 5% of our total revenue in the first quarter, which means the geopolitical risk between China and the U.S. right now, I think we are managing it very well.

So for the first quarter business review, first of all, again, our revenue suffers decline on the tapering of 7-nanometer AI chip shipments. And I believe I guess the majority of you already realized that the previous main revenue contributor, the 7-nanometer AI accelerator to the cloud service provider in North America, the shipment is tapering off, and we are expecting very limited revenue contribution from this product in the rest of this year.

And for the first quarter, again, there is seasonality for the NRE revenue. The first quarter usually is the low season for our NRE design projects. So that -- these two factors combined as the main reason for the relatively quiet revenue performance in the first quarter. And again, as I just mentioned, the gross margin in the first quarter improved to 23.2%, creating 17.9% operating income growth year-to-year -- year-on-year to TWD 45.4 million in the first quarter, which we consider this trend will keep on going in the rest of this year.

Secondly, the NRE pipeline remains strong in the North American region. So we are very proud of saying that our first 3-nanometer AI accelerator project take out in first quarter this year. And the production is expected to kick off in early first quarter next year. Until today, the whole process is -- the whole project is running smoothly, and we don't expect any delay for this schedule.

Secondly, we expect the process node migration still to speed up in 2025, we will see many 3-nanometer projects to kick off or to tape out this year. And we are going to see NRE revenue contribution from 2-nanometer projects also starting from this year. So we will keep on seeing the process node migration from 5 to 3 from 3 to 2 nanometer. And the NRE demand outlook, especially in North America keeps very strong. We keep on winning multiple projects from either start-ups or some pretty reputated companies, and we consider this trend to keep on going for the next couple of years.

And for the outlook this year, we expect the revenue remain quiet in the rest of 3 quarters. There are some reasons. First of all, again, as I mentioned, there could be limited revenue contribution from the 7-nanometer AI accelerator. And secondly, we see the 5-nanometer AI accelerator project to the North American OEM customers appears soft.

However, we think given the better revenue mix, which means we have higher percentage of online contribution this year, coupled with the better margin of the current major production contributor, we expect the gross margin in 2025 will have a relevant improvement for the year-on-year basis. Again, for the whole year, the process node migration keeps ongoing, and we do believe it fuels the NRE growth in 2025. We are still projecting a really, really strong NRE revenue growth in 2025.

Reasons are, first of all, we see multiple 3-nanometer projects picking up this year and multiple 2-nanometer projects kick off this year. And we do see the ADAS project from the China customer to be into production in the late this year.

And we want to reiterate that all important projects are running on schedule. There is no delay in sight, which means I would say, for 2025, we will have a quiet revenue performance. However, in 2026, I see very huge significant revenue growth potential in the coming years.

Thank you. That's -- I think that concludes our earnings call presentation today. Okay. Right now it's Q&A session. [Operator Instructions] . Annie Chen, please.

[Unknown Analyst]

Can you hear me?

[Johnny Shen, executive]

Yes.

[Unknown Analyst]

Management team, this is Annie from UBS on behalf of Haas. So the first question is regarding your progress on the hyperscalers 3-nanometer AI project. Could you share a bit more about the latest progress regarding the mass production schedule? And as you mentioned, like no delays seen as we get closer to the production. How should we think about like the contribution from this project when it starts to ramp? Like any reference would be very helpful.

[Daniel Wang, executive]

Okay. As for the potential contribution because, first of all, we are not allowed to mention the numbers here. And secondly, we don't consider to disclose the numbers is a good thing for our customers because the project right now is still in the mask-making process.

What I can say is, for now, everything is on schedule. Most likely, the pilot run sample will be in the market by the end of this year and the potential mass production will start in the first quarter next year. As for the revenue contribution, I can just say it could be really, really significant.

Let me give you a number for the reference. For the 7-nanometer project, the lifetime contribution to us is about [ $1.5 billion to $1.6 billion ]. And we don't know what the situation for the 5-nanometer. But according to the report I read, the scale of the 5-nanometer AI accelerator is probably like [ $3.5 billion to $4.5 billion ].

And for the 3-nanometer, I would say you can probably looking at the same trajectory of the potential scale. So for now, I cannot give you a firm number about the scheduling. But I can tell you that the whole ramping up process will be very fast.

[Unknown Analyst]

Okay. Got it. That's super helpful. And our second question would be regarding your business diversification. So like given the high potential from the 3-nanometer AI accelerator project, could you let us know like your strategy to diversify your business? And how should we think about your opportunities in hyperscalers or AI model providers? Any color on the initial engagement or progress -- like project progresses would be helpful.

[Johnny Shen, executive]

Okay. Let me take this question. Yes. In fact, we are taking out around 20 designs a year. But in order to diversify this particular customer, honestly, is very hard. To be honestly, the total ASIC contribution from this market is 95% is coming from two companies.

So in order to fully diversify this special account, we have to win others. But in terms of NAND diversification, I think we are working very hard. So other CSP in addition to these two companies are working with us very closely. And yes, we are -- even though the revenue right now is not so significant, but we are -- again, we are working with them very closely to try to win their next-generation design.

So other applications, like I mentioned before, for ADAS and eventually robotic-related application, I think we are doing very good. And also some gaming device start to ramping up. I hope we can diversify some business. But unavoidable starting from next year, this particular customer still contribute majority of our revenue.

[Unknown Analyst]

Got it. And if we could just sneak in one more. Would you mind like sharing with us, if any, progress on the 2-nanometer chipset design and your advantage or value add to your customers when migrating to like more advanced nodes?

And separately, would you be able to comment on your IP partners' progress on the IP qualification on 2 nanometers? Do you think it would be a bottleneck or like so far so good?

[Johnny Shen, executive]

Yes. For 2-nanometer design, I think we are doing -- we are doing quite well to be straight, because of our success of this N3 design, everything is on schedule. Customers are very satisfied for what we are doing. So likely, we have a high confidence to continuously provide service to this particular customer. But in terms of IP, the different customer has a different strategy. I think they intend to be open. Yes, they don't want to be captive.

So working with us along with third-party IP partner will be the right approach based on customers' demand. So there's no showstopper for -- from IP side. And I think we are not too far away from winning. Hopefully, we can deliver good news very soon.

[Daniel Wang, executive]

Okay. Charlie Chan, Morgan Stanley.

[Charlie Chan, analyst]

And also congrats on a great execution. So maybe just try to ask some tough question, right, because there are market rumors talking about your current 3-nanometer has some performance issues, some related to IP and even some talk about there are other 3-nanometer tape-outs on the same projects, which could be provided by some of your competitors. So can you please try to educate us how this -- does it make sense from industry practice?

[Johnny Shen, executive]

Okay. Yes. Let me take this question first and maybe Daniel can add some color. I think this is a very good and tough question. Yes, first of all, let us to reiterate again, there's no delay for our existing design. Everything is on schedule based on the wafer out schedule and also the prototyping and even the production schedule, everything is on track. If there is any enhanced version needed, this design definitely will be done by us as well, because as you know, all the database is owned by us.

If any enhanced version needed, the implementation, it has to be through us. If they invite to additional vendor, they have to do everything from scratch again. So again, there's no delay. Any enhanced version if it happen, will be implemented by us.

[Charlie Chan, analyst]

Daniel, do you have anything to add?

[Daniel Wang, executive]

No, no, no. We trust our engineering. I knew there is a long dispute between who will be the supplier to the North American cloud service provider, who will be the major ASIC supplier? Or is there any chance coexist to suppliers for the same project? I personally think that most of those rumors don't make any sense.

[Charlie Chan, analyst]

Okay. So if I may ask some kind of operational questions. For example, if you do want to verify the chip performance, when will be the timing you would know the chip performance? This is number one.

And secondly, since you are so confident about this 3-nanometer mass production, when will you confirm the orders at TSMC, for example, for their wafer front end and cost booking and by when TSMC can give you a confirmation for this 3-nanometer project?

[Daniel Wang, executive]

Okay. Charlie, I guess it is not good for us to disclose the detailed scheduling of the project. It violates the NDA between Alchip and our customer. What I can say is as for the production schedule, as I just mentioned, it's expected to be in the first quarter next year. And I guess many of you understand that from placing orders to the actual shipment, it takes time.

So most likely, the POs, the orders were placed to TSMC prior to the end of this year. And for the potential production scale, I guess the CoWoS allocation could be a good reference to the market. And I believe you are all having very good interest content. usually, when TSMC's CoWoS allocation coming out in June, I guess probably most of you will understand the scale of the production.

[Charlie Chan, analyst]

Great. And I do have some follow-up because in your opening remarks, you talked about 2-nanometer projects to kick off in 2025. Does that include the major customers 2-nanometer? Or are you referring to other 2-nanometer opportunities?

[Daniel Wang, executive]

No. Charlie, so for the scheduling of the next generation, I guess, it's still under negotiation. I would say take up 2 nanometer in 2025 will be a little re-aggressive kickoff.

[Johnny Shen, executive]

Aggressive kickoff.

[Daniel Wang, executive]

Kick off. Okay. Okay. Okay. Yes, kick off 2025 probably will be in the ballpark.

[Johnny Shen, executive]

Yes. So Charlie, to answer your question, other than this customer, we do have other design win for N2.

[Charlie Chan, analyst]

Okay. So is it also in kind of the AI accelerator type of product? Or Johnny, because you also attended TSMC's technology symposium, right? Or just broadly, can you share some key takeaway from your side because, for example, the ASIC market developments, ecosystem, your partnership, since you want to talk about 2-nanometer opportunity, why don't you just answer this question more broadly?

[Johnny Shen, executive]

All right. So 2-nanometer, I think the progress from TSMC is very compelling on their symposium, they mentioned about compared to the N3 stage, the N2 progress are even faster and also acceptance rate are even higher. Yes, we did see some -- a lot of opportunity, people thinking about N2, but for N2 design, other than the communication part, whoever doing the HPC, most likely will be chiplet type of approach because it's the radical size is too precious for compute.

Most of the customers intend to put all the analog mixed signal into other mask, which likely will be N3 for the -- we call IO or whatever or the IO interface. So I think the project and the potential opportunity we're seeing from N2 is all toward that direction. So I think the one design require multiple tape-out.

That's why we emphasize that we have the confidence to further grow our revenue -- NRE revenue because even with -- even one design, I think, require 2 to 3 different tape-outs. So NRE obviously will be added together. But the drawback thing is design capacity and turnaround time, especially the design capacity will increase. Doing 1 and 2 design easily require more than 100 people.

[Daniel Wang, executive]

Okay. So Hariharan, Gokul, please.

[Gokul Hariharan, analyst]

So first question I had is on your -- I think you talked about potentially diversifying into some of the other CSPs. Could you talk a little bit about how is the opportunity there given, as you mentioned earlier, most of the CSPs have one big project with a particular customer -- sorry, a particular partner. Do you see this changing? Do you start to see multiple large projects in the pipeline for each of the CSPs where you start to potentially get an opportunity to diversify?

[Daniel Wang, executive]

Okay. I would say it is correct for the ASIC business, the customer usually intend to work with the physical design suppliers generation by generation. If there's no accident or if the back-end design service provider doesn't screw up the project, usually, we do the generation by generation projects with the customer.

For now, I would say everybody knows the big 4 in North America, they are Meta, Google, AWS and Microsoft. We do see opportunity there because right now, I think Alchip is quite reputated in the North American region. Those potential customers know us, and we do have some discussion engagement with them already. However, to the capital market, we want to be conservative. We will report if there is any good news.

To us, I will say the opportunity comes out because the business model. When the cloud service provider working with companies, product companies like Broadcom, Marvell, because for their business model, they require very high-growth module. So the cost saving is not that significant to the customer. And when the customer is thinking about, oh, we are trying to build up a very, very big AI infrastructure in the future.

So the cost for computing power becomes an important issue. I think it is our opportunity to penetrate into the existing supply chain of those potential customers because, to be honest, comparing to the margins, the pricing, we are really, really more flexible than those product companies like Broadcom and Marvell.

[Gokul Hariharan, analyst]

So we have to look for your large CSP target customers to potentially develop enough front-end design capability. That's probably your entry point into those customers, right?

[Daniel Wang, executive]

So far, yes, we are focusing on the physical design part. But for the front-end design, we are building up the capability, but we are not entering the field of doing the front end for our customer.

[Gokul Hariharan, analyst]

Okay. Just a follow-up there. Would you consider partnering with somebody else who can actually do the front-end design so that you can kind of offer this as a joint kind of package to solve for this and still reduce the cost for the eventual customer?

[Johnny Shen, executive]

Yes, Gokul, let me try to answer your question. I know what's your concern. So basically, the -- like I mentioned many times, winning CSP need to be step by step. For emerger account, in usual case, our starting point is receiving the RFQ, we can moving forward. But for CSP customer, we need to work with them very closely even before receiving the RFQ.

In fact, just like Daniel mentioned before, the gross margin is their saving -- price saving is their #1, one of their #1 priority. OCSP try to reduce the dependency from NVIDIA. I think majority of their CapEx is going to NVIDIA. But if they're using the service ASIC service model from Broadcom or Marvell or other competitor, actually, the saving will not be so significant. that go through us.

I think everybody tried to copy the business model, what we are doing with our end customer. So in fact, most of them come to us, we can tell them nowadays, yes, everybody need to pick up your part. You shouldn't rely on other people to define the architect and doing the coding for you. You have to do it by yourself in order to control entire chip design as kind of a COT model. If you want to rely on other partners, why don't you just buy the NVIDIA chip? That's a straightforward approach. If they invite another ASIC vendor, still pay a huge margin to them.

I think that ruin the entire ASIC purpose. I mentioned so many times for ASIC design, customers need to focus on their architect, their differentiation and for us, we focus on the back-end production, make sure the design schedule are under control. The IP vendor can focus on their IP solution. I think that's the best business model, which our end customer is adopting. And in fact, most of our CSP are going in that direction.

[Gokul Hariharan, analyst]

Understood. That is clear. My second question is on the N2 project potential for the big customer. Could you talk a little bit about what is the kind of process that you would be going through just to educate us on like is it still going to be an RFP, RFQ process? Or do you think the project could potentially be awarded directly without too much of a bidding process? And what are kind of like the milestones in the next 6 to 12 months, given that you think you're pretty close to a kickoff before the end of this year?

[Daniel Wang, executive]

Gokul, we cannot disclose the details of the project. We can just that there are some signals for our confidence. For now, we are really, really confident that we will keep on doing generation by generation business with our biggest customer. That's what we can say for now.

[Gokul Hariharan, analyst]

Okay. Understood. Last question on the ADAS project. I think you reported that it is back on track now with mass production end of this year. Could you talk a little bit about any further ADAS projects that you have won, either follow-ons or with other customers? And also now that this is back on track, is it fair to assume that some of these geopolitical/security concerns are kind of largely behind us on the design, size of the chip, et cetera?

[Daniel Wang, executive]

Gokul, we don't have the position to comment on that potential geopolitical change between China and the U.S. However, we do think that the ADAS is not the major target by the U.S. customer. So for now, what we can do is, is being optimistic. So for the project itself, currently, we did a pretty good design job and our customer satisfaction is good. So I would say because our position in China region is quite good. So as there is any ADAS ASIC design opportunity. I would say at least Alchip is one of the major competitors for the projects. For example, if BYD is trying to build up an ASIC, definitely, BYD will invite us for at least competing for the project.

Charlie again. Charlie, please.

[Charlie Chan, analyst]

So first of all, Daniel, I know you cannot really provide the full year guidance, right, given regulators' requirements. But compared to 3 months ago, that version, do you see some updates? It doesn't need to be numeric, right? But for example, you can talk about what do you see about the crypto project because it is considered as your upside and also the major IDN. You just talked about several revised down. So net-net compared to 3 months ago, that version, do you see significant updates?

[Daniel Wang, executive]

I would say our view remains the same. In our forecast, we have the upside part and we have the conservative part. So all in all, we still consider the ballpark number will be still in the range of what we said in our last year's yearly earnings call. So we don't intend to guide up or guide down the current market consensus in this earnings call because we consider for now, at least our internal projection still point out that we were in the ballpark range for our guidance.

[Charlie Chan, analyst]

Okay. So how about gross margin? I mean the first quarter margin looks pretty good, right? So compared to 3 months ago, do you see upside to full year margin?

[Daniel Wang, executive]

We are shooting to approach like the middle 20s. However, as you may know, that our gross margin subject a lot to the NRE percentage of our total revenue. So I cannot give you a very precise number, but honestly, we are shooting for the midrange. Yes. And as I mentioned, our first quarter margin is about 20-something 23%. And we are pretty much -- we are pretty confident that for the following quarters, the gross margin is on an upward direction.

[Charlie Chan, analyst]

Okay. Great. And secondly, maybe to Johnny. So again, you approved the private placements, so this time around, what was the purpose and what kind of strategic or financial investors you want to introduce.

[Johnny Shen, executive]

Yes. At this moment, we don't have any specific target. But I think the PP -- as you know, our cash is -- has no issue. It's very fruitful. We only invite the strategic partner who can help us, who are complementary to our business or can bring more -- provide more comprehensive solution to our ecosystem. So to answer your question, we don't have any specific target yet. But the few inquiry, I think we are putting into consideration at this moment.

[Charlie Chan, analyst]

Okay. And also a follow-up from both my side and also some investors about 2-nanometer because it seems to suggest there are multiple projects or multiple customers. So Johnny, can you give us some color what are other 2-nanometer in terms of customer type, meaning startup versus CSC? Or is it the networking chip versus the AI accelerators? Can you give us some color?

[Johnny Shen, executive]

Okay. Yes, I think other than the most important customer, I think I can tell you straight the rest of 2-nanometer win is to the emerging account like startup customer, not CSP at this moment. But I think yes, we -- I welcome all the leading-edge technology, whoever does a CSP or emerger count, I think we need more reference on N2 because N2's first GAA design, the design methodology and approach are totally different. We need more reference.

And also N2 will adopt the chiplet, we need more reference. I think that's very important. Similar situation is like the [ 2.5D ] we take the most CoWoS design. And those reference really make it come because of this, we have -- we are getting to the door to almost everybody. Similar situation for N2, whatever the -- if a customer financially healthy has ambitious to do the N2, we will be the best partner for them.

[Charlie Chan, analyst]

Yes. But you also mentioned that each project easy will take like 100 people resource. You also mentioned about some kind of new technology like GAA, chiplets. So I'm not sure about the risk reward, right? So it seems like that's a resource risk you're taking. But are you sure that the emerging account will really get into the mass production?

[Johnny Shen, executive]

Yes. For N2 design, the design turnaround time capacity, it is our concern. When we're selecting the project, I think we are calculating in many different dimensions, capacity and also the design turnaround time, we're always thinking about, yes, if right now, CSP will be our #1 priority, #2 will be whoever has the volume on their previous design. #3 will be the most leading-edge technology.

[Daniel Wang, executive]

And honestly, we don't know for the emerging accounts who are doing the 2-nanometer, but we consider it is an opportunity for us and for our engineers to enhance their skills and experience in the most leading-edge process node. So we will do -- we will try to manage the resource allocation to the projects. But on the 3, 2-nanometer, we wanted to be our budget.

Okay. So is there any further questions? I don't see any people raise their hand, and I don't see questions in the message board. So I guess that's it. Thank you for you all participating in our first quarter earnings call.

And again, if you need the audio video content, you can go to MOPS or you can download from our website for your convenience. Thank you. Thank you for your participation.

[Johnny Shen, executive]

Thank you very much.