Okay, we're gonna get started with the next session. I am Tim Arcuri. I'm the semiconductor analyst here at UBS and pleased to have on stage, we have Hassane El-Khoury, President and CEO, and we have Thad Trent, CFO. So thank you to Hassane and Thad.
Thank you.
Thank you for having us.
Well, let me just start out with a high-level question. You know, since you both came to the company, you've been on a journey to have more market focus, price better to value, slim down and optimize the manufacturing network. Can you talk about where you are on that journey? Are you sort of done with the optimization and it's more about growth from here, or is there more optimization left to be done?
Yeah, look, we've done a lot, but it, the journey is always long because you-- it's never mission accomplished. So where we are, we've done a lot of optimization, what we refer to as Fab-Liter , where we divested four Fabs. We really focus our manufacturing on where our portfolio is going to be, which, like you said, is purely based on value drivers versus, you know, just mass market or Fab pillars. So we've done that transformation. Right now, it's strategic focus on the growth. However, from an optimization perspective, we talked about, in our last Analyst Day, how we are going to Fab Right. What does that mean, really, is when you have a complex manufacturing network, you know, if you have two Fabs, both running seven technologies, neither of them are efficient.
So streamlining technologies in where you want to manufacture it reduces costs across the board in both Fabs. You know, do three here and four here, for example, gets you all seven much cheaper from a cost efficiency perspective, therefore increases the margin. So that's still yet to come. Now, from monetizing on the efficiencies we've had already, you know, the four Fabs we've divested, we talked about $160 million of benefit. That will happen over the next few years as we fully exit the Fab. You know, we handed the keys over to somebody, but you still have to get the products moved to our internal manufacturing. That's when you get the benefit. So that is still yet to be reflected on the P&L.
But net of, you know, the day-to-day or the operational efficiencies we still have, we're happy where the portfolio is. We're doubled down on silicon carbide, we're doubling down on the analog domain that goes with our power domain. We have reshaped our image sensing group to be high margin, high profitability, and the focus right now is R&D, innovation, and new products.
I mean, I guess even in your, even in your, you know, legacy product portfolio, you sort of increased prices to over time, price yourself out of that market. You know, things aren't actually going away as fast as you thought. You know, maybe you didn't raise prices enough, I don't know. How does that, you know, how has that journey also influenced your, you know, how you operate going?
Yeah, you know, we always talk about, you know, pricing. We never, we don't do cost plus. We price on value. Now, value can be system-level value, where the customer sees the value in the system, and they will pay more for it because they can monetize it on their side. Specifically for these, you're right, we have said we priced ourselves out of the market, because we knew that the customer has other opportunities and other chances to buy it from elsewhere, and the margin profile for that business was not favorable or was diluted for our portfolio. So what we've done is we priced ourselves out of the market. We've walked away already from about $450 million of that business.
There's a remaining business that if we don't lose by the end of this year, it's good business. Now, why did the customer not go if we indeed priced ourself? We believe that it's not we didn't raise it high enough. It is because there is perceived value in there, even if it's not at a system level. You know, being able to place an order from onsemi and get it, high quality, always a roadmap and competitive products, tells me that that's the value that the customer is putting on it, and the switching costs are way higher than the benefit that they will get on the system level.
You know, if you get something for $0.05 and you take it to $0.10 in a $200 bond, are you really going to jeopardize that for a $0.05 saving? That's the, that's the key, that's the value, is the consistency, is the supply assurance and the high quality, which does have a price impact on it.
And you had, you know, talked about those four Fabs. You know, Fishkill has been a persistent headwind to gross margin. It's been a long journey there to bring the Fab up to your needs. What is left to do? I mean, they were making RF CMOS, and you make power analog and maybe some other stuff, and it's very different in terms of the complexity of the products versus what they were making there before. Is it really about the commitment to the prior owner falling off so that it becomes less of a drag on gross margin, or is there more optimization of the toolset that you have to do within Fishkill?
Yeah, I would say it's more about the cost structure in there, right? So, yes, it's a different product set, different technologies, but that's always been in the plan, right? So, the company had acquired this or made the definitive agreement to acquire it three years ago. We took ownership this year. We got surprised by the cost structure. The only value that is being, you know, cycle times, chemical costs, material costs, all of that. We have a large manufacturing footprint. We know exactly what that cost should be. We've got to go optimize that cost and get it into the benchmark of where it needs to get to. Now, it's a fully loaded Fab today, so you can't break it, right? So we have to be very careful.
So I would say it's blocking and tackling in terms of how we've got to go take that cost out. Of course, we're bringing equipment in and things like that, but that's always been in the roadmap. That isn't the challenge.
... we think by the time we exit 2024, we'll have that back under control, and it'll be at our cost structure and no longer a drag. So currently, it's about a 250 basis point drag on gross margins, which is significant.
Yeah.
It's blocking and tackling, we know what to do. We're on track to drive that cost out.
So if the commitment to the prior owner, if those volumes persist for longer, that doesn't, that, that's not the issue? The issue-
No, no, because, look, as a part of the transition, we've been winding up over three years, you know, prior to taking ownership. Now, they're gonna wind down for three years while we continue to wind up. So it is a defined capacity that they will take. We're doing foundry services for them. Obviously, we don't wanna do that long term. We wanna load it with high margin, high value product versus doing foundry services. So that's the process by which we'll move products in, we'll qualify new products. Everything new will be, you know, going into East Fishkill. Everything possible will be going in, but, but that's the planned ramp that we've had for many years.
Yeah, and if I talk about progress of our products going in, you know, we are already running low, medium, and high voltage. So our power products are already running in East Fishkill, qualified and shipping for revenue, part of our baseline today. And we have started sampling our image sensor within that taped out of that Fab as well. So the progress on moving products in and starting to generate revenue on these products has already in the journey, and the work that we have to do is really on the cost structure, like Thad said. As you get that cost structure back to benchmark against our own Fabs elsewhere, all these products will just get that benefit, and the margin expansion will happen. That's how we get out of that 250 basis point dilution.
Got it.
Let's talk about silicon carbide. We could talk for the, you know, rest of the time about silicon carbide. You know, obviously, your big push since buying, you know, GTAT two years ago, it's now up to low teens% of revenue in the Q4. You have a unique model because you have the wafer device and back end. Can you talk about sort of what's the true source of your competitive advantage in silicon carbide? Why are you winning, and why are customers willing to commit what is probably... I mean, you know, you're not giving a number anymore, but it's probably at least $10 billion still worth of LTSA.
Yeah, let me. I'll take it in a few pieces. I've always been consistent as we win because of the technology. Meaning, you have to have the best technology out there, and when I define technology, I talk about device and package together, because that's what the customer sees in their end vehicle or in their end system, if it's an energy system, energy storage. What that means is you need to be able to drive that efficiency. That efficiency is driven by technology, which is device and package. That's why we win, that's why we win against a lot of our peers. Now, where does GTAT play? That's the supply assurance, why we win big.
Meaning, if you're able to have the best technology that the customer can get the best efficiency on their system, and you are able to have a supply assurance and resilience to commit to that customer, that's how you get to a leadership position. If you don't have the best product, it doesn't matter if you have all the capacity in the world. Nobody, nobody's going to-- no customer is going to say, "I will sacrifice my end product competitiveness because you have a supply assurance." But having the best product and the supply assurance together, that's the vertical integration strategy that we embarked on, that's the winning platform. But it always starts with compelling and innovative technology. That's what we have. That's how we started with.
I guess, so said a different way, the whole, this perceived risk of China, who's... There's 15-- You know, some of the equipment companies talk about 15 new silicon carbide, you know, Fabs in China. You're, you're sort of indifferent to that, because even if they are successful, you could always, you could always buy their wafers and, you know, you have the best device, you have the best packaging.
Yeah. So one thing we have to be clear, when you say, you know, they're not Fabs. I know, I know, you know, it's fungible, but they make substrates. If you look at our substrate facility, it's not a clean room, it's a big kinda depot. Well, it's more than a depot, but it's not a clean room environment, so that's why I wouldn't call it a Fab. But you're right, there is capacity coming online in China, and if it's high quality and the price is lower than what we can make it internally... Because remember, it's people talk about, "Oh, it's competitive in China." Well, it has to be more competitive than what we do with the vertical integration, which today, we have the best gross margin in that business than anywhere in the world, fully, fully loaded.
So I believe we have a cost advantage. I believe we have a technology advantage. If it becomes available, high quality, and of course, we have to disregard the geopolitical risk for it, then we have a different source that we can source from, which is fine. Today, we are sourcing from China some of the substrates, because we wanna evaluate everything that's out there. First, to benchmark against, but also we said over 50% is from internal sourcing, which means that there's 50% external, not just in China, but in other areas of the world, but we always look at what's available out there. But it has to come to cost and cost advantage, really, otherwise, we have it internal.
Is there some asymptotic percentage that you're pushing toward to have captive on the substrate side?
Yeah, we've always said we wanna be about that, you know, call it 80%, give or take. And that number is defined by, of course, availability outside, competitiveness outside, CapEx. But the reason, you know, kind of that 80% plus or minus is important because we're not gonna build capacity to a peak in demand, when if you have two or three customers ramping all at the same time, and then the steady state kind of goes to a different level, you don't wanna have a peak capacity installed, because then by definition, you have underloading all the time. So when you have 80%, we can, you know, spring up externally and then have always fully loaded manufacturing, and that's the best, call it P&L performance that you can have.
So you're working on 200 mm. I think you're shipping samples at the end of next year. And is it pretty easy for you to swap out a couple parts on the 150 mm tools and just use them for 200 mm? Are you adding some automation to that factory to help improve the yields and quality? Maybe just where does that ramp stand? And, you know, most importantly, I think people have this perception that there's risk to that ramp. So can you talk about that, too?
Yeah, I'm smiling 'cause I always find it kind of funny when people ask me, "Are you gonna add automation to your Fab or to our back end?" You know, we ship about 60 billion units a year, and we have probably one of the best margins of most of our peers in the broad power space. Having said all that, you can't perform that well unless you have full automation. So from our baseline today, our Fabs are state-of-the-art Fabs. Even our six inch silicon carbide is fully automated, soup to nuts. So investing in automation is kind of a non-event because it's already automated. But to tackle the question directly, going from eight inch, you have – I would say in two phases.
You have the boule growth, which is the substrate growth side of it, and then you have anything from that on. From wafering, epi, and Fab, we're already eight inch capable. We do it currently on IGBTs. So our Fab is eight inch capable, and it's running high mass volume of eight inch. We've already been running eight inch silicon carbide in that Fab. So from a risk perspective and, you know, yield readout and so on, we have that whole thing baselined already. The conversion, you know, if you think about the CapEx light that we have to do, is on the furnace.
Mm-hmm.
We build our own furnaces. We've been building our own furnaces for the last few years, part of the GTAT acquisition. If you look at a furnace itself, you will see no difference between the six inch today and the eight inch tomorrow. Inside is gonna change, just higher diameter.
Mm-hmm
... kind of capability inside. That's where the CapEx intensity, the lighter CapEx intensity, comes in. We don't have to build a structure, plumbing, it's just the core of the furnace. That we've been doing already. So we have furnaces already running eight inch. They're running eight inch internally. That's how we baseline the Fab and flush everything through, and we're still on track to get those samples in 2024 and production in 2025.
How much of a cost advantage is that gonna provide? I mean, obviously, cost is king, but the substrate, I think, is 40%-50% of the cost of a, you know-
Mm
... Silicon carbide of a MOSFETs. So how much of a cost advantage? It isn't like going to a larger wafer for an integrated circuit. I mean, there's a significant cost advantage there. Here, there's more of a marginal cost advantage. Can you kind of talk about that?
Yeah. The focus for us on going to eight inch is really about capacity rather than cost.
Mm-hmm.
We have a very good cost structure on six inch today. You know, we talked about the margin performance of that business. You know, last couple quarters ago, we talked about the operating income for that silicon carbide fully loaded business being in the high teens. So from a cost advantage. And we're winning, like you said, you know, the LTSAs are a large number of LTSAs. So the move for us to eight inch is driven primarily by capacity increase, so we don't have to add more furnaces to get that capacity. We can go to eight inch, lower CapEx, you just get a boost in output automatically all the way through Fab. That's the main driver in the short. Over the long run, you're exactly right.
Over the longer term, eight inch will start becoming more favorable from pricing as it gets more maturity on, on the technology. But in the short term, our focus is on enabling the ramps and enabling the business that we've won with the lowest CapEx back to the, you know, what we-- Thad presented in Analyst Day. We've got one of the best ROICs in the, in the industry today, and we maintain that as a metric.
I wanted to shift to autos and maybe there were two issues last quarter. There were two different issues that I think people sort of comingle maybe, you know, one issue. There was the silicon carbide issue at the one OEM, and then, you know, number two, there was just general softness from tier ones in Europe that were not related to silicon carbide. So on those two points, really two questions. First, can you give any more color on the silicon carbide issue? It sounds like it was a U.S. OEM, maybe related to a new launch. And would you characterize it as cyclical, meaning that you get that back? Or structural, meaning that it's gone?
Yeah. Let me... So it is one OEM, I don't wanna characterize regional for the obvious reasons, but it is a single OEM. It is purely driven by demand, so it is not a supply, it is not a share loss or anything like that. Purely end demand. What we have considered, you know, what we talked on the call, is we de-risk 2024. What that means is we're not assuming that we're pushing it out and it's gonna just recover in 2024. If it does, great, that's positive. But from a risk mitigation perspective, we just took it out of the number, and the baseline is gonna be just different. Now, 2024 number didn't change because we were able to allocate that capacity, you know, that capacity, per se, to other customers that needed it.
So that's the puts and takes from that. So overall outlook for that business, even short term in 2024 and the longer term, remains unchanged for us on silicon carbide, because we've always said it's capacity constrained. So there are other customers that we weren't servicing 100% that now said, "Hey, we'll take that, because we have strength in our business." So we're able to regard the business different mix, not with that OEM, but with others. So the mix is different, but the outlook remains the same.
So you had said, I mean, you're not, you're not reiterating the number, but, but there was a time when you said $4 billion between 2023, 2024, and 2025. So the point being that, that the outlook, that hasn't changed?
It hasn't changed. What we talked about, given, you know, all of the the market outlook, we just said it's gonna be, you know, 2x the market.
2x.
You know, our growth is gonna be 2x the market. Now, if you put 2x the market on where we're gonna end in 2023, it puts it back exactly where it is. That's why I said it doesn't change. The metric we're given, the 2x market, is a different metric, but in 2024, it leads to the same number.
I also think it's worth pointing out that at that one OEM, we're still ramping. So our revenue was up in Q3, it'll be up in Q4 at that one account, and in 2024, it'll grow over 2023. So there wasn't a loss here, it just in demand is what came down.
But it's still a growth account.
But it's still growing.
In that case, or not, not just in that case, but in a case where a customer inside an LTSA says, "I, I don't need as much as I thought I did," what do you get in return? So is this a situation where you say: "Okay, fine, you can push that out, but, but, but for me to let you to do that, you have to give me something in return?
Yeah, and this is, and I'll, again, I'll cover it as a general engagement with the LTSAs. You know, I've always said in these conferences or other conference, that, you know, the LTSAs are legally binding. The minimum they get us is a phone call versus the backlog disappearing. You know, if we didn't have an LTSA, the backlog disappears, and then we'll be sitting going: "You know, what the-- what's going on here?" You get the phone call, so you start the engagement. So when we get the phone call, we figure out, is there any stranded inventory? We can't have stranded inventory. That, that would be a win-lose. So we have to work on no stranded inventory. We have to work on share. Sometimes we got share gain, you know.
We, in some accounts, we say: "Okay, we had 60% share." Well, if your denominator goes down, our share has got to go up to 80 to make the number, so we get more share. If the demand is net down and we have high share, or we are, we're a proprietary solution provider, then okay, is there somewhere else, different technology, that we're able to ramp where we're maybe a secondary supplier? Because from a customer, especially customers that co-invested with us, they want to monetize that investment, so they will move share to us. That's a win-win. So in certain cases, we did get more share in at those customers, either within the quarter, if it's fungible, or at least in the following quarter or year as we, either qualify or if we're qualified, we ramp.
Those are types of win-wins that we engage with at the customer level.
And on the point of LTSA, so, they came down from $20 billion, I think, to $18.4 billion last quarter. You know, silicon carbide also came down inside that number. But did it come down simply because customers are working off of the LTSA? There was no backfill to those this quarter because things aren't as tight as they were, and so customers just don't feel like they need to book inside of LTSAs. Is that-
Well, I think there's a couple things you got to keep in mind. One is we shipped, right? So you got another quarter behind you in terms of what you shipped off.
Right.
The other thing is, as we sign up LTSAs, they're further out as well, right? So, so that's a lifetime number, but when we get further out there, we discount those back, right? So, just not counting, not just adding them up and counting them, right? So there's a number of puts and takes there, and there's also some adjustments to the LTSAs. But I would say that trajectory is still very solid, and, you know, we announced on our last call, we're still signing new LTSAs. We have customers that are expanding their LTSAs, extending their LTSAs, or even those that didn't have them, that are coming in and saying, "I want to get on an LTSA." Even in a down market, we're seeing customers sign LTSAs.
Okay. Maybe you know, for you, Thad, you said that revenues will be down in March, you know, normal seasonal is down 2%-3%. Clearly, there's a lot of inventory adjustment happening right now. What are some of the puts and takes? I'm not asking you to guide March, but what are some of the puts and takes on March? And, you know, and relative to that, it sounds like gross margin stays in the 45-55% range, roughly, in the first, you know, half of the year, and maybe grows a bit in the back half of next year. Is that kind of the right way to think about it?
Yeah. So, look, you know, we're not guiding March, you know, but, you know, normal seasonality for Q1 is down 2%-3%. I think line of sight kind of tells us we're probably in that ballpark right now based on what we're seeing. Our Q4 guide was down about 8%, so coming off a low base. So I think that tells you where we're thinking and what we're seeing, and we also said we're gonna take utilization down further. So that's the number one driver in the short term, is utilization. So we'll take utilization from about 72% to the high 60s, and we'll manage there until we see a recovery in the market. So I think in the first half, we're probably gonna be running in that range.
You've got this headwind from silicon for EFK. You've got a couple other things, but if you step out into 2025, there's a number of drivers in margin. EFK rolls off, silicon carbide becomes accretive at that point, not just at the corporate average, but accretive. We've got the four Fab divestitures that Hassane talked about. That's $160 million of fixed cost that we start to recognize in 2024 and 2025 and beyond. So, beyond 2024, there's really a lot of growth drivers again, and assuming that we take utilization back up in a recovery, back up into the, let's call it, you know, high 70s, we peaked out somewhere, you know, 83%-85%, there's a nice tailwind on margins.
So there's a camp in analog and power that is, the lead times are still compressing, and therefore, our business is still getting worse. Bookings are in free fall, and you know, even March is biased to the downside seasonally. Then there's another camp that, well, you know, lead times have basically stabilized, bookings have stabilized, bookings have come back in the last few weeks. So there was different tones that we heard from, you know, some of your peers. So are you more in the "bookings have stabilized" camp because lead times have stabilized? Or are you more in the "things are still getting worse" camp?
Look, we're not optimizing for lead time. I'll tell you that. So our lead time really have remained on, you know, ±1-2 weeks. So from a lead time perspective, we're not optimizing for lead time because a lot of our products are, you know, proprietary products, and the LTSAs give us much longer visibility than what the backlog puts in. You know, we have LTSAs, on average, four to five years. So we know four to five years kind of what the run rate is, you know, plus some adjustments that we do in the short term. But we're not waiting for backlog, so. Because what - how do you bring lead times down? You build inventory, and the worst time to build inventory is in a softer market.
We actually did a proactive approach of we took utilization down to keep a lid on inventory until we see a recovery, rather than keep the Fabs kinda running at a virtually higher capacity or utilization just to build inventory and reduce lead time. We're taking the approach of 2024 is kind of we're not planning a second half recovery. If there is one, like Thad said, we'll turn everything on, and it will be a all tailwinds, but we're taking a much cautious approach. I don't think things are deteriorating, you know, back to your comment. I think things are stable. You know, some pockets, industrial, we talked about it before. People are asking, "Oh, we didn't talk about industrial on the last call." It didn't get worse.
It's down, but it didn't get worse from where we thought it was gonna be. So I would say things are stable, but I wouldn't say in a recovery mode.
Mm-hmm. Autos sound like maybe they are getting just a touch worse because there is some, you know, knock-on effect in Europe. Would that be fair to say, that autos are still maybe, you know, skewed to the downside in terms of the market?
Yeah, I mean, if you look at auto, I mean, you have the EV that's gonna be growing. The rest is gonna be with the market, you know, the general market. You know, we talked about interest rates kind of putting a damper on the end demand, but that's a mixed difference. You know, EV are still strong for us. So if you look at our auto number, we're actually very favorable on auto overall because of the EV.
Mm-hmm.
Right. You know, silicon carbide is going to grow in automotive in 2024. So it depends on where, where you see. I think the big range that you see in the market depends on what people feel on the overall market. You know, even the range for us for 2024, when you look at some of the reports out there, they peg to, "Are you positive on 2024, or are you extremely negative on semis in 2024?" Not company specific. That's really the range you gotta look at.
Yeah. Okay. Thad, there's, there are a couple, you know, pretty big moving parts on gross margin. We actually talked about some of those. There's East Fishkill, there's the 200 mm silicon carbide ramp, you know, utilization's still low, as you said. Can you talk through how those all play out and whether you think that 48%-50% is still a doable, you know, long-term model?
Oh, 48 to 50s, absolutely. I mean, our long-term target's 53%, which we think is still very doable. So if you, if you look at where we are today, as I was saying, the biggest impact is utilization today, right? Taking utilization down further, but yet we're holding that mid-40% floor at our margin. And I think this is because we took utilization down. We started taking utilization down in Q2 of 2022. And that's really allowed us to kind of adjust to, first, consumer and compute getting soft, then industrial, and now you know, some pockets of, of auto. As you look forward and you think about where we are, silicon carbide will be at the corporate average in Q4 for the first time. So it's been...
You know, we said on our last call, it starts with a four today, so it's got a four handle. So it'll be out there. As we go through next year and we bring additional capacity on in 2024, we think it'll be at the average, and then 2025 becomes accretive, as and we've said our silicon carbide gross margins are at or above that corporate target. You got East Fishkill that rolls off in 2024. That's 250 basis points. You've got the Fab divestitures of $160 million. We'll see most of that in 2025. We'll see a little bit in 2024, but most of it starts hitting us in 2025. And then you've got ramping of new products.
So everything coming out of R&D is all at favorable gross margins, right? So we're not doing anything that's, that's dilutive to gross margins. And we've talked about some of the opportunities that we have in the analog space to go, into higher margin categories with large TAMs. So that gives us that, that tailwind coming out of 2024, really to have that, that gross margin expansion. If you do that math right there that I just gave you, you know, we're, we're getting really close into that target as we are, right? So, you know, so we don't need, you know, the market's gonna do what the market's gonna do in the short term, but we feel really good about the long term.
Great. Well, thank you for the time. I really appreciate it.
Yeah, absolutely.
Thank you. Thanks.