FormFactor Inc (FORM) 2020 Q4 法說會逐字稿

Thank you, and welcome, everyone, to FormFactor's Fourth Quarter 2020 Earnings Conference Call. On today's call are Chief Executive Officer, Mike Slessor; and Chief Financial Officer, Shai Shahar. Before we begin, Jason Cohen, the company's General Counsel, 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 conditions, the benefits of acquisitions and investments in capacity and information technology, the impacts of the COVID-19 pandemic, the impact of regulatory changes, the anticipated demand for products, our future ability to produce products, the development of future products and technologies and the assumptions upon which 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 and 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, February 3, 2021, and we assume no obligation to update them. Also as an aside, since this is an entirely remote earnings call for us, please bear with us on any audio delays. With that, we will now turn the call over to FormFactor's CEO, Mike Slessor.

Thanks, Jason, and thank you, everyone, for joining us today. FormFactor delivered strong results in the fourth quarter of 2020, setting all-time quarterly and annual company records for revenue, non-GAAP operating profit and net profit. I want to thank and congratulate the global FormFactor team who demonstrated a combination of agility and tenacity to overcome a myriad of challenges and deliver these record results.

As we begin 2021, robust demand for our products continues, and we're adding capacity to meet that demand. However, even with the strength that produced these results, the fourth quarter had its unique challenges as evident from gross margins below our October outlook range. Shai will provide more details later, but I'd like to start our call by addressing this issue.

This gross margin reduction was caused by 2 primary factors of approximately equal impact. First, in our probe card segment, we reserved warranty costs for some early life reliability failures on a new foundry and logic product release. Since mid-November, our team has worked closely with the affected customer to resolve the issue, and this customer has qualified a modified architecture that eliminates the problem. This issue is now closed as we are shipping replacement units in volume and have implemented improvement actions in our product development and release processes.

Second, in our systems segment, we executed against an unusually unfavorable product mix as we shipped several significant lower margin orders to end the quarter. These orders contain fewer high-value features and options than in our typical mix as customers configured systems to meet a variety of end of year budget, lead time and volume purchase requirements. Some unfavorable product mix continues in our systems segment backlog, but less than in the fourth quarter. As we've described in the past, the diversified nature of Formfactor's design specific consumable and R&D-driven product mix will produce quarterly fluctuations in gross margin. Despite these short term fluctuations, our long-term trajectory towards the 47% gross margins of our target model remains intact. And as you can see from our outlook, we expect a sequential increase in gross margin for the current quarter.

Turning to market level details. Demand for foundry and logic probe cards was extremely strong in the fourth quarter. The strength is the result of new design releases in the fabless foundry ecosystem that ramped aggressively in both high-performance compute and mobile applications with continued strong growth associated with 5G handset launches. In the first quarter, these specific ramps are mostly behind us, and we do expect a slight sequential decrease in foundry and logic probe cards. On a longer-term basis, driven by 5G in advanced packaging, this market is a major growth opportunity as customers utilize FormFactor's differentiated market-leading products to meet their highly complex test requirements for millimeter wave RF front ends, next-generation application processors and high-power compute processors. In DRAM, we delivered the highest quarterly revenue of 2020 as customers released and ramp new designs in a combination of technology node migrations, 12 and 16 gigabit product ramps and the beginning of the HBM2 to HBM3 transition.

As a reminder, probe cards are consumable with a specific each chip design. And so demand is generated not just by node migrations, but also the release of new chip designs, such as 16 gigabit LPDDR5. With this strong new design activity in place across our customer base, we expect first quarter and potentially first half DRAM probe card demand to continue at levels comparable to the fourth quarter. The HBM3 transition is especially exciting for FormFactor. That's yet another example of industry adoption of advanced packaging, with customers stacking up to 16 individual silicon die to meet memory bandwidth and power requirements in high-performance compute applications. In the stack die architectures, the value of testing increases as more die are added to the stack to avoid incorporating bad component dye into an otherwise good stack. As a result, as advanced packaging schemes like HBM3 are adopted, we are seeing a substantial growth in test intensity. On top of that, the technologies required to test these devices are extremely complex and sophisticated.

The dimensions of an individual MEMS fabricated probe are comparable to a human hair, the carry power at over an AMP of current and signals at tens of gigahertz while lasting millions of contact cycles. Moreover, a typical probe card contains tens of thousands, if not hundreds of thousands of precisely assembled MEMS. This combination of increased test intensity, which expands the number of probe cards required per wafer out and test complexity, which widens FormFactor's competitive advantage is characteristic of all types of advanced packaging. As the industry's focus moves to post fab integration to offset the slowing of front-end Moore's Law, wafer test and probe cards are taking a prominent role in enabling a variety of advanced packaging schemes, like HBM stacking of DRAM Die and heterogeneous integration of chiplets and are at least partially responsible for the double-digit growth we've delivered in both 2019 and 2020. These secular growth trends support our investments in capacity with increased capital expenditures, including our 2020 purchase of a new 90,000 square foot factory in Livermore, which is scheduled to begin revenue shipments in the second half of this year.

Turning to M&A. Following 2 acquisitions in the second half of last year, our team is focused on integrating our additions. The Advantest probe card assets we acquired in July are now fully integrated into our probe card operations, and we have launched an internal program to incorporate several technology subcomponents into FormFactor's probe card road map. We are also making progress introducing the acquired mainstream NAND Flash probe card products into the global FormFactor channel and are working closely with multiple customers on first half 2021 qualification plans.

Our addition of HPD in the fourth quarter is also proceeding according to plan. We've integrated SG&A functions with the broader FormFactor organization. The combination of FormFactor's customer relationships and global footprint, together with HPD's world-class cryogenic thermal control and test expertise has enabled us to engage companies and research institutes leading in the nascent field of quantum computing in the U.S., Japan and the EU. Although we do not expect a significant financial contribution from these activities in 2021, we are excited with the long-term growth prospects, enabling quantum computing with our emerging leadership position in cryogenic test and measurement.

Finally, with record fourth quarter results and a solid first quarter outlook, we are making progress towards the target financial model we unveiled last year that delivers $2 of non-GAAP earnings per share on $850 million of revenue. Test and measurement is becoming a more important and strategic place in the semiconductor industry, driven by trends like 5G and advanced packaging. Our leadership position in these attractive markets, paired with our differentiated strategy and disciplined execution, will drive continued growth and share gains as we progress towards our target model. Shai, over to you.

Thank you, Mike, and good afternoon. As you saw in our press release and as Mike noted, we concluded the year with all-time record quarterly and annual revenue as well as non-GAAP operating profit and net profit, driven by continued strong demand in both our Probe Card and Systems segment. Fourth quarter revenues and EPS were above the high end of our outlook ranges, while gross margin was below the low end of our outlet range. FormFactor's fourth quarter revenues were $197 million, an 11% incretion increase from Q3. Quarterly revenues increased 10% year-over-year and contributed to total fiscal 2020 revenues of $694 million, an 18% increase compared to 2019. Probe Card segment revenues were $162.5 million in the fourth quarter, an increase of $12 million or 7% from Q3.

The increase was driven by higher foundry and logic and DRAM revenues partially offset by a decline in flash revenue. Systems segment revenues were $35 million in Q4, an increase of $7.5 million or 27% from the third quarter. Within the Probe Card segment, robust demand for foundry and larger continued, with revenues growing $14 million from Q3 to $123 million, comprising 62% of total company revenues in Q4, a slight increase compared to 61% in the third quarter. DRAM revenues were $35 million in Q4, an increase of $3 million in the fourth quarter and were 18% of total quarterly revenues, sales in the third quarter. As first communicated in the last earnings call, DRAM demand has returned to what we believe to be a more normalized quarterly run rate. Flash revenues of $5 million in Q4 were $6 million lower than in the third quarter and were 3% of total revenues in Q4, same as in Q3. As expected, Flash revenues continue to be lumpy from quarter-to-quarter. GAAP gross margin for the fourth quarter was $78 million, or 39.4% of revenues as compared to 43.1% in Q3.

Cost of revenues included $7.9 million of GAAP to non-GAAP reconciling items which we outlined in our press release issued today and in the reconciliation table available on the Investor Relations section of our website. The increase of $1.5 million in the non-GAAP reconciling items in Q4 as compared to Q3 is related to the acquisitions of Advantest probe card assets during the third quarter and HPD during the fourth quarter. On a non-GAAP basis, gross margin for the fourth quarter was $86 million or 43.4% of revenues, 330 basis points lower than the 46.7% non-GAAP gross margin in Q3 and 60 basis points below the low end of our outlook range, mainly due to the warranty costs and less favorable mix, as Mike mentioned. Our program segment gross margin was 43.9% in the fourth quarter, a decrease of 230 basis points compared to 46.2% in Q3, the warranty charge accounted for 110 basis points of the decrease in the consolidated gross margin.

And the remaining decrease is due to anticipated less favorable mix. Our Q4 system gross Systems segment gross margin was 41.3% as compared to 49.8% in the third quarter. The decrease of 8.5 percentage points was driven mainly by an unusually unfavorable product mix. This decrease in the systems gross margin accounted for 1.5% of the 3.3% overall decrease in gross margin. We expect the Systems segment gross margin to improve to mid-40s in the first quarter. And as we have previously said, we expect it to range between the high 40s to low 50s. As we make progress towards achieving our target financial model gross margin of 47%, margins will fluctuate from quarter-to-quarter.

Our GAAP operating expenses were $56.8 million for the fourth quarter $2.1 million higher than in the third quarter. Non-GAAP operating expenses for the fourth quarter were $48.1 million or 24.4% of revenues compared to $48.4 million or 27.2% of revenues in Q3. The decrease of $0.3 million was mainly due to onetime IT security remediation costs in the third quarter, partially offset by higher Q4 performance-based compensation and the addition of the business that we acquired in Q3 and Q4. Company noncash expenses for the fourth quarter included $7.7 million for amortization of intangible assets, $7.1 million for stock-based compensation and depreciation of $6.2 million.

Fourth quarter amortization of intangible assets was $1.2 million higher than in Q3 as a result of the acquisitions completed in the third and fourth quarters. Stock-based compensation was $1.5 million higher than in Q3 due to the timing of annual grants and the increase in depreciation of $1 million from the third quarter as a result of recent investments made in capacity expansions and acquisitions. GAAP net income for the fourth quarter was $19.3 million or $0.24 per fully diluted share compared to $22.9 million or $0.29 per fully diluted share in Q3.

The non-GAAP effective tax rate for the fourth quarter was 7.5%, down from 12.7% in Q3. This lower rate is a result of the relative increase in export revenues in our which benefit from the regulations regarding global intangible low taxes income, also known as GILTI, as we mentioned in the previous earnings call. The application of these new regulations resulted in a onetime cumulative benefit in 2020 which caused the full year effective tax rate to be 13.4%, slightly below the lower end of the 15% to 20% range we previously communicated. We continue to estimate that the annual non-GAAP effective tax rate for fiscal year 2021 will be 15% to 20%.

As a reminder, our cash tax rate is expected to remain at 6% to 8% of non-GAAP pretax income until we fully utilize our remaining U.S.-based R&D credits.

Fourth quarter non-GAAP net income was $35 million or $0.44 per fully diluted share compared to $31 million or $0.39 per fully diluted share in Q3.

In summary, EPS was slightly's above the high end of our outlook range due to the higher revenues and lower effective tax rate, partially offset by the impact of the lower gross margin.

Moving to the balance sheet and cash flows. We generated $31 million of free cash flow in the fourth quarter compared to $37 million in Q3, taking our total cash and investments to $259 million in the end of the quarter. The sequential decrease in free cash flow in the fourth quarter reflects an anticipated increase in capital expenditures. As of the end of the fourth quarter, we had 2 term loans remaining on our balance sheet, totaling $34.5 million. We invested $14 million in capital expenditures during the fourth quarter compared to $5 million in Q3, which brings total 2020 CapEx to $56 million as compared to $21 million in fiscal 2019. This increase is in line with our previous communications and chiefly reflects capacity expansion in the new building in our Livermore CapEx. We expect to continue the significant investment in capacity in 2021. And CapEx for the year is expected to be between $80 million to $100 million. As a reminder, we expect CapEx to return to 3.5% to 4% of revenues in our target financial model after we conclude these capacity expansions.

I'm also glad to report that during Q1, we went live with our ERP consolidation. These significant projects unify our ERP systems will improve our operational efficiency and help us achieve our target financial model. And the go-live is an important milestone. At quarter end, our total cash balance exceeded the debt balance by $223 million, an increase of $15 million from Q3 quarter end. The increase is mostly attributable to strong cash flow from operations, less cash used for the HPD acquisition and capital expenditures during the quarter.

Turning to 2021 first quarter non-GAAP outlook. As Mike mentioned, we expected generally strong demand for advanced probe cards with lower demand in foundry and logic urine systems, partially offset by an increase in Flash. These factors result in a Q1 revenue outlook in the range of $176 million to $188 million. Product mix is expected to be more favorable in Q1 and more -- and most of the unusual factors that impacted Q4 gross margins are not expected to occur in Q1, resulting in non-GAAP gross margin outlook for the first quarter in the range of 44% to 47%. At the midpoint of these ranges, we expect Q1 operating expenses to be comparable to Q4 due to the usual annual benefits reset and a higher R&D spend, offset by lower performance-based compensation.

Accordingly, non-GAAP earnings per fully diluted share for Q1 is expected to be between $0.34 and $0.42. Reconciliation of our GAAP to non-GAAP Q1 outlook is available on the Investor Relations section of our website and in our press release issued today. With that, let's open the call for questions. We ask that when you ask your question, please indicate if you direct your question to Michael, to me, since as you can imagine, we are still not in the same room. Operator?

(Operator Instructions) Our first question comes from the line of Brian Chin from Stifel.

Maybe just the first question. In terms of the gross margins in 4Q, were those lower gross margin products that you recognized and also expect to recognize a little bit in 1Q, were they probe systems? Because I know you obviously have a broader portfolio now of products in the system revenue component. So were those probe systems? And I guess, were they not in your prior forecast? And did they also represent a good chunk of the revenue upside that you reported in 4Q?

Brian, this is Shai. Thank you for the question. Yes, the majority of the lower margin impact came from the Systems segment. And some of these sales were not forecasted, but in a way that we always have a different mix of products, and this quarter, the mix was less durable in a very unusual way. As we said, we see less of these remaining in our backlog for Q1, and thus, we expect the gross margin to be better in Q1. But also I want to remind everybody that our margins will continue to fluctuate. We saw in Q4 of '17, a low system gross margin impacting us. And if you look at Q3 of 2020, that margin went up to 46.7%. And so this quarter, an unfavorable -- unusual, unfavorable mix. But we are confident that we are on our way to achieve a 47% gross margin of our target financial.

Got it. Before asking my follow-up, just the -- it may be exclusive or may not be mutually exclusive, but I also note China sequential revenue was up pretty high magnitude. Is there any sort of overlap there with sort of the systems? Or is that more tied into the probe card business?

Yes. Brian, it's Mike. I'll take that one. A good observation that China revenue was again up pretty significantly. The bulk of that continues to be consistent with the insight we provided last quarter. It's probe card shipments to multinationals who are directing us to ship into their assembly and test facilities in China. And so the bulk of the revenue uptick for China really has nothing to do with the Systems business. Having said that, there were some lower-margin configurations. In the probe systems business that did contribute -- they went into China, they did contribute to the shortfall in gross margins in that segment.

Okay. Maybe last question, a follow-up. To address your production capacity expansion. I guess, by 3Q end, maybe 4Q end, can you increment up your capacity in Livermore, say, 5%, 10%? Is that kind of a band that makes sense in terms of what you might be able to increment it up? And then also, maybe back to Shai, in terms of gross margins, can you quantify any drag you might expect that some of that new capacity does ramp up and how that might be allocated across the year?

Yes.

So, Brian, it's Mike. I'll start with the trajectory. As you've seen through 2020, even with all the different restrictions we've had manufacturing in California and Oregon, we managed to steadily increase capacity, and we expect to be able to do that in the front part of 2021 and in advance of bringing on the large capacity expansion associated with our new 90,000 square foot facility. Now that will ramp fairly gradually. As I said, we expect to make revenue shipments starting in the second half there. But I think it's reasonable to also assume that inside our existing footprint, we're getting more efficient and something like a high single-digit capacity increase percentage is probably not unreasonable from the existing footprint. Obviously, adding the new factory, gives us legs beyond that, but then take us to the $850 million level of our target model.

In terms of the impact on the gross margin, I think the best way to describe it is that we took all that into consideration when we put out our model. So the 47% gross margin at our target model and at $850 million of revenue, as Mike described, take that capacity expansion impact into consideration.

Got it. Kind of at the front end of that, you don't necessarily expect 50 basis point or some kind of drag in the second half of the year as some of that new capacity comes up?

Not necessarily. And we need to remember that mix still has the biggest impact on our gross margin. So we have other factors in addition to just go live. And as Mike said, we're going to do it gradually, turning on the tools and adding capacity on top of the building and the rest of the...

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

Charles Shi here. Mike and Shai, I have a question regarding your quarterly run rate here. Apparently, your fourth quarter run rate is almost -- if we annualize it, you are getting almost close to $800 million per year, which is probably less than 10% away from $850 million model. I just wonder, especially on the system revenue side, it looks like this is a record high for you. Are we looking at the new normal here? Or do we expect some of the modulation of the systems revenue going forward?

Charles, I'd expect to see modulation in all the different segments. If I back up and look at the constituent drivers of revenue for FormFactor, we've put together a pretty diversified portfolio of drivers, both through organic growth and the M&A we've done over the years. And in any given quarter, there's going to be different puts and takes associated with those.

In the fourth quarter, as you noted, we saw strength basically across the board. We had very high foundry and logic revenues driven by the fabless foundry ecosystem and by IDMs. We had the highest DRAM probe card revenue in 2020. And as you know, Systems revenue was up both organically and due to some of the acquisitions we've made over the past year or so in that segment.

So your observation associated with the $800 million run rate certainly is a reasonable one. Our outlook for Q1 comes down a little bit as some of those things come back to lower, perhaps more normal levels. But we're going to see fluctuations quarter-to-quarter. That's why we've tried to build a diversified set of revenue drivers so that we can manage those fluctuations inside each of our customers and inside each of our different segments.

Got it. Got it. So Mike, if I may, I would like to just dig a little bit more into your probe card side of the business. Definitely, you did tell us that you expect some modulation in the first quarter for the foundry logic side, given that the design release activity is going to probably slow down a little bit associated with the 5G handset. I just wonder because in the long-term model, you are sort of expecting 8% plus/minus kind of the advanced profile -- the probe card revenue growth. Given that last year, more probe card was having a double-digit growth, do you also expect that, that growth rate in 2021 to be slightly below your long-term growth rate? Or do you think 8% is still a good number for this year and going forward?

Well, the 8% was, as you noted, a longer-term growth rate. Obviously, the semiconductor industry, although collectively, we've been pretty fortunate over the past couple of years to have much less cyclicality. You're still going to have some fluctuations in demand. For the advanced probe card business in which we lead, we have had very strong growth the past couple of years. And so I think perhaps a little bit of moderation in digestion into 2021, is not a bad scenario to think about. Having said that, you look at some of the drivers like 5G and advanced packaging, and they continue to drive strong customer investment in wafer test and measurement. And so it's still obviously very early in the year, but a lot of the conditions that did drive that strong growth in the past couple of years continue to be in place. We're working hard to put the capacity in place to execute on those. But certainly, high single digits is a reasonable long-term growth rate to think about our probe card business. Fluctuations will happen quarter-to-quarter for sure.

Got it. Got it. So maybe my last question. I want to ask a little bit more about China, knowing that the domestic Chinese customers do not really have a major contribution. A lot of that is the multinationals. Do you -- over the past quarter, do you see any change of behavior, any stockpiling by some of your Chinese -- domestic Chinese customers ongoing? Any color would be great.

Yes. I mean, we're working pretty closely with a couple of those key customers, key China domestic customers, especially in the memory space. FormFactor obviously has a nice technology differentiation in memory, especially in DRAM. And so we've been quite closely partnered with some of those customers. They're ramping fairly aggressively. So I don't think there's the opportunity for a lot of stockpiling. It would be difficult for us to tell in any way. But I think we continue to work closely with those customers to monitor the regulatory environment and make sure that we can extract whatever opportunity that we can out of the domestic China customers, in addition to serving the region with the large multinationals that we're supporting there.

Our next question comes from the line of Craig Ellis from B. Riley.

Congratulations on the nice revenues in the quarter. Mike, I wanted to start with a question for you. So the systems business was particularly strong in the quarter. I don't recall at least, I don't see in my model it ever getting about $30 million. We were rounding at $35 million. At times in the past, you've helped us see some of the different technology drivers that are contributing to system shrinks. So what was it in particular that contributed to system strength? And were there -- was there or were there a singular or multiple tech themes that you're seeing as you see that kind of demand pull in the quarter?

Yes. I think, as a reminder for people, the systems business really serves leading edge R&D and development for the key customers in the industry. There's a couple of drivers to that. First of all, there's the organic piece. Essentially, the engineering probe business we acquired as part of the Cascade Microtech deal. But as a reminder, we also report FRT, our advanced packaging metrology and inspection business that we acquired in late 2019. And now HPD, the cryogenic test and measurement business we acquired in Q4. So there's organic growth components associated with some of the exciting things happening in the semiconductor industry, whether it's CMOS advancements, some of the optoelectronics pieces that are served by that engineering probe business, the legacy Cascade business, but we've also got some nice additions and accelerations through our M&A strategy that are being reported in the system segment.

That's helpful. So would it be fair to think this should be really a kind of low to mid-30s business going forward? Or for whatever reason, could it dip back more towards the mid-20s where what was tracking over a multiyear basis?

Yes. I think with the M&A additions, although HPD, for us, a little bit more futuristic in serving quantum computing. There is a contribution there, but I'm not sure it would shift the way you model much. The FRT business continues to grow nicely, again, driven by the unique metrology and inspection requirements of advanced packaging. That's a theme for FormFactor across our businesses, whether they be probe cards or engineering systems. And so going back to the mid-20 levels, I think, would be a disappointment for us because it implies some contraction among all those businesses. As we get adoption with the FRT business, as we get a little bit of momentum around the cryogenic test and measurement space with HPD and see some the slower but steady growth associated with the classical Systems business, I think it's reasonable to expect that segment to report certainly in the low 30s.

That's very helpful. I also wanted to ask a question about customers, and it's really a 2-part question. I'll keep it with you. The first part of the question is, given Intel's historic significance as a customer to the company and following some recent commentary from an incoming CEO that seems to point towards a retained significant manufacturing operation, I was hoping to get your updated views on Intel. And then with that as a takeoff point, if you could talk about some of the opportunities that the company has to expand its customer base over time, it's been a big part of the growth story over the last numerous years and what are the opportunities that you see out there today?

Sure. Our largest customer represents an anchoring relationship for the company. It's a long term engagement, well over a decade long. And we spend a lot of our resources and management bandwidth, making sure we're in lockstep with that customer. Having said that, it's a little early to understand the implications of their CEO transition. I do think, with the spectrum of manufacturing strategies that are available to them from IDM on one end to fewer fabless on the other, certainly, the recent comments would indicate that the things have swung back a little bit closer to the IDM strategy.

But I think if we look further beyond to the rest of the industry, this is basically why FormFactor has made it such a priority to serve all of the leading customers in the industry, whether they be IDMs or foundries in all of the different segments. Because we want to make sure that we're able to support these customers and have the opportunity to win business, no matter what their manufacturing strategy is. Obviously a very dynamic time in the semiconductor industry, and we want to have our bases covered so that we could have -- do have a good opportunity space.

There are a couple of key customers, both in the microprocessor space and in the mobile and automotive space, where there are some objectives that we have to go gain share. And expand and continue to diversify our overall customer base, even from the leadership position that we have.

That's helpful. And then lastly, if I could sneak in 1 more before dropping back into the queue. Shai, helpful to get the color on the OpEx intensity for the March quarter. The question is, what would the contour OpEx look like? Beyond that, are there things that are happening either with R&D or other programs may be related to just new facilities that would cause us to have to think about some incremental expense? Or would it potentially even drift down?

Sure. So if you look at OpEx for 2020 for the overall year, it was 26% of revenue. And given the growth in 2020 and the required capacity expansion and we put some -- some programs there on hold in 2020 just because of COVID restrictions and stuff like that. In 2021, we do expect OpEx to catch up. A lot of investments in R&D but also small things like travel that we hope and believe they will restart at some point. So we will continue to invest in R&D. We expect to see more leverage in SG&A making progress on that in 2021 on our way to the target model of R&D at 13% and SG&A at 12% of revenue. So I don't think we're going to be there in 2021, but we're on our way to do that.

Our next question comes from the line of Christian Schwab from Craig-Hallum.

This is Tyler on behalf of Christian. First question, for Mike probably, kind of revisiting a previous question, maybe asked another way. Your foundry and logic probe card business was very strong in Q4 here. And you said that you expect to be down maybe down modestly sequentially in Q1. So the question is, would we still expect after a strong 2020 for the foundry to grow in 2021? And if so, what would be kind of the major drivers there? I know you mentioned 5G ramps or kind of over after Q4. So if you do expect growth, kind of what are the major drivers that you'd point to?

Yes. So first of all, I certainly wouldn't characterize the 5G ramp as over. But there was -- remember, probe cards are specific to individual customer chip designs. And so if you have customers all ramping a particular design at once, that's going to create strength in that time period where you're delivering and recognizing revenue on those cards. And this is a business that's very time-sensitive, right? You have to deliver the probe cards in time for those customers to test the wafers. So we don't have a lot of flexibility on delivery timing. So you had the coincidence of both our largest customer returning to almost historically high levels in the fourth quarter.

In the fabless foundry ecosystem, a couple of mobile and a couple of high-performance compute designs. The modal designs for sure associated with 5G that ramped hard. We just see those -- that specific concentration of designs, not necessarily happening in the first quarter. I mean, our guidance is obviously not down by a huge amount and so you can see there's still significant strength there. Longer term, as we look through 2021 and next couple of years towards the target model, growth in foundry and logic probe cards is one of the fundamental underlying drivers of our path towards that model, and 5G is a huge part of it. So by no means is the 5G ramp over, we just had a bunch of designs, stack up that we had to deliver in the fourth quarter, that led to all-time record quarterly revenues.

Makes sense. Perfect. Understood. And thank you for clarifying the comment on 5G. Second question then moving over to the DRAM probe card side, strong end of the year after a couple of quarters of more digestion in the beginning of 2020. You said you expect that kind of flat Q1, maybe even Q2, or the first half. So I was wondering, can we expect kind of $35 million quarterly level be a sort of baseline for that business? Or over time, can we get back to above $40 million a quarter like we saw a couple of years ago.

Yes. The way we think about DRAM and obviously, being a supplier in the DRAM supply chain, value chain, you have to be prepared to deal with cyclicality, at least on a quarterly basis. We saw that, as you noted, in 2020, right? We went from what was a multiyear high in Q4 of '19 at about $40 million in DRAM revenue down to $20 million next quarter. That's a pretty extreme swing, but I think does represent the bands in which that business should be expected to operate in any given quarter.

2021, though, does seem to be a pretty decent year for the overall DRAM supply chain. There's going to be investments in capacity. As we told you, we see strong design pipelines from our customers that are feeding our business. I think, though, that a reasonable expectation in any given quarter is probably a $30 million, maybe low $30 million kind of run rate for DRAM with some unusual swings up and down as different design release cadences either are additive or subtractive with each other. So think about it in 35% is probably the high end of what I want to give you for an average level, but certainly starts with a 3.

(Operator Instructions)

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

Mike, maybe another geographic question for you. When you look at Taiwan, it was about 25% of the business. But I was surprised to see that TSMC was not a 10% customer. So maybe just talk a little bit about the dynamics in Taiwan.

Sure. Well, there's a lot of action in Taiwan, right? We have -- one of our large memory customers has a large production and test facility there. A lot of the outsource assembly and test capacity of the world is in there. So Taiwan revenue is not necessarily the world's largest foundry. But with them, they were a 10% customer in the third quarter. They were close to a 10% customer, again at the record revenue levels of the fourth quarter.

And given the continued momentum we have in delivering into their advanced nodes, as a reminder, this is, for us, a 10-nanometer, 7-nanometer, 5-nanometer kind of business. But we're doing a nice job, I think, of broadening that business. We've got multiple mobile and high-performance compute designs that we're shipping here in the first quarter. And I wouldn't be too surprised to see them back on the 10% list, sporadically up and down in the near future.

Okay. So in general, how often do the fab assembly and test guys buy the probes from you versus the foundry when wafers went out the door?

It really depends on individual customer and even individual projects. We'll have some customers who like to hand all of that over to either the foundry or the outsource assembly and test house, we'll have some customers that want to manage it very carefully. So it's one of the unique aspects of this business is you have a very broad engagement with all kinds of different elements of the semiconductor manufacturing chain.

Okay. Great. And moving over to the Flash business. Obviously, it's very lumpy. But I was surprised it dropped as much as it did, especially since the acquisition of Advantest technology, I thought maybe it would be a little bit more of a sustained business going forward.

Certainly, that's one of our aspirations. As we talked about that acquisition, there were 2 fundamental tenets to it. The first one was there were some very interesting technology that resided there that we felt was useful on the FormFactor DRAM and foundry and logic road maps. The second one was we felt like we may be able to make some progress in using a cost-competitive Flash technology that, that team has developed and push it into the broader FormFactor customer base, places where they really hadn't access before. The second one, I talked about both of these in the prepared remarks.

But that second element, those things take a while to come to fruition. We are in active discussions with multiple customers on qualifications for the first part of 2021. But given that business with single-digit millions revenues historically, I wouldn't expect annually, right? I wouldn't expect it to make a big impact on our Flash business until we're able to gain some traction with these additional customers.

Okay. That makes sense. And then finally, Shai, when you look at the gross margins in your target model, does that require a different product mix? Mix has a big impact on margins on a quarterly basis. But if you look over the next year or 2 to get to that target model, do you need to see certain parts of your business grow faster than others?

So as a reminder, even though it was not really reflected in the Q4 gross margin. Historically, the systems business has the highest gross margin, followed by foundry and logic, then DRAM and Flash. And of course, there is some overlap when you move from one market to another, it's not like everything in system is higher and everything is Flash is lower.

But -- so as we grow the business, when we presented the $850 million model, we talked about most of the growth coming from foundry and logic. And so this will help achieving the 47% gross margin. But together with operational excellence that we're working on, the ERP implementation project is one of them. The additional capacity will help, of course, but these are the main drivers.

Our next question comes from the line of Brian Chin from Stifel.

Just maybe a quick follow-up, probably for Mike. HBM3, SOIC, Foveros, when you kind of look at the avant-garde of advanced packaging technologies, when do you think that will really begin to favorably impact your revenue CAGR in probe cards?

I think you've seen some of it already. If you look at the probe card market in 2020, it almost certainly grew faster. I mean, we'll find out in a couple of months when VLSI rolls up their survey. But almost certainly grew faster than the historical correlation to the semiconductor industry that we've observed. That also was the case in 2019. And so I think you're seeing the leading edge of some of these things. But each of the projects you talked about, still pretty early innings. If you look at those as a fraction of wafer starts for any of our customers, they're still relatively small. Having said that, they consume a good fraction of the test capacity because of the need for high test intensity and high test complexity to make sure you've got something close to known good die, as we talked about in various -- well, most recently, the Analyst Day we did last year.

So I think you're already seeing the leading edge of this. If you look at our results and the probe card market results over the past couple of years, obviously, as semiconductor industries continues to accelerate adoption of these I think various customers have made pretty firm comments that for 7-nanometer and 5-nanometer, they view advanced packaging with these various architectures as being a big part of their innovation road map I think you'll see that probe card intensity continue to increase.

Okay. Yes, makes sense. And maybe one last other quick follow-up. In terms of -- going back to sort of the commentary around modulation of revenues and segments, yes, at a high level across the semi industry, it's hotter in some areas than others, but there's some in ability to sort of meet wafer demand. Some areas of maybe advanced, some areas of mature nodes. You guys play more in the advanced area. But based on that -- those sorts of wafer constraints, further upstream for you guys, do you see that playing in any way playing in the modulation of your revenue as we kind of walk through this year?

That's a tricky one. I think if you think about it from a bottoms-up perspective, for sure, there are individual designs with individual customers that get pulled in or pushed out depending on the priority calls they make with respect to the overall capacity they're able to secure. And so at a -- at an individual design level, you see that at work every day as our mix shifts around in the factory. But part of our strategy, again, has been to try and be exposed to so many of those different drivers that they essentially all average themselves out through the diversification of our business across all of the different puts and takes in the industry.

Our next question comes from the line of David Duley from Steelhead.

I think in your prepared remarks, you talked about NAND qualifications in the first half of this year and ramping up in the second half of the calendar year. Could you just elaborate a little bit more on that? And do you have targets for growing this business now that you seem to be engaged with a broader set of customers?

Yes. So I think I'd characterize the first part of 2021 as planting some seeds. The acquired -- our legacy NAND Flash business, our legacy Flash probe card business. Remember, we need to be fairly opportunistic, essentially skimming the high end of the opportunities that are available to us. Because it really is a higher cost platform, and we can't be cost competitive in the real sort of majority of 3D NAND. The Advantest acquisition gives us, what I'd characterize as an option to go participate there.

And so we've begun discussions with a couple of existing FormFactor customers on beginning a qualification in the first half of the year. Remember that qualifying a new architecture with a major customer usually takes a while. And so I wouldn't want to create the expectation that we'll struck the qualification box in the first half and then see skyrocketing NAND revenues. Because we also want to be careful that we're meeting all of the capacity and delivery requirements once we do get qualified with these customers. So it's a little early for us prequalification to be setting growth rate targets, but there is an aspiration for us to grow our NAND Flash market share on the back of having a much more cost competitive NAND Flash program.

Well, maybe you could just characterize it and do you have a target for market share rather than a growth rate for your business.

Well, so the NAND Flash probe card market, in round numbers between $200 million and $250 million a year spent by our customers. Some of that, you can probably consider to be captives to some Korean suppliers, so probably not a really servable market by us. But if you think over the long term, I don't think it's unreasonable that we could take our share position from what do we do, $20 million, $25 million a year, so 10%, and over the next couple of years on the way to the target model maybe get that up into the teens. Again, market share gains where you're introducing a new product to a new set of customers, not new to FormFactor, but needed that product. It's a long slow slog. And one that we're committed to, so we can make happen. But again, it's a little early to be making top line permits on.

Okay. And then one technical question for me. You spoke about DRAM and how I think it's becoming more test intensive as you stack the DRAM parts in these high bandwidth configurations. What percentage of the DRAM now do you think is stacked in that configuration? And what percentage do you think the market will move towards over time? I'm just trying to figure out because this seems like a key driver to demand for probe cards. So the more information here I can get, the better.

Yes. So HBM structures in general, maybe implicit in your question, are a pretty small portion of overall industry DRAM wafer starts, right? The vast majority of DRAM is not the stacked by high-bandwidth memory configurations. These high-bandwidth memory configurations, whether it be HBM2 or now moving to HBM3, are really targeted at the specific requirements of high performance compute, very high bus speeds, decent power management. And so it's -- if you want to come out from a revenue perspective, it's probably, I don't know, single-digit percentage of DRAM overall revenue for our customers.

Wafer starts probably a little bit higher than that because you're assembling -- you have to build a lot of wafers to assemble a 16 high stack of Die. So still, understanding it's a subsegment of overall DRAM I don't think it's reasonable to expect that all DRAM will become stacked. But we do see some growth there. And any growth there is good because of the high test intensity and high test complexity.

At this time, I am showing no further questions. I would like to turn the call back over to Mike Slessor for closing remarks.

Thank you, everyone, for joining us on this first call of 2021, and we'll talk to you, if not at some conferences between now and then, on our next conference call after completing the first quarter. Take care.

Ladies and gentlemen, this concludes today's conference call. Thank you for participating. You may now disconnect.