Good morning. Welcome everyone to this session. Really delighted and honored to have Tom Caulfield, the President and CEO of GlobalFoundries, joining us this morning. I'm Vivek Arya from BFA's Semiconductor Research team. And what I'll do is I'll start with a few of my questions, but please feel free to raise your hand if you would like to raise anything, and we'll get a mic over to you. But with that, a very warm welcome to you, Tom. Very delighted that you could join us.
Well, thank you for having me, and always, Vivek, you, you've been a big supporter for GF, since our IPO, so thank you for that.
Thank you. So maybe let's start at the top. You know, two and a half years since the IPO, and I don't think anyone would have been able to predict at that time that we would go through a demand boom and then, like, a big inventory correction. So the industry has been through a rollercoaster, and I would love to get your perspective on, you know, after coming through this cycle, right, so early in GF's life cycle, how are you feeling now following Q1 results?
So, I'll give state of the industry, state of GF, and it's an interesting twist to the question because it kinda relates back to our roadshow investment thesis, and everybody at the roadshow said, "You'll be tested in a down cycle. It looks great now." So I think we went through that test. So what's the state of the industry? If you were a CEO or a C-suite executive in the semi industry a year ago today, I don't think anybody would be predict that we would still be in the burning down inventory. You remember, 2023 started with, the sense of it was gonna be a first half down, second half back, and it was just a typical correction. You know, we're six quarters into a correction.
Put aside certain pockets of strength, like, you know, NVIDIA, AI, but the, you know, the more larger markets have been pretty muted. And while memory looks really strong now, it's coming off a really low base. And so, I don't think anybody would have predicted this. So what are we seeing? State of the industry? The good news is inventory is being worked down, not at the rate anybody thought. First, look at dollars of inventory, especially for some of our key customers, the dollars have been coming down. Little bit of noise around the days of inventory because of the revenue top line. And we think 2024 will certainly clear, whether it's 3Q or Q4, that this industry will get more to more normalized inventory levels, therefore more normalized natural demand coming through.
What have we done in the context of that? We called Q1 our bottom. Others need a little bit more time, and I think part of us being able to be confident in that is the breadth of our customer base and the breadth of the end markets we serve. We've had a beat on EPS, a beat on gross margins. We gave a guidance that was again ahead of what was projected by the street. So I think we've done a decent job as a company managing through this down cycle. And back to this, how does it feel two and a half years post an IPO? It feels like in dog years, it was like 30 years. But I'll tell you, the test was always going to be: yeah, it always works in an upcycle.
Let's see how you guys will do in a down cycle. And are you living to what the investment thesis was in the IPO? And I have to tell you, I think it would be hard to argue against that case, that we've been resilient in our financial performance. Look, this is a year, two years in a row of declining revenue, and we're gonna have year-on-year, 2-3 times more free cash flow generation. Well, first quarter alone was $261 million. Yeah, there's lumpiness. Not every quarter is gonna be that. But, you know, this is a business now that has positioned itself to be ready for the upturn and to have both the capacity and the differentiation products ready to go.
So Tom, as you suggested, the Q1, kind of the bottom, recovery from an industry perspective is uneven. So as you look at the next several quarters, which end markets are you feeling, you know, more confident that they are emerging from that downturn? Which markets are still kind of soft? And importantly, as a manufacturing company, how do you get to that, you know, magic number of 80%+ utilization? Is that something we can see this year?
Yeah, that's the magic number for sure. So let's talk about why GF can call Q1 the bottom, and where do we see kinda holding the line on revenue and where we see growth. So we think of our business in four end markets. Let's talk about the ones that we hold the line. You know, our data center business has been down, you know, since the IPO, but more or less what we predicted. And it's, we believe, at the bottom. The level we were in Q1 is gonna be roughly there. And then with the design wins we have coming in, we'll build that business back up over time. The second end market for us is home and industrial IoT.
That is a consumer-led segment that has been really hit hard, and you see some of our customers and some of their results, especially Q4 of last year, how hard that industry was hit. Now, the good news about that is for us, we think it's at the bottom for us, and we don't need it to grow for us to grow quarter on quarter. But that is the opportunity when consumer spending comes back, maybe even some refresh cycle, driven by AI, that that could be a good upside for us. But we don't need that to grow quarter on quarter for us to grow. So where's the growth coming from if it's not those two end markets? Let's start with automotive. Automotive is gonna be meaningful growth for GF this year.
Yes, in the face of all the noise you hear about automotive slowing down in unit sales. Remember, GF in 2022 did $375 million in revenue. In 2023, it grew to $1 billion. These were all new sockets, new design wins, and a breadth of applications from BCD for precision battery management, microcontrollers, infotainment. And so a lot of those sockets were just ramping in 2023, and we didn't have a lot of time to have our customers build inventory. And so even in the face of that, and look, we're almost halfway through the year, we still see, you know, big growth in our automotive business. Meaningful, what does meaningful mean? You know, mid-single digits growth.
Then smart mobile devices, you know, we're not the industry soothsayer in that, but pretty much everybody thinks that's a 3%-4% growth this year, especially in the premium end, where we participate. And so the opportunity for us to grow with the positions we already have, we see quarter-on-quarter sequential growth in smart mobile devices. So you combine Q1 alone, smart mobile devices, and automotive was, you know, mid-sixties revenue for us. And so the two areas where we had the most amount of revenue concentration, the two areas we see quarter-on-quarter growth for our company.
Got it.
So that's why we think Q1 is the bottom for us.
On the pricing side, you know, there's been a lot of concern that the industry built up a lot of capacity right now. There's still more capacity that some tier two foundries are building, but I think you are sort of holding the line, that this year you expect pricing to be flat. So how is, you know, looking at competition kinda shaping your views on... And just the confidence around maintaining this flat pricing this year?
I think our strategy, this is back to the IPO, to where it is today. We wanna be building on single source differentiated technologies. Now, what gives us the ability to do that, kind of the luxury? Is we play in a pretty big-
Right
... SAM already, right? It's $50 billion SAM growing to $100 billion over the next, you call it, 6-10 years, right? Midpoint, 8 years. So it's a really big SAM, given a company that did $7.4 billion in revenue last year. So what does that mean? It's we can be somewhat selective and grow our company nicely by not having to chase every piece of business. So what do we do? We make sure we understand the details of end markets by the actual application and device type, drive our technology roadmap in partnership with our customers, so that we can create that edge for them in differentiation, and we can drive to single source business.
You know, we've been reporting for a number of quarters now that our percentage of design wins every quarter is, you know, in the north of 90% on single source business, because that's the pipeline of the future. So when you're in single source business, a couple things happen to you. One, you can't use pricing to drive more demand because you're not second source. It's the demand is the demand. And the second is, customers already made that decision to invest the, the high degree of, of R&D they need to spend to create that part number, or to build out their portfolio around a technology platform, that they're in it to win with you. And so we work together to make sure that we're both creating the value and capturing the value.
For us to maintain the pricing in the face of a lot of capacity coming on, is to maintain our strategy, be differentiated, continue to invest in differentiation, be very selective in the battlegrounds and end markets you want to win business in, and make sure you win that business.
Got it. One thing, Tom, where we have been very impressed with GF is kind of building to demand and the long-term agreements, like the LTAs, that you have been signing with customers, right? Since very early in the process, and more recently, as the industry has gone through a downturn, right? Having those kind of underutilization payments, right, we saw that $23 million in Q3 of last year, $79 million in Q4, and I think it's like $80-some million in Q1, with some more expected in Q2. So two parts to the question. First of all, you know, how do those kind of underutilization things roll off for the rest of the year? And then just broadly speaking, how are you engaging now with customers, right, in long-term agreements?
Like, are they still a little bit shell-shocked from the downturn? Or are they not engaging in as much in LTAs? Like, how are you managing the business now, given-
Yeah
... the state of the industry?
So I'll use my catchphrase: There's no such thing as macro anymore. Meaning there's no such thing as how does a customer do something? The range of customers, the different end markets, how they feel about things varies by all the different circumstances. So let's talk a little bit about how we got into the LTAs, what they provided, the benefit to our customers and to us, and what's the future of LTAs. So how did LTAs come to be? In early 2021, customers needed us to make big investments to create capacity for them. We said, "More than happy to do it, but we need to make sure that you're committed to that capacity. We need certainty, durability, and profitability. We can't just invest. And if you really need that capacity, you know, sign up for it with us.
Right.
And it was, gladly do that. And you could see, we've, we'll talk a little bit about what that investment has positioned GF for its growth in the future in a moment. As the 2021 rolled on, we were no longer the instigators of wanting long-term agreements, 'cause typically, we, we wanted a long-term agreement just for when we had to invest. Our customers wanted long-term agreements because what they were finding, there was so much demand that they had, that there wasn't enough capacity. Every time they were coming to different suppliers, not just GF, they couldn't get as much as they wanted, and they were at the risk that the new market price was something different. So their, their mindset was: Look, if you give me certainty of supply and certainty of pricing, I'll manage my business.
And we were, I wouldn't say forced, but we were at the table creating LTAs everywhere. I think to date, over 40 LTAs, we've signed. By the way, with $30 billion of committed revenue, which has two-thirds of that still available to us. So we come into a downturn and customers need to reset. The capacity we went and invested and put on, they don't need all of that at once. And so in some cases, customers say, "Look, the end market I'm in, it's not that predictable. It was a good idea at the time, but I'm maybe better off without an LTA. How do we go reposition this LTA where it becomes more how we used to do business? And, you know, and I'll true up whatever investments you've made from me," and you'll see some of that in our financials.
Now, look, you, you've got the right numbers, but if you think about that fraction of the top line compared to our total-
Right.
It's small. And sometimes the renegotiation of the LTA means that underutilization payment's either a one-time or it's gonna be a continuous. It's blended all depending on the circumstances. But here's what we're, what every one of those conversations end with. We went into them as partnerships, we're exiting them with partnerships. Some will say, "I'll never do one again." Others, the biggest one we did ever was January of this year, right in the middle of this, right? A huge for an end market that you can guess, like automotive, would want a long-term commitment to supply. So what I think we've all learned as an industry in these long-term agreements, one, there was a bit of a feeding frenzy.
We should have all been a little bit less ambitious about where we think the industry was going, how fast, because it led everybody to wanting a lot more capacity than they needed. Oh, by the way, I could tell you, I had more disappointment, not when the customers had to renegotiate an LTA, when we couldn't commit to how much they wanted back in the day.
Right.
I'm willing to sign up for anything. You won't give it to me." And the answer: "I, I just don't have it." Imagine if we would have signed to those, how much. And so every negotiation or every partnership says: "Hey, I want—I'll take the same volume. I need more time. Here's more sockets." There's always a win-win attribute in this, and the least amount of all of that is about what's the underutilization charge, which is a financial obligation for the investments we made, and it's a fraction of the long term, the spirit of the agreement and the value of the long-term agreement.
And so again, I think the industry will learn from that where they really make a lot of sense, where maybe, if they do do them in the future, they'll be a lot less ambitious in how much they're signing up to. But they've served, I think, our customers as well as they've served GF in this process.
On CapEx and capacity, so, you know, kind of disciplined response to market conditions. I think CapEx was $3 billion in 2022, $2 billion last year. This year, it's only $700 million. Going forward, is there a reasonable way to think about how you're managing CapEx? Is it—I think at the time, the analyst state was, you know, supposed to be closer to 20%. Now you're able to run it closer to 10%. What is the right way to think about CapEx intentionally going forward?
Yeah. So again, I'm glad you started with since the IPO, because I'm gonna get back to one of the other elements of the investment thesis about GF, about its investments. So a lot of the investments we made, starting in—you talked 2022 and 2023—were an offshoot of the LTAs we signed, where we're committed to go create that capacity. So what it has done for us, we've actually put in capacity that we had capacity to do about 2.1 million wafers in the 2020 timeframe. At the end of this year, when the rest of this capacity is kind of put on and qualified, we can do 3 million wafers.
You guys can do the math on an ASP of $3,000. There's $9 billion plus 10% of non-wafer revenue. You can argue that we've put in place an engine that can do roughly $10 billion of revenue. Last year, we were $7.4 billion. You guys know what consensus is for this year.
Right.
So we have built in already the ability to use what we've already invested to grow our business, and that's why you see this year at $700 million. So what does it mean in the years looking forward? Couple things. One, you know, we won't have to invest very much for new capacity. That's different than investing to create new features and new technology within our factories versus just new wafer starts for a while, right? Until we get to that $10 billion. And then when we get to the $10 billion, what's magical about that number? What's magical about that number, you go back to the IPO. We said, when we get to $10 billion as a company, not only will we have free cash flow that's significant, we'll also be able to invest in our growth concurrently.
That's the scale we needed as a business to do both. That was before we factor in the capital efficiency we'll have when we start growing again through the co-investments that governments are making around the world with us, like the CHIPS Act and ITC in the U.S., and other things. The positioning of growth for our industry, for GF is, if you believe this industry doubles over some strategic horizon, GF is gonna have to make at least a 1.5 addition to its capacity, given that we can get to 10 without much, and then grow from there. When we start growing again, more capital efficient, and we'll still be a business that not only can invest in its growth, but produce free cash flow like we're doing this year.
Right. You mentioned CHIPS Act, so we heard, you know, GF, one of the, you know, top beneficiaries, right, of that $1.5 billion from the federal government-
Mm-hmm.
-and then another $600 million from New York State. How do you plan to take advantage of those facilities? And by the way, the CapEx for this year, is, is that reflecting some of those? Like, is it net of some of those, benefits? What, what is the right way to-
There's three components. You just mentioned two, CHIPS and New York State. There's also the ITC-
Right.
that's been in effect. And by the way, of all of those, the ITC is the winner at 25% coverage. The CHIPS bill is, you know, capped kind of at 15%, and we got kind of our fair share of that at the $1.5 billion. So how are we going to deploy that capital efficiency of the CHIPS bill that works in conjunction with the ITC? There's three phases of our build-out. First of all, to have a true global footprint, it's not just having a factory that does one thing when you have a broad range of technologies. It's having factories that service customers across a broad range of technology platforms. That's what we have in Dresden, that's what we have in Singapore, and it allows our customers to dual source globally and locally.
Putting one technology node, I don't know, say, Arizona, right, that does one thing that represents 5%-10% of your technology platforms isn't a global footprint. It's more like a global mailing address. I can't help myself. Sam's giving me stink eye over here, so. So the first part, getting back to the CHIPS bill, is to use that funding to diversify our Fab 8 facility in upstate New York, we call it 8.auto, to bring in a broad range of diversification of technologies. So some of the things we do in Singapore can be built in the US, some of the things we do in Dresden can be built in the US, and then some of the unique stuff we do there.
The second element or phase of our CHIPS bill is to, we call the modernization of our 200 millimeter facility in Burlington. It's something like $130 million, which is to convert that capacity from CMOS-based technology to wide bandgap material, in particular, GaN. And that reconfiguring and some automation and some of the technology enablement comes from that bill. And then the last part of it is, when the time is right, when we need to invest in new capacity, we'll have the lion's share of that CHIPS bill comes in the form of creating of an expansion, a doubling of the size of the output of that Fab 8 facility, but it'll be built with that diversification of technologies.
So that's how we'll be using those funds over the, you know, the rest of this decade. But do not lose sight of the ITC and how important that is, because it's 25%. It's the biggest contributor, and we've already had offsets from ITC in our financials over the previous quarters.
On competitive positioning, you know, there's a perception from the outside that there is just a lot of capacity, right, that is being built in the so-called mature nodes. So one, I was hoping you could help us kinda differentiate that, you know, all mature nodes are not created equal, right? That there's a lot of specialization, even in the so-called mature node. But you still have, you know, new capacity, right, coming out from the IDMs, as well. You have Intel, you know, wanting to partner with UMC, and then I think this morning I saw that, you know, NXP and Vanguard are planning to add or build a new fab in Singapore. So how are you looking at the competitive landscape?
Yeah, I think so. One, is there a lot of capacity coming online? Yes. But do you believe the industry doubles in a 5-10-year period, then maybe that's the capacity we need to go do, and how do you go make sure you win where you wanna win? And I think it goes back to where I started this conversation with you. We play in a $50 billion SAM that's gonna grow to $100 billion. Last year, $7.4 billion company. We wanna grow, you know, with the industry, maybe better than the industry, but more importantly, we wanna grow where we create the differentiation for our customers. So when you play in a big space, you can allow a lot of the more commoditized parts of that business are less interesting to us.
A lot of what that announcement that came out this morning was, that's TSMC-licensed technology, that Vanguard is gonna do some of it in partnership with NXP because they're looking to consolidate some of their own footprint. Quite honestly, that, that might not be a business that's very interesting to us anyhow. Is it necessary? Is it part of the $100 billion SAM? Sure it is, but it's not business we'd be interested in. And then the other end of this is, you know, all this capacity that's coming on in China. You know, to have capacity that does more of the commodity part of this market is one thing.
It takes a lot of investment, not just in tools and in shells, but in the enablement that creates that true differentiation, and for customers to invest their IP onto a platform to be able to use it and leverage it. So as long as GF stays true to its mission, we are gonna be differentiated. We're gonna drive our technology roadmap, really specific end markets where there's great durability, profitability, and where differentiation matters for our customers. We'll be able to grow to the levels we need to be as a very successful business and not get caught up into the oversupply commoditization of certain elements of our industry. That's our strategy.
Got it. So as you can guess, AI is on top of mind for many investors right now. What is GF's play in AI? I imagine more at the edge, but I was hoping you could talk about, you know, how you are looking at the proliferation of AI at the edge and how GF can participate.
I'm gonna give you three ways I look at AI. First is, how does GF participate in the secular trends of what it means, just straight up in the tactical horizon? I'll talk a little bit about how AI can impact how we run our factories as a manufacturer. And then I'm gonna talk about the most important thing, I think that's, it's how AI is gonna reshape our industry. So, my words, the feeding frenzy that's going on in data center build-out right now, and you'll hear a lot of the dialogue that, "Look, I don't know how I'm gonna use it, but I better build it, because if I don't, I won't have it and everybody else will." And these, these are large companies doing that.
I think at some point that's gonna kinda run its course because what's gonna happen is all this capacity for AI is there, and there's not enough data to run all these models anyhow. And what's gonna again, is gonna drive to the equilibrium where AI at the edge has to be where the sensing, the data is created, and the inference is run. It's too expensive to do it. Every query to go, to spend the cost to transmit the data to cloud, get an answer, and come back. And so this will be the true driver of AI, the, the edge. It's that all this capacity in the cloud wanting data to do something with it. There'll be cameras everywhere, voice detection everywhere, sensing. And what will happen at the edge? A signal will come in, it's sensing, voice, image, whatever, right? Movement.
It will take this data and run an inference engine. What should I do about it, right? It will take that data and then parse it for what's important, compress it, and send it to the cloud, where a lot of that training will take place. I think the greatest example of that is what a Tesla car has done from day one. Hundreds of sensors around the car collecting data. That data runs models for ADAS, whatever level you want to run it at. That data gets sent to the cloud, where the large language model gets refined, and every Thursday night, you get a new program on your car. Now, maybe it didn't change much. It just had a bug in it, that new version, but it's been the ultimate edge device for the longest period of time.
Now think of that being every device at the edge. I think that's the opportunity for GF. Especially untethered devices, where power is everything. Not lowest power per transistor when it's running like a data center, lowest power of a transistor when it's in the off state. You know, there's surveillance cameras we do today that are battery-operated for very famous companies, and they want that camera to last 3 years, untethered, on 2 AA batteries. So what do they do? They take our 22FDX technology, which is the lowest power used in the off state. And how does the camera work? The device works. It has an SoC, system-on-chip. 2 microcontrollers, a very tiny one, wakes up only 10 milliseconds every second, and it has an algorithm.
Depending on what it sees, it either goes back to sleep or wakes up the big SoC to tell it to do something. Get that data, someone's at the door, notify, whatever the. It's an amazing demonstration that the future of our industry is not just the lowest power per transistor in the data center, where you're doing all this computation all the time. It's the lowest power for a sensor when it's off, which goes right to the IoT at the edge, and we positioned for a long period of time our investments to be that solution. Now, how does AI change our factories?
Look, I would tell you that as strong as AI is, when you live in the physical world and you have to have a technician wrenching on a tool to put a new PVD target or a new shower head, that large language model is not gonna help you do that with less labor or less, right? The energy to run our fab is the energy to run our fab. And so I think we'll be able to use AI to maybe make a 10%-20% improvement of how we run our factories, because not everything we do is in the digital realm. It's in the physical realm, and those models don't help you there. Yeah, maybe do better real-time dispatching. Maybe we do better management of reticles and batching of jobs. We get OEE.
Maybe we can use with a little less of engineering support, where the models help engineers make decisions on depositioning. I don't see it more than, you know, 10%-20% advantage of what we do today. Where AI has big input is when it does orders of magnitude of change, and this is what I think happens to our induSy. Our industry's been driven by two things: technology and economics, and the interrelationships. The whole Moore's Law was a technology that drove economics, and that's why people invested in the next generation of technology because the cost per transition went down every generation. Well, the economics of Moore's Law are gone now, right? It's mostly now you do it for performance and power.
But when the foundry was created, it was created because if you were a manufacturer of products or a product company, you had no choice but to build your own factories, and then scale became a problem for you. And along came Morris Chang, and he said: "Guess what? I'm the scale aggregator. I'll take all your volumes, and I'll create the scale you can't do," and was born the fabless foundry model. And then the fabless model grew as aggregators yet again. Designs cost so much. An individual consumer of a microcontroller could not have their own device because the design cost was too much. So I'll become the aggregator of a fabless, as a fabless company, to aggregate all these demands to make these standard products that you system companies can use to take advantage of that. Along comes AI.
Just imagine, AI can take the cost of doing design down in an order of magnitude, order of magnitude. Now, the whole economic matter of a model of aggregating demands don't matter. Now, almost any system company can start with a product notion, what's the software I want to run? And then lead that right to a design. So there'd be massive customization of designs. I call it a super cycle of SoCs, right? That everybody will have a unique combination of capability to drive their product because they don't have to take a bunch of off-the-shelf parts to create a product, because they don't have the scale to do one, if the cost to do the design is brought down that much. I think this is what...
It's gonna change the nature of the number of part numbers that companies can run and how system companies will become more and more end-to-end, and how the role of fabless companies maybe go less about standard products and more custom products for every one of their customers. What it's gonna drive is a whole new class of customer sets for companies like GF to deal with and a much higher, higher number of SKUs to go run in our factories. And I think that's the most profound impact to the semi industry, besides driving growth, is what are the business models that will be successful?
Very insightful, and I think even Jensen said next phase, more physical AI, right? Rather than just in the cloud. So maybe just in closing, Tom, just kinda help to wrap this up.
We weren't very good to the audience. We didn't give that much time.
I think you have laid out a target of, like, 8%-12% growth, 40% gross margin. Any kind of the top two or three things that get you excited, product drivers, that will help you achieve those objectives?
So look, we are a company, when we have top line, we know how to, each and every day, make it better and better at the bottom line. That's just the maturity of our business. So I love the fact that we've made the investments to get to $10 billion. I love the fact that when we get to $10 billion, we now are in that scaled model where we have both capital efficiencies or the ability to drive free cash flow and growth and get even more efficiency with, with, with government support in some of these investments. I love the fact that we continue to win. You know, 90% of our design wins are in areas where we are single source, that we, in aggregate, are accretive to our long-term model.
And so for us, you know, it's the growth phase of our industry, the upside, we can't come soon enough. We spent the better part of two years getting ready for that, both in capacity, differentiation, and we're ready to rock and roll when this industry wants to come back.
Excellent. Thank you so much, Tom. Really appreciate your insights.
Thank you. Appreciate it.