Good day, and welcome to the Onto Innovation First Quarter Earnings Release Conference Call. Today's conference is being recorded. At this time, I would like to turn the conference over to Mike Sheaffer, Investor Relations. Please go ahead. .

Thank you, Rachel, and good afternoon, everyone. Onto Innovation issued its 2024 first quarter financial results this afternoon shortly after the market closed. 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. .

I'll now go ahead and turn the call over to our CEO, Mike Plisinski.

Mike?.

Thanks, Mike. Good afternoon, everyone, and thank you for joining our earnings call this afternoon. Strong and anticipated demand for high-bandwidth memory and logic packaging for AI devices resulted in first quarter revenue at the very high end of our guidance range, 15% over the same period a year ago.

Margins came in within guidance, and as Mark will soon outline, we expect to increase gross margin next quarter and improve further in the second half of the year. .

Before Mark begins, we'll now review the first quarter highlights, starting with the specialty and advanced packaging customers where we delivered our third consecutive quarterly revenue record, a solid 64% increase over the same period a year ago.

In the first quarter, revenue for our Dragonfly systems jumped 30% over the prior quarter, almost exclusively to support packaging capacity growth for AI devices. In addition to volume, we also project demand to increase to address new and emerging critical defects. .

For example, wafers that are being stacked for high-bandwidth memory are used in 2.5D Logic packaging, maybe as thin as 50 to 100 microns. These ultrathin wafers are then prone to the formation of small embedded micro cracks, which may propagate through the silicon, making it more susceptible to breakage.

Responding to our customers' needs, our Dragonfly platform has been enhanced with a new sensor to detect the subsurface defects at production capable speeds with orders from several customers and shipments starting in the second quarter. .

After a record year in 2023, revenue from power device manufacturers declined in the first quarter, but we continue to add new customers and expand our footprint at existing customers. We expect to see growth return in the second quarter and carrying through to the second half of the year.

Much of this demand is to improve yield to increase the factory output without increasing the number of wafer starts. .

Now turning to our advanced nodes business. Revenue did incrementally improve as expected, but remains at quite low levels. Orders to support advanced logic devices represented the largest growth and was roughly half of the revenue from the advanced nodes customers in the quarter and included both Atlas OCD and Iris planar films. .

Now I'll turn the call over to Mark to review our financial highlights and the second quarter guidance. .

Thanks, Mike, and good afternoon, everyone. Before I get into the details, as Mike highlighted, we achieved first quarter revenue and EPS at the high end of our guidance, and we generated strong operating cash flow of $57 million, representing 25% of revenue.

First quarter revenue of $229 million was up 5% versus the fourth quarter and up 15% versus the prior year. First quarter EPS increased 11% sequentially to $1.18 and up 28% versus the prior year.

Both revenue and EPS at the high end of our guidance range is due to the continued strength of our Dragonfly system to support the demand for advanced packaging of AI compute devices. .

Looking at the quarterly revenue by markets, advanced nodes, which had revenue of $27 million increased 45% over Q4 and represents 12% of revenue. Specialty devices and advanced packaging with quarterly revenue of $158 million, up slightly over Q4 represents 69% of revenue.

Software and services with revenue of $44 million increased 4% over Q4, while representing 19% of revenue. We achieved 52% gross margin for the first quarter, in line with our guidance range of 51% to 53%. .

First quarter operating expenses were $62 million, above our guidance range as we made additional investments in applications engineering within the quarter, extending our product and technology differentiation and fast emerging applications like ultrathin wafer process control that Mike mentioned as well as process tools and inspection for the litho panel packaging for the emerging glass panels.

Our operating income of $57 million was 25% of revenue for the first quarter compared to 26% from the fourth quarter. Our higher net income performance of 26% came from favorable investment income resulting from our increased cash balance, while our operating income was impacted by the higher operating expenses within the quarter. .

Now moving to the balance sheet. We ended the fourth quarter with cash and short-term investments of $741 million, achieving operating cash flow of $57 million within the quarter, converting 100% of our operating income into cash.

Inventory ended the quarter at $330 million, up only $2 million versus Q4, while we continue to ramp Dragonfly production, requiring us to procure longer lead time components. We do expect further reduction in inventory as we remain focused on inventory optimization to drive consistent operating cash flow performance levels exceeding 20% of revenue. .

Now turning to our outlook for the second quarter. We currently expect revenue for the second quarter to be between $230 million and $240 million. We expect gross margins will improve to 52% to 54%, reflecting the improvements in the supply chain initiatives discussed in prior quarters.

For operating expenses, we expect to be between $62 million and $64 million as our annual compensation elements occur within the second quarter each year. Consistent with the prior year, we do expect a decline in operating expenses in the second half as we move past these annual compensation events. .

For the full year '24, we expect our effective tax rate to be between 14% to 16%, which does not assume any impact for potential tax legislative changes that may occur during the year. We expect our diluted share count for the second quarter to be approximately 49.9 million shares.

Based upon these assumptions, we anticipate our non-GAAP earnings for the second quarter to be between $1.14 per share to $1.26 per share. .

Our focus in 2024 remains on our targeted programs for quarterly operating improvements. We have several planned productivity improvements that will take effect in the third and fourth quarters and that we expect to provide continued margin improvement.

In addition, we expect stronger metrology sales and an improving gross margin from our lithography systems as we work out the last of the very low margin initial orders. .

And with that, I will turn it back to Mike for additional insights into Q1 and the remainder of 2024.

Mike?.

Thank you, Mark. For the second quarter, we project demand for AI compute will maintain the record levels of quarterly revenue for Dragonfly systems. As mentioned earlier, we expect power semiconductor revenue to grow in the second quarter, returning to near record levels led by strong Iris's metrology adoption for planar films applications. .

In the advanced nodes, we expect revenue to increase in both logic and NAND, while DRAM revenues remain soft. The newly projected NAND demand is driven by the need for enterprise solid-state drives and high-density AI servers.

These enterprise drives have NAND stacks of over 200 layers where our Atlas and Aspect metrology have established tool of record positions at several of the top NAND suppliers. .

Also in the quarter, we are on track to ship our first JetStep system to support lithography on glass panels. We believe glass panels will be critical to realize high volume and high-performance chiplet architectures over the next 2 to 3 years.

This new system will allow customers to maximize capital investment by supporting both panel production and glass substrate R&D. Through our [ PACE ] lab, we are demonstrating the process and process control solutions required for this technology. .

In addition, we have nearly 20 companies spanning across process chemistry, substrate manufacturing and process equipment, engaging in discussions on potential areas of research and development, important to our manufacturing partners.

This new era of AI is revealing many exciting opportunities for Onto Innovation in both packaging and advanced nodes. .

Based on the traction we see with our products and the current demand drivers, our outlook for the year is improving, and we maintain our expectation for the second half of the year will be incrementally higher providing nice momentum going into what is widely believed to be a stronger year for capital equipment in 2025. .

And now we would like to open the call for questions from our covering analysts. Rachel, please open the lines. .

[Operator Instructions] Your first question comes from the line of Craig Ellis with B. Riley Securities. .

Congratulations on the very strong execution. Mike, I wanted to start just by following up on advanced nodes because it's been so quiet for so long and now we're starting to see a nice recovery.

So I wanted to understand what you and the team are seeing with respect to the breadth of the logic gains that you talked about? And then the same question for the breadth of the NAND gains?.

Sure. So for NAND, we'll start there. I mean we've always been extremely strong and well adopted in NAND customers. I think from customer penetration perspective, every NAND customer that we're aware.

Every NAND suppliers are using our Atlas OCD and several that are working on high stack 3D NAND, which we were talking about 2, 3 years ago with the introduction of our Aspect, we're adopting Aspects, but the ramp just wasn't there. .

So now we're starting to see the market demand or the market pull for these high stack devices where the Aspect and the Atlas OCD, we expect to have some benefit from the work we did several years back.

And if you recall, that's all tied to measuring multiple parameters around these elevator shafts that run up and down the stacks, [ sag while angles ] tilt, all of that. .

Going to the Logic, I think you asked about. I wouldn't say -- we're not seeing a ramp as much as we're seeing just continued adoption -- continued proliferation of the market position that we carved out when we first started talking about how important gate all around was.

I mentioned in my prepared remarks that the expansions are involving both that orders are involving both the Atlas OCD and the Iris planar films. So it's nice to see both establishing a position in getting repeat business as the pilot line start to move to low volume and then eventually high volume. .

That's very helpful. And then the follow-up question relates to Advanced Packaging and Specialty. Clearly, Dragonfly is very, very strong in AI-related hide bandwidth memory and other applications that you talked about a new opportunity. It sounded like it was related to wafer thinness and product customization that starts to ship in 2Q.

I was wondering if you could help size that opportunity and how that factors into your view for a stronger second half and potential for greater shrink in '25?.

Yes. Every time we bring up a metrology item, the question is on the sizing. And the challenge with sizing is always sampling rates. So there's a couple of knobs they can turn.

Right now, we know it's a high-value problem, and we know that these thin wafers are becoming more, they're still relatively low compared to total number of wafers, but high value, so they're tied to high-bandwidth memory, they're tied to the advanced logic we just talked about. So a lot of value in them that will justify the process control. .

So for us, we're trying to understand, okay, is this we're seeing several customers, not just memory and logic, but also some other specialty device manufacturers adopting this technology, particularly automotive.

So is this going to be more widely like a must have for every thin wafer application? Is it going to be 2 or 3 per factory? Or is it going to be 100% at certain process steps. These are all -- this is hot technology just coming out to solve an emerging critical challenge. .

And I think sizing it would be a little bit premature.

But compared to -- when I look at the interest from the customers and the risk of not capturing these defects and having entire wafers break inside a tool, which obviously losing or $50,000 wafers bad enough, but then having to stop the line, clean out the tool, requalify that tool and bring it up into production, that's also an impact. So we'll see.

But I think towards the end of the year, we'll have much better understanding of where the size of this new sensor technology could be. .

Yes. That's helpful color even without specific quantification, Mike. But just following up with my -- maybe go at it another way as we think about the year. You clearly see a stronger second half than first half.

Are we talking about low single-digit half-on-half growth mid, high single digit? Any color on the magnitude of gains half-on-half that you can help us with. .

Single digits. So when I use the word incremental, you can think 3% to 6% kind of numbers, 3% to 7%, something like this land in the middle, 5-ish. .

Your next question comes from the line of Brian Chin with Stifel. .

Ask a few questions. So it sounds like your expectation for second half improvement in WFE has somewhat strengthened.

And just to clarify, I guess, first, do you also expect packaging to be higher in second half versus first half?.

I expect it to be better than I was indicating in prior quarters. So I've always talked about what I expect to be a digestion period. Based on the velocity of orders and discussions and things like this, it could be a more muted digestion period could be low digestion period.

But right now, I'd say that I think it's still some level of digestion probably in the quarter and then reigniting a quarter later. .

Got it.

So you're saying -- when you say the quarter, are you talking about second quarter or third quarter?.

No, second quarter, I mentioned in my remarks that we expect to be maintaining these record levels. So continuing to be very strong in the second quarter, which is an improvement over the past, I thought we would be slightly down, but we won't be. Then the third and fourth quarter, there's a lot of discussions going on, but timing is not clear.

Is it all going to be in the fourth quarter is some of it going to come in. There's been a trend towards pulling stuff in. We don't know yet. Exactly where it all shakes out. .

Yes. And just -- I don't know if you are not including specialty in that, but it did sound like that was a little bit softer than maybe you anticipated in Q1.

Was that broad-based? And is your expected improvement in 2Q onwards tied to maybe catch up in some of those delays that maybe happen in Q1? And how does that impact your outlook for the year for specialty?.

I think by specialty might mean power semi because specialty, of course, is very strong because it includes the specialty in AP is one category we talked to. But if you mean power semiconductors. So power semiconductors was weak. We projected it to be weak in the first quarter, but it's reaccelerating.

And in the second quarter, we mentioned it will be near peak levels. .

And we have talked about in prior comments, I'm not sure if I mentioned it today, but we have talked to being relatively flattish to the record we set last year. So it might pull down single digits or will end up being flat, maybe slightly up, but so relatively flattish to the record we set last year in power semi.

So we're expecting a much stronger second half for power semiconductors. .

Okay. Got it. And if I could just sneak one last one in.

I know obviously, it's pretty early to talk about 2025, but even qualitatively, what sort of indications are you receiving from your advanced packaging customers about their demand in '25? And I know you're already sort of lapping record levels now in terms of peak production for Dragonfly and for that tool.

And so I imagine these discussions need to happen kind of earlier to make sure you can prepare the capacity in advance?.

You'd imagine that. I wish the customers would think that way too. We are having discussions about 2025. So we're already having those discussions, in particular for our customers supplying packaging for AI, right? So I would say it's still too early to say what that means for the entire year, though.

I would just say it's positive that they're seeing demand already. .

We're only just starting this year, and they're already talking about what's going into next year. So I'd say that's a positive indication for 2025. And then, of course, you've got the advanced nodes that we expect to improve more meaningfully in 2025. So there's some optimism for hitting on all cylinders in 2025. .

Next question comes from the line of Charles Shi with Needham. .

Mike, Mark. Thanks for getting me to ask a couple of questions. So I think the first one, 30% higher in terms of the Dragonfly revenue supporting the AI chip ramp. If my math is right, I think last year, you probably talked about the total of $230 million of the orders starting from Q2 last year to first half this year.

And if Q2 is kind of flattish, sounds like you're saying it's at the similar level, maintain the strength, you're going to deliver the entire $230 million worth of orders by the end of next quarter. .

I did notice you talked about a little bit of potential digestion ahead. But I also recall you said that there are orders you already received that will deliver in second half this year, but just not at the same level. So can you kind of help us understand this particular part of the Dragonfly revenue.

What's the run rate into second half compared with the first half levels that you're currently seeing?.

So I don't know exactly the math, but I would say your general conclusion is more or less correct. We would have burned through nearly all of the $240 million in Q2, if that was all we ever received for orders.

Of course, we've received many other orders since then, such that, a, my confidence has improved, as I mentioned to Brian earlier, I believe it's Brian, that where I talked about digestion period before, now I'm seeing less of that. There may be a slight dip, but not significant. .

So yes, so I can't really say with specificity. On timing, Q3, Q4 and all that because we just don't know yet. But I can say that the orders have continued to come in. They're coming in from all of the 3 HBM suppliers. They're coming in from AI packaging logic providers. And yes, and they've strengthened since our last call, since the last time we spoke.

.

Got it. So Mike, obviously, seeing second half potentially mid-single digit higher than first half, especially considered maybe a little bit of digestion of AI-related to Dragonfly. I think a few things you mentioned was kind of interesting. You're seeing maybe a little bit of an early data all around related to demand. .

And you also mentioned something really interesting is the NAND that you're seeing some increased demand already. So everybody is focusing on NAND recovery because it's probably the worst performing part of the WFE right now.

So I kind of wonder what's the nature of this -- if I call it, the green shoots you're seeing right now is that technology upgrades going on or some of the product penetration going on or there is a capacity addition that's about to start. So any color on NAND would be helpful. .

Yes. I believe it's more of a technology transition demand. So it's going into these AI servers that require higher-performing solid-state drives, so NAND drives. And I believe because I don't think there's been a lot of factory expansions, I believe there is a repurposing of lines and tooling that was existing to support prior generations.

So what they're doing is they're repurposing, revamping. .

And this particular, why we're seeing orders in business is because to measure this was very, very difficult. And to my knowledge, we're one of the only tools that can measure these high stack channel holes over 200 stacks of NAND. And that's a critical parameter for yield. So that's driving that part. .

I think you asked about gate all around also. I wouldn't say, I mean, it's not new -- like we've been gaining orders to support gate all around from multiple manufacturers of gate all around of that transistor process technology for over a year, probably 18 months now. So this is just kind of a continuation of that.

It's nice that it includes some Iris films as well. So we're seeing still a reaffirmation of our position and our layer share or share opportunities within this important node transition. But you shouldn't take from my comments that there's a ramp coming. I don't think there's a ramp coming. People are still preparing for a ramp. .

Got it. But we are all looking forward to that ramp. But maybe a one last thing, a quick clarification. Maybe this is for Mark. Mark, in the press release, is the advanced packaging and the specialty devices revenue was $161 last quarter, but I think I heard you're saying it's $158.

So wonder between those 2 numbers, which one is the final one?.

Yes, it would be the $158. .

Your next question comes from the line of David Duley with Steelhead Securities. .

Congratulations on nice results. My questions are going to have to do with lithography for advanced packaging. I guess first of all, there's just been a big push by foundry logic customers and now further pushed by the OSATs to expand their capacity.

And I'm just wondering if you've seen your lithography tools broaden out in either applications or customers at this point? And then I have a follow-on about glass substrates, but I'll let you answer that one first. .

Okay. Applications, I don't believe so, but we don't often know exactly what the applications are and which customers, some of our customers are serving. So some serve multiple customers and some are vertical. So that I don't know if that's expanded. It's been high-performance compute.

I would be willing to bet 80% or 100% of it is remaining high-performance compute applications. We added customers that we've talked about through the different earnings calls. .

For instance, the step where we talked about on the prepared remarks for glass that can do glass and traditional IC substrates is a new customer. And we've also had existing customers open up new lines and expand new factories and new lines with our JetStep. So that's also a nice positive for us.

As we look ahead, we're obviously expecting to start adding additional customers as the market starts to recover and high-performance compute server that really the server market starts to get healthy again. .

Now as far as the adoption of glass substrates, I thought that was something that was beneficial to your technology. And I guess I thought that, that might help some of the major GPU guys start to move to more of a chiplet or advanced packaging structures.

Could you just kind of talk about what the road map is with labs substrates and which parts of the market are going to adopt it first? And will the big GPU customers start to adopt some sort of more advanced packaging for that GPU?.

Yes. So I think it's a couple of things. So first, volume, so we have high volume for a complicated chiplet architecture, we are going to be connecting multiple die and this package ends up being quite large to process that efficiently and productively. I think everything is going to move towards panel that fits in that frame. .

Why glass is because, in addition to that, as you drive these interconnects smaller and smaller going from, let's say, 8 to 10 microns of RDL, down to maybe 5, maybe 3, you can push IC substrate. We're not quite sure. But for sure, below that, like, for instance, 1 micron type RDL, something more stable has to be used.

Right now, on a wafer basis, you can use silicon, and that's what's happening for some of the GPU applications you mentioned. .

But for going to much broader scale, something else on a panel will be more efficient. And that, we believe, is going to be glass. So glass will allow 1 micron RDL, at table. There's some heat dissipation advantages with it as well.

And so -- but there's a lot of processing challenges, including with etching, and that's part of our wafer application centers working with a lot of the overall ecosystem to identify and develop solutions for these challenges so that we can usher it, not at usher in, but help to speed the adoption of the glass substrate in a panel form for the market.

.

And this is involving developing the -- not just the lithography capability but also the process control capability. And then with through our partners, some of the other solutions, laser drilling, etching, et cetera, chemistries, resist chemistries, things like this. .

So just to summarize, glass substrates are way better for, as you mentioned, I guess, high-density interconnects.

Is that an enabling feature for GPUs to start to use that type of architecture technology?.

That would be an enabling speaking, yes, but they'd also have to have then the volumes to justify moving from wafer to panel. .

And your next question comes from the line of Vedvati Shrotre with Evercore. .

I wanted to double-click on the advanced packaging trajectory that you talked about in terms of orders.

Could you remind us what's going on with the lead times? Are they the same 3 to 6 months kind of lead times for your Dragonfly inspection products?.

Yes, I would say they're the same lead times. I mean, lead times is interesting because we build to a forecast. So we work with our customers, work on forecasting and then -- so it's not like a lead time is going from order to ordering parts to bringing it in and that kind of thing. That would be a year of a lead time.

But what we work with our customers on is about a 3- to 6-month window and that hasn't changed. Even as we've ramped, we've maintained that. .

Got it.

So when you talk about the second half and some kind of a digestion in 3Q, is that primarily based on what's in your order book right now? I'm just trying to understand, is this kind of conservatism where you're going off the order book right now, and there could be more that could be added? Or if you could provide more color on that?.

So I try to be as accurate as possible. So it's not just the order books. It's what we project will occur, and we don't have a perfect knowledge.

So could some things move -- so we shouldn't get too hung up on the Q3, Q4 because things could move around and they have been moving around as customers ramp plans get more aggressive or they struggle in other areas. And actually, nothing is pushed out, to be honest, everything has been pulled in. .

So that could still happen. But I do think -- well, at least based on our knowledge right now, we will not see the second half be, let's say, there will be a digestion. There will be a dip, let's put it that way. That's what we think right now, could it change? Could customers add more orders? Absolutely.

And discussions are happening now, and that could happen. But my best, most accurate view is what I'm sharing with you. .

Right, that's helpful.

And is there a bucket between -- or is there a different order trend between HBM versus non-HBM customers?.

I wouldn't say there's a different ordering trend. But for sure, HBM is driving a ton of our business. I mean DRAM is still pretty slow. We get incremental orders here and there. But high-bandwidth memory is driving a lot of business. The advanced logic -- the logic packaging is driving a lot as well. So... .

So the digestion is pretty well spread between HBM and non-HBM products?.

I would say it's tied to AI. So it's actually it's tied to AI very specifically. So we see power semiconductor in the second half growing, advanced nodes is growing. I think the AI is the question because it's been so strong, it's ramped so much. And I think, is Q3 going to be a little down and then a big surge in Q4 possible.

Is that going to get spread around and we see less. These are all the things that customers are working with us on. .

Right. Okay. And then I think shifting gears to specialty devices. So most of the power semi companies have started to take down their CapEx this year.

What's sort of giving you the confidence that this continues to grow through, I mean, remainder of the year?.

Well, that's the interesting dynamic with process control, is that companies, if they don't have really, really high yields, they can focus on driving more they can focus on output either by driving more capital spend or in a time like this, they can focus on driving the yields up, and generally driving the yields up means bringing in more sophisticated tools to help identify in the process where yield loss is occurring and then go and address that.

.

So that's what we're seeing right now. And our confidence comes from our backlog and the conversations that we're having with the customers. So pretty confident right now in the power semi remain strong. .

And then on the advanced nodes, how is your visibility into the back half of this year?.

I'd say it's -- the visibility is good. It's as good as what the customers have. The customers are opening new factories and ramping new lines. And as long as those stay on track, we will see what we see. But those things have moved. So we've seen delays announced by some of our large customers.

And we're optimistic that those delays are all baked in and that there won't be any more, but I can't guarantee that. But I can tell you we're in close contact, we are meeting with them frequently, working with them frequently. And so at least as soon as they know, I think they're sharing it with us. .

Right.

And so is it fair to assume that it will be kind of a sequential growth every quarter here onwards?.

Yes. For advanced nodes, I think that's pretty fair to say. Yes. .

Right. And then maybe one last one that I'll squeeze in. On the margin side, you mentioned productivity gains coming around in the back half of this year.

Could you kind of help us quantify that? Like what kind of operating expenses and gross margins we should be looking at?.

Yes. No, I mean I think from the gross margin standpoint, I mean, we're projecting to have quarter-over-quarter sequential gross margin improvement. I think as we stated before, our goal is to exit the 2024 back at levels that we've seen historically at that -- that would be that revenue level.

As we said before, plans that we have in place, it's working with our suppliers, it's working with the plans and driving through efficiencies. .

So I would say a thoughtful way to do it would be the planned margin improvement quarter-over-quarter. Operating expenses in the first half of the year is always tends to be higher for us with some of the compensation elements hitting. But again, we do see a slight decline in the back half of the year.

So I wouldn't expect operating expenses to be above the peak of Q2. .

You next question comes from the line of Mark Miller with The Benchmark Company. .

Just was wondering, you mentioned that certainly high bandwidth memory has been playing a major role.

Can you estimate in terms of your current backlog, what percent of the backlog is related to high bandwidth memory for AI applications?.

I don't have that in front of me, but I would say it's a high percentage of the backlog. .

And extrapolating, it's always dangerous to do this.

But extrapolating the next year, do you think that will be also one of the most significant constituents of your backlog? And what did that gate all around, how will that be increasing as a percent of your backlog?.

I don't worry so much about the backlog because of our 3 to 6 months in the build to forecast and that kind of thing. So backlog during COVID was a powerful, but prior 20 years, this backlog for us wasn't so interesting. Ever that big are of a good indicator of where our business was going to go.

But to answer your question, I'd say HBM is, I would say, interesting. So the concern is there a bubble happening. There's a lot of ramping of capacity for HBM. I think several players are trying to gain share and gain position. .

And the current leader is trying to maintain. So there's a lot of capacity coming on, and it remains to be seen, is that excess capacity? Or does AI continue to really outstrip projections and other players come in and that HBM memory gets all used up such that we'd see another big ramp. That part, I don't know so much.

So with that, I would say gate-all-around would be a larger -- I would expect gate all around, especially towards the second half would be a larger part of our revenue or backlog for next year. .

And then finally, you mentioned an improving margin picture.

Is that being driven by newer tools, having higher margin profiles such as your Dragonfly with the new sensor?.

That's part of it. It's also about the supply chain initiatives that we've talked about, and a lot of those were expected to start adding -- cutting in this year quarter-by-quarter, even starting last year, quarter-by-quarter. So we're seeing improvements from that, too. So it's twofold.

The new products and the new product adoptions, and that's where advanced nodes growth will start to really have another impact, but we're taking control of our own destiny. We're also addressing some of the supply chain opportunities we have through the merger. .

This concludes today's question-and-answer session. I would now like to turn the call back to Mike Sheaffer for any additional or closing remarks. .

Thanks again for joining us today. Just a quick reminder for everybody about 3 upcoming events. First, Onto management will be participating in the B. Riley Annual Institutional Investor Conference in Beverly Hills on May 22 and 23. Next, we'll be participating in the Stifel Cross-Sector Insight Conference being held in Boston on June 4 and 5.

And finally, we'll be at the Wolfe Research Small and Mid-Cap Conference in New York on June 6. A replay of the call today is going to be available on our website approximately 7:30 Eastern Time this evening. I would like to thank you for your continued interest in Onto Innovation. .

Rachel, please conclude the call. .

Thank you. This does conclude today's call. Thank you for your participation. You may now disconnect..