Unless I had a name that wasn't the right.
We have the right. We have the right company name. But we'll get that fixed. Okay. We're gonna get started. So I'm Tim Arcuri. I'm the semi and semi-equipment analyst here at UBS. Very pleased to have Tim Breen, who's the CEO of GlobalFoundries, and Sam Franklin, who's the CFO of GlobalFoundries. So, thank you to you both.
Good to be here.
Thank you.
It's great to be here. Thank you.
So let's just talk about the demand environment. Maybe you can talk a little bit about the current demand environment. How differently have the end markets played out over the past few quarters? And are you seeing anything differently than you mentioned in your earnings call? You said handsets were softer, you know, autos and data centers stronger.
Yeah. And I think we should separate. That's thanks for having us here. I think we should separate a little bit the dynamics for us particularly versus overall market backdrop. I think what we're seeing is, remember after a pretty unprecedented 2022, early 2023, you've seen for sure across broader markets that. Leave the data center aside for a second. You've seen kind of an inventory digestion, a general level of macro, weakness. I think as we've come through 2025, it has slowly improved. You know, those what were headwinds are becoming, let's say, balanced to tailwinds. You see inventories coming down consistently quarter on quarter. You see general consumer sentiment better. And you start to see a little bit of green shoots around new product launches and so on, definitely on the handset side.
So, I'd say if anything versus a year ago versus a quarter ago, better overall market demand. I think the one pocket from a market point of view that clearly everyone is talking about, data center remains really strong. Lots of speculation about how long that will go on for. Our conviction is we're very much at the beginning of that build-out phase. There'll be some, you know, puts and takes and bumps along the road, but overall that has a lot more to give. But overall very strong now and pulling for sure on a lot of content for us. I think there are markets for us where maybe the market isn't as strong, but our share story has been quite strong.
Auto has been a good example of that where we've been gaining, you know, from a de minimis business when we went public through to now quite a large business for us approaching $1.5 billion this year. It's been a double-digit growth even in a pretty flat market. That's just a story of good execution, good patience, and progression over time, so.
Great. Let's just talk about the geopolitical environment. You would seem to be well-positioned amid the geopolitical uncertainty. Your business hasn't really recovered that much. How much of a tailwind do you think the geopolitical situation is for you? And how do you expect, you know, tariffs to impact mature nodes for the,
Yeah.
Foundry industry?
So I think, you know, this journey is actually one that has intensified but didn't just start very recently, and so we've seen a trend of, you know, people being concerned about sourcing, you know, individual risk items. That was, you know, for sure exacerbated and made real during the pandemic, but I think you've really only seen this year companies set really clear policies, and so now it's no longer the case that, you know, phrases like NCNT, non-China, non-Taiwan. That's not a rare thing to hear in a supplier conversation. It's the default in most supplier conversations where if they haven't asked for it before, they're asking for a non-China, non-Taiwan supply chain, by the way, either because they themselves are doing it or because their customers are asking for that.
I think some markets led the way, automotive being a primary example, led the way to set that up earlier, but I think all markets now have some version of that, embedded in their sourcing strategies, and so that's, you know, leg one of pulling content, you know, to the likes of us and so on, but then leg two is U.S. specifically, and so now you also see U.S. sourcing as not a nice-to-have criteria but a must-have criteria in new platform decisions and so on, and I think so longer-term, very, very strong tailwind for us. We believe that will contribute to meaningful share gains in the next kind of, let's say, 5 to 10 years of the industry. I don't think of that as a short-term phenomenon that comes and goes, but more about secular share gain. It is not a panic-buying decision, right?
It's not driven by short-term policy that's gonna unwind in a short period of time.
And maybe just to add one comment to that as well. I mean, look, independent of everything that's been happening from a geopolitical point of view over the course of the last year, we actually made, you know, very strategic decisions, you know, in the 2023 timeframe to actually start diversifying our footprint, to provide even more diversity for our customers as well. Our multi-fab is becoming a great example for that. You know, if you rewind to when multi-fab was put in place, it was largely a FinFET fab. You know, we've taken the decision now to migrate in technologies on our 22FDX, our 28nm, our 40nm. And that's gonna serve all of those end markets that we have today as well as Silicon Photonics, which has already served datacom.
You know, notwithstanding those geopolitical dynamics, we've been driving towards creating this capacity, this fungibility, and this optionality for our customers for quite some time.
I guess, Tim, back to your point. Your point is really that the geopolitical tailwinds that they're still out in front of you.
I think they're very much accelerating now. So I think you see companies very publicly stating their strategy. And, you know, we have, you know, eight significant buyers of semiconductors. We announced earlier this year when we announced our U.S. plans. We had another four or five, and we announced our U.S. plans. And I think when you see some of the largest companies in the world, some of the best supply chain managers in the world commit to a, you know, like I say, non-China, non-Taiwan, or a U.S. sourcing strategy, you can start to take it very seriously, and I think, like I say, that's a good tailwind for us.
Can I ask you about your sole source mix? I think on the earnings call you mentioned 90% of design wins over the past four quarters are sole source to you. And it's been that way now for some time, actually. Yet when you report your revenue, it's two-thirds of the revenue's sole sourced. So and that hasn't changed much in the last three years. So you're winning way more than what you're reporting as revenue that's sole sourced. Is that just because of the time it takes for these designs to start to become revenue?
Yeah. I think it's very relevant to the diversification we've gone through, right? And if you think about markets like auto, which we've been very focused on, auto is notorious for a very long design cycle, but also a very long production cycle, right? And so I think we've been winning more in slower-to-ramp markets, perhaps not winning quite as much in shorter-cycle, more, let's say, dual-source type markets like mobile. And that really plays through in our revenue mix this year. And so I think as you start to see kind of those stabilize, but also the, the wins come through in things like auto, that mix of the overall revenue should increase as well.
Is there a way to think about, like, I mean, this is sort of a generic question, but is there a way to think about the margin for a sole-source piece of business versus a dual-source piece of business? How much of a tailwind can that be as more of your revenue shifts over to sole source?
So I think there are lots of different drivers of mix, of mixed shift impacting margins. For sure, the sourcing strategy. If you're getting awarded on a sole-source basis, it means the customer went, first question, does this, you know, this technology enable me to win in my market? Can I do something different that I couldn't do elsewhere? By the way, they look at the sourcing location now, again, another factor. Can I make this in one or more locations? And to Sam's point earlier, having actually optionality of saying, I can make this automotive MCU with, you know, high-performing embedded memory. I can make it in Singapore, Dresden, and the U.S. That's a pretty good value proposition. We've seen them willing to pay a premium for that.
And so I'd say, yeah, when you win on that basis, you're typically winning at higher ASPs than you might win on a dual-source piece of business and so on, for sure. And as that mix plays through the system, you obviously see an accretive kind of dynamic on our margins.
Great. Let's talk about some of your segments. So in handsets, you recently called out having made some pricing concessions for some longer-term share gains. Have we seen the positive impact yet from the share gain, or have we just seen the negative impact from the pricing? And that's still out in front of you.
I think when you look at ASP, you've seen a bit the ASPs change as a result of that. And we got, you know, some questions on that in our last earnings call. And I think, like we said, it was our initiation. It wasn't something that was done to us. It's something we actually chose to do because we said, look, as a business with a certain number of capacity corridors, what's the optimum structure for the right amount of share in that market, in that socket, not just today, but also longer-term? And so I feel in those ones, you know, we've seen both the impacts in one go, slight uptick of utilization, but also at slightly lower ASPs, but net-net accretive from an overall profitability point of view, for the business.
But again, very limited to choice customers where we are in that dual-source position where doing something like that could actually yield some incremental business that otherwise wouldn't be available.
Great. And let's talk about auto. So, I mean, you've seen very strong growth from that, you know, vertical over the past few years. How do you see the pace of activity continuing in the coming years based upon the design activity today?
So, look, auto has been a long-term story for us, and the winds we're seeing today are not things that happened, you know, a year ago. They're often things that happened three, four, five years ago in terms of that momentum. And we positioned ourselves to have a great technology suite for automotive, right? You think about all the things that automotive needs. It needs high-performance microcontrollers. It needs secure imaging. It needs radar. It needs battery management, the whole slew of content. We positioned our portfolio very much against that. We positioned the fabs as automotive-grade qualified. By the way, it's hard to do it in one fab, let alone to do it in five fabs around the world. So again, we positioned the portfolio to be really well-tailored for automotive. And I think that's playing through now in winds that we're ramping.
But actually, the design win activity has probably been, if anything, increasing, right? And I think if any industry got the once-bitten-twice-shy, you know, memo on supply security, it was auto, right? And so auto has also been a big proponent. And you've seen that in deals we've done, even with OEMs like GM, who've been a constant kind of partner. Well, how do we bring more stuff to the U.S. and more content in this part of the world? You see all the ingredients to see that, that secular share gain. If anything, our design win momentum in auto is picking up. It's more diversified, more applications, because the content in the car is growing significantly. So even if units sold out of the dealerships are not growing, content's increasing, and our portfolio fit for auto is really good.
You know, to foreshadow a little bit what we'll talk about more and more in quarters to come is we think auto is the prelude to what AI will do in the physical world, right? It's a device operating out there with increasing intelligence, sensing content, distributed compute, actuation, motor control, all of those applications, and a ton of communication within the vehicle from vehicle to vehicle into the cloud. And so, you know, all areas that will, I think, play out well as automotive becomes autonomous vehicles in different formats, delivery drones, all these kind of applications that need to be mission-critical, safe, all of these things.
Great. Let's talk about CID, comms, infrastructure, and data center. There's a lot of things going on in that segment. Yes, it's only 10% of revenue, but it's growing more than 30% year over year. So it, you know, really is adding, you know, three-plus points to your growth each year, and that can add up pretty quickly. So can you just break down the dynamics inside of that segment?
So I think there's three kind of big, big blocks we should talk about in that segment for us. Two relate to the data center. One relates to kind of more broader infrastructure, including Satcom, which we also include there. So data center story, you know, outside the GPU or the XPU in the data center, the two big problems are, you know, networking and power, right? I think very commonly, you know, acknowledges bottlenecks or challenges to data center performance. We've been very, very focused on networking with Silicon Photonics. And we can talk more about what that has been for us as a journey to build a technology really probably before it was interesting, right, for the market, but now for sure showing those inflection points.
Solving those critical, you know, networking points where copper just can't cut it anymore, for scale-out networks, now increasingly also for scale-out. We're seeing the first real growth trajectory of that Photonics offering. We're augmenting that with inorganic growth. And we can talk more about that as well. That networking piece, we see that as a very strong secular driver for us. That's a $100 million business last year, $200 million this year, more than $200 million this year. Really good growth going into years to come before we add the inorganic component on top of that. And of course, that's the Photonics part. But in optical networking, there's also other content like high-performance silicon germanium, some of these other technologies, which we've also invested a lot in that really enable that high-speed connectivity, which is essential for modern AI data centers.
So, call that block one. Block two, probably earlier in its journey, we start to see the pull-through of data center power, being a big area that basically you have a lot of difficulties converting 800 volts to 0.8 volts at the chip. Any loss is a massive loss to the system overall, a massive loss to the economics of the data center. So there'll be a lot of innovation. There's a lot of companies out there doing cool things in power conversion. And we start to play in that space. Some of our recent GaN announcements also play to that space with low-voltage GaN playing a key role in the next 5-10 years of data center power and so on, and the last kind of really good growth driver in CID has been Satcom.
And so the proliferation of Low Earth Orbit satellites, you know, this is a completely new form factor from a device point of view. We never had before the option to go to Target, buy a device that costs you a few hundred bucks, gives you 100 megabit connectivity. You can put it on your car, on your boat, in your remote, you know, cabin. You can put it in a rural area. And that's suddenly a new, you know, unaddressable market until recently where you have a massive component of RF content, right? Think of this device as a massive phased-array antenna. There's beamforming. There's all sorts of devices to steer those RF signals because it's not going to a cell tower two miles away. It's going to a Low Earth Orbit satellite 300 miles away.
And it needs to do a 100-megabit connection, not, you know, one or two megabits that we used to remember from old, old Satcom or old flights and so on. So look, these are big growth drivers where you suddenly see an inflection point of content. And we've been working to position the portfolio against that.
So you mentioned Silicon Photonics. Let's actually talk about that. It's tracking to, I think, $200 million this year, but it's expected to reach $1 billion by 2030. How should we think about that as a key driver for your business? It seems like, you know, pluggables this year, obviously, but, you know, how's your outlook on, you know, CPO and how that can help your business?
I think we've spoken a little bit about why the transition is important. I think if you go across the industry now, there's no more debate about kind of if versus when. There's a question of when for different architectures. Pluggables have become dominant. If you went into a you know AI data center today, you'll see you know optical transceivers everywhere, optical pluggable modules everywhere. They're getting more and more sophisticated, and more and more efficient. Again, that pulls content for us. That's largely the business of Photonics today for us is in the pluggables space. What we've added with the AMF acquisition is also more strength in the pluggable space. In many ways, the real inflection point comes when you can co-package the optics with you know the switch ASIC or other components in the rack.
That's when you have a lot of improvement to connectivity. You really unleash bandwidth. You deal with issues like beamforming density that literally the space you can have to put connectivity into those devices. Again, copper cannot cut it. Copper generates too much heat. Copper can only take one stream of data at a time, even with advanced kind of retimers, redrivers. With optics, you can multiplex. So you can have two streams, four streams, eight streams, 16 streams in the same fiber. So now you're seeing the real kind of secular drivers of that. For us, we invested in CPO well before, like I said, it was, it's interesting. It goes back even to the IBM days for GF. So we've been doing a lot of work on device, on material, even on packaging. How do you package this device?
And how do you do complex things like a detachable fiber connector so that the device can have a connection to a fiber, you know, multiple fibers for a board? But if a fiber breaks, you can replace it rather than have to replace the whole board. So we've systematically worked on cracking a lot of the high-volume manufacturing challenges. And so now we see the real pull-through of that as that CPO co-packaged optics adoption comes through. We've said about 2027 as a big, big inflection point for that. Our $1 billion is kind of a bottom-up number based on current engagements. I would say, if anything, maybe we're being a little bit conservative as that adoption takes place. But let's get to 500, then we can talk about where the numbers will go.
So just on that point, it would suggest that you're gaining a lot of share. I mean, obviously, it's, you know, Tower and you are the two big players. Their stock has done very well because of their focused exposure in that market. But you have a, you know, good amount of exposure relative to the size of that market. How can you better highlight that business? I mean, there's some hot startups, you know, a few of which are here, and, you know, these are your customers. Are there capabilities you're missing in that end market that you could go out and acquire? Or is it just riding the wave of the market growing and it becomes a bigger piece of your business and people start to focus on it more?
I think, look, everything, you know, the market focuses on the next problem, the next inflection point. I think we are seeing that interest now, as you say, when you have these startups with multi-billion-dollar valuations and so on. The market is clearly saying, and investors are saying that these are real businesses now that we need to take very seriously. And by the way, it's not just because of, you know, optical networking and things. It's also things like quantum computing, which also has attracted a lot of investor interest. So I think the market piece is coming. I'm not worried about that. I think what our focus is on organic execution, right, really delivering these wins. These are high-volume wins, high-velocity wins for critical applications. We need to deliver. That's a big part of it.
Our customers, you know, work closely with us to make that happen. What we have been doing on the inorganic side is saying, where can we accelerate through either adding capacity, technology, customer mix, AMF, which, you know, we acquired recently, very good fit for our business, different customer mix, right, different and complementary technology. But in Singapore, where we have a great manufacturing operation, so we can integrate very synergistically with our current footprint and, you know, adding run rate of $75 million of revenue on a base that, you know, that's quite meaningful on the base that we have for Photonics, but with significant growth potential and a modest acquisition price, around $400 million, we paid for that business. And so relatively, you know, accretive for a small acquisition like that.
And so we'll do that to build, you know, acceleration. But we're also investing, and we've announced this also recently, in design capabilities to help our customers design faster, get to market quicker, build reference designs, something we have a lot of familiarity with in other parts of our business so that they can go from concept and architecture to, you know, functioning optical engine a lot quicker. An acquisition we recently made in Egypt is a good example of a team that can do that and has been doing that as well, so.
Great. Can we talk about your non-wafer revenue? You saw strong momentum in non-wafer revenue in the third quarter. What's driving that, increasing mix and how to think about its contribution to gross margin?
Yeah. I think the strategic aspect, and I'll let Sam comment a bit on the kind of the dynamics that puts and takes. But I think the strategic aspect is we want to add more value to our customers. And so we don't have to define our role in the industry as we make what people need to make. We can provide IP. We can provide know-how. We can do NRE. On top of that, we've been doing that for some time, but we can scale it more. And so a good, you know, factor that will contribute to that in the future has affected less so so far is things like MIPS.
And so we acquired, you know, arguably one of the, you know, best platforms in RISC-V processor IP as a means to add value to our customers with an IP-type service, but also NRE and potentially also custom silicon in the future. And so you start to see, you know, additional sort of service lines that obviously come with a different set of economics. They don't require CapEx to invest to grow revenue. And they also have a different margin structure compared to kind of traditional wafer business. So I think that will, over time, drive more non-wafer revenue. But I think some of the other aspects of non-wafer, by the way, includes things like mask and reticles for new tapeouts. As your business grows and you have more momentum, you have more tapeouts, right? We've had more designers. We've had more tapeouts.
As a result, you have more masks and so on. So I think these are all good leading indicators of future revenue growth, as well. Sam, you want to add?
Yeah, sure. So, you know, you asked a question around margin more broadly. And I think what's encouraging is that these strategic initiatives, and, and MIPS is a great example of that that we've taken over the course of the last year and, and that we'll continue to look at going forward, is really creating those mixed benefits, not just from a wafer revenue point of view, but also from a, a non-wafer revenue perspective as well. So, you know, we've, since IPO, been roughly in the zip code of, call it, 8%-12% of revenue being comprised of non-wafer revenue. As you rightly say, Tim, you know, the last quarter, we were at about 12%. Our expectation for the fourth quarter is that we'll be a little closer to, to 13%. So really starting to get beyond the upper boundaries of where we've trended to be.
Now, that is really a function of where we're seeing that greater offering to our customers as we're broadening that services. You know, it's masks. It's reticles. It's recurring engineering, non-recurring engineering activities. You know, it's a great leading indicator for where we think about the growth from our wafer revenue perspective as well. And so, you know, we talked briefly on this kind of momentum of design wins and third quarter design wins were up almost 2x from a year ago. You know, that translates into tapeouts, which translates into production revenue as well. So, as Tim says, you know, we're broadening that ecosystem offering to our customers. And when you think about non-wafer revenue more broadly, it comes through at quite an accretive margin relative to our overall corporate enterprise targets.
Is there, like, a rule of thumb where non-wafer revenue usually translates to wafer revenue within two years? Like, what's the sort of general rule of thumb?
I think it's market dependent. I think it's the unfortunate answer to not make it easier. You know, take automotive, right? The automotive sequence between a design win and a tapeout and high-volume ramp, because you've got test cycles, you've got reliability cycles, can take a couple of years, you know, even sometimes three or four years, whereas a mobile product, some of the data center products can go a lot quicker. So I think it really depends, so there's a mix in there as well. I think the bigger point is the overall increase tells you more revenue is coming because, you know, it one way or another leads to the next step of high-volume manufacturing.
Just on the non-wafer revenue point, you did mention MIPS. MIPS is immaterial, sounds like, in the back half of this year, but it sounds like it could be upwards of $100 million next year in non-wafer revenue. Can you just talk about it? I know you're going to host a webinar, so you're going to probably give us more on that, but that event. But can you just talk more about the strategy around MIPS and, obviously, you're trying to, you know, recapture more of the design ecosystem with the customers. But like, where can this go?
Yeah. So I think it's really interesting. You've seen RISC-V, you know, being talked about for many years now, but did it really receive massive adoption in the market? Not so far. And the reason has been, you know, a few fold. First of all, it took time for the ecosystem to mature and to deepen for a lot of, you know, code and so on to be written and tested and validated. But what you also didn't have is you didn't really have too many, you know, scaled companies who people could rely on as a partner to build processor IP and deliver that.
MIPS had actually a lot of good offerings from a technology point of view, but they needed a kind of industry, I'd say, broad, you know, almost, you know, platform agnostic, you know, a company like GF to back them, to vouch for them for the longer term, as an owner to really see the momentum that customers could use this as a tool. Because customers, meanwhile, were struggling with the idea of saying, "I like RISC-V. I'm not so sure about relying on other providers of IP, but I can't do a full RISC-V design on my own with my internal team because the economics just don't work." So they actually want something like this.
And we were really heartened when we did MIPS that actually a lot of engagement from both existing GF customers and new kinds of customers who said, "This provides us the sweet spot we need. Now let's do, you know, a particular core, a custom project together because you're complementing the skills that we otherwise have." And meanwhile, that ecosystem is deepening even further. And obviously, we can contribute even more to that now with a bigger scale. And so I think you're going to see RISC-V play an increasingly important role. And what, you know, preview of what the team will talk about tomorrow is we think that's really, really important for that physical AI transition, right? Again, we don't want to build processors that are going to work to be, you know, a complementary or a competitor to, you know, an NVIDIA GPU.
We'd love to build a distributed compute processor in IP and cores for a, you know, robotic arm or for a software-defined vehicle zonal controller, things like that that do real-time processing out there in the field in harsh conditions. By the way, aerospace and defense, same, same driver. We see a lot of engagement from customers on this, on this front.
And so you've seen so since the deal was announced, you've seen customers come to you and say, "Hey, we were thinking about using MIPS, but we didn't really know about the ecosystem. And so now it's like, game on.
You know, great technical engagement, but they really need to build a long-term plan. Not just that they want to use MIPS now, but now it gives us an angle to say, "But listen, GF process technology can also play a role in the solution." Of course, I think there's a symbiotic relationship between, you know, the IP and the architecture and the process technology. You can tailor one to support the other. That means the GF process technology, particularly, you know, FDSOI or FDX platform, is really doing very well. It can be increasingly tailored for those kind of applications. For customers, not only do they get good RISC-V IP, but they also get a process technology that's really tailored to exploit all of the benefits of that IP.
Great. Let me, I wanted to ask about gross margin. So, earlier this year, you thought you could get to 30% by the end of the calendar year. It isn't happening, but mostly because of a slower recovery in the market.
Yeah.
So can you just talk about the levers on gross margin as we enter next year?
Yeah. So I think bigger picture, and I'll let Sam add it to it as well. I mean, we're pretty close, by the way, in Q4, based on what we've guided. I think all the factors are contributing in the right direction, perhaps not quite at the pace we expected. As you say, the recovery may be a little bit slower than we thought at the beginning of the year, but all of the tailwinds to gross margin, I think that we've been focusing on are there, right? Accretive product mix, improvement of utilization, where we have a footprint today that could probably accommodate a $9 billion top line versus the $7 billion, which is where we are today, and the fall through of that additional wafer you sell is not at the standard gross margin.
It's much, much more accretive because you're not adding a lot of fixed cost to do that. So utilization plays a very important role in that. And then, of course, you know, year on year, we work on efficient CapEx investments, efficient cost control, and cost management. And so all, I think all those ingredients are in place. If anything, I'd say that we now have other means to even improve further. Things like MIPS that come with a different margin structure, longer term, can also contribute. We're standing by our kind of 40% company goal. That's what we're driving to. Obviously, our goal is to make progress towards that in 2026. We're not going to guide 2026 now, but the goal is to make meaningful progress towards that. You want to add, Sam?
Look, we've spent quite some time talking around the fact that as we continue to remix the business, as we continue to see more diversification come through from revenue and the end markets, that we'd expect to come back as a more profitable dollar-for-dollar company. And what's encouraging is that we're starting to see that come through now. So if you look at our third quarter results, $1.688 billion, that was flat on a quarter-over-quarter basis. Margin was up 80 basis points quarter-over-quarter. But actually more pronounced is, although call it roughly 3% down year over year from a revenue point of view, margin was up about 1.3 points. So we're really starting to see this momentum come through in terms of the quality of the mix shift.
If I fast forward to the guidance provided for the fourth quarter, you're right, Tim, midpoint of that range is 28.5%. But look, upper end of that was 29.5%. So we're somewhat playing around the edges here in terms of where we expected to be at the start of the year. But general momentum is what's encouraging for us at this stage. So, you know, call that roughly two and a half points of quarter-on-quarter margin growth, you know, on what would be a comparatively to slightly below revenue in the fourth quarter implied guidance. And that's really where, as we think about this over the long term, it's kind of marching towards that continued profit fall through.
You know, all the reasons that Tim mentioned, you know, the scale dynamics, the mix dynamics, the continued growth within the various end markets, comms infrastructure and data center is a great example, right? You know, in the third quarter a year ago, about 7% of revenue. We're now seeing that in 10% of revenue for the third quarter and continue to expect that to grow. So, you know, we're quite encouraged with the momentum we're seeing from the margin fall through perspective. You know, we'd expect to continue to see that grow as we go into 2026 and beyond for all those reasons we just talked about.
Great. Last thing that I wanted to ask was you recently brought up the opportunity for more capital returns next year as the free cash flow has gotten a lot better, which is new for GF. So the question is, what would this look like, should we, and how should we think about the split between repo and dividend when you just consider that the public float's pretty low?
Absolutely. And as you can imagine, this is an area that is a nice situation for us to be in as we think about the continued flywheel of free cash generation that we've seen over the course of the last couple of years. You know, we were back in 2023, a little over $300 million. We grew that to over a billion dollars in 2024, and we expect to be on target to that roughly billion dollars in 2025 as well. Now, capital allocation obviously takes many different forms. We've taken, you know, real concerted efforts over the last 12 months to optimize the balance sheet. We've taken out a lot of our long-term debt. And so really, the way we're thinking about it going forward is, how do we look at this framework of shareholder capital allocation?
I don't want to preview anything too soon, but clearly, format is a consideration there, whether that is, you know, repo buybacks or, or standard dividend program establishment as well. So stay tuned for more on that. As we mature as a public company, this is a good situation for us to be in.
Yeah. It's a good problem to have, so.
Yep.
Anyway, we're out of time, but thank you for the time.
Thank you.
Thank you.