ON Semiconductor Corporation (ON) Earnings Call Transcript & Summary

Good morning. I'm Tim Arcuri. I'm the semi and semi equipment analyst here at UBS. Very pleased to have ON Semiconductor with us. And we have Hassane El-Khoury, who's the President and CEO; and we have Thad Trent, who's the EVP and CFO. So thank you to you both.

Thank you.

So it's been a little over a month since your earnings call in late October. We've now heard reports from the entire sector since you reported. How do you feel just about demand, how it's progressing versus what you thought when you hosted your earnings call?

Yes. I think, like you said, a few months have passed. No real change from what we talked about as far as the demand environment. Demand is overall, we see signs of improvement, but stabilization. To me, at this point, stabilization is actually a very positive thing given where we've gone. Ordering patterns have improved from where we were even 90 days ago or where we were at this point in time last year as far as what the outlook looks like. So we're looking at stabilization. We're looking at kind of return potentially to normal seasonality.

So if you talk about orders having gotten maybe a little better over the past 90 days, is there a customer base that's getting a little better? Is there an end market that's getting a little better, autos maybe getting a little better?

I think my comment is more across because where we are in the demand environment today is not any more specific to a certain segment or end market. It's more on a macro improvement or macro health. We expect the inventory drain to kind of be behind us by the end of this year. We won't be talking about inventory drain, but inventory remains at a very low level. So although customers drained inventory and that's behind us, we are at a very low level, which means that even before a demand cycle happens, there's a replenishment cycle that has not happened yet. So typically, after an inventory drawdown, you see a replenishment to get back to kind of your normalized demand. And then you have the recovery. So we have a double positive still ahead of us, and we haven't seen that replenishment. So that's what we're looking at now. And from our position, how we've been managing the company operationally, we're in the best position we could be. Our inventory in the channel is exactly where we want it. Inventory on the balance sheet is exactly where we want it. So the replenishment and the demand, we can take utilization up right away. We don't have to wait to burn balance sheet inventory and what have you. As soon as we take utilization up to match that, margin will follow up, margin expansion, 2-quarter lag, but margin expansion will happen, and we're off to the races again. So the last few years, we've managed the company operationally. We've had the fab right. We've had operational improvements as far as our footprint to position us to capitalize on whatever and whenever demand starts to go up again.

So what -- why do you think customers have yet to start to build inventory? Why hasn't this begun yet? I mean is it just macro? Is it tariff uncertainty? And what are you looking for to get more bullish about?

Yes, it's all of the above, really, is customers -- I have a lot of engagements with customers. And the common denominator is what visibility or what demand do we want to build to. So when you talk about do we believe '26 is going to be whatever you believe the market is going to be, customers say, "Yes, we hope so." But they're not going to build balance sheet inventory until they see that demand. So what needs to happen for them to commit to it? And what I mean by commit is put it in the backlog and commit to the order, not just give me a forecast because I'm not going to build to forecast and strap cash. I'd rather have free cash flow to buy back shares. So what needs to happen? One is certainty, outlook certainty. That comes from geopolitical certainty. It comes from tariff certainty. It comes from a lot of that macro uncertainty we're living through. That's what we need to see first and foremost because that builds consumer confidence. Consumer confidence builds demand, demand builds backlog. That's what we need to see. And we look at all of these metrics internally, both Thad and I every week to make sure that, one, we're not getting a false positive. One uptick or one big order or one short-term order doesn't mean a recovery is happening. But we do have signals that we're monitoring that are going to be a leading indicator for us to start building that inventory.

And are there some -- so some of your peers have talked about this sort of odd dynamic where it's the end of the year and nobody wants to build inventory because it's the end of the year, but there's some sort of risk mitigation. People are putting bookings into the backlog just because they're worried about shortages. And so maybe some optimism growing about the first half of the year.

I'm smiling because I don't sell insurance. You want an insurance policy for potential shortages and so on, then put the order and buy it, and put the inventory on your book. What I mean in all seriousness, I'm not going to start taking utilization up and building inventory just in case because then it's on my balance sheet, it's on my cash flow. And then if the demand doesn't really materialize, I have to take utilization down again. That's more harmful than having the conversation with. Look, our lead times are low, call it, about 14 weeks. Okay, you want orders in Q1, place it in December. You see what I mean there is no benefit for the customer because if the customer puts an order for, call it, June, what do you think I'm going to start building it now? I'm not going to build it. I'm going to build it sometime in April, whatever cycle time. I may have it in die bank already. You see what I mean? I don't think that's a sign that I use to change our approach to operational excellence.

Got it. So it's the time of the year or it's the time for end of the year annual price negotiations. So I have to ask about the pricing environment, how you see it?

I think it has not really changed. We're looking at overall kind of as an industry that low single-digit pricing. It's not kind of all a cliff in the January or February. It's just staggered throughout the year. But you can talk about overall for the year is low single digits.

And that plays out throughout the year basically.

That plays out through out, I would say.

Got it. Let's talk about silicon carbide. The market has been pretty challenging. You don't tell us a whole lot about the business anymore, but you do think that you -- it sounds like you think that you outgrew the market this year.

So overall, we don't talk about the business as a stand-alone business because we consider it now part of our baseline or part of our core. I don't break out every business we have. So from a market share gain, we've been very successful gaining share, one in North America. We've been very successful gaining share in China. I've been very vocal about our China market share, reaching the 50% from models introduced. And we continue to make progress with ramps in Europe. So from a market position versus the ramps that customers are having, we feel very comfortable, and we feel pretty good about where we need to be. So as that market continues to evolve, we're just going to keep continuing to grow with it and gain share.

And so you do think that this year, you gain share and you expect to gain share?

Yes.

Okay. Can we talk about silicon carbide, some of the emerging applications inside the rack? How big is this? Customers trying to optimize efficiency and power density. How much of a driver could this be?

So it is a driver. We talked about our AI revenue reaching -- well, for 2025 being at the $250 million range. And I've always said we're coming at it from the high-power side, getting all the way to the smart power stage. So silicon carbide plays a big role for us. And if you recall, we acquired the silicon carbide JFET business a few quarters ago from Qorvo. That business has done very, very well for us because JFET -- silicon carbide JFET specifically is one of the best technologies from an efficiency and power density for the UPS and the power in the rack, especially as customers start moving to the 800 volt.

And maybe we can talk about the competitive landscape for silicon carbide in China. I continue hearing investors who are concerned about some of the Chinese customers moving to more of an ODM model rather than just buying the integrated devices. And I'm sure that you get this question all the time. So can you just talk about that? Is that completely overblown?

Well, I think it's overblown for a few reasons. One is the competition in China, and I'll take it in 2 ways. One is we've been talking for 2 years about how silicon carbide competition in China is going to be so intense. Well, here we are today, and I'm still gaining share in China because I've been very consistent in my approach of you're not going to win based on pricing. What you're going to win, especially in China, where EV is really a key advantage is you're going to win based on the technology. And the technology, meaning efficiency, the number of die, if I'm able to do in 3 die, what my competition needs 4 or 5, I'm already ahead from a cost and system level. That's how we win. That's why we win. That's why we gain share in China, and it's not against the China silicon carbide suppliers, it's also against Western companies in China. So that's why we're winning. Now the transition, you talk about ODM and integrated and so on. I've talked about that transition even last year, where we have seen a transition from modules also to die. So we have a mix of both. That's not positive or negative. That's more of a cost. If they want to bring it in, that's great. They still use our die because we still have the best die. Where you need advanced packaging and differentiated packaging, those customers are still buying our modules and packaging. So it's really dependent on the customer capability and really the use case. We still tackle both. But with the transition from what used to be 100% module when this market was brand new to now there are capabilities out there just from assembly houses that can do some of these, call it, the prior generation modules, there's an ASP change. So if you look at it from a volume perspective, our volume has been increasing. That's why I referred to the market share. And I think the '25 is going to be that transition year.

And just to point out, on both those scenarios, whether it's module or die, it's favorable margin for us.

That's right.

It's just a difference in ASP.

But does it pressure utilization for your packaging operation?

No, no, because packaging from a utilization perspective and so on for these is not the overhead. We resize our packaging because our investment, and Thad has said this, part of our fab or manufacturing right approach is we only want to manufacture internally what is differentiated, meaning what we cannot get outside. So that actually fits very well with our strategy because I don't want to build modules internally if somebody else can do it. So our utilization is dependent on that differentiated packaging that I talk about, and that's still internal. That's not going outside. So there's no change to our manufacturing optimization approach because the market dynamic actually fits very well with our manufacturing strategy.

Got it. Thad, maybe we can talk about gross margin and some of the puts and takes headed into '26 with a few maybe more normal seasonal quarters, maybe even some above seasonal quarters. Utilization maybe begins to have a bit of a modest tailwind if we begin to go above normal seasonal. So how do you think about the puts and takes on gross margin and maybe how mix could be a little different next year than it was this year?

Yes. So you've hit it, right? I mean the short term, meaning '26, gross margin is going to be driven by utilization, right? And as Hassane said, we can match our utilization with whatever the market recovery starts to look like, given our portfolio or profile of inventory in the channel, inventory on our balance sheet. So as utilization goes up, every point of utilization is 25 to 30 basis points of gross margin improvement. Today, we're running at 74%. I said Q4 was going to be flat to down slightly as we're building some mass market inventory in Q3. But if you think about fully utilized for us is in that low 90%. So if you do the math on that, getting from where we are today to that maximum level, you get 650 to 700 basis points of gross margin improvement. Now you need revenue to get there, obviously, but we'll be able to match that utilization with the market. On top of that, we announced the impairments. So in Q1, we took 12% of our capacity offline. A couple of weeks ago, we announced more capacity coming offline. This is a result of all the activities we've been doing in our manufacturing footprint, our fab right initiatives to get more output out of the existing footprint. So you think about it from a depreciation standpoint for next year, it's $45 million to $50 million of benefit in depreciation just with these impairments for next year. Now there's some lag for that to hit the P&L. But exiting next year, it's almost another 100 basis points right there, if you think about the impairments. Favorable mix longer term as Treo ramps and these other products, we think we'll get another 200 basis points of gross margin benefit there over time. We believe there's more on our fab right initiatives, so another 200 basis points there. And that kind of includes our divestitures of fabs that we did a couple of years ago. So if you start adding all that up, you can get greater than 50%, which is what we aspire to be on a gross margin standpoint. So when we think about our line of sight, we see improvement. It's a matter of what this market recovery looks like. That's the #1 driver.

And just on the divestiture piece that's an additional 200 basis points. How will the timing of that lay out? Will that all be recognized next year?

No, we won't see it all next year because we've built inventory as a part of that fab transition. Now the market has come down, right? So it's going to take a little bit longer for us to burn that through. The benefit is when we burn that inventory and then we start manufacturing it in side. That's where you get the true leverage. We'll see a little bit of it next year, probably a lot more of it in '27.

Yes, that's a lot market and demand dependent. The faster we burn that strategic inventory, the faster we can start bringing it inside, which helps both from a cost and mix but also from utilization.

Got it. So all things considered back to your point about next year really will be about utilization. So if utilization is flat, gross margin should be about flat. All these other things really help more in '27 than in '26.

Yes. But I believe that if there is some recovery, we're going to match that. And I think we will see margin benefit in the latter half of the year next year, assuming that there is some recovery that we start to see early next year. Also, keep in mind, all this underutilization is noncash, right? So if you look at our free cash flow margin, we talked about the free cash flow margin being around 25% this year. That -- as that utilization goes up, gross margin improves, it improves operating margin. But free cash flow is already there.

And can you talk about maintenance CapEx? The fab network is far from fully utilized, obviously. Where -- what kind of projects are you directing CapEx toward for this year?

Basically maintenance. So we're in this mid-single-digit CapEx intensity, right? So we were there this year. We'll be there for the foreseeable future. It's maintenance. We're not having big investments. I mean I just talked about taking capacity offline, right? So we've got plenty of capacity.

And it seems to me like if the market doesn't recover, there's an argument that maybe you should further consolidate the fab network. How do you think about that?

We're always looking at it. We're not done, as I said, with Fab Right. We've done a lot on the front end. We haven't done a lot on the back end, but we're tuning up that footprint. We'll continue to consolidate where it makes sense.

Yes. But we have to also be very, very clear on there's the operational efficiency, operational excellence and strategic intent. What we've been doing are things we're doing regardless of market being up or down because those are efficiency improvements that we've had, improve cost, improve margin, improve the structure of the company. Of course, we're not looking at the market as we sit today in the short term, even in 2026 and saying, "Oh, if the market is going to be at this level, we have to take capacity out because now you're starting to stifle your strategic growth." Everything we've done so far is driven by efficiency, driven by getting more throughput out of our existing footprint because all the fab rights we've done. So they don't have an impact to our long-term trajectory of growth. So we're balancing the 2. We have to have the most optimal footprint based on where our strategic plans go. Treo, for example, we talked about Treo being a very successful launch, and we're starting to generate revenue. We're starting to get customers across all of the end markets. That CapEx is already in. That's already in East Fishkill. So we look at capacity rationalization and footprint rationalization, not as a market kind of navigating the current market environment, but more on 2, 3 years from now, do we have the most optimal capacity from a cost and throughput. And that's the decisions we've been making now.

Got it. And I get this question all the time from investors, why not just buy something to fill the fabs? And I know your answer is, well, we're not in the fab filler business. So you've said that before. But I mean, given how much capacity you do have, this might not be a bad time for you -- if something was strategic and for you to make an acquisition.

But again, it's -- we're not -- first off, your first comment is true. We're not going to make an acquisition to fill the fab. Acquisition, we're going to use our firepower to do strategic acquisitions. Now of course, a strategic acquisition that we can bring internal is very valuable as well. But practically speaking, let's say, we announce something. Well, it takes about a year or so to close and a couple of years to bring that technology in unless we already have it, which is if we already have it, why would we buy it? Does that make sense? So that approach doesn't solve a -- it's not a 2026 solution or even early 2027 solution even if I wake up today and I say, I have a deal done. Does that make sense? Part of our M&A and our discipline, of course, we look at all of the above. We look at strategic intent, technology gap, differentiation versus our financial model. And part of the synergies we can extract from a potential acquisition is can we leverage our manufacturing footprint, both front end and back end. We consider 100% all of these as part of a deal.

Got it. And can you talk about just an update on these noncore pieces of the business that you're walking away from? It's sort of come a little slower than what you would have thought. So can you talk about why that is and sort of what the update is as you look into next year?

Yes. Let me calibrate it. So in 2025, there's 5% of the revenue that won't repeat in '26. So let's call it, roughly $300 million. We've talked about these noncore exits for several years now. The way that we are exiting that business is we are pricing ourselves out of the market. And it turned out we didn't lose all that business because it got to favorable margins and it didn't disappear. The fact is this downturn has gone and prolonged, we've seen competitors now coming back and start to take some of that business. So today, that business is a healthy business around the corporate average. We're not going to chase it down as the margin comes down and the pricing comes down and that stuff. So there's about $100 million out of that $300 million that is in that category for next year. There's another $50 million to $100 million that is a part of our ISG, our Image Sensing Business that we've repositioned that business, and we've been talking about this for several years, repositioning that business into machine vision away from human vision. So we are exiting some of that business just by the nature of us repivoting that business. And then the third piece is stuff that we've been EOLing over the last several years, and that's another $50 million to $100 million. So roughly, you get to about $300 million for next year. But the noncore exit that we've talked about for several years is about $100 million going into next year. But there's about a 5% revenue that doesn't repeat in '26.

Got it. And Hassane, do you have -- when you think about gaining content in data center, do you have all the IP you need? You have Treo and that's going well, which I want to ask you about in a second. But from an IP perspective, are there building blocks that you need to gain share?

No. As it stands today, we have built the complete portfolio. So you look at it, we started with power. We complemented our existing IGBT or silicon power, whether it's high voltage or medium voltage with an acquisition of silicon carbide JFET that I mentioned. We've also announced the acquisition of the Aura Semi, which puts us even closer to the controller and closer to the core from the power pedigree. So as it stands today, and we've announced the vertical GaN as well, which is also servicing a high voltage, think about the 800 volts, but also higher frequency, much higher efficiency. So if you look at overall, the portfolio that we've built over the last couple of years, both organic and inorganic, we have everything we need to deliver on the complete power tree. And as that power tree collapses the number of conversions, you're going to start squeezing out competitors that don't have high voltage. Competitors that can go from 800 volts directly to, call it, 48 or even 12 volt, but they're very strong at 12 volt, they have to go find a solution for that high voltage. We have the whole thing. So we're in a very good spot to navigate and really capture share as the market moves more into a less conversions starting with an 800 volt.

And vertical GaN is more of a 2027 revenue?

That's right.

And can you talk about how big you think that could be and how important a piece that is?

So I'll talk about the market. So the market for vertical GaN is complementary to the other technologies we have. So if you think about -- when you have the silicon carbide brought a higher efficiency to IGBT, higher efficiency and higher power density. Vertical GaN adds to that high voltage, it adds high frequency that silicon carbide can't do it. So it's not a cannibalizing market. It's really a complementary market, which having all of that technology in the portfolio is actually very favorable from a competitive posture perspective. So the overall market for GaN in general, you can think about it as $2 billion to $3 billion. That, of course, will take time to mature, but that's where having that penetration, having that differentiated technology that as it stands today, nobody has been able to deliver to the market yet. We're already sampling and we have customers engaged. So that's the confidence we have in our technology and innovation that we're putting forward.

And let's talk about Treo. So can you give us an update in terms of how much revenue is flowing through Treo and sort of where you are on that journey?

Yes. So we haven't broken up the revenue, but what I mentioned, I think, a few quarters ago, we've already shipped over 5 million. That was a few quarters ago. That business is ramping pretty quickly. We're way beyond the 5 million units at that time. The importance of that is, I've always said, once you get the first dollar and the first customer, you've just proven the technology. Otherwise, it's kind of a cool toy that you invented. So having these customers and a broad set of customers is really the credibility of the technology and the platform we've done. Now where is that headed? We are still on track to deliver that $1 billion by 2030 of revenue. The margin, we talk about 60% to 70%. So as Thad mentioned, as the new products ramp and become a higher percent of revenue, that's going to be the tailwind for the corporate margin to get to that 50-plus gross margin. And then from a market exposure, we talk about industrial, we talk about automotive, we talk about medical. So it's proliferating across many of our existing end markets. So it's a very, very versatile technology that has proven adoption through customers designing it in and customers ramping it in production. So we're very excited about that. We're excited about how quickly we got that signal from the market. Remember, we announced that the Treo in November of last year. So in 1 year, we've done not just an introduction, but a revenue ramp and a volume ramp. That's pretty compelling.

Great. Thad, can we talk about OpEx for a second? I had in my notes at the beginning of the year, you were -- I think you were thinking that it'd be sort of 250 a quarter run rate. And obviously, you're higher than that. Can you just talk about that? I know you were a little bit above even guidance last quarter. So can you talk about OpEx and maybe you're taking costs out, but you're also adding costs from some of these strategic investments and things like that?

Yes. What we've been doing is reallocating OpEx within the company that you haven't seen. We've had huge investments that we've been making in Treo, these other platforms, vertical GaN, a lot of what we've announced, some of these small acquisitions that we've done have come with some OpEx as well. When I think about where we are today with OpEx here in Q4 is this is our new baseline. If you think about this is going -- where we're going to be going forward. So regardless of what the top line does, OpEx is going to be in this tight range here. Now you'll get the reset early next year with FICA and all that type of stuff. But this is our new baseline of OpEx. So as the revenue grows next year, I don't expect OpEx to grow. I expect it to be -- remain in this range. So if you think about it from '25 to '26, OpEx is actually going to be down. So there's a lot of leverage in that model. As revenue grows, hopefully, as margin improves, OpEx declined slightly, there's a lot of leverage in EPS.

Yes. At this point, from a percent, we expect to grow into the -- our OpEx model, long-term model.

Got it. Maybe one more quick question. The LTSA balance is down a lot year-over-year. I think it's at $8 billion or something like that. I mean, customers don't seem to be interested in LTSAs anymore. So all the orders you're getting are all in backlog, not in the LTSA, I would think. So what are the value of the LTSAs? Like why do you still even have them?

So it's a customer choice really. Some customers said, we're getting what we need. We commit to the backlog, so we're not going to renew or add. Some customers still value because look at it, if a customer is sitting here looking at and hearing some of the shortages that are caused by AI supply and so on, they go, get me in line now.

They're so happy to be in a LTSA.

Yes. So we've always said the LTSA is not a gun to the head. It's really a constructive construct that we have with our customers. And do we tackle and take business without LTSAs? Of course. So the volume or the level of LTSA dropping is purely a factor of we've been shipping against it. So as we ship against it, that balance comes down, and we're not replenishing at the same rate that we had, but that's not a here nor there because it's really a customer choice.

Got it. Well, we've run out of time, but thank you.

Thank you.

Absolutely. Thank you.