GLOBALFOUNDRIES Inc. (GFS) Earnings Call Transcript & Summary

Okay. Good morning, everybody. Pretty crowded room today. Hope [ it'll be real one ] next year. But happy to have with us here today for the first time since the IPO, GlobalFoundries. We have CEO, Tom Caulfield; and CFO, David Reeder. Welcome, guys.

Thanks for having us, Joe.

I'm Joe Moore, by the way. So maybe to start out, bigger picture. You guys had a strategic shift in focus 3 or 4 years ago that's been pretty pivotal to sort of focus on the nodes you're focused on. What will you say about the situation then? And what new conversations from customers led you down this path?

Yes. So actually, I'm about a week out from my fourth-year anniversary in the [ GFC ]. I think when we started -- when we started as a company, the world and the semiconductor industry was a lot different. We were still a compute-centric industry. And by the time 2018 rolled around, and it was really the advent of the smartphone that created the reason for everything being connected. We're starting to see a huge shift in the foundry TAM from 25% in a broad deployment of semiconductors, amongst many nodes versus with the leading edge at the time. And also, the chip, the chip, it's more like 70%, 75%. And we look at our business and said, we're so much more prepared and better off, both in our R&D investments, the differentiation we can make and the capacity we can deploy, that shifting into a bigger SAM be more concentrated and avoiding the high capital intensity of single-digit nanometer, which was more than covered. TAM is actually [indiscernible] in that single-digit nanometer space. And for us, it wasn't so much about what we weren't going to do. We're building the confidence of what we can do and win. And you don't have to look very far. We had a facility in Singapore that manufactured 40 nanometer and above. That moves the platforms, both with features like embedded memory for secure pay, RF or connectivity. And that was almost profitable facility. And so the goal was, we have a great market opportunity. We have great differentiation and continue to leverage our R&D. And now we have to make the rest of the company look like our Singapore facility. And that was during -- that ran up to our IPO last year.

I mean I feel like when you were pursuing the 7-nanometer node, you were sort of -- made you kind of a cheap alternative to other people. And I think one of the things through the IPO process that's impressed me is how much there's real sense of partnership now with you and your customers. Can you talk a little bit about that? I mean, it seems like you're booking a different kind of book of business today than you were back then.

Yes. I'll give a great example. At the time Steve Mollenkopf was the CEO of Qualcomm, and I went down to tell him about this very painful decision I was about to make. And he said, stop pretending. It's an easy decision. Whenever use you 7-nanometer, you're late, you don't have the scale to make it interesting. You're scaring me that we should not doubling down your investments where I really want to use human connectivity, which leads to the differentiation, and that's where we're really focused. I mean today, there's a conference taking place in Southern California, where we just announced a broad range of customers using our next-generation silicon photonics. Those are deep partnerships. That's where customers co-invest with us to create off of our platform, their version of differentiation. Millimeter wave, 22-nanometer FDX technology. It's going to be the best solution for millimeter wave connectivity. I love the Super Bowl commercial with Jim Carrey for the home premise connectivity. Next generation, all are silicon in that because of the connectivity. We have 30% more power range with 20% value efficiency. So we get 30% better coverage with less battery power in both handsets and the devices at the antenna driving it. And by the way, that little box gets to be smaller because we use less antenna because of our technology. That's the kind of stuff that excites us. That's the stuff why customers come and want to be single source with us.

Yes. I mean you're getting real differentiation out of these nodes. Maybe we could say -- just talk a little bit about that silicon photonics arrangement today. I noticed you had Marvell, Nvidia, you had some pretty important customers on there. Can you just talk about the genesis of that relationship and where it might go?

Well, let's talk about the industry, right, it's about $0.25 billion market SAM for us today, and we have a majority share in that. We see that 2025, 2026 going to $1 billion SAM, and we maintain that this proportion of advantage in share. And that's what the foundation of today's announcement is. It's based on 45-nanometer technology. It has the integration of RF, Photonics and CMOS to make a monolithic solution for this kind of connectivity. Starts rack to rack, but extendable to card to card. Very excited to have these partners because each and every one of them take a base platform and do a certain level of customization for their particular solution they're trying to solve. Do you have anything to add, David?

Yes. I think the only thing I'd add to that is the reason we're excited about that market is, one, because of how differentiated it is; and two, because of just the absolute size of it, but because of the margin profile that goes along with that differentiation. It punches well above its weight and well above our long-term financial model.

Yes, margins are more like the analog players.

That's exactly.

Yes. I mean strategy of differentiating at these nodes is pretty clear. Maybe talk about the growth of these nodes. I think do you have to worry that today's cutting edge becomes tomorrow's trailing engineer left out of some of those things? And what do you see as kind of the growth of the nodes that you guys serve?

We see the segments we play in 12-nanometer and above growing in the high single digits. We also see capacity being put on a little bit less than that, even with all the announcements you see. So we see there's still an opportunity for our SAM to grow at a reasonable rate. We have to get caught up in the capacity we're putting on just to get the demand in front of us. But I think the overarching view for us is that we're playing in a $54 billion SAM that's going to grow north of $70 billion. And last year, we were $6.5 billion company. We have a huge blue ocean to play in. We don't have to win in every segment where our differentiation has less value. And so for us, the growth of the SAM is about how do we continue to focus on areas where differentiation matters. How do we take these nodes that we have and continue to build the nodes platforms with features. It's foundational IP, standard libraries, PDKs, complex IP feature types, new transistor types that create this differentiation.

And then just as you made this pivot, we sort of entered a world where the nodes that you serve are the biggest shortage, the biggest bottleneck shortage. Even as we're seeing some of the substrates and some of the more advanced stuff get resolved, these -- the wafers that you serve, the markets that you serve, look like they're going to remain tight for some time. You alluded to the CapEx. Maybe just talk a little bit about how long you see this tightness persisting?

Yes, I always give myself in trouble with this one, but look, I think in the better part of the next 5 years at least, will be chasing capacity as an industry, not demand. Just for the reasons you spoke about, 40-nanometer technology last night, rest year was the most sold out node, and it's only growing. I think it fundamentally comes down to the cost per transistor. And David, maybe you can talk a little bit about this. there's no sense jumping to the next node if you're not -- don't need the performance and you're not getting a cost advantage. And I think this is why you've seen the single-digit nanometer space kind of get relegated to 25% of the market. Great transistor performance on power, not very much improvement generation over generation of performance and the cost advantage just not being there. So it becomes a very -- becomes self-limited in the applications that will use single digit. And then we have technology to where power per transistor is the ultimate measure to go create solutions for high-speed digital processing. David, anything on the transistor cost?

Yes. I think I'd just add a couple of things. I think the first thing I'd say is what's driving designs in the market? Well, there's usually kind of 3 criteria on all the designs. One is the compute power, what compute power do you need? The other one is what kind of connectivity do you need? And the other one is what kind of power management do you need? Historically, it was all about compute. And now it's really about those 3 factors. And those 3 factors could be #1, #2 and #3 in any order, depending on the actual end market application. And so when you start with that as a premise and then you look historically about what used to happen with Moore's Law. Moore's Law, you used to always get -- you got better compute performance and you got lower cost. Well, somewhere in the 20-nanometer range, call it, 28-nanometer type of range, the cost curve broke. And what that means is on a per $100 million gate type of analysis, the cost actually became flat. And then as you drop below 20 nanometer, the cost actually started to go up. And so now the end markets are looking at that and they're saying, "Will my end market pay for that cost? And oh, by the way, does my connectivity and my power, does that scale down into these nodes for that extra cost as well?" And so those things came together in a pretty meaningful way over the last several years. And so I think what you're seeing is you're seeing the market say that there are some real nodes out there that have very good performance on power management, leading edge on power management, leading edge on connectivity, and those don't sit in single-digit nanometer. Those sit in the 40-nanometer range, the 55-nanometer range, on [ ASW ] technology, the 130-nanometer range. And so it's a different application with a different end market.

And it seems like what you're describing is quite a bit different in history, also in the sense of TSMC is not serving these with fully depreciated fabs. And maybe that's partly how we got to this state of shortage is we didn't have capacity water falling down. And one of the things we heard during the due diligence process was that TSMC went out their price increase and they said that's because we have to build too fast to serve these nodes. So it seems like a pretty big difference versus the past in terms of the way people are approaching these.

Yes. That's exactly how capacity was added. The first wave of capacity was added on whatever at that particular time was the most smallest transistor size. Whatever was added was pretty much all that was. And then the second wave, third wave applications came in. And in a compute-centric industry, that was fine. Then we wake up after this complete change in the compute-centric industry to evasive deployment of semiconductors. By the way, this is always going to happen. It was just going to happen later in time until COVID-19 kind of accelerated the decade and the year of demand for semiconductors. And so an underinvestment in these areas created the situation we have now, which, by the way, that's what's changed the fundamental dynamic. Customers now sign up for capacity reservation, long-term agreements where they're willing to say, "Oh, that's my supply. I want it to take or pay for 4 years or more, fixed price, fixed volumes over this period of time." When we started that in early 2021, it was really important for us for certainty to make the investments. By the time we hit this summer, it was more important to our customers know they had certainty of supply. The only thing that could stop their business was not having certainty of supply. And so our customers push on us as much as we push on them for more of the certainty and partnership investment because supply will be the differentiator for businesses going forward, access to the semiconductor.

And if I could just add on to that, actually because I think this is a real area of differentiation for GlobalFoundries, Joe. We have a different approach to the marketplace, right? We don't just roll out here's our 55-nanometer node, here's our 22-nanometer node. We actually take an approach different from all the others where we take a market outside in approach, we say, what end market do we want to go after that looks attractive to us and our business model? What technologies are needed for that end market? And how do we go out and get talent, which we did in our pivot? We went out and actually got in-market talent that used to design into chips like automated driver assist, millimeter-wave radar. Tom's got some great examples around that. And then we brought them in-house into GF, and we said, if you were a designer designing this chip, what foundational IP and technology would you like to have so that you can win in your marketplace? And that's what we did, and that's exactly the playbook that we're executing on. And so what does that translate into? Well, it translates into 30 LTAs, more than $3.2 billion of customer funding, more than $20 billion of revenue commitments, fixed price, fixed volume, fixed duration and 2/3 of our revenue being single sourced and more than 80% of our design wins single sourced. And so it's a different approach to the marketplace that I think has helped us go after those trends like connectivity and power management versus just continuously chasing single-digit nanometer.

Yes. The topic of those LTAs comes up a lot. You have very good coverage, really almost as good a coverage as you could have for the next few years because you want some room to see if there's a upside demand. I guess what happens if there's a dislocation in demand? How durable do you think those agreements will be?

Yes. I think the biggest part of the durability of these contracts is the single source or sole-source nature of them, right? It provides a lot of balance. And the way I think about the LTA, I'll be broader on it, 3 elements of it. The first was we signed up so that we have confidence and certainty and durability in the business we want to go build and more importantly, invest in. If on the downside, we go through a cycle in the industry, I'm more on the bullish side of cycles. But if they're there, they'll be more muted and less frequent. We also know that every cycle comes back really sharp. We'll just be in a better partnership position with our customers to work through that. In the past, manufacturing assets suffered the most. They suffered because they had the idle assets sitting there that cost money. And this will just give us the opportunity in the same partnership mode we entered those contracts to figure out with our customers how do we get through that. But the thing that I think is the most -- maybe the best kept secret about these long-term agreements is it's 4-year duration. At the end of the -- every year of a contract, we get to sit down with our customers one more time. "Hey, what do you want to do for that extra year, you want to sign up again? Or if you don't, we have now 3 years to go fill that capacity with someone else." We've never had that kind of visibility as a company or as an industry, where we can plan capacity. We could plan design wins, we could plan our R&D road map to fill capacity with that kind of opportunity. It also gives an opportunity to -- if we need to make more investments to talk about what the economics need to be for us to make those investments. So these contracts allow us the certainty down the middle to do the build plan we talked about, the investments we're making over the next few years to get to our 2024 run rate. It gives us the downside protection in the event there is some kind of discontinuity or dislocation, as you said. But I think more importantly, it gives us certainty beyond the stated length of each one of those contracts.

Makes a lot of sense. Just last question on the supply kind of side of this. Briefly, the question of neon keeps coming up around Ukraine. Can you just talk to whether you see that as any kind of bottleneck?

I'm going to let the gentleman who heads the head of our procurement team talk about that. David?

So a couple of things, and I want to expand on it just a little bit just to give the team here some color. So when we were out negotiating those LTAs with our customers that we just described, we were actually going upstream, and we were securing with our suppliers, all of the things that we needed, direct materials, indirect materials, the CapEx and the equipment that we had to order, we were securing those with LTAs that were similar in nature to the LTAs we were signing with our customers, so that we had confidence that entering an environment where there's not a lot of slack in the manufacturing process, we were confident that we can deliver to our customers and sign those LTAs. So we did all of that in the background. Now specifically to the Ukraine and the Russia conflict, what we see right now is from -- whether it's the neon, whether it's laser gases, whether it's fluorides, we feel like given that our geographic diversity that's in place, we feel like we've got enough qualified solutions around the world, whether it's in Singapore, whether it's in Europe and Dresden or whether it's in the U.S., to be able to address any discontinuity of supply across those gases. Look, we're -- I would say we're somewhat insulated based upon that geographic diversity and the qualifications around the world that have already occurred and the ability for us to be able to move those suppliers back and forth. But we're not totally immune to a worldwide shortage. So insulated but not immune. We don't see any issues today around those particular gases, neon being included. I think where the challenge is today is that Germany gets 1/3 of its power for its natural gas that provides power from Russia, Ukraine. And we're not immune to that. So we have a facility in Germany that manufactures 90% of our power that's on site. So we manufacture 90% of our own electricity, but it's supplied by that regional pipeline of natural gas. And so we're working with state and local and national authorities to make sure that we can continue to get continuity of supply. But that's the risk for Germany, and that's the risk for our Dresden operation.

Is there a risk that you have to pay more for that gas or just the risk you can't get it based on outcome?

I think at this point, where we've hedged the vast majority of pricing. So from that perspective, of course, there's portions that are unhedged, but it's relatively small in the grand scheme of things. And so I think it's more continuity of supply versus pricing.

Yes. Got it. Okay. I want to shift a little bit to the public policy issues. And I know, Tom, you're very involved with the FAA. As you look at the Chips Act in the U.S. and the sort of comparable legislation in Europe, can you kind of talk about what the objectives are? And obviously, it seems like it would help you. But is it employment? Is it continuity of supply? Is it reducing dependence on Asian suppliers? Just what's the overriding objection?

Yes, the overriding objective is you take the U.S. in the '90s, 37% of the manufacturing, semiconductors in the United States today, it's 12% and declining. So the first part of the -- by the way, and 48% to 49% of the demand for semiconductors comes through U.S. headquartered companies. You have this big imbalance. If you have a demand problem, you have a supply problem. And from a U.S. perspective, it's security supply, it's security, sovereign security. It's economic security, given that semiconductors is the heart of everything we do. It took a crisis of not having enough semiconductors where we realize how pervasive they are in our lives and how fundamental our economy is on semiconductors. So it's to fix that dynamic. Now relative to the U.S., but I really appreciate, especially Secretary Raimondo and all the players that are involved with us now, they get that if they're going to do something meaningful with the funding that I think eventually will come within the first half of this year because it's in conference now, it's 2 bills that have been passed by 2 bipartisan groups. So this thing should get it make its way clear. If you're going to take a holistic approach, they're not going to say, "Oh, if we have a subscale factory in Arizona for 2-nanometer, we're going to fix the supply problem." They understand it's going to take memory. It's going to take analog, mixed signal. It's going to take the kind of foundry stuff GF does. It's going to take some single-digit nanometer investment, but they need to fix it holistically and they're prepared to do that. Now the economics part of this is really interesting. Some argue, well, look, at $52 billion, is that enough? I think $52 billion is a great start. Once you see, when you spend that money in manufacturing, the economic activity created in the region more than pays for this. And you'll look to make those investments again, you know in upstate in New York, we get constant pressure. "Hey, can we do it again?" The state of New York has great economic activity, created economic velocity of money, $1 billion in 2009, '10 time frame, huge returns for them over the next decade. So I think once we get the flywheel going, creating manufacturing capacity, the jobs it creates, the economic activity in these regions, the government will want to invest more in these kinds of facilities. So I'm hopeful we'll get this funded. It will be part of our partnership investments going forward, not just partnership in us with our customers, bringing the third party to the U.S. government or other governments around the world to create capacity.

And I wonder, I mean, the opportunity for CapEx subsidy is pretty clear. But above and beyond that, there is the sort of articulation that Chips Act of public-private partnerships and things like that. Are there opportunities for something bigger here where you're getting demand generation from it or you're getting partnerships with Department of Defense or people like that, that really need you guys?

So let me give 2 points. I know we like to use word incentive, but incentive business sounds like a return on investment. I think these are more investments because as I said, the government will see this in the employment and the way the economics around that work. The second part of this is around -- what was it?

Just the partnership...

The partnership, yes. If the world is going to put the kind of capacity that we're talking about, the CapEx deployment required to get caught up is not going to be able to be concentrated on a small group of foundries or manufacturers. It has to be spread across the whole ecosystem, including governments as well to participate in these partnership investments. And that's the only way we'll be able to deploy the capital required to get caught up on the kind of demand we're seeing. Now it could be back to an older model, which is customers and manufacturers, like you have to. It will just take that much longer. So I think this is really the opportunity for the U.S., for other governments where they say, "Look, we do want to get caught up. We don't like that manufacturing went away for this. We're going to bring it in." If the industry was flat, if there's no growth, then there'll be no opportunity. So they see this as a moment in time where they can make those investments. And then the last part of that is go create to make sure there's demand certainty. There could be incentives as part of these packages for the end users of semiconductors to be rewarded to use that capacity. That makes good economic sense. We take advantage of the capacity that's being put in there. Those models are being considered in that.

And just let me add on. We are a trusted foundry to the government and governments do want secure supply. And I think GF, you could go back in time, we have a long history of working with the government, whether it's more recently on GaN, whether it's on SiGe, whether it's on silicon photonics. We've had engagements with the government for many, many years. And as a trusted foundry, we expect to have engagements with them for many more.

Great. So time for questions from the audience. But just the question I get most often is gross margin. The past the 30%, the past 40%. 30% being relatively near term and 40% longer term. Can you talk about the levers that you get to get there and your confidence that you get there?

Sure, sure. And I'll just kind of talk to that by a year and then if we have any follow-on questions, happy to answer them. Last year, so in 2021, 2021 was really a year for GF of normalization of depreciation. The company had historically carried a lot of depreciation as a percentage of revenue, which significantly impacted gross margin, not a cash margin hit, but a gross margin hit as it flowed through the P&L and also fixed cost absorption. So last year was normalization of depreciation and fixed cost absorption. What do I mean by fixed cost absorption? Let's take our Dresden facility as an example. Dresden from a fixed cost footprint perspective, size of the building, size of the tank farm, the process piping, the electrical generation plant that we talked about, it actually has the capacity to deliver about 850,000 wafers per year on an annual basis. But in 2020, it was only shipping about 300,000 wafers per year because that's all it was tooled for. So in other words, there was white space that could be tooled, but it wasn't. So you were taking all those fixed costs, I just described and dividing by 300,000 wafers. And so with these LTAs as well as the customer prepayments and access fees, that gave us the confidence to then go and build out that fixed cost footprint and tool driven such that by the end of this year, 2022, we're essentially going to be at that 850,000 capacity for only 15% increase in cost. So now instead of dividing those costs over 300,000 wafers per year, you're dividing them over 850,000 wafers per year. So that's what I mean by fixed cost absorption. So 2021 was normalization of depreciation and fixed cost absorption. And by the way, Dresden, there's a similar extent in Malta. Go to 2022. 2022 is the year of ASP. So 10% enterprise increase in ASPs year-over-year and a little bit of fixed cost absorption continuing. That's 2022. 2023, little bit of ASP, a little bit still of fixed cost absorption and now you're starting to mix up the business. So that pivot that we made in 2018, migrating towards these end markets that I spoke about, that pivot now is increasing, it's mixing us up into these attractive end markets that will get us to our long-term financial model and beyond.

And a little bit in depreciation.

And a little bit of depreciation.

Great. So I think we have time for one question from the audience.

You guys really do this on Zoom.

Maybe I can just ask, I mean, I was sort of intrigued by the Ford relationship that you guys announced a quarter or so ago, not necessarily because it's a big announcement, but just I doubt Ford had heard of you guys a year ago, given everything that's going on. And we see that everywhere. We see every customer kind of focusing on their whole supply chain, understanding it, making sure people are investing enough. Can you just talk about what that means, the fact that you're pulling through to an auto manufacturer?

Yes. I think auto in general is an interesting perspective. Look at the carnage that was created. There was a chip shortage. Some would estimate $130 billion of lost sales already. I don't think it's -- I think it's not perishable, so they'll eventually get caught up. But still, I think it's the epitome of saying, look, I need supply chain security. I think in the beginning, there was a lot of ambitions of how much they want to do their own designs. And then reality sets in and says there's just too many chips for them to do their own designs. They'll want to do some of that. I'm talking generally automated. What they're doing now is reserving capacity. That's the essence of any of these relationships. Reserve capacity, direct your supply chain to use it. So they know their allocation is there. They'll know that they'll have the access they need to capacity. They'll make sure that they partner even more deeply to make sure the features and the technology road maps for companies like GF are aligned. And then as their supply chain, whether they make the design or have someone design for them, customer -- even if it's a standard product they'd like to buy, that they direct their supply chains, use that capacity that they reserve. And I think you can see more and more of that with automakers.

Great. Okay. Well, Tom, David, thank you so much for your time. Appreciate it.

Thanks, Joe.

Thank you, Joe.