My name is Amanda Scarnati. I'm one of the semiconductor analysts here at Citi. I'm joined by Mike Slessor, the CEO of FormFactor, and Shai Shahar, the CFO of FormFactor. We're gonna do a very casual fireside chat this afternoon. I will open it up for questions if there's any in the audience a little bit later on. Thank you, Mike. Let's just start and sort of level set where we sit today, and sort of what FormFactor is, you know, on the whole.
Yeah. Well, maybe a brief introduction. FormFactor, a little over three-quarters of a billion dollars of trailing 12-month revenue. Our businesses break into two major segments. One segment called a probe card, which is essentially a device-specific consumable used by key customers in the semiconductor industry to make sure each die on the wafer is good before they pass it downstream to the assembly and packaging processes. The unique part of this business is some very short lead times, well within a quarter, and a high degree of customization.
Again, each probe card is specific to a customer chip design, and so as they spin a mask set, change anything on that wafer drives new probe card demand for us, and is one of the key unique attributes of this business, a very short lead time with a high degree of customization. Probe cards are about 80% of our business. It's a segment where we have somewhere around thirty-five percent market share and do lead in market share. The second part of our business, about 20% of our business in what we call the system segment. These are tools not used for production, but instead used in very early development by our customers, same customers in the semiconductor and broader electronics industry, but used to characterize and improve yields on brand-new architectures and devices, both electrical and optical.
One of the neat trends we've seen in that business recently is a real uptick in the activity associated with silicon photonics pre-production, and that's been driving a lot of the growth in the systems segment. You know, if we look at the recent past, a very strong first half, near record levels of revenue, gross margin and profitability above our long-term target operating model. Our guidance for the current third quarter had about a 10% step down on the top line and more as we work through the gross margin line. I'm sure we'll talk about that in some degree of detail. Basically, a reflection of some of the real softness in both compute and mobile, that's, I think, pretty well understood across the industry now.
you know, we are continuing to invest in and deploy capital in long-term expansion. We're bullish about the long-term prospects for the industry overall and our business, but are obviously taking measures in the short term to do whatever we can to tighten our belt on the reduced levels of revenue.
Let's just kinda dive into the revenue a little bit, right? You mentioned some of the push outs and the step down in revenue. Is this really a comment on the delays in Intel moving to the next generation? What's your confidence level that these really are just push outs and not sort of delays even further out until sometime next year?
No, I think, again, it was pretty broad-based in nature. When we look at it in both compute and mobile, it was not related to any one customer. Obviously, we've got some significant customer concentration among the leaders of the industry. That's intentional and by design, right? We wanna be a supplier to the leaders in the industry, and by its very nature, that's gonna create a pretty concentrated customer profile. But, you know, as we saw, reductions in demand for certain designs, again, with a probe card being specific to each customer chip design, if a customer reduces its wafer starts associated with that design, they're gonna need less probe cards. So we did see some of that. We also saw some delays, again, at multiple customers of their planned releases of new designs.
Presumably, you know, to make sure that they're exhausting the inventory they have in the channel of their existing designs and are not, you know, putting something new behind it that then is gonna compromise their ability to sell through that inventory. You know, I think the vast majority of our customers continue to execute pretty well on their technology roadmaps. These are complicated technology transitions that are going on right now, whether it's the migration of the world's leading foundry to 3 nanometer, whether it's moving to a chiplet or tile architecture in the compute space. These are things that are really, I think, at the high level proceeding according to plan, but with some little bit of changes in timing.
Running a business that has lead times of, you know, 6, 7, 8 weeks, you're gonna see some of that volatility as they adapt their production scenarios to meet their end demand.
I think the biggest pushback I get from investors in terms of your share at certain customers is that there's potential for a greater risk of share loss at, specifically at Intel, right? Because you have the overwhelming share there. Technoprobe, your leading competitor out of Europe, has been doing a fantastic job of growing its business. How do you look at maintaining share, or even, you know, gaining share in the market going forward?
Yeah. I think part of this is natural because, you know, don't wanna be too casual about it, but all of these leading customers really need two suppliers. Except in areas like, you know, lithography or bright field inspection, there really are multiple alternatives that they can make work. You know, this is a situation at our largest customer where historically we had very high native share. It was, I think, reasonable to assume that a second supplier, in this case Technoprobe, a large and credible competitor in the foundry and logic space, was gonna be able to gain some share on us. I feel like that's largely over. I think things have stabilized in the long-term share positions that they're going to at that customer.
On the flip side, I think there's other areas where we do have the opportunity to gain some share on them in the foundry and logic space. You know, the other large fabless microprocessor manufacturer, a good example where at present we have fairly low native share, are working aggressively to try and get qualified. We have multiple units there working through the qualification process and are looking forward to growing share and revenue there. I think certainly our largest customer and some of the share loss that's happened there we view as largely in the rearview mirror. Again, I don't wanna be too casual about it, but it was sort of bound to happen in the end with a two-supplier initiative being executed too.
I think there are other areas in the industry where we have opportunity to gain share, just as we've done at the world's largest foundry.
speaking of, you know, the world's largest foundry, as they move to more outsourcing of probe cards, you've talked about sort of a 50-50 split between you and Technoprobe. Is that still what the expectation is, that you could gain 50% of that share as they continue to outsource? Are there puts and takes into how that share is divvied up?
Yeah. No, we're working hard to gain more than majority share there. There's a whole variety of product differentiation, of customer support, of delivery, which is very important in a short lead time business like this, where I think we have some opportunities to gain some share above the 50/50 sort of equal watermark. There are some opportunities there and at other places in the industry for us to gain share.
I think the big point of contention that I get a lot of pushback on is the margin.
Yeah.
The fluctuation in the margin.
Right.
You have sort of the long-term target, which you're trending towards, and you hit in many quarters, but then you can see an 8-point step-down in a given quarter.
Right.
Can we just talk about the margin volatility and if there's an opportunity to ease some of those big swings?
Yeah. I think our Q3 outlook, you know, we talked a little bit about the puts and takes on revenue, foundry and logic down, the systems segment actually being a little bit stronger and providing a partial offset for that. That I think largely, you know, a condition of the industry. To have on 10% sequential revenue reduction, to have an 800 basis point reduction in gross margin, I think legitimately people view as disappointing and we view as disappointing. Shai'll talk about some of the reasons for that, but this has been a an area that's reinvigorated a lot of our focus on gross margin improvement, whether it be operating sort of how we operate the factories, the inherent mix associated with memory versus foundry and logic versus systems. Look, I agree with the pushback.
I agree with the criticism. On 10% down in revenue, you shouldn't have 8% down in gross margin.
Right. Why it happens in Q3? We see two main reasons for that. We see revenue decreasing from over $200 million in the last couple of quarters to about $183 million at the midpoint of the outlook range. Since we've been adding capacity in the last couple of years and added to our fixed cost, this decrease in revenue has an impact on the gross margin. We are investing in adding capacity because our customers are doing that. They didn't slow down, and we need to be ready to capture our share, our fair share and more when they go online with the new fabs. That's why we're not slowing down on adding capacity, at least not in a significant way.
This step down in overall revenue is hurting our gross margin in Q3. The second part is that specifically the decrease in Q3 comes from foundry and logic. Foundry and logic, as a reminder of our margin profile, is the highest margin market within the probe card business. The systems has high 40s, low 50s% gross margin, followed by foundry and logic, then DRAM and then flash. Since most of the decrease in Q3 revenue comes from foundry and logic with the highest margin, again, it's not a surprise that we see a big decrease in the overall gross margin. How do we get back to our target model or trending back to our target model with less fluctuation?
We need revenue to grow again above $200 million, around $210 million, which is the run rate of our target model. And we need that growth to come from foundry and logic, which is the highest margin. If once we get to this high level of revenue, we're gonna see less fluctuations, but because of mix, we'll continue to see some fluctuations in the gross margin, because in any given quarter, and not only the mix of revenue between the markets we serve, but also even within foundry and logic or within DRAM, we have quite a wide range of products with different margins that can fluctuate from quarter to quarter. Along that trend line of going back to the target model, I do expect to continue to see some fluctuations in the gross margins.
Now even within that foundry and logic mix or in the memory mix, right, you talk about fluctuations in different products. Is it more sort of customer specific, where one customer might have a better margin profile than another? Or is it even within products sold to a specific customer that there's going to be fluctuations?
It's more of the latter. A great example would be if you look at one of our biggest DRAM customers from last year, the products he could have purchased from us: two types of products. One can be a one touchdown probe card, which means with one touchdown, it tests the entire wafer.
Or a two touchdown wafer probe card, which touches the wafer twice in order to test all of it. Obviously the one touchdown probe card is more expensive with higher gross margin for us. The customer, because of tester capacity, decided to ramp up actually a two touchdown design, which created an unfavorable mix for us in about half of the year of 2021. It's really design specific. That was a good example of how design specific can change, and not only even within the same customer.
We're seeing test intensities are rising across the board as we move to advanced packaging and chiplet designs and all of that. We had Teradyne earlier this morning talking about massive tailwinds in terms of their capabilities in tests and their opportunities. Can you talk a little bit about FormFactor's opportunity as test intensity increases?
Yeah. I think it is an interesting observation, right? That even over the years, customers are doing more and more wafer tests and spending more on wafer tests. For example, if you look at advanced probe card spending as a percentage of semiconductor revenue, it's gone from about 0.35% up to 0.42%-0.44% in recent years. That's a reflection of the value of screening out bad die before they go to these downstream assembly processes. That need becomes even more acute with some of these advanced packaging or chiplet or tile-based integration schemes.
Because if you spend a moment doing the composite yield math, if you're gonna break a die into 4 or 6 or 8 or even 16 pieces, and then reassemble it with relatively little rework and almost no redundancy available, you need very high certainty that each of those chiplets or tiles is good before you commit it to that stack, and that's driving up test intensity as well. Not only are we seeing this general increase in test intensity as customers try and shift more of their yield learning and testing towards the wafer fab, but we're also seeing this opportunity associated with what's in the very early innings of advanced packaging and die stacking, at least in the logic space.
Interestingly, they also talked about going from 92% yields to 97% yields in terms of improvements creates a higher cost dynamic. Do you see that with probe cards as well, where that squeezing out that extra 5% will drive greater ASPs for your probe cards, or is it not as critical in that regard?
Yeah, how you choose to exploit it, either ASPs or market share, but if you can deliver higher test yields to a customer, that's dollars that absolutely dwarf probe card costs and even ATE tester costs. Those are things. That's part of the reason why we spend 13%-15% of revenue on R&D to make a more capable probe card, one that operates at a higher frequency, one that operates with higher current densities, so that we can do a better job of delivering the capability, for example, from a Teradyne tester into the chip and having very high fidelity signals propagate back into the tester and delivering higher yields.
I think the big sort of thematic question that we've been getting from investors throughout the entire conference has obviously been on expectations for underlying market growth going into next year. Now, FormFactor sits in sort of a unique place. It's a consumable driven business, but it's not necessarily a unit driven business, and it's definitely not a CapEx driven business.
Right.
Can you just talk about how you look at sort of underlying market growth and how you're positioned, whether we have, you know, a down 5% unit driven business next year or, you know, what happens in the market environment?
Yeah. Well, I think we've gotten a lot of the same questions in our meetings today and in some of the previous conferences. I think, you know, the question of is this how long a downturn, how deep a downturn is this gonna be, an open question for people in the industry, an open question for people covering the industry. As you know, things with long lead times, like front-end wafer fab equipment, not seeing much deterioration there because customers continue to be pretty bullish about the long-term prospects for the industry, and you've got, you know, it takes you two years to get a scanner, so you don't wanna step out of that line.
If we think about some of the variables at play in the shorter term, I do think this balance between, inflation, interest rates, what the Fed's doing, it has to have had an impact on consumer spending, which we saw very directly in terms of client PC, mobile handset growth, some things like that. I think the largest variable from our perspective is that one, the macroeconomic conditions and then how does the consumer play. You know, in managing around that, we're managing it in a couple of different ways. You know, as Shai said, in the longer term, we're continuing to deploy capital in making sure that we're going to have capacity in place, when the industry growth resumes, whether that's mid 2023, early 2023, late 2023, I don't know.
We want to be able to make sure we've got the facilities and tools in place to be able to meet that demand with the technology we're developing. At the same time, you know, you're doing whatever you can, as I said in the opening remarks, to manage through the short term, whether it's our incentive-based structure, which is essentially a profit-sharing program that helps control both operating expenses and COGS somewhat. You know, making sure that we're being disciplined on, you know, headcount adds and things like that, especially where it relates to capacity. You know, you're both managing for the short-term pressure and the long-term bullish outlook, understanding that there's some uncertainty because of things like macroeconomic factors that are gonna make that timing a little bit unpredictable.
I'm gonna pause here and see if there's any questions in the room. Moving on to the DRAM business, right? That business has been fairly stable, about $35 million a quarter, across the cycle. Can you talk about what drives the stability in that business, what your expectations are for continued growth there?
Yeah. Well, it's a business that, if I back up 10 years when I joined FormFactor, it was not very stable at all back then, and is what drove a lot of our diversification strategy to drive growth in the foundry and logic space, to get into RF, to get into the systems business, and we continue to drive a broader diversification strategy. As you note, over a surprisingly long timeframe, we've seen the DRAM business kind of in between $30 million and $40 million, on a quarterly run rate basis.
What's driving that is the same thing that drives, you know, all probe card demand, and it's our customers investing in and ramping new designs, new chip designs, new nodes, move to DDR5 from DDR4, a mix of both low power and server, or low power and PC DRAM. You're seeing a lot of different things at work in DRAM, where our customers, even though, you know, the DRAM end markets are pretty weak right now, they're still continuing to invest in their roadmaps, releasing new designs, buying new probe cards, and ramping those designs, even as potentially they pull back on, overall capital spend, overall WSE. The last time this happened was actually in 2018 and 2019.
It's a really nice clean case study in how our business can be different from both the end market and wafer fab equipment spend in the segment. You know, 2018 was a very strong year, at least sequentially from 2017, for both bit growth for our customers and for WSE. It wasn't a great year, interestingly enough, for probe card spend. You go to 2019, where I think bits were probably flat, revenue was down and WSE fell off a cliff. It was the strongest year of DRAM probe card spending and a very strong year for FormFactor, and I think we may see the same dynamics at work here.
Obviously, there's a degree of magnitude associated with this, where if things get really ugly in DRAM, then customers are gonna pull back on new design investment, but we're not seeing any indication of that right now.
The other sort of interesting question I get a lot from investors is the duration of a probe card.
Okay.
You know, is it volume driven, where you're, you know, constantly changing out the probe cards? Do they last sort of indefinitely? How should we look at volume versus pricing?
Yeah. Probe cards do actually wear out, so you're making physical contact with the wafer, with the chips on the wafer, and that process and the reconditioning and cleaning of the probe card to make sure it continues to make good electrical contact actually physically wears out the probes on the card. When I first got into this business 59 years ago, it turned out that was really almost the only replacement cycle. Today, what drives a lot of the replacement cycle is actually this design-specific nature I talked about. When a customer changes a chip design, that essentially obsoletes the previous probe cards, even if it's something as subtle as changing the die size or moving some of the power bumps around for better performance.
It still is the case in longer product life cycle markets like automotive, where you have products running for multiple years, where there is a wear out and replacement cycle on an individual chip design. You know, for customers in automotive, we do go through a replacement cycle roughly once a year, right? A probe card you can think of as lasting roughly for about a year. Obviously, automotive product life cycles are much longer than yours. You do see this replacement of the same thing, of the same SKU cycle happen there.
Where you no longer see it happening is in places like mobile, where, you know, many customers are on annual product release cadences, and so if a probe card lasts for a year, you know, if they're going to a new version of their application processor, a new version of their modem chip, within a year's time cycle, that probe card probably won't wear out. It'll instead be rendered obsolete by the design change to the next design.
Great. How should we think about capital allocation going forward?
Yeah. In the last couple of years, our first priority was adding capacity to the business. We've been investing around $100 million in adding capacity mostly in our Livermore campus, and that has been the priority. Followed by continuing to invest in R&D. We are the leading provider of probe cards. In order to maintain our technological leadership, we need to continue to invest in R&D, so we've been doing that. We have a buyback plan in place. We had $50 million that we exhausted earlier this year, and our board approved another $75 million of buyback to offset dilution from stock-based compensation. As of the end of Q2, we still had about $45 million left to execute on that plan.
M&A continues to be a big and important part of our strategy. We're looking into technology tuck-ins, smaller acquisitions like the three latest ones we did in the last three years, from FRT to HPD and the recent dilution refrigerator small acquisition we did in June. We're also looking at a bigger transaction. It's something like we did with Cascade Microtech back in 2016. Something more transformative is probably gonna be an adjacent market to expand our SAM, and it is a big part of our strategy.
I guess the last question I have is on the systems business and, you know, the opportunities to drive continued growth via that business over the long term.
Yeah.
They're both organic opportunities and M&A opportunities. You know, if we stick on the organic side, you know, we acquired this business as part of our acquisition of Cascade Microtech back in 2016. This got us into the systems business and engaging with our customers in their early R&D activities. That's a business that we've been able to continue to grow, you know, at sort of mid-single digits organically. One of the nice trends we're seeing at present in that business, as I said in my opening remarks, is silicon photonics.
We've seen a nice fusion of our ability to do both electrical and optical tests and integrate the instrumentation for both of those that's really resonated with a lot of the activity going on in silicon photonics, whether it be different sensors for things like LIDAR or data center communication chips, where you're really bringing together semiconductor and optical technologies in a hybrid way. In addition to that, the systems segment has been where we focused on a lot of the tuck-in acquisitions that we've done. Shai brought up FRT, an acquisition that we did a few years ago of a small German metrology and inspection company, again, focused on advanced packaging applications.
There's some niches there that aren't very well served by the big metrology and inspection companies, and we felt like there was an opportunity there to inject some good products and technology into the FormFactor relationships with the major customers in the world, and that's really that hypothesis is really borne out, and that business has grown nicely. If you look at the systems segment and how it's grown over the past couple of years, some of the growth from those acquisitions is pretty evident as we've gone from kind of $25 million a quarter up to flirting with $40 million a quarter, and so you can see we've grown some there. More recently, in the systems segment, we've been quite active in enabling test and measurement and now infrastructure for the nascent quantum computing industry.
Again, a couple of big existing FormFactor customers, having conversations with us on how to better view first test and measurement and then some of the infrastructure for quantum computing. You know, this is an area that's several years off but plays well to our strengths. You know, thermal management, in this case, at very, very low temperatures, close to absolute zero, where quantum computers have to operate, but still making electrical contact, connecting the chips to the outside world, plays pretty nicely to our strengths. Again, two of our recent acquisitions, HPD, and then the assets of Janis ULT, to help really get us the capabilities that we need to enable the quantum computing industry and enable the key customers that are gonna drive that industry forward.
You know, although at 20% of our business, it's not a huge contributor, it is some exciting future opportunities for us, especially in an area like quantum computing, where it's an industry that's just emerging, and you have the opportunity to really create and define the business model and the customer relationships. You know, in the third quarter, if we focus back to the here and now, the growth in the systems business did offer a bit of an offset to the downtick in foundry and logic. A good example of our overall diversification strategy at work, not just longer term with things like silicon photonics and quantum computing, but even in the very short term, as we see some of the volatility in our mainstream production semiconductor business.
Perfect. We're gonna wrap up a little bit early, and I appreciate your time. Safe flight back. I know you have to bolt out of here.
All right. Thank you for inviting us.
Thanks for having us.
Thank you.