GLOBALFOUNDRIES Inc. (GFS) Earnings Call Transcript & Summary

Great. hi, everybody. Welcome back. I'm Joe Moore, Morgan Stanley semiconductor research. Very happy to have with me 2 newly mented executives in their current roles. Tom Caulfield, who's the Executive Chair of GlobalFoundries and Tim Breen, who is the CEO. Thank you. Welcome, guys.

Thanks for having us, Joe.

So you guys both -- maybe I don't know what the answer is, but you reported a good Q4, you had some positive commentary on 2025. Can you just kind of recap where we are with GlobalFoundries in terms of the business environment?

Yes, I'll talk about 2024, maybe Tim will talk about '25. So in '24, we started the year, we said Q1 of 2024 would be the low point. And then sequentially, we would -- we grow revenue, we did all that. So Q1 low point, Q4 at the high point. The thing that was sometimes gets lost is the fact that we generated over $1 billion of free cash flow, 2 consecutive years of declining revenue. It just shows the leverage in our business.

Yes. It's a really nice aspect of your business.

Yes. And then we said, look, going as we head into 2025, that year-over-year Q1 would be growth, which we've given that consensus and that the full year of 2025 would have growth. And the rate of pace of that growth really is about inventory continue to burn down and the macroeconomics of the consumer wanting to spend. We see a little bit of the uncertainty that's creeping in right now in some of the new policies. But I still think there's -- based on where inventories are the end markets we play in, that 2025 is going to be a growth year. Magnitude, again, depending on consumers. Maybe Tim, you could talk a little bit about the end markets and how we see it.

And I think the way we think about it, Joe, is think about the macro overall, what's the sort of conditions, what's happened to inventory and where are we sitting in the inventory digestion cycle? And then what's the GF idiosyncratic share dynamic? And each one is a little bit different. And so you take our 4 big end markets, auto, we've had good growth year-on-year now every year since our IPO we forecast that continuing in '25 on the same basis. That's much more a share story than it is a macro story, although obviously, there's a good story of content increase in the car pulling through. In our communications infrastructure and data center business, we've gone through the end of the transition from the GPU, CPU single-digit nanometer transition days to now what's really ramping in optical networking, power delivery, SATCOM, other areas of that market. And so what we said bottom down in '24, we see definitely growing it to '25. Mobile, relatively flat, unit growth, not necessarily huge. Our content growth holding steady, especially in the high-end handsets, so let's say, in line with market. And then IoT has been the hardest one to call, where you've seen quite a lot of inventory digestion, we've now seeing a lot of that. And so we'd say like modest growth from that base given how the inventory is behind us, including a small share gain as well.

Great. We'll focus on some of that. Maybe we could talk a little bit about the transition. So it was announced shortly before your earnings release that Tim, you'll be succeeding Tom as CEO and Tom stepping into the Executive Chairman role. Can you talk about your focus areas in the first year and outline some of the biggest opportunities? And maybe just give us for people who don't know a little bit on your background as well.

Sure. So we've had the pleasure to work together for the last 7 years in the business, and it's been an incredible journey from the company we were in 2018, very much focused on the leading-edge transition to where we are today and diversifying our end markets. I couldn't be more excited about the strategic positioning the company has today. 3 things I think about that are very relevant there. Number one is we focused our R&D now for many years on the differentiated technologies that play to those secular trends within all those markets, I mentioned, whether it's auto or the data center or mobile or IoT. And so we positioned the portfolio quite diversified and quite differentiated in those areas. Customers wanting too more with us, and that's -- there's a lot of drivers for that, including some of the need for supply security diversification and that's gone through both aspects of the cycle. And so we have definitely a high amount of traction. We also have customers we hardly do anything with today, and that's an upside for me that we can go after. And one of my goals is to penetrate some of those accounts. And the last thing is the geographic footprint, it's been an evolving story, and it's evolved from what was definitely around diversification supply security, now into thinking about how do you, as a company navigate tariffs, navigate a different trade environment. And so I think that's another tailwind that sets GF apart. The way I think about it is our strategy has changed that much in internal transition like this, you're not seeking a change of definition, but you want to change your pace. And what we want to be able to do with different -- the 3 of us playing different roles between Tom, myself and Niels Anderskouv is really accelerate the pace and driving that destination faster. And I think we couldn't be a better place to do that.

Yes, I think that's the biggest part of the transition. No one is going anywhere. We're taking on different roles, and it's the best way to use the 3 of us. Niels now becomes the COO, with Tim being the CEO. And then I get to free up some bandwidth to think about more strategic things for the company.

Great. Well, maybe, Tom, you could talk a little bit about that, the decision to move into an Executive Chairman role, what that means for you, where you'll be spending your time in that executive role?

Look, the first part of it, we'll talk Chairman first in the executive part. I'll take over from Ahmed, who's done an amazing job as our Chairman to do the right governance and the things that protect shareholders and how we manage our company. But on the executive side, there's things that really need more time to focus on. First, when you think back to 2020, with the new administration and how important semiconductors started to become every one's radar. We spend a really important amount of time and an important resource for the government, understanding the industry and how to go -- create that supply chain security, diversification and local sourcing. And so me getting the opportunity now with some bandwidth to be constructive with the U.S. government and the administration is an important part of that mission. The third thing for most companies is the kind of the horizon 3, 5 to 10 years out, we need to spend more time with universities and research institutes to leverage the work we do there to help steer that innovation towards GF. And it's an area I've always had a passion and now I'll have a little bit more time to spend on that. And the third thing is we've kind of earned the right over the last 7 years or so to be a little bit more strategic in this industry. The things that we can do to build on our company, not just organic but inorganic. Typically, those things just take more time and effort, and we need to create that bandwidth. And that's why I view this organization that will create bandwidth for both Niels, Tim and I to be able to do that.

Well, congratulations on the role. And I think -- congratulations really on, as you said, the last 7 years, the position you were in chasing EV without scale to really pursue it, doing all kinds of things that maybe weren't customer-friendly before you came into the role, and you guys have really done a good job of kind of cementing your role in the semiconductor economy.

So the operative word is you guys, Tim and I started this 7 years ago. And the amazing thing about it is we both got to touch so many parts of the business just because they needed it. And Tim has been a copilot this whole time. So a very natural transition for us to be doing this.

Great. Good to hear. Thank you. So let me talk about geopolitics. You guys talked about the importance of that with the change in administration, focus on tariffs everywhere. The supply chain geographies are back into focus maybe with a different slant than we had into the CHIPS Act. Can you talk about how GlobalFoundries is positioned to take advantage of that, and just in general, how you think about geopolitics for you?

Yes. Look, I think as Tim highlighted it before, our footprint is more and more becoming an advantage. And we seem to focus a little bit on the U.S. part of that and we should. But remember, we have great government relationships and footprint in Singapore and Germany, and we need to make sure we nurture those relationships equally. When I think about what's happening in the U.S. and what our customers are telling us, it first starts with -- if you have a global footprint and it's a mailing address. It doesn't do anything for your customers. So what does it mean? You need to diversify all your factories. So you have the opportunity to multi-source for them. And we spent the better part of Tim's tenure as the COO doing that diversification. It wasn't about bringing new things in as much as it was, making sure that every platform or a large fraction of our platforms can be multi-sourced in the different regions. But I think about what the U.S. is trying to do right now in bringing semiconductor manufacturing, into this country, it needs 2 things. When I think of the ITC and the chips bill, it was about creating global competitiveness to add capacity, but it didn't solve the demand part. And at a very high level, maybe a simplistic view, I think the tariff part now creates the demand to match the capacity. Because what's the sense of building capacity, if no one is incented to bring the demand, the taps could be the incentive to bring the demand. I still think you'll need some of that funding to make the cost competitiveness level the playing field versus adding capacity elsewhere. It's a little bit of a fluid situation. But again, net-net, chips, ITC, create capacity, tariffs to create demand and have those 2 come together.

And maybe if I follow-up a little bit every time I see you in this forum and other places, I've made a big deal out of these automotive OEM relationships that you constructed because automotive OEMs didn't know who GlobalFoundries was before the shortage of a few years ago. Now the shortage is ease is there's still that awareness of the supply chain? Are people still very focused on what things are being built and their ability to procure in all types of environments?

Yes. I think it went from there wasn't enough supply and they needed awareness for that. Now with the geopolitical [indiscernible] they say, well, now I got to make sure I have the supply in the right spot, not just where the total less. And what we're finding, and Tim coins the phrase all the time about the silicon awareness. Look, we have this whole transition is going to take place in AI, how is AI going to change our industry and the cost to do a design and the barriers to enter for OEMs and for Tier 1s and other system houses to decide to do their own designs, because AI will democratize that cost structure. And those are the things that we're playing for. Leveraging the fact that there'll be a broader set of customers and a broader set of needs of where they want their stuff built. This is the kind of the future for GF.

Great. Maybe we could talk about China as an opportunity. I feel like people over the last few years have focused on the CapEx and what happens with China local sourcing. But it seems to be also an opportunity for you guys. Can you talk to that a little bit?

I mean we have a relatively small amount of China business today, and that's why I think of it actually has an upside more than a risk for us. And there's sort of 2 drivers of that. One is you have plenty of Chinese customers who have global aspirations in terms of their businesses. And for them, they honestly think about supply diversification in the same way the rest of the world thinks about the other way around. And so they talk about non-China and non-Taiwan sourcing. In a most cases, they're actually diversifying away from Taiwan and away from domestic manufacturing and saying, good for me, the U.S., Europe and even Singapore be a good solution for my business and think about the auto OEMs who have global ambitions, but also industrial companies and others, a handset makers and so on. So number one is we're an avenue for that and that is driving some good business our way. But the second piece of it is we have customers who sell significantly into China for their products, particularly in the automotive space. And they're saying, I do want to have some manufacturing locally in China, and that's where we're exploring partnerships where for a specific product, a specific technology for a -- very much a specific customer, can we establish a local partners to make that product and that product only. And I think we feel there's some good win-win opportunities there where our customers get what they want in terms of local sourcing, but we're not going to enable a broad kind of technology license to a bunch of companies who will then compete with us, but really have made those specific technologies where they have that low demand. And those 2 things by the way reinforce each other because with that, you have more engagement on the ground, and you can build on your business. So I'm actually quite bullish on the opportunity for us in China given the base we're starting from.

Yes. And the key is the answer for us isn't to go create additional manufacturing footprint in China. Better to do it in the partnership [indiscernible].

Okay. Great. And then just in general, the global footprint that you have, I mean, I think there's an obvious sort of advantage from being a U.S. manufacturer, but you mentioned you have facilities in -- addressed in Singapore, I think, still coming in France. Can you talk about the opportunities that come from those additional geographies?

Yes. I think it comes back to we could build what our customers want on the technology platforms where they want it for a lot of different reasons. The key is to make sure it's diversified. So you say, well, if I want to source in the U.S. well that's great. I have a factory, but it doesn't build what you want. And so we have the ability working with European Union, Singapore government to continue to add modular expansion. Look, if we were -- if we had to go create a global footprint today, that incremental cost of going greenfield is you -- to that first wafer might as well be a gold bar, right? Now if you can take scale, you already have and then add capacity first within the 4 walls, and then maybe add new modular space and clean room space. It's much more capital efficient and your return on those investments come much quicker. Especially when you add capacity, that unlocks existing capacity you have because the first tools you'll put in your modular expansion, are pinch point tools. So for us, when we think about our global footprint going forward, it's not about creating new areas of adding capacity. It's about expanding the existing facilities and keep driving scale there, U.S., Singapore and in Germany.

And Joe, if you remember in the last upturn, people wanted capacity, but they also wanted it yesterday. And so when you got a footprint like that, your time to market is much, much shorter and because we spent this time in the last few years, it's not been easy about cross qualifying the fabs, transferring technologies. We have a lot more optionality about where to build, so we can hit the right market moment as quickly as possible. So I think preparing for the upturn has been the word of the day for the fab because we know this industry will go through the cyclicality. The question is how can you respond quickly when it does.

And the first response for us right now is kind of our utilization. We have roughly, call it, 3 million wafers worth of capacity depending on mix, with productivity improvements, things that we have in place, we have time to grow into that capacity as customers forecast maybe higher demand in that continue to add capacity above that.

And that effort to be able to build stuff in multiple geographies. Does that -- the customers have to recall? Is it -- how difficult is that? Can you just give a sense.

Well, the best way to do it is upfront when you do a product qual, just do 2 fabs. It's always painful to have to go back and do the workover again because then you're tying up engineering resources to do a redundant thing rather than a new thing. And so upfront, our customers are being a lot more thoughtful about dual qualify. And we need to get to a point to them, once it's dual qualified, we'll just decide where we'll build it for you. You don't have to decide which location you're on the building.

I mean, other foundries have multiple locations as well, but it seems like that's -- I don't hear other people offering that quite as much as maybe you are.

I don't think they do it at the same level of diversification. I think they're just -- they're more locations than they are diversification.

Yes. okay. Got it. So maybe talk about growth, where do you see growth coming from the next several years? I mean, at the IPO, your focus was automotive, and you've definitely executed to that in one of the more challenging automotive environments. Can you talk about the next few years and what markets you're looking for growth?

Yes. I think we're obviously very bullish about the long-term growth drivers across our end markets. I mean the transition of the auto space hasn't fully happened. It has a long way to go in terms of client growth. Regardless of the pace of electrification we see all the content that we're playing in, whether that's sensing, imaging, battery management, all of these areas are going to be driven by what's happening. So this age of the software define vehicle definitely is a tailwind for us longer term. And again, we're starting from a relatively small base of share. So we have plenty of upside, I think, in auto growth. The data center, obviously, that's where the industry is obsessed right now with the amount of CapEx going into the data center. And while we're not playing in that CPU, GPU arena, we're focusing on all of the problems around it. And the 2 big problems that everyone sees is, number one, how to get my power in the data center as efficiently as possible and avoid 5% to 10% losses all the way from the grid to the chip, and that's a big value driver for the hyperscalers and for those investing in data centers. But then also obviously getting data to those GPUs and so on that are hungry, they sit idle half the time and more than half the time, because they can't get the data that's there and some of the innovation we've done initially in areas like TIAs and drivers, but now into pluggable and linear optics and now into copackage optics, really provide a good road map for data delivery in the data center for more efficient processing and lower power. And so data center, we think long road of growth and an area we're actually probably ahead from a technology point of view. Obviously, there's a lot of work to do to get those architectures mature and see its transition. So I'd say data center, very, very exciting. SATCOM, we've seen a very strong growth there. Think of it as covering the remainder of the world with satellite connectivity in a cost-effective manner, obviously a number of players have driven that transition. That plays to RF, it plays to amplifies, areas we know very well from our RF franchise. So we see good tailwinds in that, which we're heavily engaged on. And then mobile, I don't think we're ready to count mobile out as a form factor. And we think probably the device ecosystem will change, but the needs will remain the same: connectivity, low-power compute, display, control and so on. I think about form factors like glass is, I think about -- we're not done with kind of fully building our personal ecosystem of devices, and that could drive big inflection points of content growth for us. And we continue to focus on, again, where can we be differentiated to win those sockets as they come to market. Some of them will look like the phone socket, but some will look completely different like the display in linear glasses, for example.

Maybe if I could follow-up on that. I remember back when we first met Tom, I would sort of describe GlobalFoundries as like a legacy node company and you said, hey, we're trailblazing in these nodes. We're doing new technologies. A lot of the things you just mentioned pretty exciting stuff. Do you feel like others are as focused on areas like co-packaged optics or wide-band gap different materials and capabilities?

You made a point is about how you define innovation. The industry spent the best part of its first 40 years, innovation was the size of transistor. It doesn't mean it's not, it's an important part of the mix. But the innovation we bring in adding embedded memory for security, adding connectivity for range and quality of voice on your phone. These are all the things that we spend our $600 million a year on R&D to be better with the customers. We were meeting with a customer yesterday was talking about they're going to win because they're going to give better performance at the endpoint of range of sill towers. And so the most important thing their customers are telling them, you got to keep me from having to build more infrastructure right, and make sure that you have the ability to have the power app and the LNA work at the limit of the range of my infrastructure. And this is the only way we can do this is with your technology. There's no other solution for it. Those are the stories we like to hear because those are the ones we've invested in. And the key for us is the measure if we're differentiated, it means we have single source business and still a high fraction of our design wins and our funnel are on things we've invested with differentiation, so customer needed to make their products winners.

I think the other part of it, Joe actually to build on it is the innovation is not just process technology innovation. It's also how the design and technology interact. And you'll see more from us around think of it like the next iteration of design enablement. How do we provide much more reference circuits for our customers to do things faster and learn and accelerate. We're seeing that in GaN with the acquisition we made there and so on. So that will be an area that we'll continue to innovate as well.

Our customers are actually asking for more of that content and building blocks in that enablement. And that becomes another as Tim said, an area of differentiation.

Well, and I saw that at the time of the IPO already, you were developing as kind of more of a partnership culture than what had existed a few years earlier. And I guess maybe segueing to the long-term supply agreements. At the time of the IPO, you had several years of coverage, obviously, it's different when there's oversupply versus the shortage that we had. But how are you finding the long-term supply agreements? And how much -- how important is that to your business going forward?

So first start with, why did they come? They came because customers wanted certainty of supply and certainty of price. Our customers were driving more of that than we were. And the reason we were interested in it is because we were making investments to create the capacity they needed. So it's this win-win partnership. At our peak, we probably -- we had in excess of $30 billion of long-term revenue under those agreements. We sit here today, we're around $15 billion because they amortize it. Most of those will put in place in 2022 and some have been renewed. So they think the era of LTAs is over, we're only kind of halfway through them. Now some of them needed to be restructured and rethought through given the dynamics of the marketplace, but they always created a framework for how do we go use this to create more business, more longer term and durable business for GF. When you think about who wants LTAs today, it's customers that have sockets that last 10, 12 years. Think of it the automotive industry. They win a socket; they're going to need that. And so certainty for them is not next design cycle, next year, I could be designed out. I have to be careful about what I signed up for, they're thinking about a long-term road map. They know they're designed in for 10 years. And so they're the ones that are very interested in not only creating new ones, but extending the ones they have as they work through it. And a lot of our automotive business is tied to the long-term agreements because of the nature of that business.

Okay. Great. Maybe talk about capital deployment. You've had a lot of free cash flow generation. As you said, it's one of the strengths of the model that it becomes much less capital intensive when things low. So you generate a lot of cash. Can you talk about uses of cash and how are you thinking about those priorities?

Job number one keep generating cash. Job number two, keep generating cash. The deployment of it, we have to be smart, there always comes in 3 flavors. One, what's the best use of your cash on your balance sheet debt versus cash. You've seen we've done 2 buyback of debt just because it was -- it made good business that's a good use of our cash. The second one is, is there a way to give it back to shareholders. And then the third one is keeping dry powder so that you can -- you build your business in an inorganic way. And I think for us, that's how we think about our cash. Number one, generate it. Number two, continue to generate it. And it's a good problem to have, especially where Tim and I started 7 years ago. It was in fact your firm that lent us $500 million for 365 days just to get us through the tough period and sitting on cash today is a little bit of a luxury for us. But we'll make sure we always do with the right thing for our shareholders and keeping a certain amount of that free for us to be nimble if opportunities come across is an important part of that strategy as well.

In any sense of what our inorganic growth might look like? I mean is that just be fabs, would that be capacity or capability? Just how not looking for specifics, obviously, but just how would you think about inorganic?

I think it's about handy bolt-on capability that's relevant to your customers in the end markets that are growing disproportionately higher. What are we missing? What ingredient are we missing? Are customers asking us for more reference designs? Is there something we can do on that front that would bolt on and accelerate that.

The theme will always be highly customer-centric, something that customer is driving for and how do we accelerate our differentiation in the area. And like I say that the definition of differentiation is changing, including some of the softer aspects because customers want to focus their dollars on software and other things they want to do. So they want actually more innovation from us on things like IP and so on. So you'll see more in those areas.

Last year, we bought a company called Tagore, and it was all about bringing in reference designs. Our customers wanted us to have to build on. Now we could have spent a better part of 3 years doing that ourselves versus just bolting that on. Those are the types of things we look for. And it was the customer that led us to that acquisition and what they wanted to see from us.

Okay. Great. So before I open to the audience, so anything else do you want to leave us with today that we didn't capture in the questions.

I let you go first, and I'll...

Sure. I think look, we're obviously very excited about it. The industry goes through its chapters. I think the long-term tailwinds are very strong. And I think that gives us a lot of confidence that we can leverage this model that we built from a customer perspective, differentiation perspective to grow. I think we're very happy also with the financial model that we can generate free cash flow as we grow. We can accrete margins as we grow. And so this is an area now where in that flywheel mode. And obviously, we have to execute through that. But we couldn't be more excited about positioning of the company.

Yes. Look, first, it starts with revenue. Well, where does revenue start? The sockets we're winning now. These are the sockets that become topline growth for us. The beauty about our model is the flow-through is outstanding, high fixed cost business, even in a down from our high of $8 billion of what we did last year, right? Huge ability to generate free cash flow. So if you look at from our design win funnel that's going to drive our revenue growth and what does that mean in the financial performance of the company, it's super exciting. What we need is a little bit of the economy and consumer demand to pick up to make the electronics industry hot and exciting again.

Great. Well, I appreciate that. We'll see if we have any questions from the audience. We have one over here.

What's your [indiscernible] and foundry and particular [indiscernible]?

Can you just read the question for the webcast because we didn't have a mic.

Yes. So look, for us, RF [indiscernible] is GaN. We think there's the right combination of digital performance and power or RF you could bring. If for us, we don't see a good play in [ SiGe ] because there's not a lot of digital you could bring to the mix. So it doesn't -- there's not a lot of margins you could all provide in a foundry fabulous type of model. So our focus is [ SiGe ] and gallium nitride in those materials. And we have leadership technology in both of those, some of it in 200-millimeter and some of it in 300-millimeter.

And a lot of RF capability in CMOS as well, right? You have a lot of RF capability in CMOS.

Yes, yes. [indiscernible] CMOS.

[indiscernible]?

Not at this time, it's typically on the laser side, and that's more of a -- for us, that's more of a buy than [indiscernible].

Thank you for that question. Any other questions up here in the front.

So you mentioned your LTAs. You mentioned auto is a big part of that. Just wondering, sort of last year 2024, what proportion of your auto revenues would have come from LTAs?

Yes. I'm trying to do a little math in my head. The lion share, I would say.

And is that going to be true this year as well?

Yes. Generally. It's a market were customers desire for certainty. And given the long product cycle is motivating them to have and extend LTAs as well. In fact, we extended 1 LTA in that space early last year as well for that reason.

And the more I think about it, I can't think of too much of that business that wasn't on the LTA, again, because of the nature of it.

It seems like those customers are behaving a certain way when lead times get short, but there's still the memory of what happened 2 years ago and still a desire to avoid that ever happening again.

There's a memory and then there is the acting on the memory, and I think that's a mixed bag. It's very bespoke to the particular customers and how they think about their supply chain.

But what you don't have, Joe, is you don't have a socket switch win and losses that drive uncertainty for them. So they can commit, they have to think about inventory, but they don't have to think about, I may lose next year, in which case, what do I do with this, right?

Yes. Any other questions? Sorry, one in the front.

How you think about that [indiscernible] CHIPS Act impact, how you think about that [indiscernible]? And then the second question -- first question, just CHIPS Act impact, how you think through it? And second is how are you thinking about all the AI A6, I think it's been publicly talked about your relationship with Groq and other companies, how you're thinking about that market?

Two very diverse questions. So let's talk about the -- and I touched about it before. Look, it's a very fluid situation. But to say what's the endgame. The endgame is U.S. wants to have more semiconductor capacity for a lot of good reasons, economic, supply chain and sovereign security, you need to do two things, I think, to have that happen. You have to create competitiveness on the capital investment required to create that capacity and you need demand. And I don't think one or the other is enough. I do believe you need CHIPS and the ITC to create the capacity, and the tariffs is to kind of create the dynamics to make the demand want to come home. And I think as we get through a little bit of the fluidity of the situation right now, that's where this thing will land. We'll find that the combination of the two is you'll get the outcome you're looking for. And maybe there are some days where CHIPS looks like it's not as in favor or not. I think it's just too early, that over time, understanding of how demand and capacity you need to play off one another, we'll let this thing land in the right spot. Remember, ITC is 25%, CHIPS is 15%, so there's still big incentives for capacity. And then your second question, I wasn't quite sure.

Yes, so the [indiscernible].

Yes, [indiscernible] accelerate. Look, why don't you take that? And then I...

There [indiscernible] two categories , right? I mean Groq is a very specific case. And we're very excited to see what they're doing. And they've picked their niche cool inference and so on, and I think they're actually building a pretty good play there and good luck to them. That's a product that we've design a long time ago together, and we're making for them. I think our broader AI content will more likely be what I mentioned around the data center, power data delivery and so on. We're obviously not going to compete on the GPU, CPU and kind of core AI accelerator market. However, you'll find AI workloads at the edge and at the endpoint, which needs, for example, low power and you have other issues, connectivity integrated and so on. Those are areas that are more in our wheelhouse. So there's plenty of growth in AI. And we get the question a lot about what do we think about things like DeepSeek is, it just proves that model innovation is only just getting started. Model innovation, especially smaller models, means more applications running on different devices, especially that edge endpoint, which we think is a good tailwind for us.

Just building a little bit on your question. When we think of companies like Groq or PsiQuantum, these relationships that started 6, 7 years ago that are actually getting to the marketplace. Now the life cycle in our industry, we get used to sometimes the product life cycles are quick, but some of the semiconductor build-out of startups getting to real value, you have to -- you need to have a wide net all the time because you never know who's going to be the big winner in this. We think we have the best partnership with the best opportunity for quantum compute and PsiQuantum, Groq another example, started 6, 7 years ago, right, with even a little bit different business model, not knowing where the world is heading and now, they're their player in inference. So for us, it's to make sure we're constantly engaged with start-ups and give them disproportionate amount of attention so that they stay with us. And some won't make it, but what they do is they become big customers in the future.

Questions? We have 45 seconds left. So we'll just wrap it up there. Thank you, guys. So much. Congratulations on your role. Thank you.

Thank you.

Thank you.