Onto Innovation - Q3 2024

Executive Summary

• ONTO delivered Q3 revenue of $0.252B, GAAP EPS of $1.07, and non-GAAP EPS of $1.34, near the high end of guidance; gross margin improved to 54% GAAP and 55% non-GAAP, with record operating cash flow of $67M (27% of revenue).
• Demand was driven by AI packaging and a recovery in advanced nodes; inspection revenue hit a quarterly record led by Dragonfly, and power semiconductors set a quarterly revenue record.
• Q4 guidance implies continued momentum: revenue $253–$267M, GAAP EPS $1.04–$1.19, non-GAAP EPS $1.33–$1.48, GM 54–55%; guidance strength was partially offset by ~$10M JetStep lithography pushouts due to customer capacity timing.
• Strategic “tuck-in” acquisitions (Lumina Instruments and K&S lithography business) expand inspection sensitivity below 100nm and accelerate panel lithography, with a combined target of up to $100M annual revenue in 3 years and accretive within 12 months—an incremental growth catalyst.
• Management highlighted strengthening backlog, positive 2025 setup, and confidence to outgrow WFE again given elevated process control intensity in gate-all-around and AI packaging; short-term trading catalysts: continued AI packaging strength, advanced nodes ramp, and integration of tuck-ins versus near-term lithography timing headwinds.

What Went Well and What Went Wrong

What Went Well

• Quarterly records in inspection and power-semiconductor revenues; “we executed well...setting quarterly records for inspection revenue” and “record quarterly revenue from power semiconductor customers”.
• Advanced nodes revenue grew >30% sequentially; films metrology traction and gate-all-around investment improved gross and operating margins; “advanced nodes customers grew more than 30% sequentially”.
• Cash generation and margin expansion: record cash from operations of $67M (27% of revenue) and non-GAAP operating margin at 28% at high end of guidance.

What Went Wrong

• ~$10M JetStep lithography shipments were pushed out by customers; reschedule timing remains unclear, creating near-term revenue timing risk.
• HBM capacity additions muted into Q4 despite logic packaging uptick; management is awaiting tangible orders before turning more constructive on HBM ramp.
• Q3 OpEx exceeded guidance high end due to accelerated R&D ramp; though management aims to hold OpEx flat or better in Q4, the R&D spend lever is elevated near term.

Transcript

Operator (participant)

Ladies and gentlemen, good day, and welcome to the Onto Innovation Q3 earnings release conference call. Today's conference is being recorded. At this time, I'd like to turn the conference over to Sidney Ho. Please go ahead, sir.

Sidney Ho (VP of Investor Relations)

Thank you, Lisa, and good afternoon, everyone. Onto Innovation issued its 2024 Q3 financial results this afternoon, shortly after the market close. If you did not receive a copy of the release, please refer to the company's website where a copy of the release is posted. Joining us on the call today are Michael Plisinski, Chief Executive Officer, and Mark Slicer, Chief Financial Officer. I'd like to remind you that the statements made by management on this call will contain forward-looking statements within the meaning of the federal securities laws.

Those statements are subject to a range of changes, risks, and uncertainties that can cause actual results to vary materially. For more information regarding the risk factors that may impact Onto Innovation's results, I would encourage you to review our earnings release and our SEC filings.

Onto Innovation does not undertake the obligation to update these forward-looking statements in light of new information or future events. Today's discussion of our financial results will be presented on a non-GAAP financial basis unless otherwise specified. As a reminder, a detailed reconciliation between GAAP and non-GAAP results can be found in today's earnings release. Let me now turn the call over to our CEO, Mike Plisinski. Mike.

Michael Plisinski (CEO)

Thank you, Sidney. Good afternoon, everyone, and thank you for joining us today. Overall, we executed well in the third quarter, with revenue coming in at $252 million and setting a new quarterly record for inspection. In fact, we're on pace to nearly double our inspection revenue this calendar year. We also improved our gross margin to 54.5% and operating margin to 28%.

This resulted in record cash generation from operations of $67 million. Mark will soon discuss these highlights and our outlook for Q4, which was negatively impacted by over $10 million in JetStep lithography push-outs due to customers' capacity needs. But first, we'll review the third quarter highlights, starting with our specialty device and advanced packaging markets, where AI packaging revenue led the inspection business, with growth in high-bandwidth memory offsetting a little less than projected decline in 2.5D logic packaging.

Looking ahead, we expect to see increases in volume for logic packaging, as well as an increase in capital intensity for process control to address the growing complexity and need for higher process yields. This includes new demand for our front-end metrology systems, particularly for films and acoustic metrology. In fact, advanced packaging was one of the largest markets for our metrology business this quarter.

Revenue from power devices was the second-largest market and also set a quarterly record. Growth came from both metrology and inspection process control systems. Our power semiconductor customers continue to focus on driving yield improvements, especially with challenges associated with transitioning to larger wafer sizes, even as end demand remains temporarily muted. We expect this focus on yields to continue into next year and at least sustain this record level of revenue.

Inspection has clearly been a strong driver for us, and we're expanding our core inspection technology with the tuck-in of Lumina Instruments announced earlier today. Lumina is a small company with a very rich background in laser-based inspection technologies used in unpatterned wafer and emerging panel applications. Their patented technology will allow us to simultaneously scan top, bottom, and subsurfaces with sensitivities below 100 nanometers for silicon carbide and gallium nitride applications.

We believe this technology will also be important for inspection of glass substrates and carriers used in 2.5D and 3D advanced packages, where detecting surface defects, buried inclusion defects, and residues on the silicon or glass core are important to yield. This new capability is complementary to our patterned inspection technologies with no overlapping capability. And as a result, we expect the new applications will expand our SAM by $250 million annually in the next three years.

In addition to Lumina Instruments, we announced the acquisition of the lithography business from Kulicke & Soffa. With this tuck-in, we add an incredibly talented team with over 200 man-years of lithography experience, 24 issued patents, and eight more pending. Based in Eindhoven, we believe this team and technology will contribute to the acceleration of our JetStep lithography roadmaps and extend our competitive differentiation.

We expect the combination of these two small tuck-ins to be accretive to earnings within 12 months and generate up to $100 million in annual revenue in the next three years. For reference, revenue today is negligible. While we strengthen our opportunities in the specialty and advanced packaging markets, we also see recovery from the advanced nodes. As expected, we saw growth in logic, DRAM, and NAND in the quarter.

In addition to our strong position in OCD metrology for these markets, we're seeing solid traction with our film metrology. This year, we're on pace to grow films metrology by over 50% versus 2023. Now, I'll turn the call over to Mark to review our financial highlights and provide fourth-quarter guidance.

Mark Slicer (CFO)

Thanks, Mike, and good afternoon, everyone. As Mike highlighted, we exceeded the midpoint of our revenue and EPS guidance, executing towards the high end of these ranges due to better-than-expected demand for advanced packaging for AI devices, gate-all-around investments in advanced nodes, and stronger software and services within the quarter. We achieved another record operating cash flow of $67 million for the second straight quarter. Operating cash flow yielded 27%, representing more than doubling of operating cash during the same period last year.

Third-quarter revenue of $252 million was up 4% versus the second quarter and up 22% versus the prior year. The third-quarter EPS increased 2% sequentially to $1.34 and up 40% versus the prior year. Looking at the quarterly revenue by markets, our biggest market remains specialty devices and advanced packaging, which was down slightly from Q2 with quarterly revenue of $161 million and represents 64% of revenue.

Our biggest sequential increase was advanced nodes, which had revenue of $42 million, increased 32% over Q2, and represents 17% of revenue. Software and services with revenue of $49 million increased 5% over Q2, representing 19% of revenue. We achieved 55% gross margin for the third quarter at the high end of our guidance range of 53%-55%, driving more than 100 basis points improvement over the second quarter and over 300 basis points improvement since the beginning of the year.

Third-quarter operating expenses were $67 million, exceeding the high end of our guidance range as we accelerated our ramp in R&D investments within the quarter, extending our product capabilities in integrated metrology and technology differentiation to expand our 3D metrology for advanced packaging applications. Our operating income of $70 million was 28% of revenue for the third quarter, compared to 27% from the second quarter.

We achieved quarter-over-quarter operating margin improvement, with three consecutive quarters totaling approximately 300 basis point improvement since the start of the year. Our net income performance, also 26% of revenue, was supported from favorable investment income resulting from our increased cash balance.

Now, turning to the balance sheet, we ended the second quarter, sorry, we ended the third quarter with cash and short-term investments of $855 million, achieving operating cash flow of $67 million and converting 100% of our operating income into cash. Inventory ended the quarter at $308 million, down $12 million versus Q2, and achieving five quarters of sequential decline. We expect further inventory reduction of another $8-$10 million for the fourth quarter as we—

Operator (participant)

This is the operator. Please stand by. This is the operator. Please stand by. We are experiencing technical difficulties. Please stand by. I'll place you back on a music hold while we reestablish the feed line.

Mark Slicer (CFO)

We are holding for the Onto Innovation. We are just experiencing a brief interruption in today's conference. Please continue to stand by, and the conference should resume momentarily. Thank you for your patience.

Operator (participant)

You may continue the conference, sir.

Michael Plisinski (CEO)

Okay, well, thank you very much. Hopefully, everybody is still on the line, and I'll finish my prepared remarks, and we'll go to questions, so, in summary, we're aligned to several diverse end market drivers, and we're well-positioned to leverage our portfolio of inspection metrology and software to solve our manufacturing customers' high-value problems.

Through close customer collaborations, we have many exciting new product launches, such as 3D bump metrology, which we recently delivered to a leading memory manufacturer, and void inspection for wafer bonding applications that we expect to ship this year. In addition to the organically developed technology, our recent tuck-ins further enhance both our portfolio of synergistic technologies and the markets that we can pursue.

Combining the outlook for the end markets we're serving with our new product opportunities, we expect another solid year of growth in 2025, and that concludes our prepared remarks. Lisa, please open the call for questions from our covering analysts.

Operator (participant)

Thank you, sir. If you'd like to ask a question, please signal by pressing Star One on your telephone keypad. If you're using a speakerphone, please make sure your mute function is turned off to allow your signal to reach our equipment. Again, Star One to ask a question. And our first question comes from Brian Chin with Stifel.

Brian Chin (Managing Director)

Hi there. Thanks for letting us ask a few questions. And also, just FYI, Mike, I think where you picked up versus where Mark may have left off, I think there might have been a break there in terms of some of that content, but just FYI. Anyways, back on script here. TSMC effectively ran out of space to expand its CoWoS footprint this year, but the demand is very high.

I was kind of curious, what do you currently see as timing for when that activity could pick up again? When you combine that with the visibility you have on gate-all-around expansion, how confident are you that Onto revenue will show further improvement from existing levels moving into first quarter or first half of next year?

Michael Plisinski (CEO)

Good question. I'd say highly confident. We're confident in revenues growing from here as we move into the first half of next year. And it's driven by not just gate-all-around, but we also mentioned DRAM capacity expansions that we're seeing to support both the enterprise server starting to pick up a little bit, but also the lack of capacity due to supporting all of the HBM growth. So we see both.

And as far as the TSMC, or sorry, yeah, well, you mentioned TSMC, the CoWoS expansion, they have been very aggressive at adding the capacity. We did mention on the prepared remarks that may or may not have made it out there that we expect the fourth quarter to see a fairly significant increase, while the HBM maybe is a little more muted in the fourth quarter from an AI packaging perspective. So they're already starting to find space to add capacity. And we expect that to remain fairly strong in the first half.

Brian Chin (Managing Director)

Okay. Got it. And that's helpful. And so maybe a little bit earlier customer readiness from that standpoint to take equipment, it sounds like. And then in terms of that $10 million lithography delay, was that customer-driven? Any other sort of color behind that? And when has that been rescheduled to?

Michael Plisinski (CEO)

We're not clear on the reschedule, so that's still being discussed. But yes, it was customer-driven based on their needs. The tools are ready to go.

Brian Chin (Managing Director)

Okay. Got it. And maybe just kind of one more question in broader strokes. KLA last night on its earnings call, in addition to sort of being pretty upbeat on process control intensity they're seeing at TSMC as that customer shifts from pilot to high-volume production. They also expressed a lot of confidence based on the higher process control intensity that they would outgrow WFE in 2025.

And so when you look at, again, that high process control intensity, both for gate-all-around expansions as well as CoWoS, advanced packaging, maybe HBM, I guess how much confidence do you have in Onto's ability to outgrow WFE again in 2025 like you did in 2024?

Michael Plisinski (CEO)

It depends on what you're expecting WFE to be, but if it's in this 5%-10% range, which is where I think most of the consensus is landing, then we're highly confident in outperforming those numbers for the same reasons, and I did talk about increased process control intensity, especially in the area of 2.5D logic or AI packaging based on the complexity of the process as well as the needs for much better yields.

These are very expensive devices, and any yield issue across any of the products is going to drive a pretty expensive loss, so yeah, there's a lot of process control intensity, and yeah, we're seeing that as well, and they're still learning. So a lot of the capabilities of our Dragonfly with the many different sensors, we see the customers working with us to combine different sensors to find the solutions and metrologies that don't exist today in any other tool. So there's a lot of learning that's going on through our collaborations with the customers.

Brian Chin (Managing Director)

Thanks. I appreciate that. I'll reach out for any follow-ups. Thanks.

Operator (participant)

Our next question comes from Vedvati Shrotre. Please go ahead.

Vedvati Shrotre (VP)

Hi. Thanks for taking my question. So the first thing I wanted to ask about is last quarter, you had talked about volume purchase agreements for gate-all-around nodes. I think they were roughly $120 million. Can you give us an idea or a sense of how that splits out between customers, given that some of the leading-edge customers are now facing issues with their gate-all-around transition? So has your visibility on those VPAs changed? Is there any conversation changes where you may not get that whole $120 million for next year?

Michael Plisinski (CEO)

Not to any major degree, no. In fact, we continue to work off some of that VPA. There's still quite a bit left for 2025, and our backlog continues to strengthen and look relatively good across the board. So no, I would say that, yes, there's certainly some movement by some customers, but our position remains strong and growing or strengthening, I should say.

Vedvati Shrotre (VP)

Is it primarily because of the leading foundry customer being strong? Is that a way to think about it?

Michael Plisinski (CEO)

That's one. But we did mention that the number we talked about were two customers, and both look still strong.

Vedvati Shrotre (VP)

Okay. Understood. So now, maybe on changing tracks on the HBM kind of a ramp, what's the visibility you have on the HBM capacity additions? One of the things Teradyne pointed out on their call is they're seeing HBM capacity additions could be muted next year, as in the growth for Teradyne's HBM revenues could be muted next year. What is your sense or visibility into how that HBM piece of the business goes into or the trajectory of that business into 2025?

Michael Plisinski (CEO)

I also mentioned that in my remarks around, we're seeing quite an increase, a doubling in capacity from the 2.5D logic side. Last quarter, we talked about HBM increasing. That alone would drive an increase, an expected increase in HBM. Then, of course, you have an additional intensity and additional number of HBM around each GPU for the new latest devices.

That said, we also see, let's say, we're not seeing movement on HBM expansion yet. I echo that. That's what I mentioned in the prepared remarks is that it's not clear yet, but we would expect that some kind of expansion would have to follow to support all of that new 2.5D logic that's coming on board.

My guess is there could be still some conservatism by the players and still trying to understand who's going to win what share from, let's say, the major driver, the major end customer, and so they're careful with the capacity expansions. That's my guess, but I don't know, but we are seeing that muted behavior from HBM right now.

Vedvati Shrotre (VP)

Right. That's fair. And then the last one I had was on the power semiconductors. Could you help us understand sort of the size of that revenue opportunity for you? And given that we're seeing a big downturn in the auto market, but your inspection revenues continue to be strong on the power side, it'd be great if you could provide what's the disparity there, what continues to drive your revenues versus we're seeing CapEx cuts across the board?

Michael Plisinski (CEO)

Yeah. I mentioned it the last couple of quarters. It's all about driving yields and higher yields. And I mentioned there's some wafer transitions, whether it's silicon carbide going from 6-inch to 8-inch or GaN moving from 8-inch to 12-inch. That has an impact on yields and sort of creates a requirement for additional process control or better process control capabilities.

So we're seeing some of that. And then I think it's pretty well publicized that the yields in general are not that high. So customers tend to want to, if they want to increase output or prepare for increased output, they can focus on yield instead of adding just capacity and throwing away. So that makes them more profitable when they do actually ramp.

Vedvati Shrotre (VP)

So how should we think about the opportunity for you as in kind of the size of the business or any color there would be helpful?

Michael Plisinski (CEO)

We don't break it all down there, but it's becoming one of our top markets behind AI packaging in the specialty and advanced packaging market. I think it is number two in that space.

Vedvati Shrotre (VP)

Thank you. I'll get back in the queue.

Operator (participant)

We'll move to our next question from Edward Yang with Oppenheimer.

Edward Yang (Managing Director and Senior Analyst)

Hi, Mike. Hi, Mark. Congrats on a great quarter. Just wanted to drill down a little bit deeper into your outlook into 2025. You're expecting continued growth there and outpacing WFE. But obviously, this year, you're looking for revenue to grow about 20%, and that's well a multiple of how much WFE grew this year. So thinking about all the different puts and takes, what are the things that give you confidence or less confidence? And can you size up again? Are there any reasons why growth should meaningfully accelerate or decelerate from your 2024 run rate?

Michael Plisinski (CEO)

I would say, since I said we'd grow, there'd be a meaningful acceleration. And that would be in advanced nodes for sure. That's been bouncing along the bottom. And now we've talked about gate-all-around opportunity expansions, but we now also are starting to see and gain more confidence in DRAM, additional DRAM capacity and growth there driving our advanced nodes. So we see that staying growing quite nicely.

The AI packaging for sure, gate-all-around and the process control intensity and volume increases there is creating opportunities. We'll see what happens with the HBM. Does it stay muted or not? If not, I think there'll be some significant growth there. I don't see how it could stay as it is if the CoWoS essentially doubles. So we'll see. But that's that. And then we talked about the power semi.

In there, I said we would at least sustain these record levels in the prepared remarks, which implies we would expect to exceed and set new record levels. This would be a floor for us. That will be another growth driver for us as we look at 2025.

Edward Yang (Managing Director and Senior Analyst)

Just to clarify, when you say accelerate, I mean, do you mean accelerate off of a 20% revenue growth rate?

Michael Plisinski (CEO)

Yeah. Good point. No, I'm not trying to imply we'd grow above 20%. No. So I should have used the word not accelerate.

Edward Yang (Managing Director and Senior Analyst)

And one of your larger foundry customers signed an advanced packaging deal with an OSAT in Arizona earlier this month. Does this have any relevance to your order book? And again, can you speak to the broader ability of your customers to place tools at this point? It seems like things are loosening up a little bit. Is space still a bottleneck?

Michael Plisinski (CEO)

Things are loosening up. So there's new capacity coming up. But even signing that deal, there's timelines to transfer and to qualify and to bring in tools, etc., which some of it's in process. But there's not an immediate launch, right? So yes, it's all part of the breaking of the current bottlenecks in the CoWoS capacity.

There's also actions being taken by that large customer as well internally. And there's been discussions about the Innolux purchase. And in my prepared remarks, I did mention Q4. We see a nice big uptick from the gate-all-around. Sorry for the 2.5D logic packaging, and we expect that to maintain pretty healthy in the first half.

Edward Yang (Managing Director and Senior Analyst)

And just a final question maybe for Mark. I saw the SG&A ticked up a little bit sequentially. And is that a good run rate going forward, or was there any extra spending in there that impacted the quarter?

Mark Slicer (CFO)

Yeah. I mean, I would think our goal is to hold total OpEx in line or better than Q3. So I'd use that run rate from Q3 into Q4 in my prepared remarks, which might have been abbreviated. Our goal is to drive offsets to the cost of the tuck-ins within Q4. So our goal is to stay at those levels for Q4 or better.

Edward Yang (Managing Director and Senior Analyst)

Thank you.

Mark Slicer (CFO)

Thanks.

Operator (participant)

We'll move to our next question from Mayur Popuri from B. Riley Securities.

Mayur Popuri (Senior Research Associate)

Yeah. Hi there. I'm actually on for Craig Ellis. But you mentioned packaging pickup in the fourth quarter. Is that kind of related to the increased complexity needs as we move into this RDL-based packaging, or is it more to do with just general volume increases? And if it's to do with this RDL-based packaging increase, is that a trend that we can expect to continue as RDL picks up over the incumbent?

Michael Plisinski (CEO)

I think that the complexity increase is from two things, a couple of things. One is, yes, they're introducing some new processes that they've talked about. But two is yields have room to improve. So they're also looking at areas that could be impacting yield and how can they measure? So you can't fix what you can't see. So how can we help them see what is impacting yields so then they can make the adjustments and fix it? And that dynamic is also in play.

And that's where I mentioned the incredible breadth of capability we have on the Dragonfly platform to bring to bear different types of sensors and metrologies and inspection in order to combine that data and see things that you wouldn't see on any single tool. So that provides new insights into the yield opportunities to improve.

Mayur Popuri (Senior Research Associate)

Okay. Yeah. Right. No, that's a great answer. So I have a question on the flies, the Dragonfly and the Firefly. Obviously, they're really capable tools in 2D metrology. But you've also pointed out before how they're incredibly capable in the 3D inspection space. Do you see that kind of picking up share? I know you mentioned one memory customer who wanted it for these 3D inspection processes. Do you see that 3D inspection aspect to these tools picking up?

Michael Plisinski (CEO)

So there's localized 3D capability, which is very powerful. And we use that for high aspect ratio 3D, very high precision metrology. That is part of the 2D applications. What I was talking about was 3D bump metrology. And that's still early stages. So it's too soon to predict how big or how much that could be. It depends on adoption rate and how well we do in production.

The tool we shipped is an evaluation tool. So they'll now, based on the data exchanges we've had and all the wafers we've run for them in our facility, they now want to take the tool on site, prove it in production, and then hopefully we start to see revenue. And that's probably a three- to six- to nine-month-on-the-outside type process.

Mayur Popuri (Senior Research Associate)

All right. Gotcha. Yeah. I hope that goes well, and then about those volume purchasing expectations now.

Michael Plisinski (CEO)

We expect. Sorry. I was going to say we do expect to ship additional tools in the fourth quarter to additional customers for evaluation.

Mayur Popuri (Senior Research Associate)

That's evaluation on 3D?

Michael Plisinski (CEO)

Yep.

Mayur Popuri (Senior Research Associate)

Okay. Great. Great. And again, about those volume purchase agreements. So we were talking about last quarter. I forgot exactly who asked this question, but there was some talk about how these might convert into kind of larger agreements in the future. Is there any progress there in terms of kind of converting these initial agreements into perhaps larger partnerships going forward?

Michael Plisinski (CEO)

I think what I might have said is that we would expect perhaps additional revenue through the year. And right now, they're still working on this. So I think they got to cut through this, and then we'll see what happens in the second half of the year. But that's still my projection. If I was going to bias it, I'd bias it towards they're going to need some additional tools in the second half versus not.

Mayur Popuri (Senior Research Associate)

Okay. Thanks so much. Yeah. That's all I have. Thanks for talking to me.

Michael Plisinski (CEO)

Yep. My pleasure.

Operator (participant)

And our next question comes from Charles Shi with Needham.

Charles Shi (Principal and Senior Analyst)

Hey. So you guys got cut off the big chunk of the prepared remarks we actually didn't hear. So maybe, Mark, can you kind of repeat what the Q4 guidance line by line was in your prepared remarks? Because I think it's kind of important. If you can repeat for us, that would be great. Hopefully, this doesn't count as a question.

Michael Plisinski (CEO)

No. Thanks. Yeah. So I mean, I'll just start out at Charles. I'll just start out at inventory. So inventory ended the quarter at $308 million, down $12 million versus Q2 and achieving five quarters of sequential decline. We expect further inventory reduction of another $8 million-$10 million for the fourth quarter as we project to be below $300 million as we exit 2024, which will be a $50 million reduction from our peak of 2023 inventory levels.

As we look at the fourth quarter, we currently expect revenue for the fourth quarter to be between $253 million and $267 million. We expect gross margins will be 54%-55%. With our inventory still above our target level, this is delaying our ability to cut in these supply chain cost reductions as we continue to prioritize the burndown of existing component levels.

For operating expenses, we expect to be between $66 million-$68 million as we look to hold OpEx flat or better versus Q3 as we optimize R&D to minimize the cost impact of the tuck-ins we announced earlier. For the fourth quarter, we expect our effective tax rate to be between 15%-16%. We expect our diluted share count for the fourth quarter to be approximately 49.8 million shares. Based upon these assumptions, we anticipate our non-GAAP earnings for the fourth quarter to be between $1.33 and $1.48 per share. So Charles, you can ask your real question.

Charles Shi (Principal and Senior Analyst)

Yeah. Thanks. So I'm trying to connect the dots here, Mike. Over the last quarter, let's say three months, two out of the, let's say, four leading-edge customers you have have had a pretty tough, I mean, news coverage, their struggles and the potential pressure on CapEx. But we know that your Litho, which unfortunately has been quite often a downside contributor to your quarterly earnings for the past one year, one and a half, has a lot of sales tied into those two customers.

Am I connecting the dots right here? Because I would be thinking maybe some of the push-out or maybe it sounds like it's more like a, I mean, it's delayed to an unknown date. It's really tied to these two customers. Is it anything to do with the CapEx cut that they could be going through?

Michael Plisinski (CEO)

So I don't know which two customers particularly you have in mind. It doesn't matter too much. I think if you look at the substrate market where the lithography tool plays, there was massive, let's say, bottlenecks that several of the enterprise server customers, manufacturers, complained about publicly that they were supply constrained by lack of substrates.

And so there was a really, really aggressive expansion through 2022, maybe a little bit into 2023. And then, as we all know, the market's really softened, especially for enterprise high-performance compute. And NVIDIA's AI is the big engine now, and that's on a wafer basis. So that capacity, that excess capacity, is starting to be picked up. And we see a little bit of pickup, but it's still off of kind of a low base.

You can see that reflected in the comments I made about DRAM and the strength we're seeing now in DRAM. And that's driven by some of the enterprise or hyperscalers and some enterprise compute warming up. So I'm not sure that answers your question, but I'm not sure what you're trying to get at.

Charles Shi (Principal and Senior Analyst)

Yeah. Okay. So it sounds like you think the push-out is probably more of the cyclical factor at play rather than anything that's structural. I mean, those two customers are probably having more of a structural problem than a cyclical problem. That was what I'm trying to figure out.

Michael Plisinski (CEO)

Yeah. I think it's more, yeah. Go ahead.

Charles Shi (Principal and Senior Analyst)

Yeah. So the other thing I do want to talk to you about really is the AI packaging business. I think last quarter you talked about maybe second half this year, roughly 10% below the first-half level, combining 2.5D and the HBM. And based on what you said, it sounds like in Q3, HBM was okay. 2.5D was down a bit Q4, 2.5D coming back up, but the HBM a little bit more muted. But do you still view that minus 10%, half over half, the right number? Is there any upside or downside to that number so far based on what you see?

Michael Plisinski (CEO)

Yeah. That's a good question. It was in my prepared remarks, which, of course, no one seemed to have heard. But I did say that it's cut in half. So things have, if I had said 5%-10%, it's about half of that now as far as the down goes. It's about half the decline that we originally projected.

Charles Shi (Principal and Senior Analyst)

It's roughly 5% down compared with the first half level. And the first half of 2025, because, allow me to finish this question, you kind of said that you expect there will be higher than second half of 2024 level based on the order intake, based on the customer indications. Do you still feel like that's about the right, like first half of 2025, higher than second half of 2024, but still have to wait and see if it can exceed the first half of 2024 level?

Michael Plisinski (CEO)

This is for AI packaging specifically?

Charles Shi (Principal and Senior Analyst)

AI packaging, yes.

Michael Plisinski (CEO)

Yeah. I think for Logic, it's going to be relatively healthy. So maybe at the same level, I'd have to double-check. But the real question mark is the HBM piece. As we mentioned, we see that muted right now. Though when we look at the expansion on the 2.5D Logic side, it's hard not to expect expansion on HBM to keep up.

Charles Shi (Principal and Senior Analyst)

Got it. Thanks, Mike.

Michael Plisinski (CEO)

You're welcome, Charles.

Operator (participant)

Our next question comes from Mark Miller with The Benchmark Company.

Mark Miller (Equity Research Analyst)

Congrats on yet another good quarter. I was just wondering if you gave us a feeling for what you're expecting in China and Korea next year?

Michael Plisinski (CEO)

Korea, we can say I mentioned DRAM and the DRAM growth. So you could guess that Korea would participate in that. China, we expect, I mean, we're already relatively de-risked in China. So we're around the 10%-15% range. And I would expect to be in that same range, maybe. Yeah. I would expect to be in that same range.

Mark Miller (Equity Research Analyst)

10%-15% of sales from China next year.

Michael Plisinski (CEO)

Yep.

Brian Chin (Managing Director)

Thank you.

Michael Plisinski (CEO)

You're welcome.

Operator (participant)

And our next question comes from David Duley with Steelhead Securities.

David Duley (Managing Principal)

Yes. Thanks for taking my question. My first question is on the NAND market. Your big Korean HBM customer also plays in that market. I think they were talking about their SSD business being up 20% sequentially and 430% year over year. Lam's talking about a big upgrade cycle to move up in the number of layers, so we're not seeing new wafer starts added, but we're seeing a big upgrade cycle, and I was just wondering how you might participate in that.

Michael Plisinski (CEO)

We see NAND growing for us in 2025. And on a percentage basis, it would look very impressive. But it's still off of a very small base. So we don't see NAND recovering. So it's probably, as we mentioned a couple, I think now two quarters ago, it's really the high-level, high-stack NAND to support AI devices and AI server farms, the high-speed data. So that's essentially what we see.

As far as the high-stack and then the more layers mean a lot more of our process control, not as much. So the capital intensity, there'll be a couple of extra steps in there. And that's where the aspect metrology comes into play. But we don't see any massive increases in, let's say, our OCD metrology as a result.

David Duley (Managing Principal)

Okay. And my second question is kind of around the high-bandwidth memory market. I realize your customers aren't giving you a lot of visibility, I guess, into when they might expand the capacity. But when you think about, I think you've highlighted this, the number of chips per GPU is probably going to double with Blackwell versus Hopper. You got them going from stacking eight to 12. And you also have Micron ramping up.

And I think Samsung just announced yesterday or the day before that they're close to signing their agreement with NVIDIA as well. So I'm kind of curious why you wouldn't be much more positive about the growth in that in-market given all the unit volume growth and more customers coming online. But anyway, maybe you could elaborate a little bit more.

Michael Plisinski (CEO)

Well, what makes me positive is orders. So I see all the activity. And I like our position. And we're trying to expand our position with the work we're doing on the 3D metrology, so going after more, let's say, wallet share. But we're not seeing the orders yet. And as I mentioned, I think earlier, that there's some conservatism with these customers.

If everyone's ramping and qualified, they may not know yet what share they're going to have and how much they want to expand in order to serve that share. I'm sure NVIDIA is working them all against each other. So I don't know if that's just a guess. But yeah, when I start seeing orders, I'll get a lot more confident. What we can do is look at the model and say, "Hey, the capacity we see is not matching the demand that that 2.5D upgrade or expansion is going to need," so something has to give.

David Duley (Managing Principal)

You keep highlighting how all the CoWoS capacity expansion should mean that HBM capacity expands. I think I understand what you're saying, but could you just elaborate a little bit more on that?

Michael Plisinski (CEO)

If the markets were at perfect equilibrium and we're going to double the 2.5D Logic side, and we're saying that for each 2.5D Logic, the amount of HBM around it is going to increase, let's say, a factor of two, like the number you used, that's a 4x increase in HBM that would be required. That's just if everything was in equilibrium.

So now you have to say, "Okay. Well, some capacity was added. Not everybody got cut into, let's say, the NVIDIA supply chain early on. So who's going to win? How much excess capacity is there?" I mean, we try and model this out. But to us, no matter how we look at this, it looks like some capacity expansion is going to be required.

David Duley (Managing Principal)

Yes. Okay. And two final questions. What are your lead times for your HBM inspection tools? And the second question is, a lot of this CoWoS capacity that's going to come online is not necessarily going to come online at TSMC. If you listen to ASE, they're ramping up as fast as they can as TSMC's partner to expand CoWoS. And there was another question earlier about Amkor, but that's a couple of years out, I would think. But do you benefit from capacity expansions at the Taiwanese OSATs to the same degree that you would benefit from capacity expansions at TSMC for 2.5D packaging?

Michael Plisinski (CEO)

If they run the exact same process, then yes. That's yet to be determined. So we are benefiting. We are seeing engagement. We are getting orders. Obviously, not to the same degree right now as the leader that you mentioned. But they're also nowhere near, I mean, they're not even ramping yet, right? They're just starting to ramp. So I would say that remains to be seen. But again, yields are yields. It's hard to believe anyone's going to have better yields or better process than TSMC. So my guess is we'd see at least an equivalent process control intensity.

David Duley (Managing Principal)

Okay. Thank you. Oh, and the lead time.

Michael Plisinski (CEO)

Lead time. I was not going to answer that anyway, but I would say we're looking at three months or so. It's definitely increasing. The volume's gone way up. But as we've always mentioned, we build to a forecast. To the extent we have good forecast data, we can adjust lead times. But things are ticking out a little bit because of such strong demand we have right now.

David Duley (Managing Principal)

So with that kind of short lead time, obviously, if a customer came in and want a bunch of tools, you have the capacity to meet that order.

Michael Plisinski (CEO)

Yeah. We work hard to make sure we do. I mean, no one expected us to have to double the capacity output for Dragonflies this year. And yet that's essentially what we've done. So yeah, the team's outstanding at getting creative, reducing cycle times, leveraging our supply chain partners, and making sure we serve our customers. And as we've commented before, we have the capacity within our manufacturing to do that.

David Duley (Managing Principal)

Excellent, you guys. We appreciate the excellent execution.

Michael Plisinski (CEO)

Thanks, David. Thank you.

Operator (participant)

We'll take our last question in queue from Brian Chin with Stifel.

Brian Chin (Managing Director)

Hi there. It wasn't really a question, but what I was going to suggest or maybe just put out there is that I appreciate Mark repeating the complete fourth quarter guidance. And I was going to ask Mike, if you had substantive commentary after Mark's guidance, I think we missed pretty much all that. So if there was something there, it might be worth repeating. If not, then no bother. But I just wanted to throw that out there.

Michael Plisinski (CEO)

Sure. Yeah. Brian, we're just aligning to where his prepared remarks was cut to where he picked up this one. Okay. So essentially, I had said that demand for process control and AI packaging, gate-all-around power semiconductors, remains quite strong. Specifically, with AI packaging, we see improvements over our prior second-half 2024 projections. And this I've already mentioned, so that I'll skip.

And I mentioned that's helping to offset that added growth in the AI packaging is offsetting the $10 million push-out that we had expected from the lithography. So in fact, it would have been a significant beat. And then I mentioned that the market leader in AI logic packaging recently announced a doubling of 2.5D logic capacity for next year. Though not yet certain, we would expect to see orders for supporting HBM memory to also improve to support this growth in logic. Again, something we discussed.

I mentioned that the growth in high-bandwidth memory has taken a meaningful amount of capacity away from standard DRAM, as HBM requires roughly three times more wafer capacity. This, in turn, is contributing to an expansion in advanced DRAM to support a recovery in enterprise servers and investments by hyperscale customers, which we expect to see or benefit from more meaningfully in the first half of 2025. Yeah. I think that's the essential message. I tried to bring that all in when I answered some of the questions.

Brian Chin (Managing Director)

Okay. Yeah, yeah. I think we're able to incorporate some of that. No, I appreciate that. And maybe one last question against that. I know you don't dictate your customers' intake and demand and shipment timing, but to the extent that you kind of can have some, I guess, modulation here where kind of one customer is bigger, another customer maybe subsides for a quarter or two or whatever the case is.

I guess that alleviates sort of your manufacturing upward pressure on your manufacturing footprint. To the extent you may have HBM stronger in the same period that CoWoS is strong, do you have that ability to flex upwards in terms of higher output in manufacturing?

Michael Plisinski (CEO)

Yeah, we absolutely do. I mean, we're not even running full second shifts, let alone third shifts. So that alone, if we made no other improvements, would allow us to significantly increase capacity. So we absolutely do. There's other things we're working on.

I mentioned working with supply chain partners. So we're moving some of the less skilled or some of the sub-assemblies to partners where we can take that off and free up the floor and free up our higher-trained technical people to focus on the more difficult integration. So yeah, we definitely have the ability to serve the customers and their needs as they grow.

Brian Chin (Managing Director)

Yeah, and maybe just to clarify one comment you made. I think you said backlog continues to strengthen, so it sounds like you're running a positive book to bill with orders ahead of revenue across the business.

Michael Plisinski (CEO)

I don't have it exactly in front of me. I just mentioned the backlog has strengthened. We don't really report on it. I knew people would be concerned around or asking about the VPA and what does that mean, how's that being worked down. In fact, we continue to grow our backlog even as we work through that VPA. To me, that was a comment just to indicate we still see strong demand and not much softening, at least in the areas we're focused on right now.

Brian Chin (Managing Director)

I imagine even the fact that you have that VPA and HBM is a portion of that. I guess that does give you some comfort that some of that activity is still on the come next year.

Michael Plisinski (CEO)

Yeah, for sure. I didn't say it would go to zero. HBM is still going to be there. I just think it could be even stronger based on the demand-supply models we have between the 2.5D Logic and the HBM. So hopefully, there's some upside we can talk about in future quarters. Yep.

Brian Chin (Managing Director)

Appreciate that. Thank you.

Operator (participant)

Ladies and gentlemen, this concludes today's Q&A session. I'd like now to turn the call back to Sidney Ho for any additional or closing remarks.

Sidney Ho (VP of Investor Relations)

Thank you. We will be participating in a number of investor conferences throughout this quarter. We look forward to seeing many of you there. A replay of the call today will be available on our website at approximately 7:30 P.M. Eastern Time this evening. We'd like to thank you for your continual interest in Onto Innovation. Lisa, please conclude the call.

Operator (participant)

Ladies and gentlemen, this concludes today's call. Thank you for your patience and your participation. You may now disconnect.