Welcome, everyone. Good afternoon. Welcome to this session with the management team from On Semiconductor. I'm Vivek Arya. I cover semiconductors at BofA Securities, and really delighted to have Hassane El-Khoury, our CEO, and Thad Trent, CFO from On Semi. I'll go through a list of my questions, but please feel free to raise your hand if you would like to bring anything up, and we'll make sure to get a mic to you. But a very warm welcome-
Thank you. Thank you.
Hassane and Thad. Really appreciate you being here. We'll go through the kind of the near to medium term, but, Hassane, I was hoping you could just maybe let's take a step back, right? I mean, both, you know, yourself and, and Thad have been there for roughly about 3.5 years, right, or so. And another, right, aspect of that transformation has been this Fab Right strategy. I was hoping to tell us where onsemi is in that journey, how much more is still ahead of you.
Yeah, so we've made a lot of progress over the last, call it, three years or so. But we're not done. You know, these things take time. Obviously, we've made a ton of progress. I'll highlight some of the things, but this period where we're in today is a testament to what we've done already to build confidence in what's yet to come.
Right.
So if you think about it, what you said is true. You know, the company's story has been deeply cyclical. You know, does okay in good times, and then I would call it craters in bad time. So how does that different than today? You know, over the last few years, we talked about the structural changes we're doing. The structural changes we've done on the product portfolio, we walked away from $475 million of high volatility business in a downturn. So that gave us more sustainable and predictable results as we sit here today. We've divested four fabs to give us more of a floor on when there's underutilization to get those fixed costs out.
So net-net, we re-redirected the company into high growth areas that even in the period we're in today, those areas still showing growth. Maybe not as strong as we expected, like silicon carbide and EV, but still growth nevertheless. So if you put all of these together, the company's going from-- If I take apples to apples, we're at 65% utilization today from a manufacturing footprint as it stands today. And in the, at the beginning of COVID, we were at 65%. In the financial crisis, we were about 65%. In the last inventory correction, call it 2017, 2018, 2019, right around there. The difference is, structurally, our margin floor is 45% today. At that same utilization rate in those prior scenarios I mentioned, the gross margin used to be low 30, 31, 29%-31%.
So apples to apples, from an impact of a softness reflected in utilization, we are a very different company. That's what the work we've done, led us to hold that mid-forties floor. Moving forward, however, there's still a lot of opportunity because to remind everybody, our long-term margin target is 53%. So how do we get from here to there, given that now we've established that baseline, that new baseline of a structurally predictable, financially predictable company? And that's with a lot of the growth. So utilization, we'll start with that, given that's where we are, 65%. Our sweet spot utilization is, call it, 80-83. Every point of utilization between 65 and, call it, 80, is 15-20 basis points of margin. So market recovers, utilization goes up, all that go up. Revenue goes up with it.
The growth goes up. We, backfill, or we utilize, our silicon carbide capacity with EV. Silicon carbide margin, become accretive to the corporate. And I mentioned we divested 4 fabs. There's about $160 million of benefit from those divestitures that we will benefit in the 2025, 2026 timeframe as we fully exit the fabs. You know, what does that mean? We sold the fab, we handed the keys to somebody else, but we still buy those wafers from those foundries. As we bring those wafers in, we benefit from our cost. That's about $160 million that will be over next year. That's without even talking about mix with our new products. So if you add all this stuff up together, you're already, you know, within spitting distance from our long-term target. That's what's yet to come.
Got it. And also as part of that, kind of the fab-right strategy, what is the progress of the East Fishkill fab, right? How does that kind of fit in your journey towards this 53% gross margin?
Yeah. Part of the mix and how have we been able to grow the company, change the product footprint, yet divest four fabs, which we divested four fabs in the height of the shortages. So how do you do that? Is by what we call the fab right, that you referred to, which is having that footprint there, how do we partition the products? And how do we partition it to minimize and optimize the CapEx deployment to get a better ROIC? So to give you a simple example, we ramped our silicon carbide 4x from 2022 to 2023, $200 million to about $800 million. But our CapEx, we didn't build a new fab. So how do we do that?
We took our IGBT, our silicon power from our Korea fab, we put it in East Fishkill, and in that transition went from 8- to 12-inch. So you got a fab utilization on East Fishkill. You have an 8- to 12-, so capacity increases based on output, and then we moved silicon carbide into our existing power fab in South Korea, and we ran silicon carbide. We did not buy a shovel nor build a new fab, yet we entered a market, quadrupled our revenue to $800, meaningfully quadruple. It's not like $1-$4 million-
Right.
It's 200 to over 800, and we didn't build a new shelf. That's the CapEx light that comes with our brownfield investment. That's going to expand that concept to the fab-right, which is: where is the best place to deliver and build our products that minimizes CapEx, improves cost, and gets us time to market much faster? So a brownfield versus greenfield strategy for expansion, 40% cheaper and about 2 years faster time to market. That's the reason we capitalize on the silicon carbide market as a, quote, unquote, "newcomer" back in 2021.
Let me give you a little more color on the East Fishkill as well. When we closed that acquisition, you know, a year and a half ago, the dilutive impact at that time was a surprise to us. It was 250 basis points, and at that time we said, "Look, it's blocking and tackling. We know how to run fabs, the fab-right strategy, and we just had to get the cost structure right." On our last earnings call, we talked about we now have parity in the wafer cost with our other fabs. So we've taken that cost out, right? We've improved the cost structure, improved the cycle times. We've now taken the cost out, we're at parity. The next step is it's a 300 millimeter fab, so we got to get better than that, which we'll do.
We're on the, we're on the curve to get there. But today, the dilutive impact is the foundry business that we're doing for GlobalFoundries, and that's 100 basis points for the remainder of this year. That rolls off next year as well. So in addition to what Hassane was talking about, the utilization, you have that 100 basis points gross margin improvement that you'll see in 2025. So the market recovers a couple of these other things, and we've got, you know, margin expansion coming in the future between 2025 and 2026.
Got it. Does that change that, the underutilization or the utilization of that fab if GF moves out of there?
Sure, we've got to fill that up, but that's in that 15-20 basis points improvement.
Got it.
Yeah.
Okay, thank you.
The way we're filling it, obviously, is we moved the IGBT. We have some of our new products from the image sensor that we already are running in there.
Right.
Our first new product is sampling out of that fab. We have the medium voltage FETs already there. In our last Analyst Day, we talked about our new push in the analog mixed signal.
Right.
New products that Sudhir talked about for the power tree, for automotive, industrial, and cloud, those are all going into East Fishkill. So as we wind down the GlobalFoundries, low-margin foundry business-
Right
... we're actually replacing it with new products at 60%-70% gross margin. So you lose the dilutive impact, but you're replacing it with a much better swing on margin, and that's what's gonna get that trajectory. That's the margin expansion from the new product.
Right. So I wanted to talk about the two big parts of the business. So let's start with industrial, then we'll get to, you know, a lot more exciting stuff in automotive. So maybe industrial, help us unpack what is in your industrial business right now, kind of what are the big applications there, and then where are those big ones in their state of inventory adjustments?
What's in industrial? Everything that doesn't fit anywhere else.
Got it.
So industrial is pretty much the breadth of it, but representative of industrial, I look at our industrial business as really two sections. You got the traditional industrial, and you got the secular industrial. You know, traditional industrial is what I referred to when I first called the, you know, the softness in industrial, you know, power tool. It's more everything in industrial that's closer to the consumer. It's more, you know, spend driven, you know, power tools, more closer to the consumer side of industrial.
White goods type, yeah.
White goods, all of that. That's driven by, you know, market, that's driven by GDP, that's, you know, consumer confidence. Then you have the secular group, which are megatrend driven. You have the renewable energy, energy storage, factory automation. Those are trajectories of growth. So if you split the two, and I comment on both of them, the traditional, we started seeing that softness in Q4 of 2022. So we were one of the first that called the softness in what I called it, the consumer side of industrial, again, back in Q4 of 2022. And there are reasons for that, is, you know, we have the LTSAs. I've always said, if the LTSAs don't get us anything, they're gonna get us a phone call. Well, we got the phone calls.
So we started taking utilization down, and we adjusted for it. There was still strength in the second... You know, there's the build-out of the energy infrastructure and so on. So that started getting softer later. So that's where it is. Now, where we are today, there's stabilization. So the traditional part of industrial, stable, there's some green shoots in there, but it's stable. You know, the rate, if I look at KPIs, you know, rates of cancellation, pushout, and so on, it's has slowed down and stabilized. Again, like I said, some green shoots. Medical has been-
Right
a growth for us in our analog business. And then the other one is still having some softness, but again, slowing down from a softness perspective, I would call that stabilization. So that is kind of how we look at the industrial business. So when people say, you know, now, "Oh, well, you know, your industrial quarter-over-quarter didn't go as much as some of our peers," but you gotta, for us, you gotta go back to, like, Q4 2022, Q1 2023, and measure from that. Because we started taking it down, and we decided to take utilization down way ahead of everybody else.
Got it. Some of your, you know, industrial and analog peers have suggested, sort of a, I wouldn't call it a hard bottom, but at least some bottoming and some recovery over the next several quarters. I think you have been more conservative, Hassane, if, if I'm right about calling for a bottom in, in that market. What is preventing you from calling for some bottom if you are seeing some of these KPIs start to improve?
It's simple. I don't like to be wrong. You know, at the end of the day, the people that are calling a bottom are the same people that called a growth in 2024, at the end of 2023. Guess what? 2024 wasn't. Now they're saying, "Well, yeah, but the second half." Look, there's two things before I, for me, to call the bottom, I need to see two things. One, I need to see the demand pick up, the order patterns picking up. That is the first signal for us to say, "Okay, there is now ordering coming up, therefore it's an indication." But the second and more important thing for us to call a recovery, which is market-driven, is you ship to your customer, they have to ship through. You know what I mean? The end demand, consumer purchasing demand, has to recover.
Right.
for us to call a recovery. Does that make sense?
Right.
Otherwise, we'll see the orders, I ship them out, get a good quarter, inventory sits at the customer, then they reduce their order patterns again. Okay, that's one great quarter. I don't measure in quarters, I measure in long-term view, which is how we run the business, is not on a quarter to quarter. We manage it on the health and the sustainability of a result that we wanna achieve. So therefore, it needs longer time. So those green shoots has to play out on end demand versus just sitting in inventory. We're not there yet. That's why I say it's too early to tell. But as we... that market develops, if stability is sustained, then by definition, you'll start getting kind of that market demand picking up.
But I have to see both before I call the bottom, and by then, I'll be on the other side, on top.
Any specific parts of industrial where you think things are still pretty bad, and where do you think those?
I don't think there are parts of industrial that are bad, I would say, right? I mean, yeah.
Alternative energy is still soft, right?
It's soft. We don't play in the residential-
Right
... side of it. So for us, it's, you know, there are areas that are soft, like that set, but there's not an area that's kinda like a-- it was when it first started, where it's like a steep decline, peak to trough. We ended up about 30% or so. So that, I don't think there's that. There's more of a tapering off.
Let's talk about automotive. So last year, auto production was up 9%. You obviously grew, you know, better than that. This year, auto production, from what we're hearing, is either side of, you know, flat, it's not really. So what does that tell you about how your auto business can do this year?
Sure. Auto business overall, I think you can say, you know, call it flattish, depending on where we end. But with if you take the where the Street has us, just take that and project with 1%, now you have a growth in silicon carbide. So I know there's a lot of headlines about, you know, EV. No matter what you think about EV, EV is gonna grow, just a different rate than what everybody thought, depending on what you thought walking in the year. So if you take that growth that we talked about and where the Street has us, the rest of the non-SiC business is down mid- to high-single digits.
About it.
Yeah. So that's kinda where you can think about it. It feels right, you know, given where a lot of our peers are. It depends-
Why isn't content helping you on the non-SiC side?
Because it's not... Remember, it's not the content inside, it's how much of it was in inventory depletion. So it's not a one-to-one to what the content is. There's an inventory digestion period, which is you're under shipping, basically.
Right.
You're not a 1:1 until you get to that level.
Right.
So that's the dislocation that we have in 2024. And that has to work its way out before you start getting to, you know, a SAR plus for content.
Got it. Do you have – if you set aside silicon carbide, Hassane, do you have a greater sense of comfort in automotive versus industrial? 'Cause, you know, you have worked in the automotive industry. Do you think you are getting dependable signals for them, or you think we should be prepared for another leg down in overall automotive demand?
I hope not. Look, I think from the patterns and so on, I think both our customers, the OEMs and ourselves, we have a pretty good handle on the ordering patterns. And it's not, you know, what I see in the backlog. It's the dialogue we've been having with customers, where, you know, when we're talking about engagement, and remember, the LTSAs are legally binding unless we both come to an agreement to change them.
Right.
... Having to go through that agreement is where the alignment comes in, where we have to feel comfortable that, you know, whatever the win-win is, it's sustainable. So what does that look like? Design and activity is still healthy. Every OEM that had intent on launching a new platform for EV is launching it and launching it when they expected. So none of that thing, which, that leads me to believe it's temporary. You know, the-
Right
... that correction or that, that inventory digestion or even the, the contraction in, growth in EV is temporary because no plans have changed. I would be more worried if people say, "We're canceling 3 platforms." Okay, that's a different approach, but no OEM has backed off on their EV strategy. Some said, "We're not gonna be 100% by 2030. It's gonna take us longer." Not a problem. We never bet on a 2030 as a, you know, black-and-white date. I've always said we'll be maybe half, half penetrated by the end of the decade, give or take a few years. Well, today it's take a few years. It's the lumpiness is always part of our plan, so nothing that has happened has changed that, and from a customer perspective, nothing has changed from that. You know, I was at the Beijing Auto Show.
It's all EV, and most of it went to 800 volts. So the trajectory that we're going with our roadmap at 1200 volts supports exactly where every OEM is. So again, no OEM has changed their strategic intent on EV. That tells me that this is temporary, and it's market-driven.
Got it. How fast do you think the silicon carbide market can grow this year?
Well, I tell you, we're gonna grow 2x, whatever that number is.
What is X? That's what I want to know.
I'm not gonna call the X, because, again, I have to see what the end, what the denominator-
But if you're saying nobody's canceling orders, everyone in the EV side-
What you don't know yet is how much they're gonna sell through. Back to my demand question. You know what I mean?
Right.
It's. I know what we're shipping. I know the platforms we're in. I can see the cars in the showroom. How much is it gonna sell through? That's depending. The denominator is what is really, really needed here. I know what we're shipping, and I need to make sure that it will be sustainable.
How do you know it's 2x?
How do I know it's 2x? It's based on the sockets position. If there are 100 cars being built, we're in 50. If the market is down, we're still in 50 of 100. You see what I mean? But if that's 50 units or 100 units, we're still gonna get 25 or 50 of them. That's the 2x. You see what I mean? So that's where we know we're gaining share, we know what designs we've captured, we know who the OEMs... Remember, in Q4, I said that the second half of this year is going to be driven by the growth in European models. You know, back in our earnings in Q4. Those are gonna start ramping. We're in what, in the models that will drive 2x market.
Right.
The question is, what is the number that is gonna go and not the penetration rate? So that's what the 2x is based on.
Got it. Makes sense. Medium to longer term, are you concerned about the amount of silicon carbide capacity and the competition? Because, you know, we have Wolfspeed who is ramping a lot of capacity. They are expecting to get big grants from the CHIPS Act, right. I think STMicro announced, you know, more fabs. I'm sure that Infineon is also ramping, and there's, of course, like a whole bunch of people in China. So are you not concerned that there is going—there could be overcapacity and a glut of silicon carbide over the next few years?
Look, I'm not concerned enough to change what we're doing, and let me explain why. First off, building the fab doesn't mean that you have something good that goes in it. We've been winning today, not because there's constraint or not. We're winning because our technology is better. Customers make decisions on who to design in based on the evaluation on the bench because they want cars that are winning in the market.
Right.
So maintaining that, as long as we're able to support our customers and have advanced technology that's better than all of our peers, we're gonna keep winning. I don't care how big fab. The only difference is they're gonna have an empty fab, and we won't. That's the-- they'll have a bigger empty fab than we don't. Overcapacity only works in commodity. The fear of overcapacity only works in a commodity business, where the more you make, you use pricing, and you just flood the market. We're not there yet. Silicon Carbide is a proprietary value-driven product, not a volume-driven product. When volumes go up, price doesn't go down because the more percent efficiency that we offer customers, they can monetize it on the battery side.
To give you an example, if our silicon carbide is more expensive, but it provides a 10% better efficiency, the customer will say, "Drop in the bucket. I'll pay you more. Give me the 10% efficiency. I'll go save-- That 10% is worth way more in dollars if I take it out of the battery than your silicon carbide." You see what I mean? So from a system level, when you want to talk about low-cost EV, it's not about, you know, giving a bonus to your procurement guy because they got 5% off the silicon carbide die. It's, you got $5,000 off the battery because you overpaid for a silicon carbide module. That's the economics of low-cost EV. It's not optimizing, you know, nickel and diming. That didn't work out for the auto industry overall.
They have to look at it from a more dramatic approach, which is, what is the BOM of the car versus the BOM at the local? That's why our engagement with the OEM is more important now than it ever been, because those decisions are made at an OEM level, not at a tier one level.
Got it.
That's where we're winning.
Vivek, I think, you know, everybody gets hung up on EV, you know, growth rates, right? What is getting overlooked is the penetration rate of silicon carbide in the EVs. So if you look at the market today of what's in production, it's about 25% penetrated, silicon carbide in EVs. If you take out the number one player in the world, it's 10% penetration. So the opportunity is not only the growth rate in EVs, but the penetration rate within the EV.
Got it.
That's a huge opportunity. When we look at the design wins, silicon carbide is outstripping silicon. Low-end cars would be silicon, but, you know, anything medium, high, have sometime, some form of silicon carbide on them. So you can think about those, those trajectories. The other piece that everybody's overlooking is that penetration.
You know, there is talk of low-priced EVs coming to the market. How is onsemi positioned with that? Because what I find unique about you guys is that, right, you have access to both, as you mentioned-
Mm-hmm
... both the silicon side with IGBTs and silicon carbide. So you can mix and match combinations depending on market requirements.
Yeah. If you recall, in Analyst Day, you know, Simon showed the penetration pyramid with technology. We work from the ultra-premium. It's 100% silicon carbide, four independent wheels, all the way down to full IGBT, one motor, to IGBT and silicon carbide hybrid module, and anything in between. So being able to provide the technologies, we're agnostic, versus a pure-play silicon carbide, where they have to go on an account, and they have to sell silicon carbide, right? In certain areas, silicon carbide cost is a disadvantage for the performance. If you don't need the performance, IGBT is fine. So why sell silicon carbide then? We are able to provide a cost-optimized performance premium based on the cost based on the vehicle. Because whether I sell silicon carbide or IGBT, same for me.
One area that onsemi hasn't spoken about as much is the AI and, and the data center, right? We have to talk about AI and, and we-
We have to.
So, what is ON's exposure? Is this a market of interest? Is it a big enough market for you to go after? Like, can you do it organically?
Mm-hmm.
... do you think something inorganic might be needed?
Yeah, if you look at, you know, again, a couple of years ago, or in May, actually, as early as May, we talked about the power tree for auto, industrial, and AI, and cloud, right? And we talked about where we play. Our strategy is, first, we're walking in. You know, today we announced a silicon carbide, at for the cloud-
Mm
... at 650 volts. Now you're talking about the power rail coming in. And a T10 trench FET for the internal power tree. So our approach for this market is, you know, it's very good adjacency from auto and industrial. You know, where auto is 800, 1200 volts, that's a 650 volt. Still benefit from all of the technology and efficiencies that we have to provide in the car. Now we're able to provide that in the cloud, which is as important when you talk about efficiency. There's not enough power coming into these servers.
Right.
The efficiency matters because you have smaller size, smaller footprint, and more efficiency from a power conversion. So we've introduced our new generation, silicon and silicon carbide that tailor those. As you look forward on the roadmap, what is yet to come is the analog mixed signal that we talked about from controllers and drivers that go in. But once you have the power switch-
Mm-hmm
... then adding that cross-selling to it through the controllers and the drivers and, and everything else becomes a cross-selling opportunity versus a penetration opportunity. Same play we've done in auto and industrial.
Do you think this segment becomes interesting enough that you actually start to break it out and not just put it in other?
It's interesting enough from a strategic perspective, but from a scale, you know, automotive, you're talking about platforms of... You know, when you talk about LTSAs in auto, you know, the numbers are in Bs. Every single one of them is in $1 billion over life. So we'll see how big that market becomes before we start breaking it out, because we also don't wanna lose the focus on, you know, the value creation forward-looking from the auto and industrial strategy that we've outlined. So we're gonna be able to strategically address all three, but, you know, what we talk about in the level of detail, like we have been with auto and industrial, you know, that we'll have to decide on.
Got it. And then finally, in closing this on, you know, as we look at the second half, everyone is interested, excited. I think you're taking a little more measured view of second half recovery. How are you feeling about second half recovery right now?
I don't know. Every day I wake up, I feel differently.
How are you feeling today?
I think at the end of the day, it's what I said on the call. We're running the company in a, or an L shape. And people are like... We came up with that, you know, on the call, because to me, it's, you know, down, we expect it to be flat. And why do I feel comfortable with that? So if you ask me how I feel, I feel comfortable with that because it is the most balanced of... If things don't recover, we are in the best position to get through it. You know, our inventory, 109 days, very low. Utilization is 65%. Our balance sheet is healthy. You know, all financial metrics and inventory metrics are how we've been managing. If I'm wrong and there's a recovery, utilization goes up, margin goes up, revenue goes up, inventory burns.
It's all tailwind. I'd rather have, be wrong, and the cost of it is all tailwinds, financial, rather than plan for a recovery, and if it doesn't happen, now I have to burn inventory. I still have to take utilization down. It's all negative. Our balanced approach puts us in a much better position to very quickly react favorably across all financial and operational metrics should the recovery happen versus plan for it and not happen. That puts me in a much more comfortable position as it relates to the uncertainty in the second half.
Makes a lot of sense. Thank you so much, Hassane. Thank you.
All right. Thank you.