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

Sure. So we're very excited about our partnership with NVIDIA. Again, this is just a testament to customers really wanting to shore up their supply. As Doug said, the imbalance is unprecedented. So this multiyear strategic partnership has really aligned a technology road map and also enabling our U.S. expansion. So we have not disclosed specific products. It also is not indicative of a revenue stream. It's actually the partnership in order to provide that supply security. We have shared that as the services are provided to NVIDIA, that will then be applied to their services over time. So depending upon the production levels, the timing, the speed, that arrangement could be anywhere from 5 to 10 years. But it really is a testament and the confidence to Amkor and their confidence in us being able to develop this technology. We've already proven that technology. We have engagements in production today and expanding in this critical U.S. market to enable an end-to-end supply chain is what we're really achieving together.

Yes. And then I mean, help frame for us, like how much bigger do you think this could potentially be than the $1.5 billion I realized that -- still very early days. We're still away from revenue, but 5 to 10 years. It's a big time frame, might be dimensionalized for this -- that for us and help us understand sort of relative to the Arizona capacity expansion. I'm sure it had some impact.

Yes. So it's really showing that they're an anchor customer and that they are committed to us expanding in the U.S. So that $1.5 billion is really going towards the $12 billion of expansion. That's how I would frame it. It is not necessarily a revenue stream or what it could potentially be. But by having that anchor customer and having that partnership to develop the next-generation technologies, that's what's going to really grow Amkor. And as Doug mentioned, we'll keep our eyes on whether further expansion may be needed in the U.S. .

Maybe I can just add another piece to that. So the technologies that we're going to be using for a majority of our customers in Arizona are new technologies, they're not licensed technologies. They're technologies that we have developed in our Korea location that we're already ramping to HBM volume. So it's a transfer. So it will shorten the cycle to be able to bring our customers up in Arizona because we're essentially transferring no technology, no processes, no yields. And it's also -- I know we talked about the $1.5 billion from NVIDIA, but it's really not a single customer discussion. We have multiple customers with known technologies. They are securing allocation. And as current products ramp in Korea, those products will come up in the United States and Arizona. New devices, next-generation will come in after that. And then as those ramp, they get true qualification to ramp for production. That's where the revenue really starts coming in the U.S. But because we're leveraging existing technologies and existing processes with existing relationships with these large customers, we expect the United States or Arizona to come up much faster than it would be if it was a external transfer of technology or a brand-new build as a greenfield that we're starting from scratch.

Okay. Let's talk about the competitive landscape for a second. In compute specifically, TSMC was very early with their CoWoS product. You've been a fast follower with your 2.5D packaging. You've seen strong engagement with high density fan-out. Now Intel seems to be making some good progress with their EMT product that enables the larger panel sizes. So how should investors be thinking about Amkor's market position? And where do you think you have a particular right to win?

So I think it's safe to say that the compute market, in particular, is a fast-growing market. And the ability for labs, foundries, whoever is doing advanced packaging, for example, cannot keep up with the level of demand. So we believe that the market opportunity is growing and it will continue to grow. As I mentioned earlier, we've been doing this for a long time. We have known technologies, and we believe we're well positioned in that when a customer has very expensive silicon, limited memory supply, limited substrates, they need to go to somebody -- to a company that can produce high output yield. And otherwise, they're not going to be able to make their output demand. So we feel that we're positioned quite well in this area. Certainly, there's new technologies that are being looked at. We continue to path find and look at those technologies also. And we will be in position to support the market when the market is ready for an HBM solution. It really -- it would be foolish for the market not to look at different options of technology because -- again, because of the constraints. But at the end of the day, it has to be manufacturable. It has to be high yield, and it has to have low-cost options. And this is somewhere we're -- we've been involved with and we continue to be involved with, with the largest companies in the world.

Got it. And maybe just finally touch on the automotive end market. Help us understand where you see that business trending given some of the stronger trends you saw there in Q2 and you guided for Q3?

Yes. So automotive is a nice story. We saw really quarter after quarter of reduction in automotive. And certainly, there was supply constraints or not constraints, but there's actually oversupply in the market. But automotive is our second largest market of growth for Amkor behind compute. So we're seeing a number of different reasons. One, as you look at automotive, really the complexity of the packaging needed for automotive is increasing. So ADAS and infotainment, electrification, those are all driving advanced packaging solutions. So that puts us in a very good position. Also, our wire bond mainstream business that we've had supporting automotive for a long time, that is also recovering. So we're seeing a good momentum on our automotive. We expect that to continue. And really, it comes down to automotive is a different market than some of the others in that the device lifespan is much, much longer. So it's harder to qualify. It's harder to make sure you secure sockets. But once you're qualified, once you have reliable partners, those parts could last decades. And so we see a lot of momentum both on our legacy business, but also on the new technologies that are driving more advanced compute within an automobile. So that's a very nice trend we're seeing on automotive.

Great. Now I wanted to kind of shift back to the sort of factory footprint side of things for a second. You sort of precogently outlined the rationale for the Arizona build-out, proximity to TSMC, but there's other customers there, too, including Samsung and some others. Maybe talk about beyond the U.S. manufacturing, how important is proximity to front-end wafer manufacturing as part of the strategy and specifically in Arizona?

Yes, I can start on that, and then if you would like to add, Doug. So as you know, we are in very close proximity to TSMC in Arizona. When we were looking for the right footprint, we didn't only look in Arizona. We did look nationwide. But ultimately, it provided all the right ingredients with respect to land, infrastructure, community, workforce, et cetera. And then I would say the cherry on top is we're within 6 miles or so of TSMC. . So that does provide a lot of value with respect to feedback loop, cycle time, logistics, et cetera. So there is a significant amount of value in proximity. But as Doug mentioned, we are not exclusive to one foundry. We can support multiple foundries and other foundries in the U.S. are able to use Amkor and our customers want options as far as how they're going to manage their supply chain. So I would say it's important. It does provide value, and we're going to be able to support all front-end partners at our Arizona location. Anything to add?

No, it's a good point. I mean if you just look at silicon, a foundry makes silicon, ships it. We process the silicon, and we put it in a part in a device. And so that structure, it doesn't maybe matter as much. But when you look at processing partners and being able to take silicon, do some processing, maybe send it back, have them do some processing, send it back, do additional processing. The close proximity makes a significant difference. Going into a truck 5 miles down the road versus going on a plane across the world, that could benefit the supply chain for sure.

Okay. Great. And then in terms of what else is happening in your factory network, you talked about the rebalancing of making space in Korea basically moved from SiP to Vietnam, as you said before. What other changes should -- I mean, I think that caused a little bit of questions from investors this past quarter. So going forward, what other changes should we expect as you rebalance the factory network.

Yes. So it's all about centralizing our products, right? So we develop new technology in Korea, and then we want to make sure that we have economies of scale for efficiency, profitability, et cetera. So that's really what drove the strategic decision that has been planned for -- since the beginning of Vietnam to centralize our SiP business. And it takes time. Our customers need to move those products. That's what's creating a little bit of noise in this transition during Q3. As far as the rest of our factory network, we are continuing to expand outside of Arizona and Korea. We do have programs that are being expanded in Taiwan, which is a very important market for advanced packaging, but that's within the footprint. We also are expanding our advanced packaging capacity in Portugal. And Europe is also a very important region where customers are requiring regionalized support. And so that trend of geographic flexibility is we're seeing everywhere, not only in the U.S., but in Europe, in Asia. So all of those things are what's contributing to our decision. But no other major, I would say, changes in our footprint at this time.

Got it. Okay. Then at your Investor Day, you laid out financial targets for 2030, pretty ambitious targets, revenue over $11 billion, gross margin, 22%, earnings per share, $5. So maybe walk us through some of the elements and how you expect that revenue growth story, in particular, to sort of unfold off the larger base business you've got this year?

Yes. So that growth up to the $11 billion plus, there's really 2 contributors to that. The primary contributor is what we're seeing as far as the growth in AI and our compute business. That is, as we mentioned, over 30% in '26. We see similar growth into '27. That is the primary driver for that growth up to $11 billion. The second factor will be bringing on the Arizona facility. And we have shared that time line as far as when we'll begin production in '28 and how that will scale and expecting that to be at full capacity by 2030, that will add about $1 billion to that growth for our revenue. The announcement for Phase 2, the timing for Phase 2. We anticipate breaking ground by the end of next year. That will then be a 2-year construction period. So we'll begin producing in 2030. So I would say there's some upside to 2030, but it's all dependent on the speed of the construction and the production ramps of our customers.

Got it. Okay. Now at the same time, on the gross margin side, you've talked about some of the dynamics of dilution as you ramp Arizona. I think you said you're also absorbing the start-up costs into OpEx initially until products get qualified, then they hit gross margin. So maybe help us understand the moving parts between OpEx and gross margins for '27 and '28.

Sure. So this is the same framework that we had when we opened our flagship K5 facility as well as Vietnam. When construction completes and depreciation begins on the building as well as other preparations for workforce, and production, those costs are in operating expenses. And so we would anticipate that we would start to see that -- those prep costs in '27 after we finish our construction of Phase 1. We have shared that, that could have a dilution factor of 1% to 2% on operating income margin. Once we begin to qualify a program, those costs then move into cost of goods sold. So they'll still be in our op income margin, but they'll transition probably into '28 into COGS, and therefore, you'll see some of that dilution in gross margin. But that's going to improve as we scale to build the scale of the facility, and then we will have an accretive performance in 2030 once we pass that midpoint, which we believe the transition will be in '29 and then into 2030 at full scale.

2030 is when the gross margins for Arizona get accretive.

Yes. They'll start to be accretive exiting '29. But for the full year, it will most likely be neutral. .

Excellent. Then finally, just thinking about some of the free cash flow and CapEx dynamics investors should be watching for over the next 18 months or so. Sort of remind us of how you're thinking about those issues, especially in the context of Phase 2. And then how you think about buybacks and debt paydown over the, say, the next few years?

Sure. So we did announce that we have increased our investment in Arizona from $7 billion to $12 billion. that's really centered around refining our estimates around Phase 1 as well as the scale of what we're planning to build for Phase 2. So that $12 billion is a multiyear investment. Obviously, we've given our '26 guide. We're not going to be giving '27 guide or whatnot. But to give you a flavor for how to think about that investment in the next '26, '27 is the finishing the construction of Phase 1. Beginning in '28, you're going to have a little bit of a stack. You're going to have the equipment coming on for Phase 1, but we're also going to have the construction of Phase 2. So I would see that the years '28 and '29 will be peak CapEx outlay of our investment in Arizona. And then when you get to '29 and '30, that will just be the equipment for Phase 2. So to try to bring some magnitude to that, for the $12 billion, we would anticipate about 10% to 15% of that in '26 and '27. That will step up to about 20% to 25% in '28 and '29. And then when you get into 2030 and 2031, that will step back down to 10% to 15%. So it's a bell-shaped curve is how you would want to think about our investment in Arizona. So over time, with that investment, there will be some pressure on free cash flow. But we do see that as Phase 1 starts to ramp and becomes fully utilized in 2030, that's really going to provide an earnings engine that will help support the onboarding of Phase 2 as well. You asked about buybacks and priorities. We did announce a buyback program. We expect to modestly execute on that really to offset dilution to stock-based compensation. And then as I think about debt priorities, our current debt levels are reasonable. Our leverage is reasonable. We will continue to have incremental debt capacity as our business expands, and we will keep our eye on our long-term goal of 1.5x or below for leverage.

Excellent. I think we're -- with that, we're basically out of time. But thank you very much, Megan and Doug for being here. We appreciate it.

Thank you.