GLOBALFOUNDRIES Inc. (GFS) Earnings Call Transcript & Summary

Good morning, everybody. Let's get started with the next presentation. We're excited to have GLOBALFOUNDRIES management here. We have Ed Kaste, who's the VP of Product Management; and Sukhi Nagesh, who's the VP of Corp Dev and IR. So we'll start at a high level with me asking questions. And then if anybody in the audience has a question, just raise your hand, get my attention. We'll get a mic over to you so the people on the webcast can hear, and don't know what that noise is.

So why don't we start with the macro question. Just talk about the demand environment. Clearly, there's areas that have been weakening in the consumer side of the equation and some areas that have stayed stronger. You guys have good visibility across a wide array of markets, but even most particularly on some of the consumer applications with your smart mobile device business. Just talk about from an end market perspective, what you guys are seeing in demand right now.

Yes. Well, thank you, Ross, for having us here. Done to see some of the old friends here. First of all, I think we just reported and guided a couple of weeks ago, for our Q2 and Q3, no real change since then. I think what you'll see is you'll see us show continued wafer shipment growth this year, this quarter and this year. And except in some of the low-end consumer markets, I think that's pretty telegraphed, right? I mean we mentioned this in our earnings call as well, specifically on the smart mobile device end market that we have, we have really strong growth in the 5G premium tier segment that we are benefiting from. That's being offset by weakness in the low- to mid-tier end market in Smart Mobile. Our PC exposure is very small, so we're not seeing anything there at all. But we're seeing growth in all of the other end markets we play in, IoT, home and industrial IoT. I think Ed will talk about that a little bit. Automotive, comms infrastructure, all of them are showing pretty robust growth for us. So if you take all that into consideration, I think we're pretty much on track.

And how does -- big question people have is, over the years, people would use foundries is kind of flex capacity for those companies that were fab-light in their business model or even the fabless companies that one of the benefits of being fabless was they didn't have to have the burden of the cost in regular times but they could hit the brakes and downturns and the fixed cost burden would fall into the foundry side of things. That business model seemingly has changed a bit given all the shortages we've had, and so I'll get into the LTA dynamic specifically in a minute. But before we get into that, how do you believe that balance of power between the foundry and the customers has shifted if at all? And is it just a reflection of the cycle? Or is this something that's more of a sustainable shift in business models?

We think it's more structural. And the reason being, it needs to be more of a shared partnership model, and that's how we've approached the company and the strategy and how we want to approach our customers since the pivot in 2018, right? I mean it's -- we've got to work with our customers through mutually agree -- mutually acceptable agreements for wafer shipment for a period of time. That wasn't the case prior to our pivot. Maybe we had 1 or 2 long-term agreements before. Today, if you fast forward, we have 36 long-term agreements with customers, that constitute about $27 billion in revenue over multiple years. And plus, by the way, we have $3.6 billion of prepays from them. So it's real skin in the game type partnerships that we have with our customers, and so that's structural.

So talk a little bit about on the LTA side, how they're working. We've seen some examples recently of chip companies writing off long-term supply agreements. One could have been with you even though I don't want to name any customer. Another one most surely wasn't given the node that they operate on. But if the fabless companies are starting or fab-light companies are starting cut or to write down the LTAs, is that evidence of them actually working from your perspective? Or is that something that you believe is starting to see some cracks in that partnership that you just described?

Yes. So before I answer the question, maybe, Ed, you want to answer something about some of the LTA examples that we worked on this year?

Sure. Yes. So it's been a busy couple of years on the LTA front, and this summer has really been no exception. I've been close to a few of them. I'll just reference, too, in the IoT space, where we've been working with customers serving industrial and general purpose microcontroller market, specifically for connected microcontroller products. These are next-generation platforms where the design win was awarded a couple of years ago. But as they've matured their product development and getting ready to ramp, we're now inking LTAs with them, one in Europe, one in Asia in China. And what's noteworthy about both of these is their products are not yet in the market. These are next-generation platform that will be ramping in the coming years, and yet they've still leaned in to commit in both cases for your LTA to lock in that supply. And what makes it particularly sticky is they're using all of the differentiation that's available in our platform, the strong RF performance, the very low-power digital, the embedded nonvolatile memory, Taken together, that is also a potent and unique combination in our technology, which makes it sticky, which motivates them to make sure that they're going to have the capacity to ramp their products. So both of those examples I've been close to. One is fully signed, and the other one will close in the quarter. But that type of activity continues at a rapid basis.

On the [indiscernible] side of things, sticking with it. We had an example of customers [ one less ] in the near-term perspective is just that they come to you and have the discussion about how that can be adjusted. As opposed to in the past, they would just cut the orders, and that would be it. And it's not as simple as people just saying, you're going to enforce a take-or-pay and stick the customer one way or the other because that's really not -- that might be good for you in the near term, but that's not a good example of a partnership either. So talk a little bit about some of the levers that GF pulls in these negotiations.

Yes, I know that's a very good point, Ross, and our partnership [indiscernible]. That was why [indiscernible] that going back in time when we had one customer back in the day. We had a long-term agreement with them, a way for supply agreement with them. We would always work with them in partnership whenever they had [indiscernible] for mutually-acceptable outcomes. In the case, demand is higher than they expected or demand is lower than they expected. And there's multiple options we provide to our customers. It could be an extension of their LTA today. It could be pricing adjustments. It could be new design wins. Lots of different options or a combination of any of those. So -- and every customer is bespoke. And what we really are focused on is working with these customers or to come up with mutually-acceptable solutions. Now keep in mind, even though what you've heard in the market from some of our customers, there is -- today, there is no customer that we have on LTAs that's in breach of their LTA. So we have ongoing discussions all the time with our customers, but there's no breach right now.

I believe you told me, not only is there no breach, but there's nobody where you're enforcing the take-or-pay which -- that might be synonymous with the breach side. But it's not as if you're getting paid upfront for things you're not delivering either.

Look, this year, we are meeting all of our obligations to our customers, and our customers are taking off of the required wafers that they've asked for, right? So there is no breach.

Right. When you think about the negotiation, I think Dave, on the earnings call and then our call back, talked a little bit about just mechanically how you adjust things, and you can give somebody less volume but at the same price. It seems like you're protecting the price side of the equation more so than anything else, and you'll give people a little bit of leeway on the volume side. Is that true in a broad sense as much as possible?

In a broad sense, yes.

And the logic on that is driven by what?

Well, I mean, obviously, we want to protect our margins. right, first and foremost. And that's important for us from where we -- mainly because of where we're coming from, right? And so I think it's important for us. I think that's -- I think we pretty well understood. Our customers understand that as well. So that's where it's coming from.

So the final question is kind of LTA-driven is in a recession or a downturn in semis, there's always the worry of fungibility between different foundry partners. Now you guys have a ton of different more specialized processes, a lot less relatively speaking, of the bulk CMOS side of things. Talk about how much that is lending itself to stickiness customer relationships or the flip side if people are going to be worried about that fungibility with other traditionally kind of Asia-based folks on more bulk CMOS. What percentage of your business would you say, independent of LTAs, would potentially be exposed to that sort of substitution effect?

I mean, I think today, this year, I think we have a chart in our Capital Markets Day deck, nearly 75% of our revenue is differentiated. That's meaningful. That's increased about 10 points from 2020. And about 65% of our revenue is single source today, right? Our design win -- about 90% of our design wins today is single source. As these design wins turn into volume over the next few years, we're going to see a bigger and bigger portion of our revenues being differentiated revenue. It's sticky single-source revenue. And so we don't do a whole lot of commodity-type business. Most of our business with customers are purpose-built specific for a particular application from customers. And they know, for example, it's not we don't do a whole lot of standard CMOS, right? We do embedded nonvolumemory-based CMOS, RFSOI, FD-SOI, silicon germanium, certain photonics, GaN type applications. All of these are very highly specialized, differentiated technologies for very purpose-built solutions, and so it's not going to be something that can be easily replaced by just going to another foundry. Maybe, Ed, you can give a few examples in some of the customers we work with in some of our differentiated technologies.

Yes. So across the end markets, right, the specialty is at play. So in the handset, most people think of the RF. Certainly, we have a strong presence in the RF front-end module. But also as the data rates increase from 4G to 5G, then the features that accompany the user experience with those higher data rates also call for a feature-rich silicon, whether it's the high-resolution displays or the higher resolution imaging, the cameras that we use, which all have requirements that lend themselves to feature-rich semiconductor solutions. And then there's a kind of a direct application to how those solutions apply to other markets, right? So IoT requires the same similar connectivity. Also, IoT devices don't have keyboards so they need the touch display and the audio input, right, the voice processing, gesture control, right? All of those things are the human machine interface, right, which is a -- the last kind of connection point between the analog, the real world which is analog and the compute world which is digital, and so that also lends itself to feature-rich silicon. And then in the automobile, it's all of the same themes. The automobile as it goes connecting to the cellular network. Similar portfolio is our mobile. Inside the cabin of the vehicle, there's a mini IoT network where you're connecting a handset and audio devices and entertainment. So in many ways, the mobile phone places the trail for these capabilities, and then they have direct applicability to other end markets where we're growing.

So generally, when you talk about the process technologies you guys have, very well suited for RF connectivity, those sorts of things and then exceedingly low power is kind of how I would summarize those. Any feature set that you guys offer in your more specialized processes that's something beyond -- I know you have some photonics and silicon.

I would say the one that you didn't mention that's important is this analog capability and higher voltage capability with our VCT technology, which have some of the most efficient power conversion and delivery figures of merit. And beyond the VCT technologies, we're investing in new technologies like gallium nitride, which service that space. So that plays in this human/machine interface, the analog capability. And then at the higher voltage regime, it plays in things like electrification for the automotive.

So I want to come back to the end markets here at a bit, but one last topic I want to hit on kind of the structural change in the industry. We talked a little bit about the partnership and the LTAs with your customers. The other side that's clearly changed is the partnerships with government entities. You guys have kind of blazed the trail and moved ahead of most of your peers in that with agreements in Singapore that are already in effect, newly announced ones in France with the French government and STMicro and Kroll. And then obviously, there's the U.S. CHIPS Act and everything you've played a role on with that. So talk a little bit about giving people an idea of just how much savings can be accomplished with these partnerships. And I believe you have a slide in your Capital Markets Day that kind of walked us through that. But just to remind people, what's the magnitude of savings that you can get? And what's the mechanism by which you get recompensated?

Yes. That's a good question. Now we have been focused on this whole concept of public/private partnership. And the reason for that is to have better, balanced economics between us and our customers, right? And so with that being said, on the government -- we play in Singapore, we play here in the U.S. and in Germany. In all those locations, we have very strong relationship with the government entities. Let's just stick the recently announced announcement with STMicro in the French government. That's about 620,000 wafers total new capacity that's going to be added. GF will own about 58% of it. So roughly about $4.2 billion or so in CapEx, gross CapEx. We expect about half -- more than half of that to come from the French government. And that would be about $500 million or so coming from customer prepays, which leaves the remaining to us -- for us. Now contrast that to how capital was deployed prior to the step of new model. That was 80%, 90% of the capital expenditures would fall directly on GF. There's a big divergence in how things were done in the past and how we're moving forward now in partnership with governments and customers. That's one example in Europe. I think it will be something similar here in the U.S. once the entire funding mechanism for the U.S. CHIPS goes through. Right now, it will take about 3 months for the Commerce Department to write the rules of disbursement on how U.S. CHIPS will fund projects specifically here. We're fairly confident we'll get our fair share because we have customers lined up to -- for expansion projects here in the U.S., who want to have steady supply surety for the next decade. And so we're going to work with them in a model that's very similar to what we just did in France.

And so the French example where it costs you basically half as much as it otherwise would have. Upfront, that's great. Longer term, we've heard of other companies that have foundry aspirations talking about some kind of project finance recently where things can be accretive upfront. But longer term, then they get dilutive because you have to pay people back. They're not going to do this for free. Yours is a little bit different than that. It's -- I wouldn't describe it as project finance. But is there anything at the back end, other than the prepayments from your customers that, obviously, they will -- they're not going to get -- they're not going to pay as much to you longer term because they're paying you near term. But other than that, is there any other kind of longer-term ceiling on the upside that you guys have because of people paying more upfront?

There isn't any ceiling per se. We looked at project financing as well, and we had multiple companies, including the company that you mentioned. It will make sense if your -- if you have surety in your volume in the future, right? I mean if you -- if there is the high level of confidence that you're going to fill fabs with wafers in the future, you could potentially take that risk. And I think from our standpoint, our return on invested capital with the model that we're going with is really good for us. It works out for us, so I think we're going to stick to how we're playing it out.

And how does it work lastly? And I know the payment mechanism in the U.S. hasn't been fully borne out yet, but there's also state incentives. The state of New York, I think, has about $0.5 billion incentive program, which you guys will get your fair share or more than your fair share of given your footprint there. Are there also states beyond the government -- the federal governments that are also kicking in? Or is that just a New York-specific dynamic?

Well, I think wherever we have a footprint, if there are local governments that will participate or help out, we will definitely benefit from that. I think New York state is a perfect example. We have had help and subsidies from the New York state local government in the past for our multi-facility. So I wouldn't be surprised if this time around, they also provide some kind of incentive program.

So last thing on these government incentive programs, I think people -- the payment mechanisms, as I've said multiple times, are yet to be fully hammered out, but people think there will be a gross CapEx number and then a net CapEx number. How does that then flow through hypothetically to the income statement side of things? What does it do to your gross margin? And how do you think it flows through? Is it a tax incentive so the tax rate is lower? Just what are the levers that don't put magnitudes on? But generally speaking, how does it flow through on the income statement?

Yes. I think it's still to be determined. But I think if we just look at it, most likely outcome is going to be a direct offset to depreciation cost on [indiscernible]. So that would be a net positive to gross margin.

Got you. All right. Well, let's move beyond all of those kind of structural dynamics. And Ed, let's talk about some of the product-specific side of things. So why don't we go into the area that I believe is a little closer to what you focus on first, the home and industrial IoT market? That's about 15%, 20% of your business, but it's also the fastest-growing part of the business. Talk a little bit about what's the differentiated technology that GLOBALFOUNDRIES brings to the customers that are addressing that market.

Yes. So yes, it is a key area of focus for us in the second quarter. That end market was about 17% of our revenue. The growth was 70%. That growth is driven by a number of things. But of course, the first among them is the explosive unit content growth as IoT deployments become very real. Just 2 years ago during the pandemic and probably accelerated by the pandemic, the number of units of IoT connections surpassed for the first time the number of all other connected devices combined. So that includes the smartphone, the tablet, the laptop, the PC. And it's not turning back. If you look at the unit growth rate of IoT, it will far surpass that of smart mobile devices, and it's an area where GLOBALFOUNDRIES has real strength. That IoT solution is very constrained, right? They operate on tiny batteries. And batteries that unlike a mobile phone, which you're willing to charge every day or a car you plug in every night, some of these IoT deployments are in hard-to-reach areas, and the expectation is that the battery will live for a year or 2 years or more. So low power processing is at a premium. Especially as in these applications, you're trying to push more intelligence into those applications, which calls for more compute horsepower. It has to be done at the lowest possible power point. The next thing that it really pulls upon is the RF connectivity because these devices only work if they seamlessly connect. They stay on the network. And again, some of them are in hard-to-reach areas. Every, say, 50% increase in a linear range increases the coverage area for an IoT deployment by 225%, and that's really important in a home where you have a central signal and then your IoT devices are in a factory or in an agricultural setting, where these things are being applied. So the RF capability is really important. And then the third attribute is the embedded nonvolatile memory. Up until 28-nanometer, there was -- this was some version of Embedded Flash. Below 28-nanometer, the same. Solution space is different. So we have MRAM and resistive RAM and other novel nonvolatile memories, and these are important because they are the code storage and the devices can take over-the-air updates and refresh their operating systems, which lives in the nonvolatile memory. And then being able to combine all of those capabilities, lowest possible power, strong RF capability, embedded nonvolatile into an integrated solution, right, because the form factor is very constrained as is the cost. Being able to put all of those together, and that's where, for example, 22 FDX, the FD-SOI technology in our portfolio really shines because it has leadership capabilities for all of those things in a fully-integrated package. And so I referred to those 2 LTAs that we had in progress this summer for IoT companies, they're both leveraging 22 FDX solutions with embedded nonvolatile memory.

So that was perfect. That was going to be my follow-on was translating the feature set that you said, the low-power processing, connectivity and embedded memory to the process technology that GLOBALFOUNDRIES offers. So you have the FDX side of things. Just talk a little bit about how that differentiates and the fact that you need multiple different sorts of process technologies before you put that all together into a single package, if it's a [ FET ] instead of a SoC, those sorts of things. And how you differentiate from other foundries that people could choose to go to? What leads them to choose GLOBALFOUNDRIES when they're choosing that foundry partner?

Yes. So it's really the strength of having the capabilities that I described fully integrated into one package. You can get the RF front end in a discrete power amplifier. You can get the digital processing and a bulk CMOS process in the larger geometries, you can get it in embedded flash process. But to have them all coming together in a 22-nanometer geometry, it's really a sweet spot for the IoT. Going down to the next node, across the board, that would be a FinFET solution. It doesn't play as strongly in IoT because the design costs are higher; the masks costs are higher; the cycle times, the turnaround times are higher. There's 10 more masks in FinFET process than there are in our FDX solution. And given how broad the IoT space is, all of the deployments are different, and there's many different players. It's not like the handset market dominated by a few. There's just a very broad landscape of customers in IoT, which means a higher part number mix, more SKUs, more designs, more tape-outs. And again, there's a higher barrier to moving those into a FinFET technology even if it had all of those capabilities because the design cost and masks at cost, the cycle times are all higher. That's why we really have enjoyed 22 FDX as a sweet spot for this market.

I think the diversity of that end market is going to suit you well in that the concentration and the volatility that concentration can create sometimes to the good side, to the upside. But in times like today, where people are worried about demand, that concentration can be -- have the opposite result. So good fragmented market. And just remind me, in your longer-term forecast that you talked about, this is supposed to be the fastest-growing segment for the company. What was the CAGR you guys talked about for that going forward? And again, it's about 17% of sales now. How much do you think it will be 3, 5 years down the road?

I think 3, 5 years down the road, it will be greater than 20% -- between 20%, 25% of our business, right? So if you look at our -- if you look fast forward and look at our business, I think we had a chart in our Capital Markets Day deck. We expect all of our end markets to grow but some to grow significantly faster. IoT and home and industrial will grow one of the fastest. Automotive will be growing, I think, one of the fastest as well. Comms infrastructure will grow. We expect mobile to grow but not at the same level as some of the other end markets. So we think maybe a few years from now, mobile is probably about 45% of our overall business. But home and industrial end market will be close to 25%. Comps will be higher, and automotive will be significantly higher. The other area I think maybe you could talk to, and Ed can give us some more color, is the automotive space, where we're extremely, extremely excited about some of the design wins we've had. And that business is on track to hit a $1 billion run rate by end of next year.

So that was actually where I was going to pivot to next was -- yes, and I think you said on the last call, it was either $800 million or $1 billion run rate exiting next year. And I think our model has that something like closer to half of that, so definitely an upside area versus our model. You talked about your design wins growing substantially in the last 12 months. I think 15% of your total design wins are automotive versus automotive revenues being currently about 5%. So a lot of positives happening there, and it's probably translating into LTAs. Talk about the correlation between the design wins and LTAs in the segment. And when will that all kick in to drive growth in a market that's great secularly but does usually take some time to ramp?

Yes. So there's no real correlation between design wins and LTAs, right? I mean, I think from a process standpoint, we have a design win, and it takes a while to get the design win. To convert that design win to LTA also takes some time. So there's no -- it doesn't mean -- if you're in a design win, doesn't mean we have an LTA. There are 2 discrete processes almost. And not all design wins get converted into LTAs, right, and so that's why you won't see that correlation between design wins and LTAs. But maybe, Ed, do you want to talk about some of the LTAs that we've worked with or design wins in the automotive state?

Sure, yes. And they're loosely correlated, but there can be an offset in time or a lag. And I think in the automotive market, that lag is the longest because the design cycle, the qualification cycle, all the robustness testing and certifications or automotive safety that have to be undertaken take a longer time. And the LTA phase is usually closer to the production ramp. And so for that reason, the offset between a design win if there will be an LTA in automotive can be a longer span. To give you a very good example, one of the LTAs that is ramping the most deeply right now and over the next 3 years in automotive, it's actually a design win that we achieved in 2012, a full 10 years ago. And this is using a 40 embedded nonvolatile technology. It's the product that's ramping the fastest in our Fab 1 facility today and supported by a multiyear LTA where that ramp will continue for several years. So that's just one example where it's a major piece of our automotive business. It's an older design win and a new LTA that we're ramping capacity for now.

Have you started to see design wins or interactions, engagements directly with OEMs? I think that's another kind of structural shift that's happened in the automotive side of things where after a couple of years of shortages, the OEMs kind of pulled an end around on the Tier 1s and said, we'll have the relationship with the foundry side and the OEMs will be the ones along with the foundry to dictate the capacity to a given chip. So is that a structural shift in your relationship with your customer and your customers' customer?

It sure is. So it was precipitated, I would say, by 2 things. One, this well-publicized supply chain crisis in the automotive market, right? The first phase of that crisis when it started was the automakers trying to deconvolve their supply chain to even know where the escalation calls should be placed, right? It took quite a bit of time because it was a complicated supply chain with a lot of intermediaries, indirect communication, foreplanning. And in fact, that's one of the architecture points that led to the -- how severe and deep the supply chain crisis was. The second piece of contributing factor is this ensuring our localization of supply. It's not just ensuring for U.S. interests. But certainly, there's a strong automotive OEM presence in Europe, and they're also happy to see automotive supply coming out of our factory in Dresden, Germany. So the supply chain crisis and this localization of supply have driven much more significant direct OEM relationships from the foundry. Right around the time of our earnings call, in fact, we hosted -- we cohosted a CEO Summit in Washington, D.C., which had attendance from both CEOs from Ford and GM in addition to many other CEOs and industry leaders. And that's a dynamic that we just would not have seen, say, 5 years ago. And in all of these cases, they're looking for 2 things. One is to kind of tactically fix to the greatest extent possible, maximize the use of available capacity. But then more importantly, put in place structural changes so that we can ever get back into this position where automotive makers have tens of billions of dollars of impact can't ship automobiles because of semiconductor content. And that's leading to these more collaborative discussions on road maps and how we could partner more closely together, not only on supply, but on technology. The other artifact of the model that was in place before was there was a disconnect between development needs for the industry and the supply chain that was in place. And if you think about how quickly now the automotive market is moving towards autonomy and electrification and connectivity, the technology needs are really evolving quickly, and they can't afford to have that very distant relationship with where the technology is coming from. So all of those things are contributing to a good environment for global foundries, and you can expect to hear more good things to come in that space.

Yes. It seems like that's probably the best evidence of this being a bit of a structural change with the relationship as opposed to something that's just a reaction -- a knee jerk reaction to more of a cyclical type period.

I agree.

So in the last 30 seconds we have, really quickly, just one financial question. The gross margin side of things is obviously key to the earnings power of GLOBALFOUNDRIES. Gross margin has gone up from kind of high teens to high 20s. It's had great tailwinds. Wafer pricing has been one of those. As you, Sukhi, look forward, what do you think the big key steps to go from the high 20s to your 40% target will be?

Yes. Look, I think we're executing steadily towards our target, right? That's the main key point we want to make here. We have about 300 basis points of roll-off from Fishkill that's going to help out for next year.

Not all in the one quarter.

No, not in one quarter, spread out through the year. We have continued to roll out some of the high depreciation costs that we had in multi-rolloff, that's going to help us out. And then we're shipping higher volume, more units next year. And our business is going to mix up. So if you add all of that together, we're on a very steady path to achieve our 40% margin target.

Great. Well, guys, we are exactly on time, so thank you so much for your discussion. Always good to end with an on-track comment, especially about gross margin. So we thank you very much for your time.

Thank you, Ross.