This conference call transcript was computer generated and almost certianly contains errors. This transcript is provided for information purposes only.EarningsCall, LLC makes no representation about the accuracy of the aforementioned transcript, and you are cautioned not to place undue reliance on the information provided by the transcript.
spk03: Thank you and welcome everyone to FormFactor's second quarter 2024 earnings conference call. On today's call, our Chief Executive Officer, Mike Slessor, and Chief Financial Officer, Shai Shahar. Before we begin, Stan Finkelstein, the company's VP of Investor Relations, will remind you of some important information.
spk00: Thank you. Today the company will be discussing their P&L results and some important non-GAAP results intended to supplement your understanding of the company's financials. Reconciliations of GAAP to non-GAAP measures and other financial information are available in the press release issued today by the company and on the investor relations section of our website. Today's discussion contains forward-looking statements within the meaning of the federal securities laws. Examples of such forward-looking statements include those with respect to the projections of financial and business performance, future macroeconomic and geopolitical conditions, the benefits of acquisitions and investments in capacity and in new technologies, the impact of global, regional, and national health crises, including the COVID-19 pandemic, anticipated industry trends, potential disruption in our supply chain, the impacts of regulatory changes, including the recent U.S.-China trade restrictions, the anticipated demand for products, our ability to develop, produce, and sell products, and the assumptions upon which such statements are based. These statements are subject to known and unknown risks and uncertainties that could cause actual results to differ materially from those expressed during this call. Information on risk factors and uncertainties is contained in our most recent filing on Form 10-K with the SEC for the physical year and the December 30th, 2023, and in our other SEC filings, which are available on the SEC's website at www.sec.gov and in our press release issued today. Forward-looking statements are made us of today, July 31st, 2024, and we assume no obligation to update them. With that, we will now turn the call over to FormFactor CEO, Mike Slater.
spk05: Thanks, everyone, for joining us today. FormFactor's second quarter revenue, non-GAAP gross margin, and non-GAAP EPS all exceeded the midpoint of the outlook range we provided in May. The expected strength in our probe card business drove second quarter results as we recorded sequential increases in all three probe card markets. DRAM probe card revenue reached record levels, with revenue from high bandwidth memory, or HBM, doubling for the third consecutive quarter to nearly 75% of our DRAM revenue. In the current third quarter, we're experiencing steady overall demand for form factors products, as we continue to make progress towards our target financial model. Our second quarter results and third quarter outlook illustrate the unique value of FormFactor's diversification strategy, which differentiates us from our direct competitors with a broad lab-to-fab product portfolio across Foundry and Logic, DRAM, and Flash probe cards, together with our system segment products. This portfolio enables us to compete for business across diverse demand pools at all major customers, producing relatively stable results during last year's downturn and the top line growth we're delivering this year. Our revenue growth is being driven by exposure to expanding areas like high bandwidth memory and DRAM probe cards and co-packaged silicon photonics and systems. enabling FormFactor to grow even as we await refresh cycles in important high unit volume end markets that drive Foundry and Logic probe card spending, like mobile handsets and client PCs. FormFactor's third quarter outlook also demonstrates several unique features of probe card demand. As we often note, probe cards are consumable that's specific to each new chip design, meaning we benefit from both technology node transitions and from the release of new designs on existing nodes. Together with our broad market and customer exposure, this produces a more stable demand profile than the more volatile demand cycles that characterize capital equipment. This difference is particularly evident in the current environment, as form factor stable third quarter outlook contrasts sharply with the sequentially weaker third quarter outlooks offered by several companies in the test and assembly capital equipment sector. Turning now to market and segment level details, as I noted, we set a record for DRAM probe card revenue in the second quarter, driven by the sequential doubling of high bandwidth memory revenue layered on top of steady DDR5 new design activity. To put HBM's torrid growth into perspective, in the first half of 2024 alone, we doubled the HBM revenue we delivered in all of 2023. In addition, our second quarter HBM revenue is greater than form factor's total quarterly DRAM revenue in each quarter of 2023. This activity is a direct result of the large hyperscaler investments in generative AI infrastructure and our customers' corresponding ramps of HBM capacity and output. HBM, which is a stack of eight, 12, or even 16 individual DRAM dye assembled with advanced packaging processes like through silicon vias and thermal compression bonding continues to offer a powerful example how advanced packaging is driving our current results. It also foreshadows our long-term opportunity as we benefit from the increased test intensity and increased test complexity that is inherent in advanced packaging architectures. In the third quarter, demand for DRAM probe cards continues to be robust, and we expect a slight increase in revenue from the second quarter record. We do, however, expect a shift in our DRAM product mix, with sequential increase in DDR5 shipments and a reduction in HBM shipments. We've mentioned in the past that our customers often have one to two quarter periods of digestion following a strong quarter of shipments. With lead times of less than a quarter, our short-term visibility remains challenging as always, but given our customers' continued strong investment in HBM capacity, we expect the third quarter pause in HBM growth will represent a temporary digestion phase, and the growth will resume in one to two quarters. Shifting to the Foundry and Logic probe card markets, we delivered the expected second quarter sequential growth, driven by the seasonal ramp of new mobile application processor designs and stronger probe card demand for client PC and server microprocessor designs. In the current third quarter, we expect Foundry and Logic demand and product mix to be similar to that achieved in the second quarter. Advanced packaging processes like Foveros and Coaos are being increasingly used to architect Foundry and Logic chip designs. As with HBM and DRAM, This disaggregation of a chip into subcomponent chiplets, or tiles, increases both test intensity and test complexity compared to an equivalent monolithic chip. The increase in test intensity is driven by the need for our customers to probe and test each component chiplet prior to stacking, and then to probe and test the multi-chiplet stack at various points during the assembly process. The resulting increase in probe card use per good die out is the same dynamic that is driving the strong probe card spending by our customers for HBM DRAM and is the source of relatively strong mid-year founder and logic demand despite the lack of recovery and important end markets like mobile handsets and PCs. At the same time, the technical requirements for probe cards for Foundry and Logic designs built using advanced packaging processes are significantly more demanding than for standard unstacked products involving higher test speeds and more challenging thermal and power specifications. As we've seen with HBM and DRAM, our differentiated ability to meet these performance requirements will drive both market share and profitability gains as the adoption of advanced packaging in Foundry and Logic continues. In the system segment, we expect a slight sequential increase in third quarter revenue as customers continue to engage us to solve the most complex electro-optical test and measurement challenges in areas like quantum computing and co-packaged silicon photonics. In co-packaged optics, which is poised to revolutionize chip-to-chip communication in the data center by significantly reducing power consumption at high data rates, We're collaborating with key customers in the early stages of the lab-to-fab transition from R&D to low-volume production. In these collaborations, customers are deploying FormFactor's turnkey electro-optical measurement systems, built on our CM300 and Summit 200 engineering probers, together with our proprietary Pharos optical probes. These highly differentiated optical probes enable either surface or edge coupling to the photonic dye, with unrivaled coupling efficiency, providing our customers with higher yields and shorter test times. As silicon photonics matures and moves to high-volume production in the coming years, we expect that our leadership positions in combined electrical and optical tests will provide a new growth vector for both our systems and probe card businesses. In closing, we're excited about both our strong second quarter results and our solid third quarter outlook. as the accelerating adoption of advanced packaging drives increased demand across form factor's lab-to-fab product portfolio. Longer term, we're confident in the growth prospects for form factor in the industry overall, driven by the fundamental trends of semiconductor content growth and advanced packaging innovations like HBM, chiplets, and co-packaged silicon photonics. As is evident from our recent results and outlook, these are trends where form factor is well-positioned. We're confident that our investments in R&D, capacity, and talent will further enhance form factor's market leadership. This will enable us to achieve and then surpass our target model that delivers $2 of non-GAAP earnings per share on $850 million of revenue. Shai, over to you.
spk01: Thank you, Mike, and good afternoon. As you saw in our press release, Q2 revenues were $197.5 million. $2.5 million above the midpoint of our outlook range, and non-GAAP gross margin of 45.3% or 0.3 percentage points above the midpoint of the range. These, together with OPEC slightly lower than the midpoint of the outlook, resulted in a non-GAAP EPS at the top end of the range. Second quarter revenues increased 17% sequentially from the first quarter and increased 26.7% year-over-year from our Q2-23 revenues. The upside versus the midpoint of the outlook range was due to higher revenues in our probecart segment. Probecart segment revenues were $167 million in the second quarter, an increase of $30 million, or 22%, from the first quarter. The increase was driven mainly by higher Foundry and Logic and DRAM revenue. The system segment revenues were $30.7 million in Q2, a $1.3 million decrease from the first quarter, and comprised 15.5% of total company revenues, down from 19% in the first quarter. Within the prop card segment, Q2 Foundry and Logic revenues were $104 million, a 19.5% increase from the first quarter. Foundry and Logic revenues increased to 52.5% of total company revenues, compared to 51.4% in the first quarter. DRAM revenues were a record $58 million in Q2, $12.1 million or 26.5% higher than in the first quarter, and rose to 29.4% of total quarterly revenues as compared to 27.2% in the first quarter. Within DRAM, HBM revenue almost doubled from $22 million in Q1 to over $43 million in the second quarter. Flash revenues of $5.1 million in Q2 were $1.1 million higher than in the first quarter and were 2.6% of total revenues in Q2, as compared to 2.4% in Q1. Gap gross margin for the second quarter was 44%, as compared to 37.2% in Q1. Cost of revenues included $2.5 million of gap-to-non-gap reconciling items. which we outlined in our press release issue today and in a reconciliation table available in the Investor Relations section of our website. On a non-GAAP basis, gross margin for the second quarter was 45.3%, 6.6 percentage points higher than the 38.7% non-GAAP gross margin in Q1, and 0.3 percentage points above the midpoint of our outlook range. The increases compared to Q1 was mostly the result of higher gross margins in the prop card segment, while we expect fluctuations quarter over quarter, mainly due to product mix changes, achieving this 45.3% gross margin in the second quarter validate our progress towards our 47% target financial model non-GAAP gross margin at annual revenue of $850 million. Our ProbeCard segment gross margin was 45.1% in the second quarter, an increase of 8 percentage points compared to 37.2% in Q1. Our Q2 system segment gross margin was 46.2%, an increase of 0.9 percentage points, compared to 45.3% gross margin in the first quarter. As compared to Q1, approximately two-thirds of the increase in non-GAAP gross margin is attributable to the higher volume, and the remaining third relates to a more favorable product mix. Our GAAP operating expenses were $69.4 million for the second quarter, as compared to $61.7 million in the first quarter. Non-GAAP operating expenses for the second quarter were $60.9 million, or 30.8% of revenues, as compared with $52.3 million, or 31% of revenues, in Q1. The $8.5 million increase relates mainly to higher performance-based compensation. Company non-cash expenses for the second quarter included $10.2 million for stock-based compensation, $0.6 million for the amortization of acquisition-related intangibles, and depreciation of $7.4 million, all similar to the first quarter. GAAP operating income was $17.8 million for Q2, compared with GAAP operating income of $21.3 million in Q1. Non-GAAP operating income for the second quarter more than doubled to $28.5 million compared with $13 million in the first quarter, an increase of $15.6 million or 120%, demonstrating the leverage in our operating model. GAAP net income for the second quarter was $19.4 million or $0.25 per fully diluted share, compared with a GAAP net income of $21.8 million or $0.28 per fully diluted share in the previous quarter. GAAP-need income in Q1 included a gain of $20 million from the divestiture of our Chinese subsidiaries. The non-GAAP-effective tax rate for the second quarter was 15.4%, 1.7 percentage points higher than the 13.7% in the first quarter, and we continue to expect our annual non-GAAP-effective tax rate to be between 14 and 18%. Second quarter non-GAAP net income almost doubled to $27.3 million, or $0.35 per fully diluted share, compared to $14.3 million, or $0.18 per fully diluted share, in Q1. Q3 APS was $0.17 higher due to significantly higher revenues and higher gross margins, with relatively flat OPEX as a percentage of revenue. Moving to the balance sheet and cash flow. We generated free cash flow of $14.2 million in the second quarter, compared to $19.7 million in Q1. The main reasons for the decrease in free cash flows was increased working capital, attributable to the revenue growth, partially offset by lower capital expenditure spending of $5 million. We invested $8.4 million in capital expenditures during the second quarter, compared to $13.4 million in Q1. There is no change in our previously communicated expected capex range for 2024 of $35 to $45 million. At quarter end, total cash and investments were $366 million, an increase of $8 million from Q1. At the end of the second quarter, we had one term loan remaining with a balance totaling $14 million. Regarding stock buyback. During the second quarter, we used $2.9 million to buy back shares under the $75 million two-year buyback program that was approved in Q4 2023. At quarter end, $53.5 million remained available under that authorization. As a reminder, the main purpose of the share repurchase program is to offset dilution from stock-based compensation. Turning to the third quarter non-GAAP outlook. We expect Q3 revenue of $200 million plus or minus $5 million with a slight increase over Q2 coming from DRAM and systems. Within DRAM, we also expect a mixed shift with a higher percentage of BBR5 versus HBM revenue. Third quarter non-GAAP gross margin is expected to be 43% plus or minus 150 basis points. The expected decrease in non-GAAP gross margins in the third quarter at the midpoint of the outlook range on slightly higher revenues relates to a less favorable product mix, with DRAM revenues as a percentage of total revenues expected to increase, and HBM revenue within DRAM is expected to decrease. At the midpoint of these outlook ranges, we expect Q3 operating expenses to be $61 million plus or minus $2 million, similar to Q2. Non-GAAP earnings per fully diluted share for Q3 is expected to be $0.31 plus or minus $0.04. A reconciliation of our GAAP to non-GAAP Q3 outlook is available on the Investor Relations section of our website and in our press release issue today. With that, let's open the call for questions. Operator?
spk03: To ask a question, you will need to press star 1-1 on your telephone. To remove yourself from the question queue, you may press star 1-1 again. We ask that you limit yourself to one question and one follow-up. to allow everyone the opportunity to participate. Please stand by while we compile the Q&A roster. Our first question comes from the line of Craig Ellis of B Reilly Securities.
spk04: Thanks for taking the question and congratulations on the execution, guys. Mike, I wanted to start just by following up on a point you made with regard to high bandwidth memory. I think you indicated the company expected that after some digestion in the third quarter, growth could reaccelerate there. I was hoping you could just share some of the either customer interaction or things that you're seeing that lend confidence to The reacceleration and any time it would be helpful. Thank you.
spk05: Yeah. So, Craig, good question. Just to level set everybody, you know, we again in the second quarter doubled HBM revenue to 75 percent of our DRAM revenue. So a really significant contribution from that. As I think most of you know, it's pretty concentrated with a single customer, although each of the three major DRAM manufacturers do have contributions in there. And it's not unusual when we ship at that scale for a customer to have a period of digestion where they then use the probe cards for a quarter or two. You can see this historically through our major customers as they're in their cadence quarter to quarter that we report for 10% customers. So what gives us confidence that this is that digestion? I think a couple of things. If I look at it from a macro level, you know, continued hyperscaler investment in generative AI, which in turn is driving our customers, our DRAM customers to increase their capacity and output of HBM3 and 3E currently, and beginning to do development on HBM4 and pilot production. And really in conversations with those customers, there's no pause in that. They really are continuing to invest very heavily in this pretty lucrative part of the DRAM market. again, driven by hyperscaler investments. Our direct visibility, as most of you know, is pretty limited. We operate with lead times of less than a quarter. And so in terms of POs and backlog, we can't see much past the end of the third quarter here. But when we look at the conversations we're having with these customers, when we look at the investments the hyperscalers are making, it does seem like it's pretty reasonable to attribute this Q3 pause, still HBM, still at pretty healthy levels, to a digestion period.
spk04: That's really helpful, Mike. Thanks. And then the follow-up question is somewhat similar. Three months ago, you noted that a Foundry Logic customer, formerly the largest customer for this last quarter where it's now DRAM, would at times in the past go through a period of digestion after a really strong quarter. Are you seeing signs and is it based in the guidance that we will see that digestion in the third quarter or how are you thinking about the risk that that could happen at this point?
spk05: Yeah, Foundry and Logic overall we see in the third quarter pretty comparable to the second quarter, both from overall levels and customer and product mix. And I think this is really another proof point associated with how advanced packaging is driving our business. You see major customers, certainly in compute, but also to some extent in mobile, beginning to adopt advanced packaging, in fact, more beginning to adopt advanced packaging. And we see that across the compute sector. Technologies like Foveros and Coos, even as the PC market remains, PCN market remains somewhat muted, we're seeing pretty solid demand for probe cards because of the increase in test intensity and complexity driven by advanced packaging in these sectors. So third quarter, pretty similar to second quarter in Foundry and Logic. from an overall customer and product mix perspective, again, driven by advanced packaging.
spk04: That's really helpful. And then just the observation on that, Mike, would be since those heterogeneous dye-based products are such a small part of mix now, but over time will become a majority of mix, that would seem to be a real nice long-term secular tailwind for the business.
spk05: Yeah, I think across the industry, it's a really interesting and compelling secular opportunity for those of us exposed to advanced packaging.
spk04: Thanks, Mike. Good luck in 3Q.
spk05: Thank you.
spk03: Thank you. Our next question comes from the line of Charles Shi of Niedemann Company.
spk09: Hi, Mike. Good afternoon. Thanks for letting me ask two questions. The first one, It's about DDR5 versus HBM. I think prior to this quarter, the way you characterize the DRAM pop-up market was if you back out HBM, the standard DRAM or DDR pop-up cards were still run rating around that, I would say 20 million-ish, the downturn level, basically. Looks like the next quarter, the September quarter, you're expecting a Some were similar DRAM total revenue, but since you talk about mixed shift away from HBM, more to DDR5, would you characterize maybe this is the beginning of the standard DRAM recovery, or this is maybe just one off quarter? I just want to know the sustainability of the DDR5 probe card growth from here. Yeah.
spk05: It's an interesting question, and as you might imagine, we're having Similar discussions internally and with our customers. I think if you look at the overall non-HBM DRAM market, we're characterizing it as DDR5, but that's a pretty broad brush that's low power server and compute. You know, it is seeing a significant step up here in the third quarter to levels that are comparable to the previous cyclical highs. Now, you know, when we see, you know, where HBM is, we said there's a bit of a one quarter digestion period. And that's not unusual. Remember, probe cards are specific to each customer chip design. And so as customers move their wafer starts in back and forth between technologies and different products for them to optimize their output for their market, we can see these pretty significant swing between designs and market submarkets and submarkets. Whether this is the start of a DRAM upturn, I think, is still to be seen. But it is encouraging. If you've seen DRAM spot pricing, if you've heard the narrative from our customers on the overall DRAM market improving, it certainly may be the beginning of an upturn. For now, we just don't have the visibility to say so, whether it is or not. But nice to be, again, operating our DRAM business, a very important business for us, up at record levels.
spk09: Thanks, Mike. I want to have a follow-up on the FoundryLogic side of the business. It's up in Q2. FoundryLogic was up in Q2 by quite a lot. But your largest microprocessor customers, the revenue you do disclose on a quarterly basis didn't really go up. I mean, as much, I mean, it doesn't explain all the incremental you're seeing in the FoundryLogic side. And so this kind of a mid-year strength, mind if you clarify a little bit of what's driving that outside of this microprocessor company? And more importantly, going into next quarter, who is driving that incremental growth in the FoundryLogic probe card?
spk05: Yeah, we often see this mid-year strength in mobile application processors. If you look at phone release cadences and work back to the timing where customers are going to need probe cards for this, that's exactly what we're seeing right now. And we talked about it on the last call with the anticipated second quarter strength, which materialized being associated both with microprocessors but also mid-year releases of these mobile application processors. That's a theme that we see carrying through the second quarter. Again, we see Foundry and Logic having a similar level and profile in the third quarter as in the second quarter. And so, you know, beyond that, again, visibility pretty limited, but we're pretty happy with how we're executing in the broader Foundry and Logic market. Always some improvements, always some share to gain. But that's, you know, both of those factors driving our Foundry and Logic strength in the middle part of the year.
spk09: Maybe a quick follow-up to that. Do you see a little bit of a mixed shift in Foundry logic probe card from Q2 to Q3? Thanks.
spk05: No, it's very similar, Charles.
spk09: Thank you.
spk03: Thank you. Our next question comes from the line of Tom Diffley of DA Davidson.
spk02: Yes, good afternoon. Thank you. Mike, I was curious, when you look at the high bandwidth memory ramp that you've had this year, how much of that was driven by just increasing volumes of high bandwidth memory versus a design change? And then if there is a big shift to the next generation design, what does that do to your business?
spk05: Yeah, as we've said, probe cards are a design-specific consumable, and so there's these two pieces. There's the release of a new design, But there's also the number of wafer starts and number of test cells and therefore probe cards that need to run on this design. I'd characterize the growth in the second quarter of the HBM business being concentrated across, let's call it in round numbers, 10 designs that are really the high runners driving the growth. And so that could be one chip. Remember HBM, a stack of mostly eight high DRAM, but there's a base die, there's test insertions for the eight core die, and then there's also a test insertion when the customer's done stacking. Each of those three sets of probe cards is a unique fleet of probe cards. And so that's part of the test intensity increase associated with HBM in particular and advanced packaging in general. But, you know, more than a handful of designs really driving the strength.
spk02: And then when you look at the business next year in 25, do you think it's pretty well split between the three main suppliers? Are you still going to be very leveraged to one of the players?
spk05: No, I think, you know, as I said, we currently see contributions from all three manufacturers, although we're over indexed to the leader in HBM market share. You know, a fundamental tenet of our strategy is to make sure that we're a key supplier at the leading edge to all major customers. And so, you know, qualification work is ongoing of both of them. We're in a position to compete for that business. As, you know, HBM3 and HBM4 come out, those are opportunities for our customer share to show and for us to go compete for new business as well.
spk02: Great. And the final question, Shai, when you look at the margin gap, guidance for the next quarter. Is it simply the decrease of high bandwidth memory as a percentage of product match issue?
spk01: Well, it's a combination of DRAM revenue overall as a higher percentage of revenue and within DRAM, the shift from HBM to DDR5. So it's both of them.
spk03: Okay, great. Thank you.
spk01: Thank you, Tom.
spk03: Thank you. Our next question comes from the line of Christian Schwab of Craig Harlem.
spk06: Great. Thanks for taking my questions. So just again on high bandwidth memory, we can kind of back into the math, pretty modest revenue from the other two big guys. As we get into 2025, do you think you have three meaningful customers um in that space or do you think that there's the potential for uh as the other two guys get up to speed um that they you know begin to take share um i'm just trying to size what your your expectations and 25 and 26 are for for high bandwidth memory as the other two guys come to market and have talked about being sold out for next year yeah
spk05: Well, remember, if we back out a little bit, we're a key supplier to all three major DRAM manufacturers in DDR4, DDR5, and we hope to be for HBM as well. Currently, one of them owns the lion's share of HBM shipments. And so That's why we're over indexed to that customer. But again, you know, as, as Tom asked, we, we do expect, um, as they, as the other two DRAM manufacturers gain share in the transitions from HBM three to HBM four, work their way through some pretty significant technical and yield challenges in ramping this up. We do expect that share to balance out a little bit. Um, So as we look into 25, the transition to HBM4, certainly exciting for us. It raises the speed requirements for the probe card. So again, a notch up in complexity. And as customers go to stacking 12 and even 16 high die, that obviously is going to increase the overall test intensity and complexity in testing that overall stack. So we feel like we're in a pretty strong competitive position. even as the share starts to balance among all three DRAM manufacturers.
spk06: And to follow up to that, on the high bandwidth memory for transition, you know, given, as you said, the speed and then increased stacking capabilities, I would assume that there should be some pricing power on those probe cards, not all the way, obviously, to foundry and logic levels, but, you know, potentially higher than where we sit today. Is that fair?
spk05: Yeah, that's fair. And one of the reasons, right, you know, we touched on it in a question a couple of minutes ago, but the mix shift away from HBM towards DDR5 inside DRAM at these high levels certainly is a gross margin headwind. And so you can connect the dots there. our ability to build probe cards that are highly differentiated and provide a lot of value to our customers in testing at speed and across multiple temperatures in a very complex mechanical situation. You know, you're testing a whole 300 millimeter wafer of the stack die across a temperature range of over a hundred degrees Celsius. And you can imagine the thermal scaling challenges associated with that are significant. And, That's why an HBM probe card is worth more.
spk06: Great. And then my second question, we've got two quarters calling it roughly $200 million, which is kind of the target model. Do you guys have a set of expectations when you plan on updating your target model?
spk01: Sure. I'll take this one, Christian. So we are two quarters of $200 million, but the target model is a little higher than that, right? If you take the H50, it's about $210, $212. So what we want to do is what we have done in the past. Once we have a quarter or two at this model run rate, and I agree we're getting closer to it, but we're not there yet, then we'll set up a date, schedule a new analyst day, and then work on publishing the new target model. So we're not there yet, but stay tuned.
spk06: All right. Sounds great. Thank you, guys.
spk01: Thanks.
spk03: Thank you. As a reminder, to ask a question, please press star 1-1 on your telephone. Again, that's star 1-1 on your telephone to ask a question. To remove yourself from the question queue, you may press star 1-1 again. Our next question comes from the line of Brian Chin of Stifel.
spk08: Hi, thanks for letting us ask a few questions. Mike or Shai, did you comment whether you expect sort of that, you know, the next shift you're seeing in the DRAM business to persist in the Q4? I guess that's kind of the first question because, you know, maybe it's like a 150 to 200 basis point headwind or something relative to what it was in 2Q. And so that's, I guess, the first question.
spk05: Yeah, Brian. Mike, I'll take that. I think in terms of timing, I'll again caveat this with our visibility, our direct visibility doesn't really go much past the end of 3Q here. We're operating with short lead times. We certainly have forecasts from our customers, but the dramatic shift in DRAM towards production capacity for DDR5 is something that's a relatively dynamic situation. Whether it persists or not is something that Probably we need a few more weeks or months to really get a read on from our direct backlog. Having said that, if you do look at the underlying DRAM market, DDR4 and DDR5, low power server and PC, they do seem to be strengthening. And so it wouldn't be all that surprising if there was a shift of some capacity between towards those markets, and we saw a bit growth in the non-HBM parts of the market. A little bit early to tell, and again, our visibility doesn't go that far into the future, but some of the fundamental factors are in place for a DRAM recovery, a general DRAM recovery.
spk08: Yeah, I was even thinking, you know, Samsung seemed pretty confident last night that their HBM3 shipments, you know, they aren't the dominant player, but they would accelerate uh, substantially in the back half of the year. So maybe it's not three Q, but I imagine that could factor into your outlook for HBM, uh, at some point in the second half. That's right. Okay. Um, and I don't know if anyone asked, but in terms of the, I missed this part, but the China increase Q and Q, was that mainly just multinational than any kind of additional color on, on what drove that?
spk05: Yeah, no, that, that wasn't asked. So a good topic to touch on. Yeah. Reminder. So in Q2, um, a little over 10% China as a percentage of revenue. And inside the Q3 outlook, we are forecasting a moderate increase, but it'll still be in the team, nowhere near the 40 to 50% some of the WFE suppliers have for China concentration. You know, remember, we've been pretty proactive in managing this situation. We divested our China operations and then formed an exclusive distributor ship with the with the purchaser of that business. And we're opportunistically serving the domestic China business as best we can in the face of some pretty significant geopolitical headwinds, both from direct export controls and the China industry's semiconductor response to stand up its own domestic semiconductor industry. So we see a moderate increase. I don't see it as a theme where we're going to get a significant contribution from the China business, again, because of these almost structural geopolitical headwinds.
spk08: Great. Let me just kind of go off topic here for one last question. Energy efficiency, usage, bandwidth, these are obviously key themes in data center. Understanding that silicon photonics is an area the company has invested in, where are we on the adoption curve for co-package optics and how significant of a market opportunity could that be in maybe a couple of years?
spk05: Yeah, we are very, very early. As I said, just inching out of the lab where we've been engaged with customers for years in co-developing the fundamental measurement technologies like our Pharos optical probe that allows customers faster test times and better overall yield because it's better signal-to-noise and coupling efficiency. But we've got a handful of tools inside production sites that are in the very early stages of pilot production. As you note, though, this is one of the ways to help solve, it's not going to entirely solve, but it helps solve a pretty significant problem associated with data centers, and that's energy consumption. Going to silicon photonics on chip-to-chip communication in the data center has the potential to impact the energy usage, energy budgets for these data centers. And that's something the broader compute industry is going to have to confront. as data centers become bigger and bigger users of electricity worldwide. So we view silicon photonics as something that's a when, not an if. I think the timing, you know, probably at the earliest, late 25 into 26, where we start to see moving into the knee of the curve of adoption and running significant volumes inside the foundries worldwide. Okay, great.
spk08: Thanks, Mike.
spk03: Thank you. Our next question comes from the line of Robert Mertens of TD Cohen.
spk07: Hi, this is Robert Mertens on for Krish. Thanks for taking my questions. You mentioned the better than expected gross margins in the probe card business, largely due to higher volumes and maybe around 200 basis points or so attributed to the mix. Was the positive mix largely the high bandwidth memory increase or more of a general mix throughout both the Foundry logic and memory and markets?
spk01: So if you refer to Q2, we were right on spot, right? So gross margin was 45.3% versus midpoint of the outlook of 45%. So I'm not sure what miss you refer to.
spk07: Sorry, just in terms of the gross margins, I thought I heard two-thirds driven by volumes and a third driven by mix.
spk01: That's the increase from Q1. So Q1 gross margin, if you recall, was in the high 30s, 38.7. So the increase from Q1 to Q2, from the 38.7 to the 45.3, two-thirds of it relates to volume and about a third to mix. And that change in mix is more HBM, more sound real logic. Okay.
spk07: And then could you just provide a little more color on the progress with qualifications of MEM-based probe cards for GPU tests and maybe what the size of the market opportunity could look like?
spk05: Yeah, I think sizing it is interesting. You know, maybe to get everybody on the same page, as the largest GPU manufacturer transitions to advanced packaging, you know, primarily co-ops at the world's largest foundry, that requires MEMS probe cards, right? The legacy probe cards they use for monolithic GPUs no longer work. And so we are in a qualification process there. I'd say our major competitor has been ahead of us and has got the initial business, kind of what we've been more focused on HBM and that opportunity. But it's a strategic imperative for us to qualify there here in 2024 and make some progress. I do think, again, relatively difficult to size the opportunity. But as you heard from some of the memory manufacturers, there's an interesting asymmetry associated with some of these AI products where the amount of memory in the package is almost 10 times that of the GPUs. And so for unit-driven businesses like ours, HBM becomes a much more compelling opportunity. That's not to diminish the importance of qualifying during this important transition of GPUs to needing MEMS-based technologies for probe, but we like, obviously, the exposure to HBM as well.
spk07: Got it. Thank you.
spk03: Yep. Thank you. I would now like to turn the conference back to Mike Schlesser for closing remarks, sir.
spk05: All right. Thanks, everybody, for joining us today. We've got a couple of conferences in late August and early September that we hope to see you at. Until then, take care.
spk03: This concludes today's conference call. Thank you for participating. You may now disconnect.
Disclaimer