FormFactor, Inc. (FORM) Earnings Call Transcript & Summary

Hi, everyone. I'm Madeleine Jenkins. I work on the tech hardware equity research team at UBS. And it's my pleasure to have Mike Slessor and Aric McKinnis from FormFactor. I think Mike will start with a brief introduction of the company, right, and then we'll head into Q&A.

Great. Well, thank you very much, Madeleine, to you and the UBS team for hosting us. We've had a great conference with a lot of meetings. So happy to finish up here. To get everybody up to speed on FormFactor and make sure we're all on the same page, we're a company in the semiconductor supply chain that provides test and measurement equipment. We operate in 2 segments: one called probe cards, which is about 80% of our $0.75 billion annual revenue, the other called engineering systems. Probe cards are used at the end of the wafer fab to make sure chips are good before they go to the downstream assembly processes. And we've seen some significant growth in that business overall. I'm sure we'll get into some of the details driving that. But as the industry has started to adopt advanced packaging techniques, things like chiplets, high-bandwidth memory stacking, TSMC's CoWoS, that's driven a rather dramatic increase in what we call test intensity, the use of wafer testing to make sure these chiplets are good before they go to the downstream assembly processes. We operate in a variety of subsegments inside probe cards, and we do lead in market share overall in advanced probe cards worldwide. The other 20% of our revenue, the Engineering Systems business, really focused on helping customers develop their next-generation technologies. And one of the real highlights of that business at present is enabling co-packaged optics, which are set to displace or at least partially displace the existing switches in data centers because of the much higher speed and lower power usage, lower energy usage associated with photonics. So that probably gets us mostly up to speed on form factor. I think to get everybody sort of sharing what the topics we've covered during the conference are, we've got a lot of exposure to high bandwidth memory, right? And HBM, obviously, a key enabler of all these generative AI investments and NVIDIA's GPU road map. So lots of interest in that, that I'm sure we'll cover as well as co-packaged optics and where that's going from a top line perspective. There's also been a lot of interest around our gross margins and profitability and the recovery back towards our target model, and Aric will talk about that.

Perfect. So maybe if we go segment by segment, starting with logic and foundry, what's your kind of current assessment of the demand in that area?

Well, inside probe cards, if you look at our financial statements, we report these 2 segments, Probe Cards and Engineering Systems. But inside probe cards, there are some subsegments or submarkets that we often give people, almost always give people visibility to. And our largest business is foundry and logic. This is probe cards for testing chips like microprocessors, application processors, GPUs, different XPUs, but all the way through RF subcomponents as well. And so it's a fairly broad-based business. And I think like most suppliers in the industry, our demand is being driven -- certainly, the growth in our demand is being driven almost exclusively by generative AI and the investments in high-performance compute in the data centers. That's also the case in memory, which we'll get to in a second. But if I look at foundry and logic, it's an area where our business classically was driven by mobile handsets and PCs. For many years, Intel was our largest customer. That's changed, although they're still a significant customer. We've seen our growth really be driven by, again, these investments in silicon-enabling generative AI. That's true in logic and foundry. It's also true in DRAM. We've got a variety of initiatives underway as we continue to advance our technology road map based on our advanced MEMS technology. We use a derivative of microelectronic mechanical system technology to build the probes on a probe card. And these are remarkably complex structures, right? We test multiple chips at once, each with 50,000 to 100,000 contacts, all about the size of a human hair and they have to do this reliably for millions and millions of contacts. So you can think about probe card as a fancy connector in between the test equipment built by Advantest and Teradyne and the wafer. But inside that connector, there's a great deal of complexity. And in foundry and logic, we're seeing that complexity in those customer challenges really accelerate, higher speeds, higher power, all the things you see on our customers' road map as they continue to progress down Moore's Law, but also take advantage of advanced packaging.

Perfect. And then maybe moving on to memory. You're saying that a lot of investor interest, I'm sure. How do you see that market next year? And what sort of revenues are you running out for HBM?

Well, memory splits into 2 subsegments. One, probe cards for testing DRAM devices, the other probe cards for testing NAND devices. We don't participate very much in NAND. So I'll focus most of my comments on DRAM. And DRAM has been a real growth driver for us, primarily because of high-bandwidth memory and HBM. As you'll notice in our financial filings, SK hynix has become our largest customer. And when we reported Q3 and gave Q4 guidance about 1 month, 1.5 months ago, we created expectation that we would again set another DRAM -- quarterly DRAM probe card revenue record. Some of that's driven by continued strength in HBM. We're in the midst of the HBM3 to HBM4 transition with our customers and have actually are shipping in the third quarter and fourth quarter more HBM4 probe cards than we are HBM3. So we've achieved that crossover at this point. But we're also seeing some strength in the non-HBM DRAM part of the market. Applications like DDR5 are responsible for the sequential growth in our DRAM probe card business, kind of an interesting dynamic. And I think for people who follow the DRAM industry, if you look at spot pricing and contract pricing associated with modules like DDR5, I think it's pretty easy to understand why our customers are starting more wafers on those. An interesting time where there's a set of capacity constraints across the industry and demand really starting to outstrip supply, not just with HBM, which I think is a well-understood dynamic, but now also with data center DRAM and data center DDR5 reaching some pretty critical shortage conditions. These are conditions we haven't seen for a long, long time in the industry.

Perfect. And then another kind of very exciting growth driver, co-packaged optics. Maybe could you just talk a bit about the demand that you're seeing there?

Yes. So we're in the very early innings of co-packaged optics or CPO. But this is an area where we've been collaborating with customers for the better part of a decade. It's primarily been in this engineering systems business, where we work with customers in their very early development and help them transition from what we call the lab to the fab. We now have multiple systems installed in fabs around the world doing pilot production for the initial volume CPO devices. And we think we're going to start to see at least medium volume production in 2026. These are really exciting growth area for us and a great example of how early engagement with our customers in their labs allows us to really understand these challenges and then drive them into the production arena where the volume scale and our revenue grows. And I think from an organic revenue growth perspective, other than the markets we participate in today, like foundry logic and DRAM, CPO is poised to be a really interesting long-term growth driver for us. And we continue to invest to keep the competitive leadership position that we've cultivated over the last decade or so.

Perfect. And then maybe moving to gross margins. I think you did 38.5% in Q2. How do you see that progressing to your target model? And are there any impacts from tariffs?

Yes, great question. Gross margins is one of our key priorities that we're really focused on. And for those that maybe don't have as much context, our target model, which we established in 2020, is 47% gross margins at around $850 million of revenue. We -- if you look at our trajectory over the past couple of years, we have not been operating at those levels and hence, the focus. And so what are we changing here? There's a couple of things that we're really focused on. And I guess I'd put it into 2 buckets: operational efficiency and really financial discipline. Neither of these are rocket science, if you will. It's focusing on the fundamental cost drivers in our business. And we started laying some important groundwork probably a year to 1.5 years ago and how we structured our global manufacturing organization. We are the product of multiple acquisitions over time. And our structure 1.5 years ago kind of reflected that where we had this more business unit-oriented structure. That led to generalists running our manufacturing organization, people that were not specialized necessarily in high-volume manufacturing. As we've grown, we've come to acknowledge that, that's probably not the best way to progress forward and therefore, made some important changes in that space created a core manufacturing function and have brought in some real experts from outside the company who are very experienced with running high-volume wafer manufacturing. And that has led us to start making progress in the important areas of yield and cycle time. And if you think about the effects of those improvements, there's a variety of important benefits to driving down cycle times and improving yields. First and obviously, you get more good product out for every unit that you put in, right? And so that results in a lower unit cost of your products and therefore, is supportive of higher gross margins. You also decrease, if you will, your cycle time. And so -- and you also -- and that enables you to be more responsive to your customers. And you just make more efficient use of your overall resources. So again, not a rocket science concept here, but really focusing on this, it's an area where maybe we lacked focus in the past. So in our Q3 earnings call, we talked about this path that we're on. We've continued to make progress in this area. And that's what's leading us from 38.5% gross margins to 41% in Q3 and the guide that we provided for Q4 at 42%. We expect to continue to progress on that road map, incrementally improving gross margins over the course of 2026, ultimately taking us up to our target model profit levels, barring tariffs, which didn't exist back in 2020. Unfortunately, they're a reality in our space today.

Perfect. And you mentioned the kind of new Texas fab. I just wondered what's the timing of that? What CapEx do you expect to spend? And then what revenue do you expect to generate from the fab?

Yes. So our Texas fab is a pretty exciting opportunity for us to, one, create potential for business expansion, additional capacity, but also, again, returning to this gross margin improvement initiative. It's a move into a lower-cost region for us. The time line for this is basically over the course of 2026 and 2027. We will be building out and ramping that site. And that will involve investment over the course of 2026 of around $140 million to $170 million. We expect that once that site is up in full production after the end of 2027, that it will also be accretive to gross margin. Important to note here that we aren't depending on that improvement, that incremental improvement from Farmers Branch for purposes of our current target model attainments that is really enabling our future road map beyond the current target model.

Perfect. And maybe if we can kind of go back to AI. So what traction are you currently seeing at the GPU manufacturers?

Yes, yes. It's -- the major GPU manufacturer is a significant customer for us, but not directly on the mainstream GPU. That's been an area where we've been working with them and their foundry to qualify over the past several years, and have made excellent progress and expect to be able to compete for that business as we go into 2026. As I said, it's part of a broader engagement with that customer. And given where the growth in the industry is coming from, a really key strategic initiative for us to round out the relationship in supplying that customer. Key part of the supply chain, both for them, both directly and indirectly through HBM, obviously, but a piece that we're probably not quite as focused as the gross margin improvement, but pretty close.

Perfect. And then maybe also on the custom ASIC side, maybe talk a bit more about that opportunity in the next few years?

Yes. Custom ASICs are interesting in that we provided an update for people, I think it was in the second quarter, where we won a multimillion dollar opportunity and seen some revenue growth associated with that. But by and large, custom ASICs have not required advanced MEMS probe card technologies to date. And if you back up a few years, that was true of GPUs as well at lower densities, lower power, lower speeds, you can use a legacy probe card technology, don't need the advanced MEMS probe card technology. And our customers obviously are going to go with the lowest cost -- lowest overall cost solution. As ASICs advance and start to approach where GPUs are today, we're pretty confident they're going to need these advanced MEMS probe technologies, and we're engaged with all the major ASIC vendors to make sure that we're ready in there in the same way we're now on the cusp of driving some GPU business now that it needs our advanced MEMS technology. It also goes to the element of why we're making such a large investment in Texas. When you look at all of these fundamental growth drivers and how we have the company positioned we're running pretty close to capacity right now. Our Q4 guide was for $210 million. That would be an all-time record for FormFactor. And given all the elements I've talked about, HBM, GPUs, custom ASICs, we believe there's a lot of future growth potential in front of us, but we need the capacity in place to be able to serve it.

And maybe then on -- also on the smartphone and PC kind of end markets, how do you see those into next year?

Yes. So it's kind of the tale of 2 markets, right? I talked about the growth being really driven by generative AI. And historically, our business was driven by mobile handset growth and PC growth. That's really changed. Although mobile handsets, all the application processors have an annual refresh cycle. So we do get new business there. One of the interesting things about the probe card business is there's no reuse of probe cards designed to design. They're essentially consumable when consumable driven by the obsolescence of old designs and the ramping of new designs. And so you've seen our foundry and logic revenue certainly move in lockstep with some of these annual refresh cadences on the silicon associated with big mobile handset launches like the iPhone. But from a secular growth perspective, it's been pretty flat, right, tracking again the overall handset market, which has also been pretty flat. Similar story in PCs. For a while, we were holding out hope for a major PC refresh because high unit volumes associated with client PC drives a lot of probe card spending. But it does seem like PCs are lasting longer and longer, and there's no compelling application driving us to buy new PCs. So consequently, part of why we've shifted a lot of our R&D focus and a lot of our customer focus to make sure we've got great exposure in that overall high-performance compute generative AI supply chain.

And I know you've kind of touched on this already, but the competitive landscape, kind of how do you see that? And what are the areas would you say you're kind of stronger than your competitors?

Yes. It's an interesting competitive landscape. In each of the segments we talked about, we have one primary competitor. And what we've tried to do as part of our strategy is create scale and leverage across the combination of these served segments. In particular, we use the same basic MEMS fabrication technology to build the probes that go into our probe cards for both foundry and logic and for DRAM. And that's one of the core enabling technologies of the performance we're able to deliver. There's some other -- so we have a competitor in foundry and logic, an Italian company called Technoprobe. We have a primary competitor in DRAM, a Japanese company called Micronics Japan or MJC. In the systems business, we have a Taiwanese competitor called MPI. But by and large, each of these served segments are 2 supplier markets, right, where the competitive dynamic continues to be pretty rational. We all understand that we need to make large investments, drive our R&D road maps forward to be competitive and make the capacity investments we talked about. But we also understand that none of us is going to run away with the market. Our customers need 2 suppliers for business continuity and balance. And I think it's led as in many other subsegments of the semiconductor supply chain like inside WFE -- it's led to a pretty balanced and healthy competitive situation.

And can you kind of give us a rough sense of your market share at the 3 DRAM for HBM specifically?

Yes. Well, DRAM for HBM is a bit of an estimate at this point. As I said, and this is all in our public filings, SK hynix our largest customer. And I think most people understand that they have very strong market share as our customer in HBM bits produce. And as a consequence, that's a sweet spot for us and is driving a very strong market share position for us in HBM. Given where we focused on things like high-speed test, where we have a strong competitive advantage and customers need that performance, I think HBM4 it's been well publicized is a big step-up in speed from HBM3 and HBM3E. So we've invested a lot in that high-speed capability and it really differentiates us for HBM, which has driven a very strong market share position in high-bandwidth memory probe codes.

Perfect. And maybe a final question on HBM. Obviously, there's a lot of chatter about potentially kind of changing the packaging going forward. I know it's probably quite a few years out. But does that change kind of how you see that market at all?

I don't think it changes how we view the market. It's one of the reasons why we do spend 13%, 14%, 15% of revenue on R&D because -- these -- the change to a 1-year road map cadence in high-performance compute has really accelerated the need to partner with customers and other suppliers and innovate. And so changes in packaging technology, maybe HBM goes from solder micro bumps to hybrid copper bonding. That's something where we're developing copper probing technologies so that we're going to be able to effectively test those devices. But just like the co-packaged optics story, those are things we're co-developing now with customers to be ready when and if those insertions happen. And I think that's one of the things that really characterizes our engagement model with customers is a very, very close coupling of our engineering team with their engineering teams to make sure we've got solutions ready when these big road map changes occur.

Perfect. And maybe a more general question. Are you considering M&A at all?

Yes. As Aric said, we're really a product of M&A. And I'll let Aric talk about capital allocation in a minute. But FormFactor is built on a set of acquisitions. We continue to believe that M&A thoughtfully and carefully done to expand our served markets is a great way to deliver growth and shareholder value. But if you look at the things we've done recently, they've basically been a set of small technology tuck-ins. In the systems business, we've acquired a footprint in test for quantum computers. That's a market that certainly is nowhere near commercialization today. Well, depending on who you talk to, maybe it's closer than some other people think. But compared to the mainstream semiconductor business, it's still in an embryonic phase. I think that -- and I continue to think I've been of this opinion for many years, that back-end suppliers like us need to consolidate. If you look at what happened in the front end to create the WFE suppliers of scale, that was largely driven by M&A and has created a much healthier ecosystem for front-end WFE suppliers. If you look at the back end, our customers are now relying on us to drive this advanced packaging road map forward. whether it's things like HBM, TSMC CoWoS, the bar is being raised on our ability to innovate, our ability to produce at scale. And I do think the collection of companies nominally our size, sort of between $0.5 billion and $1 billion in annual revenue could benefit from some consolidation and scale.

Interesting. And Aric, would you like to touch on...

Yes, regarding our deployment of capital, I would say over the next year to 2, we're going to be clearly focused on our expansion that we're doing in Texas. That's going to be where we're really focused. But we will continue to be opportunistic as it relates to technology tuck-ins and potential opportunities like that. And it's one of the benefits of having a strong balance sheet like we have, the ability to be opportunistic, see those opportunities and take advantage of them.

Perfect. And maybe for the last couple of minutes, kind of what's your vision for FormFactor over the next kind of 3 to 5 years? How big can it be?

Right. Well, as I said, it's founded on these market share leadership positions in our served markets. And we think if you look at starting at the highest level, there's a consensus view that the semiconductor industry is going to reach $1 trillion sometime either late this decade or early next decade. And I think when that view was put forward a few years ago, there were a lot of skeptics. But given the growth we've seen, given the silicon content associated with high-performance compute and the growth that's driving, I think most people who follow the industry and are in the industry feel like we're a lot closer to that than we might have been 4, 5 years ago. So if you look at a $1 trillion semiconductor industry, and what's spent in our served markets, our served markets are actually growing faster than the overall semiconductor industry, partially because of this increase in test intensity associated with applications like CPO, advanced packaging, HBM. We've seen a real increase in what we call test intensity, which means our markets are growing faster than the semiconductor market. If you then peel inside of that, we're working very hard, although we do lead in overall market share to gain share. If you look at our served markets, we're in round numbers, somewhere around 30% market share across all served markets. So there's clearly pockets of opportunity to gain share in there as well. You put all of that together, and it's not too hard to at least do the math to see the vision of FormFactor doubling from here. Think of it as $1.5 billion in revenue, given all those growth drivers. But it's one of the reasons why we're making big investments in expanding our capacity. We've got great products, great positions, just look at our HBM leadership position and very strong customer relationships. Those are things that we're working hard to basically continue to grow and strengthen while expanding our ability to manufacture at scale by adding a Texas facility to our current revenue footprint.

Perfect. And if you...

The only thing I would add to that and do it more profitably.

Hence, the focus on gross margins, right?

Perfect. And on that note, I guess, unless there's anything either of you want to add, maybe we can kind of wrap up a few minutes early.

No, I think we covered it. Thanks, everybody, for attending, and thanks again to UBS for hosting us.

Of course. Thank you, Mike and Aric.