GLOBALFOUNDRIES Inc. (GFS) Earnings Call Transcript & Summary

Good morning, and thank you for attending JPMorgan's 20th Annual Technology Investor Virtual Forum here at the Consumer Electronics Show. My name is Harlan Sur. I'm the semiconductor and semiconductor capital equipment analyst here for the firm. Very pleased to have the team from GlobalFoundries here with us today: Tom Caulfield, Chief Executive Officer; David Reeder, Chief Financial Officer; and Sukhi Nagesh, Vice President of Corporate Development and Investor Relations. For those of you that don't know, GlobalFoundries is the third largest semiconductor foundry in the world and leader in specialty and pervasive manufacturing technologies targeted at segments such as analog, power management, RF, wireless and wired networking connectivity, targeting comm infrastructure, mobile, IoT, automotive and industrial end markets. And so gentlemen, thank you very much for joining me this morning. I'll kick it off with the first few questions. And then for those clients that have questions, feel free to chime in on the digital conference book. So good morning and happy new year to the team.

Hey. Thanks for having us, Harlan, and happy new year to you, your family and everybody calling in today.

Happy new year, Harlan. Glad to be here.

Yes, absolutely. Tom, first question for you. I mean demand across almost every end market this year looks strong. Given the supply-demand gap and your discussions with customers, and in some cases, your customers' customers, when do you expect supply and demand to start to normalize for the semiconductor industry? Is it second half of this year? Is it 2023? And what are the end markets that you see that are going to be particularly strong this year?

There's a lot in there. So let's talk tactically and then a little bit more strategically. I don't see any relief in 2022 in the size of the gap between supply and demand and capacity. Investments are in flight, but we became so far out of whack between supply and demand, it's just -- there's a time element to this. But every time I'm in these conversations, it always feels like you think it's digital. It's either broken or fixed. It's either a gap or not gap. To me, it's kind of a continuum. It is a question of how far off is supply-demand. Is it 10% mismatch? Is it 30%? For what I see from our customers, and I think we have a pretty good vantage point, we serve a broad set of applications, a broad set of customers. It's hard to imagine over the next 2 years, things getting where, oh, we don't talk about supply issues anymore. I think it gets better. I think capacity comes on. People are making contingency plans. But just from every conversation I have, I'm still in escalation calls, meaning customers calling, "Can you help? What can you do more? Can you squeeze more capacity out for us? If you can't, can you prioritize these wafers over these? Can I anticipate in some more expedite just so I can kind of bring some of that into the first half from second half?" Still having an outsized number of those types of calls, which gives me an indication that our customers are seeing very strong demand. And again, I think if this is an industry that we all believe doubles in the next 8 to 10 years, it's just not enough to get caught up to today's demand. We have to continue to invest in a very thoughtful and pragmatic way to make sure we capture that growth going forward. David, anything you'd add to that?

No, Tom, I think that was well said. I mean, I guess, if I was going to add anything, it'd just be a proof point. I think a proof point to where we see the supply-demand imbalance was shown by the amendment to the AMD wafer agreement, wafer sales agreement, right? They added another year to what was already a 3-year agreement. And that, I would say, is pretty common across our customer base where our customer base is continuing to make prepayments and continuing to secure capacity out in time.

So from a maybe more near-term tactical -- and we've been getting this question from investors with some of our other companies. Despite your views, there are some concerns around the potential for overordering and maybe some near-term easing. But from where you sit -- I know you talked about escalations. But has the team seen any other sort of indicators, pushouts, cancellations, rescheduling of orders of that sort?

Those are the things we look for. We look at [indiscernible] for pushout or cancellation, rescheduling. The only changes we see is customers saying, "Well, given that mix or given what you can supply, I want to remix." Maybe it's richer for them. It fixes near-term needs, but nothing that is any signal that says, "Well, we have enough of this, we want to hold off." By the way, I have to tell you, maybe this one thing is I could use a little less pressure on the demand side. These are gut-wrenching painful calls to not be able to give customers everything they want when they need it.

On your long-term agreements -- we'll get into AMD in a little bit, but the team has mentioned that customer prepayments on the last earnings call increased to $3 billion, right, from $2 billion, $2.5 billion when you guys were on the road, reflecting customer willingness to continue to commit to future capacity. Additionally, your long-term agreements were well north of $20 billion with 25 customers, as you mentioned, on the call. Obviously, you recently extended your LTA with AMD by $0.5 billion. I'm assuming that you're seeing more customers that are extending their LTA with you. But I don't know if you can just give us an update, maybe total amount of LTA value through 2025 and maybe total number of LTA signed.

David, why don't you take the first part of that?

Sure. Yes. Look, I don't think anything has really materially changed from what we told you at our earnings call. Obviously, we continue to work with our customers on those long-term agreements. We'll update you once a year, most likely at an Analyst Day on how we're progressing. I mean we'll provide color periodically like we did in our last earnings call, where we said we've gone up more than $0.5 billion from our road show, which was a month prior. But I think at this point, suffice it to say that those 8-K amendments that you see coming out, those will be proof points for you. And then, of course, if other things that are really material come out, then you'll probably hear us comment on those as we go forward.

Yes. And let me just segue a little bit on that. Every one of these contracts is a very important piece of it, and it's about giving visibility even beyond the length of these contracts. And it's essentially in the first year anniversary of a long-term contract to the parties to sit down and understand if they want an extension to add another year. And over the first 6 months of this year, first 7 or 8 months of this year, a lot of these contracts will hit that trigger and we'll be able to understand where our customers are and with their outlook and continuing to want to take reservation against that capacity. And so you're seeing some of that happening even before the 1-year anniversary of the contract, as David pointed out with the AMD contract. So those will be things that when we get towards our analysts in the second half of this year, we'll be able to give an update of who's come in and looking to do some extensions on their contracts.

I thought the recent extension of the LTA was AMD from $1.6 billion through 2024, now extended to $2.1 billion through 2025 or an incremental $0.5 billion. I mean, I think, this is a great example of how a leading-edge semiconductor company still has a big portion of their mix on pervasive or specialty semiconductor manufacturing, right, whether it's high-speed I/O tiles for their heterogeneous tile-based compute architectures. And in the case of AMD, in addition to that, as with some of the other leading-edge players, they also have a large embedded processor business, right, targeting auto, industrial, retail markets, again, that don't require leading-edge compute solutions. Do you guys believe that these leading-edge companies view the embedded markets as just as much of a driver for their businesses as their core compute focus and therefore, a continued strong growth opportunity for the GlobalFoundries team?

Yes. There's a lot to unpack there. I'll start and pass it over to David. So one, I think you're right. I think if you're a processor-centric business, whether it's graphics or CPU or both, you want to make sure you have great products in that and then you'll look to open the aperture to how do you put companion chips along with that, that complement the overall function. When you're talking about a little bit more specific, and I think there's really 2 key messages in that. This is this idea of, hey, my processors today are 2-chip solutions. And why is that? I think it's the proof point on how difficult it is to leverage single-digit nanometer other than when you need it for the absolute compute function because every generation of technology, you always integrated the whole chip in the next generation of technology. And here's one we're starting to create a 2-chip solution. You don't do that for any other reason other than economics are forcing that. And so the economics is better and lower cost without suffering performance to take I/O analog off the CPU processor and just let the computation intensive part of the chip staying single-digit nano. And what it says, it means when we start to go below 10-nanometer, a lot of applications are not going to find the value of that because the cost per transistor goes up and they only want to put technologies in that, that require that absolute performance. And so, well, you'll get migration of applications down the nodes in our industry, I think it kind of stalls at around 12 to 10-nanometer when you're not getting the value proposition. And there would be no reason in the world to have tiles or chiplets if there wasn't an economic reason to have a 2-chip solution when you could have had a single-chip solution. So I think that's what you're having there. You have two things, companies that are traditionally processor-centric want to broaden their portfolio and participate more in the solution. And the second thing is they'll need a broader range of technologies, not just the latest node if they're going to be competitive, not only in their processors, but in addressing those markets. David, anything?

No, well said, Tom. Well said. I wouldn't add anything.

We're seeing an interesting dynamic with some of the partnerships that you've formed recently, right, partnerships with Ford and BMW, right? These are your customers' customers. They're coming to you to work directly with you to, number one, secure supply for their chip suppliers. But some of them have also talked about potentially doing some of their own chips in-house as well. And you already have a lot of other end customers that you do business with directly. Cisco, Microsoft and Amazon are a few direct customers that I can think about as well. Do you see that more -- whether it's cloud titans, auto OEMs, consumer device OEMs coming to the table to talk to you? And how many of these discussions are you having? And how do you see this benefiting GlobalFoundries and the industry sort of longer term?

Yes. I'll take one big step back. I think the fact that at the end of the day is the idea of silicon awareness across the world and in particular industry has really risen to the level maybe it should have always been. Capacity and manufacturing was kind of not taking for granted, but it was always there. And all of a sudden, you find there's not enough of it. And so if you're a business is suffering from lack of capacity, you will take more of an interest in, okay, where does it take place? How do I make sure that -- shame on me, I won't let this happen to me again. And so the relationships we're seeing are evolving. And you pointed out the contract with the partnership with Ford, they really take the form of, I'd say, 3 elements. The first one is, "Hey, how do we get better visibility and see if we can work together to help this near-term issue in the next 6, 12, maybe 18 months. The second element is about how important differentiation is in silicon if we're going to become more silicon aware and GF and others, we want to make sure that what you're doing in your technology, the significant amount of money we spend on R&D is being targeted to areas that differentiate for us. So having a closer relationship on making sure our technology alignment meets the features and needs that our customers want, in particular, the automotive. And then the last thing is around, I would call capacity reservation. Let's make sure, whether we want to do our own designs, use the capacity or direct our supply chain to use the capacity reserve. Let's make sure in the future, we've allocated on the right nodes with the right company with the right features. We've allocated and reserved capacity for our needs. We don't have to design to use it. We can direct our supply chain to it. And those relationships can be as simple as capacity reservation long-term agreement. It could be joint partnerships and manufacturing to create capacity. It could take a broad range. But I think that's the general trend we're seeing and what we're doing in our business is to go tactically how can we help, how do we align technology road maps and then how do these end users reserve capacity to know that they'll be allocated in the future that their volumes will be addressed in the future.

On the supply side, the team is ramping its Dresden and Singapore and Malta fabs through 2023. How should we -- how should investors think about what programs are driving more of the capacity adds in 2022 and then in 2023? And how should we expect or think of supply growth as we move through this year and next year?

Hey, David, why don't you start with this one?

Sure. Well, Harlan, as you mentioned, we are growing kind of all of our major sites in the U.S., in Europe, in Germany and in Singapore. And so let me just kind of unpack those, maybe starting with Germany, with Dresden first. Dresden, we had a big fixed footprint that hadn't been fully tooled. And so what you're seeing from 2020, 2021 and 2022 is you're seeing us tool that footprint. And as we tool that footprint, we're growing from about 300,000 wafers to roughly 825,000 wafers on an annual basis. And so that's what's occurring there in Dresden. And of course, we're picking up the benefits of scale and absorption as you would expect as you build out a fixed cost footprint. We're doing something similar to a lesser extent in Malta here in the U.S. And then, of course, we're building the new fab there in Singapore on our Singapore campus. So we're expanding all of those sites. So what's driving the expansion of those sites? Well, what's underpinning that is the growth that we're seeing across all of our end markets. So all of our end markets are growing in a pretty material way, with really the exception of our personal compute market, which we were planning -- it was planned for that market to become smaller and then stabilize. And so it will decline again year-over-year '21 to '22. And then it will kind of stabilize from that point forward. So we expected that market to decline and then we expected the automotive market, the IoT market, the comms and infrastructure market and of course, the smart mobile market. We expected all those markets to not only overcome the decline from personal compute but to fuel the growth that we're seeing in 2022 and beyond.

On that note, the team is in the midst of a CapEx investment cycle, '22 through '23, $6.5 billion of aggregated CapEx to increase the manufacturing capacity. My concern would be that because we cover the semi-cap equipment companies, given the extended lead times and supply shortages for some of your semiconductor equipment vendors, have you guys seen any delays or pushouts in the time line of capital equipment shipments or deliveries?

Look, nothing material. I think one of the advantages we had is we started to plan this capital deployment in fourth quarter 2020. So we got our orders in early, so we're kind of first in line. Now that doesn't mean that our suppliers bottleneck a particular component or not, and we move schedules around and maybe they pull one tool in, another one comes out, but nothing yet that is a significant impact. And by the way, that also includes the recent news out of Berlin about ASML's factory fire. The preliminary things we're hearing is the kinds of parts that are produced in that 1 of 8 buildings where the fire took place are not relevant to the equipment. We either use spare parts for or use for new builds.

Good to hear. Good to hear. From a technology perspective, one of Global's differentiators is the portfolio of compound semiconductors and silicon photonics. In fact, you've got a strong position in high-speed silicon germanium devices. You recently added GaN technology for power devices, and you've got a strong position in silicon photonics. On the compound semiconductor side, you're already shipping high-speed silicon germanium, obviously. Help us understand when you're going to start offering GaN-based technology to your power customers. Longer term, any plans for adding new capabilities like gallium arsenide or indium phosphide?

So let me do a little bit on silicon photonics because I think it's a really important market, and it really plays to our strength, and then I'll get to the GaN stuff. We -- in the 2023, 2025 time frame, we plan to start having production of -- a significant production of the silicon photonic. Now this is a segment that requires a high degree of, call it, customer enablement because customers all have different designs and architectures. So partnering and actively collaborating with these partners for GPUs or switch companies is an absolute requirement. And so for us, it all starts with enablement. And we start with probably one of the best versions of an electro-optical PDK, physical design kit. It allows our customers to simulate either monolithic designs, which are technology offers or multichip designs where they can actually design in our technology and integrate their own components and then be able to model and simulate the performance. This really allows for the quick turnaround time of designs and not have to build hardware test and build hardware. And so we offer the whole monolithic solution where we have RF capability, CMOS and photonics in one integrated solution. We also provide the opportunity for customers to mix and match different components where they could bring -- we could bring in a PIC and they want to bring in some CMOS. But importantly is that we've designed the technology so that the interconnect, the copper pillars, the copper pads are integrated so that it could bring high-speed RF and the photonics can all be integrated for our customers. We've enabled this technology platform, what we call passive alignment of the fiber test. What does that mean? It means you don't have to -- for every component aligned to fiber, we have where it actively aligns, which means it's a highly manufacturing process for the assembly of these very complicated chips. And then lastly, our photonics road map doesn't end to where we are today. We continue to build on it. We have in our pipeline, 100-gigabyte device technologies will be integrating heterogeneous integration of packaging or others called [indiscernible] and TSVs and things of that nature that are all in our pipeline. What I love about it is all being designed custom in sense for our customers and integrated with their technology road maps. And then on the GaN side that you spoke about is I would say we have 2 segments that we're using GaN 4 and 3 applications. The first is on the mobility side for RF, and it starts with the power amplifier for our infrastructures. And that's our partnership with Raytheon, and we expect production of that in 2025. Now in the later part of the decade, we'll be taking the same GaN technology and building it on our road map for what we do today in RF SOI, when it needs to switch over to GaN for the switching function in front-end modules, that's the build on our that technology road map. And then the other application is for GaN using it in [indiscernible] for fast switching power converters. And again, that's in 2025. So it's well and complementary to our SiGe technology. It aligns with the market sectors we want to play for both in mobility and RF and for power management. With regards to GaAs and indium phosphide, we have no plans to do any of that at this point.

Okay. Well, no, I appreciate that. And I appreciate the update on your silicon photonics. We wrote a report not that long ago. I think our view is that the silicon photonics opportunity, and you guys are -- I think we believe that you guys are partnered up with 2 guys that are leading the market on silicon photonics, but that's looking to be about a $1 billion market opportunity for your 2 lead customers in the 2024, 2025 time frame. So hopefully, that should trickle down and be a good book of business for the Global team on a go-forward basis. On the financials, for the first half of this year -- first half of last year, you drove 12% gross margins. On the December quarter guide, you're targeting 19.5%. So full year, about 15.6%, which is about 50, 60 basis points better than what we were modeling, both for Q4 and the full year. Mix improvement impact wasn't expected to really kick in until this year. So what's driving -- what drove the better -- what is driving the better gross margin profile in the second half of last year? And how are you seeing the trajectory of gross margin improvements through the next couple of years, especially as you have great visibility on volumes, pricing, cost downs and so on?

Well, look, I think I'll start with where you finished, which is we have great visibility in our business, right? We spoke briefly at the beginning about those LTAs, more than $20 billion of LTAs, more than $3 billion of customer prepayments. And so that visibility then gives us the confidence to then not only do the capacity expansions that we've talked about, but also have a very predictable and a methodical way to grow our gross margins over time throughout the life of these LTAs, which are averaging more than 4 years. And so with that as a backdrop, 2021 was really a story about normalization of depreciation and then fixed cost absorption. So we had these large fixed cost footprints in Dresden and in Malta. You've seen us really start to tool out in a material way that footprint in Dresden. And we're getting a lot of fixed cost absorption that's occurring as we tool out this existing footprint. And so 2021, depreciation normalized, and we started to get some really material movement on fixed cost absorption. So spreading essentially the same fixed cost over more units, if you will. And so that was the story of 2021. 2022, we're still going to get the benefit of some of that fixed cost absorption. And so we're going to get benefit in Dresden, we're going to get benefit in Malta. We're going to get a little bit of benefit in Burlington and as well as in the Singapore 200-millimeter facility, but you get a little bit of fixed cost absorption that still occurs in 2022 and you get a meaningful step function increase in pricing. So enterprise ASPs, as we've mentioned before, more than 10% ASP increases on an effective basis for the enterprise in 2022 over 2021. And so you get this ASP increase. It doesn't all happen January 1 because those LTAs are layered throughout the year. But as those LTAs start to ship, you get the ASP increase, you get additional fixed cost absorption. So that's the story of 2022. When you move into 2023, you have the full year impact of some of the ASP increases. Again, not all of them start January 1, '22. You get the full year impact in 2023 of those ASP increases. There's a little bit -- just a little bit left of fixed cost absorption to go a little bit more of depreciation roll off to go. We get the benefit of the East Fishkill transaction that occurs at the end of 2022. And then towards the end of 2023, you really start to see the business start to mix up. So as we mix into these higher-margin businesses, you start to see a little bit of that start to bleed into the second half of 2023. And then, of course, you're into 2024, and now you're really starting to mix up the business on a go-forward basis. So at a high level, pretty summarized level, that's how I describe the gross margin walk.

Yes, I'd be remiss. Dave and I get to tell this great story for the company, but there's 15,000 employees that -- I want to tell you, the things they've had to overcome to continue to grow our capacity, deliver it to our customers each and every day, make these flow-throughs happen, it's just amazing. This is a company that I think our DNA is we just know how to execute. And so when we get certainty in our business, he knows how to get it done, and it's a really impressive thing to watch for David.

Can you guys give us -- we're running out of time here, but I really wanted to ask this question. Can you guys give us an update on the U.S. TRIPS act? $50 billion of subsidies, $3 billion per program. We can see the GlobalFoundries team getting $3 billion of subsidies just to fund the Malta, New York expansion over the next few years. You've got a great government relations team. So what's the status update on Congress signing off on the bill? And is it fair to assume $3 billion in subsidies over a multi period of time to the GF team?

So let me start in reverse. I think we'll get -- it's my version of what's our fair share because our fair share has to be whatever makes the economics work to go create capacity. We state that we're only going to make investments that give competitive returns in invested capital. And so first and foremost, the economics for that to work. Yes, as a pure-play foundry, addressing 75% of the market, we play an important role in the world for semiconductor and in particular, the U.S. So yes, we should be in the mix. The key is to get it funded. And so I want to talk about 2 elements of that. Our view is this will get done. There's just such a high level of bipartisan support. In fact, I was in Washington second week in December and meeting with members from both parties. Our odds on guess right now bet is part of the funding the U.S. government, which is the Ombudsman bill that will go and get funded mid-February to fund the government, so there isn't a shutdown. The CHIPS bill, whether it's you seek over some other form, will be in that to fund that, and that starts the process you talked about. But don't forget, there's another bill called the FABS Act that fits the Build Back Better. That's a 25% -- up to a 25% reimbursement for any construction or CapEx that's used to create manufacturing in this country. Now that's back has -- the Build Back Better has to be funded. I think that will also get done as a high priority, high focus to get that done. That's not one of the elements in that bill that's contentious right now. So assuming Build Back Better gets funded, CHIPS gets funded, you have really 2 ways of getting access to make the economics work. You get to 25% rebate on what you spend, you get grants or however that's going to work in the U.S. government. They're complementary. If you don't get one or the other, they both come into effect. And so I think Q1, February in particular, the CHIPS bill with the Ombudsman funding of the U.S. government and then seeing when Build Back Better gets funded, I think it's got to happen in Q1 if it's going to have any kind of impact in the near term.

Perfect. Well, we're just about out of time. Great insights, Tom, Dave, Sukhi. Thank you for joining us here today. It looks like 2022 is shaping up to be a very strong year for the GlobalFoundries team. So best of luck, and thank you for participating.

Thank you for having us.

Thank you so much, Harlan.