Amkor Technology, Inc. (AMKR) Earnings Call Transcript & Summary

All right. Welcome. Sorry to read this so many times. But for important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your sales rep. So with that, I'm Joe Moore, Morgan Stanley Semiconductor Research. Very happy to have with me today from Amkor, Kevin Engel, Chief Operating Officer.

Right. Good morning, everybody.

Thank you. So let me just jump right into questions. I mean your historic seasonality has been about 45% of revenue in the first half, 55% in the second half. At the latest earnings call, you talked about a little bit more of a second half skew. Can you just talk to the trends that are driving that view?

Yes. So a couple of things going on there. So first, we would look back to what we have talked about on some of the other earnings calls. We had a socket loss for the iOS ecosystem. We can talk more about that later. But that definitely is giving us a muted first half. And as we go into the second half and start to recover that socket for the next generation as well as hopefully some help from other markets, which we can talk about, that will definitely give us a higher seasonality for that second half.

Okay. What are major customers telling you about their order outlooks for this year? And just maybe give us a contrast with what we've seen in the last couple of years.

Yes. So I think if we go by market, so if we think about communications market, as an example, I'd say that feels about normal. So no major deltas there. When we think about the compute market, again, obviously, with AI, still a lot of bullishness there. So compute, along with PCs, a little bit of moderated strength. So I think the compute market feels a little bit better than the last several years. When we think about automotive, that's really kind of 2 different segments we think about there. We think about the more advanced type applications, which are the ADAS and safety functionality, a little more strength in that market, whereas when we think about the more mainstream products, like MCUs, that type of application space, still a story of inventory correction and when that's going to get consumed. So I think that's an area that automotive in general, we see flattish. And then we think about the consumer market, obviously, that's pretty heavily dependent on what the consumer does from a purchasing perspective. But again, that feels about normal from what we've seen for the past couple of years.

Okay. Helpful. And then probably the most asked question we got after your earnings is also probably the most difficult for you to speak to directly, but you mentioned the socket loss in the Apple ecosystem or the iOS ecosystem. How do you think about that in terms of what milestones should we look at to determine your progress around that? And just any help you can give us to sort of navigate what you're seeing there?

Yes. So I think there's a few pieces. I mean if you look at the cell phone kind of heartbeat or the cycle there, every year, you're going back through next-generation products, in some cases, new packages, in some cases, reusing existing packages, but you have to rebid and continue to stay hungry for each generation of product. If we look at this scenario, a little bit different in that the customer was migrating to a different architecture. And they -- we worked with them basically to talk about Amkor being involved in the next-generation architecture to make sure we had a long pipeline there, and that left us out of the current generation. So in general, we feel very well positioned with that customer. If we look actually at our '23 to '24 revenue, even without socket loss, we grew with that customer. So we still feel pretty good about it.

Okay. All right. That's helpful. You've also talked about a new wearables project that you started ramping last year. Can you give us a sense for the durability of that growth?

Yes. So there -- just like every new product launch, there are cycles there, there's the buildup, and then how much of the sell-through there is. So we still see that as looking pretty strong going into this year. And then the consumer market in general, we're projecting kind of low single-digit growth trajectories. I think it's important to just highlight that Amkor in general plays more in that premium tier wearable space where we can differentiate a little bit more.

Okay. Helpful. Maybe looking at some of the markets that you talked about, starting with compute. Can you kind of clarify your comments on compute growth ticking down a little bit this last earnings? How much of your compute business -- how much of that was related to China and just general sense?

Yes. So a couple of dynamics there. So one, again, was the China dynamic. We had a couple of products that we were working with customers to launch. And as we started seeing some of the new restrictions that came in place towards the end of the year, that put the dynamic with the customers where they put some of those projects on hold or canceled. So that created a dynamic to where some of these ramping products slowed down. The other aspect was the next-generation GPU we were working on. We -- the transition into that next generation was faster than we or the customer had anticipated. So the current generation, we saw that muted. Still have some legs going into this year, but definitely muted compared to what we were anticipating. If we look forward, we definitely see a strong pipeline for additional products ramping. Some of those going into the next-generation technology. If we look at the GPU today that we're working on, that's a 2.5D type technology or similar to CoWoS-S. If we move forward into some of these products that we're launching, they're more RDL -- organic RDL-type products. So those are exciting for us because that's our first launch of that new type of technology.

Great. And I know, again, probably hard for you to be overly precise about the AI GPU part of the business and the advanced packaging there that catalyst like product. But can you just give us a general sense for how important that is for multiple year growth prospects for Amkor?

Yes. So when we think about AI in general, it was very important for further growth. And I think there's a lot of legs left there. When we think about AI and data center in general, not just GPUs, we're talking CPUs, XPUs, switches, retimers. So that whole segment for us is going to continue to grow.

Yes. Okay. And then I guess, can you talk about your top couple of customers in this space? And maybe with your primary customer, can you talk about prospects for future platforms? And then maybe give us an indication of how big the second customer is?

Yes. So let me think about it from a perspective of -- when we talk to that primary customer that everybody is interested in, we're looking at their total product road map and where we participate, we feel, again, across multiple segments, not just the GPU market. I think we feel very good about our pipeline with that customer specifically. When we think about the other customers, again, we look at road maps, and these are from IDMs to fabless companies to OEMs, and the road map for all these customers is pointing towards more advanced packaging in the future. So we definitely feel very well positioned in that space.

Yes. And I guess a lot of those advanced packaging technologies now when you talk about CoWoS-like product or those kind of chip stacks, it takes you into a little bit of a different domain than you've been in. Do you compete with TSMC? Do you help TSMC? Like how should we think about that as a competitive dynamic? I think of your competitive set usually being other test...

Yes. So I think there's a few dynamics there. Obviously, we've had several press releases with them related to some technology alignment as well as partnerships in their U.S. manufacturing. So in general, we definitely view them as a partner and a strategic partner from that perspective. When we think about the technology space, obviously, we're continuing to upscale our technologies. We view TSMC as the technology leader, and we are a fast follower. I think we've seen that with the CoWoS-S type technologies, our 2.5D. Now we're seeing that with their CoWoS-R, which is similar to our SWIFT technology and then going into the future, the silicon bridges. So I think we feel pretty good about that positioning.

Yes. Okay. Automotive, you talked about kind of a mixed environment. We just had NXP here talking about kind of a lack of visibility. That's generally the sense. It's maybe not approaching the worst point, but maybe not visibility when it recovers. Can you talk to that? And can you talk to the business mix that you have, your visibility into that?

Yes. So I think that's similar to what we see. If we talk to that customer base, and the customer base is very broad, I'd say the same thing. We think about the mainstream type packages, MCUs, more legacy-type devices, definitely still a story of inventory. What's -- when is that inventory going to get burned off. Obviously, for us as a service company, first, the customer has to burn off the inventory and then start rebuilding again. So that visibility is still unclear. When we think about the more advanced packages, definitely a slightly different dynamic. I mean we did see a slight drop coming into Q1, but at least that segment, we see a little more legs to growth going into the year.

Okay. And at the latest earnings call, you mentioned a split of your automotive business, roughly 60% ADAS, 40% infotainment, I believe. Can you talk to the growth rates of those 2 businesses and the prospects that you have?

I think that was 60% advanced and 40% mainstream or maybe vice versa, and not just ADAS. So again, when we think about that advanced segment, that's ADAS plus infotainment, connected car, digitization, that type of application. And then again, the mainstream would be more of the MCU market. So that -- if you look at that forward, we do anticipate that, that advanced segment will continue to grow at a faster rate than the mainstream business.

Yes. And can you talk about the advanced package types that are being...

So again, I think you have to look -- one thing I always try to explain a little bit is when the industry talks about advanced packaging now, it's kind of been skewed a little bit towards TSMC advanced packaging and the CoWoS-type technologies. Traditionally, Amkor has always talked about advanced packaging as flip chip, flip chip CSP, flip chip BGA, other type products, along with these applications that TSMC is talking about. So we look in that connected car space. It's things like flip chip BGA, MCMs, flip chip CSP, a little bit of SiP, those types of package types.

Okay. Helpful. So I guess, thinking about CapEx and maybe the strategy there, we're sort of moving from an environment where -- and I don't want to get too political, but we're moving from a subsidy environment to more of like tariff concern kind of environment. But it seems like the goal is still the same, that stuff -- there's a desire to have things manufactured in the region where it's being consumed in the U.S. You guys are investing in Arizona, which I have kind of not thought of the U.S. as being necessarily the cost structure that you look for in kind of packaging. But can you talk about the thought behind that facility, the benefits that you guys see from it and maybe in navigating uncertainty in the political environment? It still seems like a pretty big benefit to have that capacity here.

Yes. I mean, so from a general perspective, if we look at the supply chain and the logistics. So there's more front-end fabs here, there needs to be more packaging in the U.S. as well. So I think that's the first baseline. Obviously, the cost here between labor and other things is elevated. So that's obviously something we're working very closely with. You can imagine customers don't want to pay more. In some cases, maybe there's a little bit of opportunity there, but you're trying to balance a cost structure versus logistics and a supply chain concern. So if we look at the way that that's transforming, how do we manage that cost delta, subsidies help. That can be one way. Obviously, tariffs can basically do the same thing. So I think either one of those mechanisms can help level set some of that. We're focused on, obviously, a lot of automation. That helps to manage the labor cost as well. And then the other aspect would be to look at how we can load the factory to higher utilization consistently. We talked about -- we already talked a little bit about our seasonality first half versus second half. If we're able to have a consistent high utilization, that also helps the cost structure. So I think in general, that, along with, again, now the recent announcement from TSMC to go even bigger in the U.S., obviously, other fabs in the U.S. So we have -- we feel there's enough of a scale from a front-end fab perspective to drive that packaging in the U.S.

When you said customers don't want to pay more, which I guess, on the one hand, is obvious, but is there a trade-off between wanting supply chain certainty, wanting things to be manufactured in a region where at an OEM level, the customers feel comfortable, is there a willingness to pay for that? Or is there just...

I'd say, yes. I mean there's -- it's always a put and take, right? What -- a lot of customers will look at that as a blended cost, what percentage do you do in the U.S. with a higher price and how they blend that and what's the right ratio based on what their end customers can afford. So I think that dynamic will continue to shape up.

And still a pretty -- it's an exciting announcement, but still a pretty small part of your CapEx. I think you guided for CapEx of $850 million this year, of which you said 5% to 10% is going to that facility. Can you talk about what that looks like going forward? Do you -- does the capital intensity go up as you...

Yes. So first, let's level set that our normal capital intensity is in the low teens. So I think that's kind of the benchmark. And then if we look at ramping up and building out our Vietnam facility, we were able to maintain that low teens capital intensity. So for the U.S., you're right, so this year, relatively light, 5% to 10% of that $850 million. That's breaking ground, doing some other things. If we look into '26 and '27, that spend will go up. And depending on what the overall market does, if we look to '26 as a growth year in general or a recovery year, that can be a dynamic to where it could push our capital intensity a little bit higher going into '26 and '27.

Okay. All right. And you mentioned Vietnam, you're up and running there now. Can you give us a sense for where you are in that ramp and how much revenue you can ultimately support out of that facility?

Yes. So it's a big facility. I mean we have 1 module of 4 completed, and now we're filling that as we ramp capacity there. So we're ramping our memory devices as well as some SiP products. So that ramp is going very, very well. The yields are reaching or exceeding our Korea facility. So I think we feel very good about that progress. If you think from a scale perspective, again, over time, it will be one of our largest facilities from a footprint perspective and full build-out. So yes, a lot of revenue growth potential there in the future.

Yes. Okay. Great. Test, can you talk about your test business? How are you thinking about tester purchases in 2025? And it seems like there's some availability issues with certain testers. Just can you talk generally to the prospects you have...

Yes. I mean lead times are always something we watch and not just with testers, but with other assets depending on the market dynamics. So I think that's one check box. When we look at the test business in general, there's a few things that we try to do. We're very focused on our turnkey flow. So if we're doing, as an example, assembly, but no test, we're constantly looking at how do we bring value to the customer either through logistics or reutilization of existing assets or other ways. Same thing if we have just a bumping product, but we want to go after probes. So we're really trying to make sure we capture much of that value for every product that's out there. And then when we look at that investment, there's a good bit of investment in the test area. If we look at that $850 million of capital intensity or capital spend this year, about 70% of that is for assets, for equipment. And if we look at that equipment, the 3 areas that are the primary spend are for advanced packaging, SiP for specific devices and then test would be third. So it's definitely a significant area of investment. And that's mostly related to some of the next-generation platforms with the key 2 suppliers there.

Okay. Okay. Helpful. Maybe talk to financials a little bit. Your gross margin in the second half was a little lighter than what people were expecting. You had material content on the SiP underutilization in Vietnam. Can you just walk us through the puts and takes there? And you've sort of talked about getting back to 20% gross margin at some point. What does it take to get to that?

Okay. Yes. So I think first, it's important to point out that for '24, our margins were slightly better than '23, and there's a couple of puts and takes related to that. Obviously, cost optimization is a key area, making sure we're controlling the cost structure. The other puts and takes are related to -- we had some FX tailwinds as well as you're probably all aware, we had a modification of our depreciation cycle, which also is a tailwind, and that helped to offset some of the ramp costs related to Vietnam. So I think we feel pretty good about that. Obviously, with the first half being muted, as we talked about, but we expect for the full year to be flat to slightly up. So that means our second half will be much stronger, as we mentioned earlier. So if you look at the second half compared to some of our previous quarters, as an example, like Q3, Q4 '22, those were high revenue, high utilization quarters where we were up in that high teens-type margin levels. And I think what we can model is that for second half of the year, we're anticipating those types of margin levels.

Okay. Okay. Can you talk about the business mix? And I think you've mentioned some of the legacy businesses can actually have higher gross margin at times. Just how do we think about business mix is impacting that?

Yes. I think, in general, I would just think about that as it's all utilization. We have -- if we look at that automotive space, as an example, we have some factories that are very heavily concentrated in automotive. Today, they're very underutilized. And as that business comes back, that will definitely help the margin in general. Across any market or any business, there's different margins here and there. But in general, if we're able to fill those assets, that really benefits us.

Yes. And I know -- I mean, you guys have made big investments into advanced packaging and the differentiation that, that gives you. What are those negotiations like? Are you able to, you think, extract more gross margin over time from that is from the fact that you have more capability that maybe is a little bit more differentiated.

Yes. I mean I think there's a few dynamics there. Number one, the capital intensity is higher. So you really need to look at that from a margin perspective. But in general, you're starting to whittle down the supply chain options. When you think about the Tier 1 OSAT space, there's really only a couple of companies, maybe 3, that can offer these types of advanced packages. So yes, you can get some benefit out of that. So I think that's a key driver is what that space looks like compared to when you're dealing with Tier 2, Tier 3 supply chains with a lot more optionality.

Yes. Okay. Let me stop and see if we have questions from the audience.

I just wanted to ask about silicon photonics, co-package optics, just from like a customer road map perspective, just at a high level, what you're seeing there?

Yes. So a few things. I'd say pluggables are already out there. So I think we're working with in the pluggable market. When we think about CPO, still a relatively fragmented type market with a lot of different offerings, a lot of different package structures. So we'll see how that consolidates over time. But I'd say it's definitely an area that we see a lot of interest from our customers. We're looking at how we can support that overall different package structure. And to me, it's a matter of timing. We've been talking about CPO for a long time. And I think we're -- between the power and the data center and other things, I think there's definitely -- that's an area that will grow pretty soon.

It seems like this AI is an accelerant to all these things that we've been talking about for a long time, but it's coming. It seems like it's coming much [ faster ].

I mean, to me, it's power and electricity in general across everything is a key driver.

Yes. Makes sense. We had a question here.

Can you talk about your opportunity for HBM and memory? What are the kind of easy aspects of kind of getting into the space? And what are the challenges that you face for that?

Yes. So for memory in general, we participate in NAND and some other areas. For HBM, obviously, we're taking packaged HBM and putting them into packages. At this point, we're -- we continue to evaluate whether or not it makes sense to go into actual NAND or go into the HBM space, working with some of the supplier, the customers in that dynamic. But I think that's still a ways out for Amkor.

I guess a similar question. But maybe stepping back to your Arizona facility. I think in the past, you said it's a partnership potentially with TSMC. Now recently, they announced 2 of their own advanced packaging facilities. So how do you see that dynamic play out?

So a couple of things. So number one, this facility was never envisioned to be captive to TSMC and/or Apple, and we had press releases with both definitely important key customers. When we look at TSMC's trajectory, obviously, as part of the CHIPS Act, they talked about 3 facilities, first facility done, second facility wrapping up. That will progress forward. The recent announcement was another 3 front-end fabs. 6 fabs is a tremendous amount of scale. So for us, we view that as just continues to show that front-end capability and back-end capability in the U.S. needs to grow significantly. So their announcement doesn't change our direction today at all.

I get that. And then I guess a follow-up. So in terms of their type of packages, you said there's a bit of a differentiator between what OSAT business used to be advanced packaging and what they put in the market. And I think the pushback has been that scale is important for OSATs to be able to drive the business to get the margins. So is that changing now? You're thinking around, for example, upgrading your cleaning facilities. Hybrid bonding is a term that they often use. Is that something that is becoming interesting now because of these huge investments?

Again, let me -- I think your comment on scale is right. The application spaces for the most advanced, the volume levels are typically relatively low, and they're obviously more expensive. So I think what will happen over time with things like hybrid bonding, CoWoS-L, those types of packages as they proliferate into the broader market, broader customer base, then that's an area for the OSATs to get into those markets and participate.

No. I'm going to wrap up there. Thank you so much.

All right. Thank you, everybody. Thanks.