Thank you, and welcome, everyone, to FormFactor's Second Quarter 2025 Earnings Conference Call. On today's call are Chief Executive Officer, Mike Slessor; and Chief Financial Officer, Shai Shahar. Before we begin, Stan Finkelstein, the company's VP of Investor Relations, will remind you of some important information.

Thank you. Today, the company will be discussing GAAP P&L results and some important non-GAAP results intended to supplement your understanding of the company's financials. Reconciliations of GAAP to non-GAAP measures and other financial information are available in the press release issued today by the company and on the Investor Relations section of our website. Today's discussion contains forward-looking statements within the meaning of the federal securities laws. Examples of such forward- looking statements include those with respect to the projections of financial and business performance, future macroeconomic and geopolitical conditions; the benefits of acquisitions and investments, including acquisition of manufacturing facility; anticipated industry trends; potential disruptions in our supply chain; the impact of regulatory changes, including tariffs and changes in export controls; the anticipated volatility in demand for products; our ability to develop, produce and sell products; and the assumptions upon which such statements are based. These statements are subject to known and unknown risks and uncertainties that could cause actual results to differ materially from those expressed during this call. Information on risk factors and uncertainties is contained in our most recent filing on Form 10-K with the SEC for the fiscal year ended December 28, 2024, and in our other SEC filings, which are available on the SEC's website at www.sec.gov and in our press release issued today. Forward-looking statements are made as of today, July 30, 2025, and we assume no obligation to update them. With that, we will now turn the call over to FormFactor's CEO, Mike Slessor.

Thanks, everyone, for joining us today. FormFactor reported sequentially stronger second quarter revenue that exceeded the high end of our outlook range due to higher-than-anticipated growth in our probe card business. Despite this revenue strength, non-GAAP gross margin and overall profitability fell short of our outlook, mainly caused by an unfavorable shift in product mix and unforecasted ramp- up costs for a second HBM DRAM customer. In the current third quarter, we expect to deliver revenue comparable to the second quarter and slightly higher gross margin and operating profit. Before we dive into segments and market level details, I'd like to spend a few moments reviewing the year-to-date. FormFactor's business continues to be driven by 2 dominant themes: advanced packaging and generative AI. As front-end-driven Moore's Law slows, the innovation and performance offered by advanced packaging and chiplets allows our customers to accelerate their road maps and deliver spectacular performance improvements in compute and memory. These innovations, like the stacking of DRAM chiplets to produce HBMs that are then integrated with GPUs in multi-reticle CoWoS packages, co-packaged optics and even chiplet-based processors deployed at the edge, are enabling the transformative capabilities of generative AI, fueling forecasts of semiconductor industry growth to $1 trillion early in the next decade. As the leading supplier of probe cards and systems that ensure the quality and performance of each individual chiplet and the stacks of chiplets in the advanced package, FormFactor is uniquely positioned in enabling these innovations. We continue to be excited by our growth prospects as advanced packaging drives increased test intensity and test complexity, creating increased demand for our products. At the same time, we acknowledge that our recent financial results and especially gross margins have not reflected our unique market leadership position. There are multiple reasons for this gross margin…

Thank you, Mike, and good afternoon. As you saw in our press release, Q2 revenues were $195.8 million, $0.8 million above the high end of our outlook range, and non-GAAP gross margin of 38.5% was at the low end of the range. These, together with OpEx slightly higher than the midpoint of the outlook, resulted in a non-GAAP EPS of $0.27, $0.1 above the low end of the outlook range. Second quarter revenues increased 14.3% from the first quarter and decreased 0.8% year-over-year from our Q2 '24 revenues. Probe Cards segment revenues were $162.1 million in the second quarter, an increase of $25.6 million or 18.7% from the first quarter. The increase was driven by higher revenues in all the markets we serve, most notably in foundry and logic and DRAM. Within the Probe Cards segment, Q2 foundry and logic revenues were $100 million, a $14 million or 16.7% increase from the first quarter. Foundry and logic revenues increased to 50.8% of total company revenues compared to 49.8% in the first quarter. DRAM revenues were $57.1 million in Q2, $8.2 million or 16.8% higher than the first quarter, an increase to 29.1% of total quarterly revenues as compared to 28.5% in the first quarter. Within DRAM, HBM revenues increased $7.4 million from $29.5 million in Q1 to $37 million in the second quarter. Flash revenues of $5.5 million in Q2 were up $3.1 million from the first quarter and were 2.8% of total revenues in Q2 as compared to 1.4% in Q1. Systems segment revenues were $33.7 million in Q2, a $1.1 million decrease from the first quarter and comprised 17.2% of total company revenues, down from 20.3% in the first quarter. GAAP gross margin for the second quarter was 37.3% as compared to 37.7% in Q1. Cost of revenues…

[Operator Instructions] Our first question comes from the line of Brian Chin from Stifel.

Maybe firstly, just -- sorry if I missed this, but how much residual customer or HBM4 product cost is still embedded in the third quarter gross margin guide?

I'm not sure I understand the question. The -- you're talking about the ramp-up cost when we...

Yes. Ramp-up cost, yes. Anything that hasn't already been reflected in second quarter.

Got it. Yes. There is no additional ramp-up cost assumed in Q3 for this HBM customer or others.

Okay. Got it. So you're comfortable that, that's behind relative to the product ramps for that customer.

Correct. As we said in the prepared remarks, we resolved the issue with the customer in the third quarter.

Got it. Got it. Maybe just on the broader business trends relative to how you talked about third quarter across your markets. Surprise to nobody, but consumer PC phone markets continue to feel somewhat underwhelming. I know you don't want to guide 4Q per se, Mike. But as of today, do you think that trend of higher sequential DRAM systems and lower sequential logic/foundry could persist in the 4Q as well? And any thoughts about how that kind of trend impacts gross margins?

Yes. So if we look at -- and we're not going to guide Q4. But if we look at the current trends, clearly, our business and the industry overall is being driven by 2 fundamental trends: one, advanced packaging; the second, generative AI. And clearly, there are a couple. I think in growing the foundry and logic business, we're also taking the approach of not just waiting for recovery in mobile and PC but doing qualifications for things like GPUs, the hyperscalers and custom ASIC. And in fact, there are some contributions of hyperscaler custom ASIC in our second quarter results, where we've delivered significant volumes of one of those custom ASIC chips. And so I think I agree with you. The PC and mobile markets remain tepid. I don't think pinning foundry and -- strong foundry and logic secular growth on those is a particularly wise move. We did see some second quarter seasonal strength, but shifting our focus to grow the foundry and logic business in areas driven by generative AI is where we're putting more and more resources. I think we'll continue to see DRAM grow driven by HBM strength. And systems really levered, as I said in my prepared remarks, to co-packaged optics and, to some extent, quantum computing as that continues to develop. All told, that is -- continues to be a challenging product mix for gross margins, which is why we're taking the steps we outlined in the prepared remarks to improve gross margins and get back on track to our target model 47% level.

Got it. And then maybe last question. As you begin to ramp up with your largest customer HBM4 and then presumably with second and third customers HBM4 as well, do you see a better -- that's wafer test intensity, but also maybe a margin potential out of those products. And how significant is that sort of Q2 into Q3? And where do you think that goes in the next couple of quarters as a percent of the HBM business?

Yes. it's still a pretty balanced demand profile between HBM3 and 4, but definitely increasing trend on HBM4. And as we said, I think, in the last call, we expect that crossover to happen sometime in the latter part of 2025. HBM4 does have, at least for certain test insertions, more challenging test requirements, what we refer to generally as increased test complexity with things like higher speeds, different temperature and scaling ranges. So HBM4 does offer the opportunity, at least for some of these high-speed insertions, for us to deliver more value. And when we do deliver more value, higher yields, higher performance envelope, we do get compensated for that by our customers. So HBM4, at least part of the test insertions, we do expect to have higher ASPs for the high-speed cards.

[Operator Instructions] Our next question comes from the line of Craig Ellis from B. Riley Securities.

The first question I wanted to ask is just a follow-up on gross margins to clarify that I've got all the pieces correct. So Shai, for the third quarter 40% guide, there isn't any customer HBM start-up costs in there. But did you say there were any headwinds from tariffs? And given the level that we're at, is there anything happening with competitor pricing at these levels or any other one-offs that might be adversely affecting gross margin?

So I'll say again, yes, we don't expect additional ramp-up cost in Q3. So the 40% gross margin assumes no repetitive expenses like we had in Q2. When it comes to tariffs, what we said is that the 40% assumes 1 to 1.5 percentage points headwinds from tariffs as they stand today. There are some things that might increase tariffs on August 1, and there are other rules that might be imposed by the government. And that's what we say that if these things will actually happen, the headwinds might increase to up to 2 percentage points. When it comes to competitors' pricing, we don't see significant changes in that. And it goes back to Mike's answer on the previous questions. It all depends on the value we deliver to our customers.

Got it. And then the second question is for Mike. Mike, I was hoping that you could just step up to a higher level and talk about the business and how it performs toward target model parameters given that, that was an emphasis in your commentary. And specifically, the question is that we're only about 7% or 8% from target parameter revenue levels on a run rate basis, but we're 700 basis points away on gross margin. So if you were to bin out the 3 or 4 things that can help us bridge that gross margin gap, how would those prioritize? And what would be the relative contribution?

Yes, Craig. So I think an absolutely fair observation and one that we tried to point people to in the prepared remarks. We recognize that we're just shy of the revenue levels, at least on a quarterly run rate basis of the model, yet a long way away from the gross margin levels. There is an element of volume that is helpful for gross margin. There's also a variety of different operating cost reductions that we are continuing to execute on. I'll remind people that around a year ago, we completely changed the operating organizational structure and have brought in some very talented people from the outside with strong semiconductor experience. We're now gaining momentum and improving our operations and lowering our operating costs. And so there's some short-term elements there as well. And then finally, the initiatives around growing our foundry and logic market share with things like GPU qualifications, with things like addressing the hyperscaler custom AI business as we continue to see a pretty lackluster PC and mobile environment. Those are roughly all 3 equal components as we think about bridging our way back up to the 47% of the target model at $850 million in revenue.

And our next question comes from the line of Tom Diffely from D.A. Davidson.

Maybe for Shai, when you look at the ramping of the new facility in Texas, what do you think the impact will be, if any, on the P&L over the next few quarters? And then once up and running, what is the -- what do you think the long-term impact or benefit to margins will be?

Yes. So we acquired this facility about a little over a month ago, right? We're still working on our plans, and it's been only a few weeks since we closed it. We will provide updates as we make progress, as I said in the prepared remarks. One of the reasons, I think it's a good reminder, that we acquired this manufacturing facility is that it is located in a lower-cost region. So we are committed to the target model. The Farmers Branch facility is part of our medium-term process as we expect lower cost of operations in that region, including environmental-related cost, labor cost. And also, as I said in the prepared remarks as well, we are seeking state and local incentives, which would lower our cost as well.

But we won't see an impact on the P&L until after it's up and running?

We will provide an update on that as we make more progress.

Okay. And then getting back to the start-up costs, would you expect to see start-up costs at some point down the road for your third high-bandwidth memory customer?

Tom, it's Mike. I'll answer that one. I think the headline for the second quarter results and what we're trying to convey is we're now shipping in volume to all 3 major HBM providers. They've all been customers for many years, primarily on commodity DRAM, but we're now qualified and shipping in volume. The issue with one customer in the second quarter was really associated with a single design. And as you know, probe cards are consumable specific to each chip design. In customizing our architecture to meet that specific design at this customer, our engineering team made a choice that we had to go back and change. We couldn't reach one of the critical specifications for that customer. And that's sort of the fundamental root cause and reason for this charge in the quarter that we're calling a start-up cost. We don't see anything like that at that customer with other designs or with any other customer. The issue has been resolved. Unfortunately, it did have a gross margin impact. But as we continue to build share at those other 2 HBM manufacturers, we don't see anything more like this. It's part of ramping up and learning some of the subtleties of delivering into a specific application with each customer. But now that we're at volume with all 3 of them, I don't expect -- and have resolved the specific issue associated with that one design, I don't expect anything else like this to come along.

And our next question comes from the line of Charles Shi from Needham & Company.

I'm actually viewing the ramp-up cost issue that happened in Q2 with another HBM customer, probably I see that as a good problem to have. To me, it feels like you guys are more deeply involved in that particular customer's ramp. So if I go back maybe like 1 year- ish or maybe a little bit longer, I thought that you guys initially had pretty low market share assumptions with that particular customer. But how have things been evolved? And let's say compared with your -- the HBM leader, the market share you have there, how is your market share at this current -- this particular customer being -- I mean, is it actually getting closer to the market share you have at the leader? Or any way you can quantify that would be great for us to think about your HBM ramp going into the future.

Got it. Well, I appreciate the glass is half full approach to the ramp-up cost issue. Obviously, we aspire to execute more cleanly than that, both for our financial results and in delivering to our customers. But nonetheless, it is indication of strong participation in a second customer's HBM4 ramps. I'd say we've got a lot of share opportunity there, right? If I contrast it to our #1 customer, where we have a very strong share position, based on a strong historical partnership with that customer that goes back essentially decades as well as some great collaboration on aligning our technology road maps, that's a very strong share position. And as I've said in the past, I don't expect to duplicate that share position at the other 2 DRAM manufacturers. Having said that, if you want to view an entitlement share position as maybe a 50-50 split, we're a long way from there. So that opportunity is in front of us. We definitely need to improve our execution, as we've shown. And we need to increase our capacity, which is part of the optimization we're doing in our existing footprint, but is also longer term behind us deploying our balance sheet to buy the Farmers Branch facility and begin to ramp it up.

Maybe another question. I think it wasn't really being addressed anymore. Your microprocessor company, the customer, microprocessor customer, they are going through quite a lot of restructuring recently, restructuring, cost cutting, probably a lot of rethink of how they're going to do manufacturing, how they're going to do product development, et cetera. I don't think you actually mentioned about the revenue trend for that particular customer. It was nice to see that customer's contribution actually went up in Q2. But what's the thought going forward from here and especially the implication from the restructuring that is currently underway? What would be the implication based on what you see today?

Yes. Well, obviously, there is a lot going on at that customer. If we look at what we're executing to now and the dialogue between the company, there's been no significant change. But obviously, you see some revenue volatility in our results with that customer. And I don't think that's a reflection of any kind of share volatility. I think that's a reflection of changes in programs, changes in priorities and changes in demand. Longer term, if I back out a little bit, this is why it's so important for us to continue to make progress around our strategy of being a supplier to all the leading customers in the industry. They're all going to go through headwinds and tailwinds and ups and downs. And we want to make sure that we're a supplier to all the major customers in the industry so that we're somewhat insulated or at least buffered from these ups and downs of an individual customer. It's why it's so important for us to be now shipping in volume to all 3 major HBM providers. It's also why it's important that we continue to make progress, and we have made progress in the second quarter on qualifying and shipping in volume with the large fabless microprocessor company. So different elements, no big change on the ground with that customer, but certainly, many different scenarios, which could be either positive or negative that we're working hard to insulate ourselves by making sure we're exposed to the alternative demand streams.

And our next question comes from the line of David Duley from Steelhead Securities.

I was wondering, first off, if you could just talk a little bit about what you might expect from the hyperscaler customers or the GPU customers in the second half of the calendar year. You talked about having some, I think, hyperscale revenue in the quarter that just ended. I'm wondering if we're going to start to see the GPU guys use advanced probe cards and what the qualifications are, scheduling are around that. And are you seeing -- also the other area of excitement for some companies is the networking segment. I'm just wondering what you're seeing in these AI markets in the third and fourth quarter.

Yes. So if we look at -- let me start with the hyperscalers and then we'll move to the GPU piece. I think we've got pretty good exposure overall to networking. Although for us, it's not the biggest business. Some of these higher volume, high test intensity markets like the GPUs and the custom ASICs really are where we see the biggest probe card opportunity. So having said that, we did have a nice contribution, a multimillion-dollar contribution in the second quarter from one of the hyperscalers and custom ASICs, where we delivered one of our advanced probe card technologies to test one of their custom ASICs. We continue to focus on what really is a new segment and application space for the industry, right? A lot of these hyperscalers are launching their own chip design and chip test teams, and that's a new activity where us and other members of the test ecosystem are focusing a lot of our application engineering and R&D resources. So seeing some nice progress there. On the GPU front, we are continuing to make progress at qualifying for GPU testing, primarily through the foundries as the foundries are responsible for both the wafer sort and then the eventual advanced packaging associated with this. There has been adoption by the large GPU manufacturer of advanced probe cards. Unfortunately, that went to our competitor first. And I think if you follow them, you understand that. But we're rapidly catching up and qualifying and now moving into a phase of volume pilot production, where we expect to generate revenue around that application from that customer in the second half.

Okay. And then I'm a little bit curious as to, with a flat outlook in Q3 to the Q2 revenue, why you would be bringing on additional capacity in Texas. I understand it's lower cost, but maybe help us understand what your capacity utilization rate is at your current facilities and what the thinking is of buying another facility with -- obviously, I think people expect more growth from you guys, but is there a reason you bought Texas now? Or what's the rationale?

Yes. I got to paint a longer-term picture for you because I agree, right, if you could bring capacity online instantaneously, you wouldn't really do it now with a flat outlook. But if I look at our prospects longer term -- and let me paint a much longer-term or multiyear opportunity for you. If you look at the overall industry, I think most consensus forecasts have the semiconductor industry growing to $1 trillion sometime early in the next decade. So 5-ish years out. If we look at our markets, you've seen them grow faster than the industry. Test intensity is going up. And HBM is a great example where HBM test intensity -- basically 1% of HBM revenues are spent back on probe cards. That's a significant step-up from the industry average. And so if we put that math together, with even some modest share gain aspirations for us, it's not too hard to see doubling the size of FormFactor. And we need to have capacity in place. It takes a while to do these things. The purchase of Farmers Branch was motivated by a couple of things, certainly, that long-term opportunity, but also it's kind of a perfect fit for us. And there were a handful of facilities like this. And by handful, I mean, 2 in the U.S. that were the right size, equipped to -- with the right facilities, the right clean room capabilities. So this was really an opportunistic buy for us to get us the optionality and future capacity in place that we're going to need arguably 2026, 2027, but you've got to start that now if you're going to be executing and qualified with that facility with our key customers over that time frame.

Okay. And I'm going to take one other shot at the second half revenue kind of levels. I know you don't want to guide to Q4, but is there any reason to expect that the current HBM ramp that we're seeing at the 2 major customers who are part of your customer mix would slow down for any reason in the near term given what we've heard from their big customers and the hyperscalers? I mean it just seems -- so anyway, I'd love your commentary there.

That's right. And again, I want to reemphasize that we are shipping in volume to all 3 -- shipping HBM3 -- HBM probe cards to all 3 major DRAM manufacturers. So we're beginning to diversify that business. But we also took pains to point out that there is going to be some quarter-to-quarter volatility. The direction is clear, right, the investments by the hyperscalers, the increased DRAM bit intensity associated with HBM coupling to the GPUs. It's clear that, that's a strong growth market, and we expect that business to grow. But even as you've seen through the HBM3 to 3 now into 4 trajectory, where our revenue was dominated by a single customer, the leader, there's some volatility in there. So no question, the long-term trend line is up. That doesn't mean every quarter is going to be sequentially up. I'm hoping and expecting that as we bring all 3 HBM customers on and their share balances a little bit, that, that will diversify the business some and smooth things out. But the secular growth in HBM is something we completely agree with. It's part of the reason why we're making some of these investments in capacity like Farmers Branch and improving our overall outputs and yields.

[Operator Instructions] Our next question comes from the line of Christian Schwab from Craig-Hallum Capital.

Most of my questions have been asked, but just a follow-up on the hyperscaler GPU, custom ASIC business. It's good to do a few million dollars in the most recent quarter, but as kind of an initiative to help grow foundry and logic and improve gross margins in aggregate for the company -- I mean, how long of a time frame do you expect that to be more of a significant contributor and scale number?

Yes. I think it's a great question and one that we're debating internally because there's clearly some different crosscurrents in what GPUs are going to be used for, what merchant GPUs are going to be used for, what custom ASICs are going to be used for. But I think as we look into the latter part of 2025 and into 2026, we're working hard to make sure we're exposed to both of those trends. I updated you on the GPU -- gave you some updates on the hyperscaler piece. To a large extent, our business in the hyperscaler custom ASIC is going to depend on the success of their product road maps. And I think a lot of that is still a to-be-determined state. Having said that, it is a great opportunity to potentially diversify the foundry and logic business further and move its demand drivers away from things like PC and mobile, which seem to be stuck in pretty unexciting growth times.

And our next question comes from the line of Gus Richard from Northland Capital Markets.

Just on the tariffs, 1.5% is a reasonably good-sized impact. And I'm just wondering, first of all, is it your imported raw materials? And then if that's the case, what can you guys do to diversify your supply chain to minimize the impact?

Yes, you're right, Gus. This is a significant impact on our gross margins. It's because we are importing goods into the U.S. So what we do in response is we are working closely with our vendors on different scenarios. It's really on a case-by-case basis. But as we said in the previous earnings call, we are taking kind of wait-and-see approach as we evaluate various tariff scenarios before committing to any significant change to our manufacturing footprint and our supply chain. And we're also looking into maybe getting drawbacks options. But we need to keep in mind the semiconductor supply chain is complex, and we are developing our supply chain to mitigate the impacts of the tariffs as much as possible while acknowledging the negative impact on the gross margin as a result of the tariffs under the different scenarios.

Got it. And then my follow-up is on silicon photonics. I think in the second half, we should be entering into pilot production, at least in some programs. And I was just wondering if you could give us a little bit of a status on where those ramps are and what those pilot programs are.

Yes. I touched on it in the prepared remarks. We now have multiple systems installed that are indeed involved in pilot production, primarily at the foundry for our large driver customer. The work now is really on making sure this technology is hardened and ready for high-volume production as we move into 2026. So we've got multiple tools running production levels they've never run at before, right, pilot production levels, though, with 1 or 2 designs. But really, a lot of the work and investment going on right now, both by us, our customers and our partners, is to make sure that we're all ready for what's anticipated to be a significant ramp as we move into 2026. As you probably know, CPO is one of the more compelling -- silicon photonics and co-packaged optics or CPO, one of the more compelling ways to reduce data center energy consumption, which I think in any reasonable scenario is going to start to turn out to be the limiter in AI growth. You just can't get the power for data centers to build out. And so I think solving -- helping solve this part of the equation of the actual energy consumption, power usage of the data center is a pretty exciting place for us to be.

Thank you. This does conclude the question-and-answer session of today's program. I'd like to hand the program back to Mike Slessor for any further remarks.

Yes. Thanks for joining us today. If you take a look on our Investor Relations website, we're doing several conferences in late August and the early part of September and would hope to see you there. Until then, take care.

Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.