All right. Good morning, everyone. Welcome to the 28th Annual Needham Growth Conference. My name is Charles Sheehan. I'm the semi-cap analyst at Needham. Joining me on stage here is Onto CEO Mike Plisinski, CFO Brian Roberts, and IR Sidney Ho are also sitting in the audience. So, Mike, once again, thanks for joining me today.
Pleasure to be here.
I heard that we're going to finally have a fireside chat this time.
That really bothered you.
You want to have a few slides to go through first. The next stage is yours.
Yep. Thank you very much, Charles. So I just have about four or five slides just to kind of level set everybody on where we are, especially with this market and what happened during the year. So just to start, of course, the safe harbor. Whoops. Remind everyone what I'm about to share is based on my best knowledge, yada yada, subject to risks. Look them up, please. Now let's go into it. So I think everybody's starting with AI, right? AI is driving our industry right now. You've seen NVIDIA's comments about a 40% CAGR over the next few years. That's all on the demand side. Of course, the supply will be driven for us. It will be driven by the expansions of factories, things like this. So super great backdrop for the industry.
What you may not realize is how tied we are directly to that AI supply chain. If we look at just the 2025 revenue, about 61% of our revenue came from customers directly supporting the AI supply chain. A very tight tie. Because of that, and because of the exuberance or the more positive tone our customers are having in the last few months even, we're starting to see an outlook for 2026 that is about at least 10% stronger in the first half than our second half of 2025. That's an improvement over the last few months where we said, yeah, it'll be stronger. We kind of implied single-digit kind of numbers. This is above. Then above that, we have the SDI or the Semilab benefit. Additional growth on top of that.
We still believe that the second half is going to be stronger than the first half. Now where do we play in that AI value chain? Several areas listed here. I think none of these are a surprise, but we just wanted to level set everyone. As far as the gate-all-around, gate-all-around node, we have the strong OCD position. We have growing position in the common films. Of course, opportunities to grow in the integrated metrology. Then you can see throughout these different markets where we're strong. Leading in inspection for advanced packaging. Of course, the 2D macro inspection leadership with new opportunities now in 3D metrology, which we're qualified at two of the three HBM memory manufacturers. Good progress there on the 3DI. You can see right through here.
But I did want to call out the co-packaged optics. Even though it's very small, I think there's really strong demand from the market for bringing this kind of technology online. And there we have a very big position, not in revenue, but in the types of products serving co-packaged optics from specialized inspection to metrology and unique opportunities in 3DI metrology. And by the metrology I mentioned first, it's primarily films. So films metrology and then 3D metrology. So great opportunities there. And then that's all started through, let's say, late 2024 through 2025. Recently, we've expanded our opportunities for 2026. And we're looking at critical films, something we've talked about.
We think that's going to be more of a play in 2026 with our critical films, our Iris G2 tool, as well as the 3DI metrology, which has been qualified at OSATs, HBM manufacturers, and co-packaged optics, so photonics manufacturers. We think that's going to continue to grow into this year, and then charge metrology, something new. That came from the SDI acquisition. All told, these opportunities add about another $1 billion in opportunity with critical films being the biggest, but let's go into charge metrology, so this is a new capability that we added with the SDI, and we talked about covering what synergies we see from the SDI acquisition, which we'll do in more detail at the next earnings call, but I just wanted to give you a quick view of what this new technology is going to do for us.
When you think about chiplet architectures, we're involved in everything tied to the interconnects. The measurement, the printing through our lithography capability and inspection of interconnects. That's what we do today. Whether it's RDL, TSVs, bumps, the interconnect technology for process control, we're very strong there. What's happening in heterogeneous packaging or chiplet architectures is that residual charge. If you have a die and it goes through some plasma dicing or plasma etching, it can pick up a charge. That charge can stay on the die. Typically, designers have circuitry to dissipate these charges. When you're going to chiplet architectures, the designers are saying, no, the package has to have that dissipation because we don't know what kind of chips you're going to be connecting together.
So customers like TSMC have a very big concern about taking these potentially charged die, putting them on an interposer or on another package, and then having that charge go through, as you can see here, through the interconnects and damaging the die next to it. And there's no way to measure that except through this technology. So we think there's a big growth opportunity here as chiplet architectures become more and more prevalent. The other area is in power semiconductors. Not super hot right now like AI, but definitely a secular growth driver, at least in our opinion. Here, for compound semi-manufacturing, epitaxial is one of the most critical steps for the process, so for the device. So the Epi process aligns the crystalline structures. It's one of the biggest determining factors for the performance of that chip when it comes out through test at the final stages.
So we do a lot of inspection and analytics on the crystalline structures. But we don't know which one's going to electrically fail and the customer doesn't until the very end. With this surface charge metrology capability, we're able to actually take our inspection data, feed it into them, into this tool, the charge metrology, and measure in those areas for the electrical performance of those chips, of those defects to determine whether or not it's going to be a killer. And that's way up front in the process. So it's a significant advantage for our customers. So this is another whoops, I didn't hit the plus. So this is another area of growth with this new charge metrology capability we have with SDI and how it fits in nicely with our portfolio. Now, I mentioned inspection. I'm sure everybody's interested in inspection. What makes the Dragonfly so special?
Why does everyone bug me about Dragonfly? Just kidding. So it turns out when I made this slide, my own team didn't know why we named that thing the Dragonfly. So in nature, the Dragonfly has the most eyes of any creature in nature, 30,000 eyes. And their reflector, they can capture multiple different wavelengths. They have different sections. These Dragonflies have different sections for high-speed acquisition and that kind of thing. So very diverse capability, which for our Dragonfly is very similar. So the Dragonfly capability we have today is much more than 2D inspection. Over the years, we've added a tremendous amount of sensors that solve a variety of applications for customers, both known and unknown. So applications that customers didn't realize they had until they start processing and ramping and then say, holy moly, we don't know how to measure underfill.
And how do we get this data? Well, we have three different sensors. When you combine those data streams, we can give them exact accuracy for that. So you can see here the number of new applications that we added in 2025 just on the Dragonfly G3. So pretty powerful, pretty compelling. And that's why we've seen less of a loss or less of a drop in the Dragonfly business as we expected at the start of the year. Now, everybody's probably interested in an update on the new Dragonfly platform. So this will preserve all of the sensor capability that we had before, but significantly enhance our resolution. And it's a ground-up system you've heard me talk about on earnings calls, literally from the platform it sits on all the way up, the vibration isolation.
You can see here the slides a little off, but you can see here the performance improvement over the prior generation. There is significant improvement in both throughput as well as resolution. Those ovals show you the resolution, where it is in a resolution perspective, lower is better, and where it is on throughput. Then the competition. We get a lot of questions around the competition. The competition is somewhere in between that. When you think, is this a me-too product line? No, it was designed to go after front-end applications. It was designed to be significantly more than what the packaging market needed because, frankly, a packaging market didn't need it some 18 months ago. We have a pretty compelling position here. The other thing everyone's asking is, did we ship the tools we said? Yes, we did.
And in fact, we now have a purchase order from a fourth customer that we expect to deliver in this quarter, as well as several other systems that we expect to ship in the quarter. So already, we're seeing some significant traction on this new platform based on its capabilities, et cetera. One more. I know you're anxious to grill me. I'm not anxious. Last slide. So we talked a lot about revenue, about growth, about opportunities. But we haven't lost sight of the fact that we have significant opportunities to improve our operating margin. And it starts with gross margin, goes right down to the bottom line. We've done three things to help improve that. The first is the extended factories. So we talked about our move, our very aggressive move to shift manufacturing over to our Asian partners.
We've achieved, I think we said on the last call, about 50% of our production is now overseas. Target is around 80%. We should be able to achieve that mid next year or mid this year, sorry, so in the next few months, next two quarters, so very good progress there. We're at the point now where all products are being built over there, so now it's a matter of optimizing and ramping the suppliers, and it's going quite well. The other area is because of these new products, these new applications, we're able to deliver more value for our customers and, of course, share in some of that value creation, so we have higher margins on the products, on the new products coming out. And then we've always said that the SDI benefit, the Semilab benefit, is margin accretive out of the box.
So both from a gross margin perspective as well as an operating perspective. All told, without counting any of the increased benefit from growth in our core products, these three areas will result in at least a 30% improvement in our operating margin, in our earnings, net income for this year, 2026. And that is my last slide.
All right. Thanks, Mike. All right. Let's go back to one comment you made 90 days ago. A lifetime. Yeah, a lifetime away. You said 20% more AI packaging tool opportunities. But you were characterizing that 20% as its initial discussion. And over the last 90 days, do you see that 20% more packaging tools, AI packaging tools, that discussion has turned into firm orders? And are you ready to raise the number, maybe to higher?
No. So we just ship the tools. They need to be qualified.
Then we can raise. So the discussions were around potential needs, helping us to make sure we had our supply chains ready to be able to deliver what could be expected. Three months ago, we were talking single digits for first-half growth. Obviously, we're seeing some significant improvement to that. So you can assume that there might be some benefit there on the packaging side. But we wouldn't change any of the numbers until we get through the qualifications, which are three to six months. We have to look at the ramp timing. Those are also changing. And we need to look at those insertion points.
Great. Thanks. Maybe go back to the other thing, the first half of 2026 versus second half last year, that single-digit sequential half, not exactly half over half, but sequential single-digit growth. Now you're expecting 10% +. Some of that is packaging.
What's the other part? What's driving the other part of the raise?
Yeah, advanced nodes. So it's primarily advanced nodes and packaging. So I don't have a breakdown. But I know gate-all-around was certainly a big one. We've seen some spending there. Packaging from HBM perspective, the Dragonfly demand is strong. The HBM orders we talked about from a 3DI perspective, one of those customers we've gotten the VPA, so we have some visibility now. The others we're working on still. So I think the growth we're talking about is pretty much uniform, but still driven a lot by AI, both front end and back end.
Got it. Got it. So maybe we'll touch upon each of the elements you just mentioned. Let me start from the CPOs. As I understand that the Dragonfly G5 qualification is very important. Can you kind of remind us where you are?
I think you mentioned something, but it's more at a higher level. You talk about the fourth customer, but we specifically would like to know what exactly at the leading foundry where the qualification has been going on and what's the next milestone, and for CPOs.
Yes. Yep, so what we've said publicly is three to six months, which is a pretty accelerated qualification period driven by the customer. We won't share any details beyond that now, but internally and with the customer, we've worked through the exact timing, the exact layers, the exact specifications they want to see demonstrated in order to prove out our process, so that's going to all add up to be in that range we just talked about, and each earnings call, we'll be able to provide clarity as to our progress through that. Hopefully, we're on the earlier end of that.
But three to six months is the comfortable place to be.
So still some part around the middle part of this year.
Yeah. We should be through those qualifications by the middle part of this year, if not sooner.
Is it fair to say that the 20% AI packaging opportunity, you will be more comfortable committing to that number once the qualification with this particular customer is done?
Yes. That would be a very fair statement to say.
Okay. Thank you. The other interesting development around CPOs is OSAT. The two OSATs seem to be building out more of the CPOs, more involved in CPOs. And I know you guys started talking more about the OSAT side of the AI packaging business.
Can you kind of talk a little bit more about what's your market position in OSAT, the two OSATs, leading OSATs, and how do you feel about the growth of the OSAT business, particularly in AI packaging this year, maybe even a little bit longer term?
So I think there's a couple of areas there. So there's the outsourcing of capability from TSMC to the OSAT. So TSMC can focus on areas that they're going to consider more profitable, so some harder areas. That's driving some growth. And then there's the, let's say, other customers of TSMC that don't feel like they're getting enough attention, et cetera, et cetera. And they are looking at some OSATs to pick up some packaging capability. That drives another piece. And that's some different technology. We're benefiting from both, given that we're a strong 2D supplier to the OSATs for decades.
The strength of our tools is generally more on the high end, and all of these examples I just gave, that's driving OSAT business today, are high-end applications, so the higher resolution, the higher precision, the capabilities of ClearFind, et cetera. In addition to that, we talked, I think, two quarters ago, maybe last quarter, about being qualified now and getting volume orders from OSATs for 3D metrology, so bump metrology, which we didn't have before. That was typically a competitor's area of strength, so the new 3DI is proving compelling, not just for the high-end latest HBM applications, but also for OSATs, and that's based on its throughput and the fact that we can add it to our Dragonfly and provide a complete total solution.
Stay on logic for one more question.
One area I think you guys have talked a little bit less, I would say, over the last year was your lithography business, the panel-level packaging lithography, that your JetStep system. But we're always still paying a good amount of attention to that particular area because your indirect customer, as they go through their transformation, their turnaround, we think it's probably going to be a little bit more positive this year and maybe going forward. And especially, we're hearing more about EMIB getting more of a traction among the AI packaging applications. So mind if you walk us through where you are, how you think about the litho business, and maybe not just a litho, maybe Dragonfly business, at that particular customer directly or indirectly as well. We won't speak specifically to one customer because they tend to get mad about that.
But what we'll say is that from a level. Well, from a process control perspective, let me step it back. From a panel perspective, for sure, we see the markets heating up. We see a lot of excess capacity that had been there for three years starting to be taken up. And we're getting much more traction with customers looking at, let's say, funded investments now versus theoretical investments. So that's a positive. Since then, so in the last three years, we've also seen customers recognize the importance of inline process control. So as the panel market has evolved and gotten more sophisticated, moving down in lines and spaces, the need for inline process control and the capabilities of Firefly, which was significantly more expensive than their existing tools, which they just used for a spot check at the end of the line, that's become more critical.
We're seeing the metrology capabilities on the Firefly, like those sensors we talked about, as more interesting to them even than the inspection. There's a lot of opportunity in panel for both our process control as well as the JetStep lithography. What makes a JetStep so interesting? JetStep, with its very wide-field optics, is able to handle large packages without any stitching. That's a key point for our customers. The other advantage, and which makes it so expensive, is the wide-field optics are also very high resolution, so near one micron resolution, printing image resolution. That's unique in the industry. No one has that capability proven and delivered.
So that's an area that our PACE Lab has been really doing a great job bringing in customers to run samples, to learn from the partners we have there, as well as our own equipment there, both in the substrate as well as glass, which glass we didn't talk about. But that's the next evolution, or maybe the first. We'll see. There's kind of two camps there. But for sure, we see the panel growing in opportunities. Enterprise servers are where they came from. But we're seeing AI-driven opportunities from our customers as well.
So you talk about over the last three years, right, that the excess capacity was built probably during COVID for the panel-level packaging advanced substrate side of the business. Where do you see today in terms of that capacity digestion?
Are we closer to a point maybe the demand could pick up in the next 12-24 months?
Yeah, definitely 12-24 months. Based on the conversations we have with customers, there's real plans from now. Remember, the lead times are long on the steppers, which is good. But there's real conversations happening for funded expansions, not hope, happening now, so actually over the last six months. And hopefully, we'll start to see those discussions turn to firm orders and then expansions into next year. This year will also be OK.
Got it.
But Firefly is the kind of upside. That's a tool where we're gaining a lot of traction. It's much cheaper. So it's easier to bring in. They can apply it to their existing lines, including R&D and pilot. And again, the recognition how important inline process control is to these customers is becoming eye-opening.
Got it. So maybe let's talk about memory. I think memory, it's two things for you, right? One is HBM packaging. The other is the front end, DRAM. And also, I would like to talk about NAND as well.
First on HBM packaging, I think you were relatively early to call a little bit of a capacity overbuild. I think you caught that last year. That helped us. What's your view right now? Are we nearing the end of that digestion, or still some way to go?
So everyone knows a part of that digestion was tied to Samsung. So you can look and see if Samsung's been qualified by NVIDIA, who's driving the biggest lion's share of HBM adoption right now, and they're out. So you hear multiple comments around that. I'm convinced that that will happen. So that'll happen. Those tools will get digested or utilized.
And Sidney has highlighted the ASIC guys are getting more aggressive. They can win some business there. It's not all tied to NVIDIA. So. And Samsung is well positioned for some of that. So I think we'll see that occur. But based on the discussions with the HBM customers, I think there's meaningful levels of expansion, at least at the other two, that suggests we should see a nice year of growth in 2026.
Great. Growth. That's what we want to hear. Oh, and you asked about the front end.
Did you ask about the front end?
Not yet, but please. The DRAM and NAND, the front-end business, advanced node business. Yep. Where do you see them? I think you mentioned about clean room constraints. But what's your current projection right now?
When do you think the business could pick up again?
The reason why I said again was, I believe, first half of the year, memory was great. Then roughly in the summer, it got a little bit quiet. Then come back a little bit again towards the end of the year. So that was the trajectory of your memory business, especially on the DRAM side, right? What do you think how this year will play out for DRAM? And then maybe I'll ask you about NAND as well.
I think from what we're seeing, it'll be more second half, kind of mid to second half weighted, based on factory expansions, et cetera, on the front end side. For HBM, we're already seeing demand growth there. The question really is around the magnitude. So when you look at these big numbers, about 40% CAGR, 20% growth, then you kind of bring that down to how much factory space is available.
And you start to say, geez, it can't grow that much. There's just not enough capacity out there. That's what we're working with customers on now, trying to understand. They're looking also how we free up, how do we reallocate, et cetera. For us, any expansion is bigger now than it was, as we've added more common films. OCD has always been very strong, so that's a very high position there. But we've added more common films. We've added more integrated metrology, so higher share, more applications there. So any spending there should be outsized benefit to us than it was certainly two years ago or even last year.
Got it. Let me ask you about China. Your China exposure is relatively light compared with almost all your peers.
I noticed.
But it's probably good in one way, but it can be bad in the other way.
Good in a way, it provides protection against geopolitically driven policy volatility, let's say. But we want both things, right? Of course. We want protection in bad time, but we want you to be participating in the upside in the good time.
So how should you think about your China business? I think you started to talk a little bit more about China. But can you shed a little bit more light on how you plan to grow?
So I think there's two new elements to the strategy. One is the addition of the Asian factories. So by moving manufacturing outside the U.S., we've significantly de-risked ourselves in the mind of our Chinese customers. And so we've seen much deeper engagement from the executives, even visiting our PACE Lab, in looking at how they can expand with us.
In fact, one of the executives that visited us made a comment, "Why haven't we been buying from you? You used to be our best customer. You tell me." So I think that that's going to be a positive side benefit of moving overseas. The other piece was the SDI. SDI has a pretty strong position in Chinese manufacturers. Some of them are different than the customers we had. So we'll be able to leverage those installed bases to bring in some of our additional products and offer a broader portfolio to the customers in China. And because especially the surface charge metrology is so unique, you can only get it from us, that's going to be a great door opener, just like Echo, the old MetaPULSE was.
All right. I think let's take some questions from the audience, please. Very interesting presentation. Thank you.
Given the tight capacity that we're seeing particularly in things like HBM, you obviously bring tremendous value to your customers. So can you talk about how they quantify the value you're bringing and how you think about the price increases that you can justify to them?
Allow me to repeat the question for webcast. So the question is about the value, how you quantify the value you are providing to areas like HBM, and are you able to raise prices? So great question. It all comes down to step one, can you see and solve the problem that they're having? So can you see the defects of interest? Can you make the measurements that are critical to them? And then, if you can, and do it reliably and repeatedly, then the question is, can you do it with a better COO?
And COO is going to be a price and a speed, the combination. So that's where we typically have a good advantage. So we might not be the cheapest price, but we can be much faster. And the new Dragonfly G5, if you saw it, it's almost double the speed of a G3. It's a significant improvement over G3. And its sensitivity goes down. So we can see much more. Plus, we have all those other capabilities, those other sensors. So I think in that case, and the case of our metrology, the Iris platform, the critical films, it's a matter of making that precision. But then we'll be hopefully faster, but definitely at a different price point than their alternative today. So the COO will be much better.
What's the difference between the old version of Dragonfly versus the new, so you quantify it in terms of?
We've said it'll be meaningful improvement to margins.
OK. Fine.
Thank you.
So between price and COGS. Yeah. One last question, please.
Thanks, Mike. Jeff Feinberg, what do you think is the key to your sustainable advantage being ahead here at Onto Innovation? What's the key differentiation that you think is sustaining your advantage?
Yeah. I think it's understanding the customers, the markets. I think it's because of that understanding, we know a lot of the challenges they have beyond the simple inspection piece. That's one lane, but they have many lanes they have to address when they're trying to release something through packaging, through the final steps. So I think that's one of our advantages. 2D inspection has been an advantage. With this new platform, it'll continue to be an advantage. So one of our competitors has a tremendous breadth in 2D inspection, tons of technology.
A lot of it's not as applicable to a packaging world, and where we need to go, we understand, so the next step after this, we're already working on, so I think we have pretty high confidence in that it's not going to go down to 10 nanometer needs in packaging, but all the filtering out noise, dealing with high warp, dealing with crack, all these things are different problems that need to be solved on the same tool. That's, I think, a strength of ours, and customers want to work with us, and they want to see an alternative.
All right. Thanks, everybody. That's the end of the session. Thanks, Mike. Thank you very much, everyone.
Thank you, Charles.