Earnings Call Transcript
Formfactor Inc (FORM)
Earnings Call Transcript - FORM Q3 2023
Stan Finkelstein, VP of Investor Relations
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 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 crisis including the COVID-19 pandemic; anticipated industry trends; potential disruptions in our supply chain; the impact of regulatory changes including the recent US-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 and could cause actual results to differ materially from those expressed during this call. Information on risk factors and uncertainties is contained in our most recent filing on Form 10-K with the SEC for the fiscal year ended December 31, 2022 and in our other SEC filings, which are available on the SEC’s website at www.sec.gov and in our press release issued today. Forward-looking statements are made as of today November 1, 2023, and we assume no obligations to update them.
Mike Slessor, CEO
Thanks everyone for joining us for FormFactor's third quarter earnings call. Stronger than anticipated demand for Foundry and Logic probe cards, coupled with record revenue in our systems business, produced third quarter revenue near the high end of our outlook range. This, together with higher than expected gross margin and leverage on our operating expenses, produced non-GAAP earnings per share above the top end of our outlook range. We continue to operate in an overall demand environment that remains relatively stable at the levels we've experienced throughout 2023. The moderate sequential decrease in our fourth quarter revenue outlook range is due to the reduction in system segment revenue from the sale of FRT to Camtek, which we completed today and weaker foundry and logic probe card demand due to a short-term reduction in customer spending, partially offset by stronger DRAM probe card demand. While we continue to operate in a cyclical downturn and see no near-term acceleration of demand, we're encouraged by broader signs of bottoming in our most important high unit volume end markets like client PC and mobile. In this environment, we continue to carefully balance short-term results and long-term investments with disciplined cost control that prioritizes quarterly profitability and sustains our strong balance sheet. We're maintaining our investment in R&D for new product innovation and competitive differentiation, especially in our product roadmap for advanced packaging applications like chiplets and tiles, high bandwidth memory and co-packaged optics. As you'll hear from Shai, we've completed the majority of the long lead time facilities and equipment capacity increases required to reach our target financial model and have slowed our CapEx to ensure a better balance between our capacity and anticipated long-term demand. Turning now to segment level details. Foundry and logic probe cards, our largest business, delivered higher than expected sequential growth in the third quarter as key customers ramped several major designs more aggressively than originally forecast. This inter-quarter acceleration provides insight into FormFactor's demand profile. Because probe cards are device-specific consumables customized for each individual chip design, increased wafer starts for a specific chip design can drive incremental demand for the corresponding probe cards. In the foundry space, we experienced strong third quarter demand for probe cards testing two major high-performance compute designs. In these HPC applications, FormFactor’s differentiated vertical MIMS probe card products meet demanding power and speed requirements with best-in-class test cell uptime and productivity, delivering significant value to our customers. We do expect a smaller contribution from foundry probe cards in the current quarter as our strong third quarter shipments are utilized to ramp fourth quarter foundry wafer production volume. In the microprocessor space, we delivered on higher demand for probe cards testing a major tile-based client PC design. Like the high-performance compute foundry example I just shared, probe card demand for this design accelerated within the third quarter as our customer expanded their wafer start plans to meet increased end customer demand. This client PC microprocessor is assembled from multiple tiles or chiplets into a single die using leading-edge advanced packaging processes. This is a great example of how advanced packaging is driving FormFactor’s business. As we've noted in the past, advanced packaging integration schemes drive both higher test intensity, which expands the number of probe cards required per good die out, and higher test complexity, which raises the performance for each probe card. Probe card architectures like FormFactor’s MEMS technology are essential to meeting these challenging performance requirements on short lead times at a compelling cost of ownership. In memory, as expected, we experienced stable overall demand for DRAM probe cards while flash remained weak. As in the second quarter, DRAM was driven by DDR5 as well as third-generation high bandwidth memory designs, which are enabling the rapid growth in generative artificial intelligence, and we expect a similar HBM-rich DRAM revenue profile in the current fourth quarter. The positive impact of high bandwidth memory is evident in the product mix profile at one of our largest DRAM customers where probe cards for testing HBM comprise more than half of our revenue to that customer in the third quarter. As in the tile-based microprocessor example I discussed earlier, the need for sophisticated probe cards to test HBM illustrates how advanced packaging is driving FormFactor’s results. As a reminder, each HBM chip is a stack of 8, 12, or even 16 individual DRAM die assembled with advanced packaging processes, such as through silicon vias and hybrid bonding. To ensure high yields of the stack DRAM chip, customers probe and test each component DRAM die prior to stacking and probe and test the multi-die DRAM stack at various points during the assembly process, leading to a substantial increase in the overall probe card intensity for good die out. In addition, the technical requirements for HBM tests are significantly more advanced than for standard unstacked DRAM products, involving higher test feeds and more challenging thermal scaling specifications. FormFactor's MEMS-based SmartMatrix probe card architecture meets these advanced requirements, providing significant value to our customers and differentiation for FormFactor. We believe our superior performance capabilities will drive both market share and profitability gains as HBM continues to grow, driven by the accelerating adoption of generative AI. Shifting now to systems. As anticipated, our systems business delivered record revenue in the third quarter as we shipped systems that were pushed out of the second quarter. This business continues to be driven by strong demand for our market-leading test and measurement products for early development of applications like co-packaged silicon photonics, quantum computing, and advanced packaging metrology. Co-packaged optics enabled by silicon photonics remains an important and exciting driver for FormFactor's current and future business. We're engaged with major customers in the transition of silicon photonics from early R&D to low volume production. In the fourth quarter, we plan to ship a CM300 silicon photonics system to a leading foundry to support low volume production. As silicon photonics production volumes increase over the next several years, our product roadmap delivers both systems and consumable probe cards to test these electro-optical devices and improve yields, enabling our customers to seamlessly transition from the lab to the fab. Moving forward, we do expect a short- and medium-term reduction in system segment revenues from the third quarter's record levels due to the sale of our FRT metrology business. We grew FRT substantially following our 2019 acquisition of the business, and I want to thank all of the FormFactor team worldwide for the contributions to this growth and value creation. Operationally, the transaction enables the FRT team to leverage Camtek's established scale and expertise in inspection and metrology to deliver the next stage of growth. Financially, the transaction maximizes FormFactor shareholder value by realizing a robust return on our investment, while also allowing us to focus our resources on strategic initiatives and businesses where we have market leadership and significant scale in enabling advanced packaging like the tile-based microprocessor and co-packaged optics examples I shared earlier. Shai will update you with more details later, but our capital allocation priorities of focused reinvestment in R&D and capacity together with M&A and share repurchases to offset dilution are unchanged. M&A remains one of the primary pillars of FormFactor's growth, and we continue to evaluate a broad funnel of targets that add complementary technologies and expand our served addressable markets. In closing, we continue to operate efficiently in what we see as a relatively stable near-term demand environment across our diversified product and technology portfolio. Longer term, we remain excited and confident in the growth prospects for FormFactor and the industry overall, driven by the fundamental trends of semiconductor content growth and exciting innovations like chiplets, high bandwidth memory, and co-packaged silicon photonics. These are trends where FormFactor is well positioned as an industry and technology leader, and we're confident that our investments in R&D and capacity position FormFactor to emerge from the current cyclical downturn a stronger and leaner competitor, enabling us to achieve our target model that delivers $2 of non-GAAP earnings per share on $850 million of revenue.
Shai Shahar, CFO
Thank you, Mike, and good afternoon. As you saw in our press release and as Mike mentioned, Q3 revenues were near the high end of our outlook range and non-GAAP gross margin and non-GAAP EPS were above the high end of the range. Third-quarter revenues were $171.6 million, a 10% sequential increase and a year-over-year decrease of 5.1%. Q3 revenues were $0.4 million below the high end of our outlook range and benefited from strong revenues in both probe cards and system segment revenue. Probe card segment revenues were $128.3 million in the third quarter, an increase of $13 million or 11.3% from Q2. The increase was driven by higher foundry and logic and flash revenues, partially offset by a decrease in DRAM revenues. System segment revenues were a record $43.2 million in Q3, a $2.6 million increase from the second quarter and comprised 25.2% of total company revenues, down slightly from 26% in Q2. Within the probe card segment, Q3 foundry and logic revenues were $96.4 million, a 17.7% increase from Q2. Foundry and logic revenues increased to 56.2% of total company revenues compared to 52.5% in the second quarter. DRAM revenues were $27.4 million in Q3, a $3.1 million decrease or 10.2% lower than in the second quarter and decreased to 16% of total quarterly revenues as compared to 19.6% in the second quarter. Flash revenues of $4.5 million in Q3 were $1.6 million higher than in the second quarter and were 2.6% of total revenues in Q3 as compared to 1.9% in Q2. GAAP gross margin for the third quarter was 40.4% as compared to 38.7% in Q2. Cost of revenues included $2.5 million of GAAP to non-GAAP reconciling items, which we outlined in our press release issued today and in a reconciliation table available in the investor relations section of our website. On a non-GAAP basis, gross margin for the third quarter was 41.8%, 1.2 percentage points higher than the 40.6% of non-GAAP gross margin in Q2 and 0.3 percentage points above the high end of our outlook range. The increase as compared to Q2 and the upside versus our outlook range were a result of higher probe card segment gross margin. Our probe card segment gross margin was 38.5% in the third quarter, an increase of 2 percentage points compared to 36.5% in Q2. The increase from Q2 is due to the net effect of three main factors. First, higher revenues contributed to a 2.1 percentage points increase. Second, improved factory utilization at higher production levels contributed to a 0.3 percentage points increase. Partially offsetting these two positive factors was a 0.3 percentage points related to a slightly less favorable product mix. Our Q3 system segment gross margin was 51.8%, basically flat with a 52% gross margin in the second quarter. Our GAAP operating expenses were $66.6 million for the third quarter as compared to $61.6 million in the second quarter. The two main reasons for the increase were $1.8 million higher stock-based compensation related to timing of grants and a higher grant date share value and transaction expenses of $2.1 million related to the sale of the FRT business. As previously disclosed, we entered into a definitive agreement during the quarter to sell FRT to Camtek for $100 million in cash, subject to customer repurchase price adjustments. The transaction closed today and the outlook and the results of operations for Q4 will include FRT for the month of October. Net proceeds from the transaction after adjustments, expenses and taxes are expected to be approximately $95 million. Non-GAAP operating expenses for the third quarter were $54.5 million as compared with $52.1 million in Q2, declining to 31.8% of revenues from 33.4% in the prior quarter. The $2.4 million increase relates mainly to higher performance-based compensation. Company non-cash expenses for the third quarter included $10.8 million for stock-based compensation, $1.6 million higher than in the second quarter, as I said earlier, $1.3 million for the amortization of acquisition-related intangibles, which was $1.1 million lower than in Q2 due to certain intangible assets reaching full amortization, and depreciation of $7.8 million similar to the second quarter. GAAP operating income was $2.7 million for Q3 compared with GAAP operating loss of $1.3 million in Q2. Non-GAAP operating income for the third quarter was $17.3 million compared with $11.2 million in the second quarter, an increase of $6.1 million or 55%. GAAP net income for the third quarter was $4.4 million or $0.06 per fully diluted share compared with a GAAP net income of $0.8 million or $0.01 per fully diluted share in the previous quarter. The non-GAAP effective tax rate for the third quarter was 12.2%, 200 basis points lower than the 14.2% in the second quarter, mainly due to discrete items arising from normal differences between estimated and actual US taxes. We continue to expect our annual non-GAAP effective tax rate to be between 13% and 17%. Third quarter non-GAAP net income was $17.3 million or $0.22 per fully diluted share compared to $11.2 million or $0.14 per fully diluted share in Q2. Moving to the balance sheet and cash flows. We generated free cash flow of $16.9 million in the third quarter compared to $2.1 million in Q2. The main reason for the increase in free cash flow was a decrease of $14.6 million in capital expenditures. We invested $5.9 million in capital expenditures during the third quarter compared to $20.5 million in Q2. The decrease in CapEx in Q3 is partially due to timing of payments, but it is also aligned with our intent to slow down capacity expansions to ensure capacity does not significantly outpace demand as we have completed the majority of the long lead time facilities and equipment investments required to reach our target financial model. There's no change in our previously communicated expected CapEx range for 2023 of $55 million to $65 million. Overall, at quarter end, total cash and investments were $249.4 million, an increase of $9 million from Q2. As of the end of the third quarter, we had one term loan outstanding with a balance of $15 million. Regarding stock buyback during the third quarter, we did not purchase shares under our $75 million two-year buyback program approved in Q2 2022. On October 30, our Board of Directors approved a new two-year $75 million share repurchase program, which is in addition to the $18.6 million that remains available under the existing authorization. As a reminder, the main purpose of these share repurchase programs is to offset dilution from stock-based compensation. Turning to the fourth quarter non-GAAP outlook. We expect Q4 revenues of $165 million plus or minus $5 million. At the midpoint of this outlook range, Q4 revenues are expected to be approximately $7 million lower than in Q3 with half of the decrease related to the sale of FRT metrology business at the end of October. We also expect lower foundry and logic revenues in the fourth quarter, partially offset by an increase in DRAM revenues. Fourth quarter non-GAAP gross margin is expected to be 41% plus or minus 150 basis points. At the midpoint of these outlook ranges, we expect Q4 operating expenses to be $51 million plus or minus $1 million. Non-GAAP earnings per fully diluted share for Q4 is expected to be $0.20 plus or minus $0.04.
Operator, Operator
Our first question comes from Krish Sankar from TD Cowen.
Krish Sankar, Analyst
Mike or Shai, the first question is you said that December quarter has more DRAM in the mix but you also said it's more HBM. I was under the impression that HBM probe cards' gross margins are pretty strong. So I'm just wondering, wouldn't that help overall gross margin or is it still below foundry and logic?
Shai Shahar, CFO
It's accurate that typically, HBM gross margins or HBA DRAM gross margins are higher compared to other DRAM designs, falling on the higher end of the spectrum. However, they may still be lower than the gross margins for foundry and logic. We are anticipating a decline in foundry and logic revenue in Q4, while expecting systems revenue to remain stable. Taking all of this into account, coupled with a lower overall revenue projection, we are forecasting a Q4 gross margin outlook of 41%, which is slightly below the 41.7% we achieved in Q2.
Krish Sankar, Analyst
And then a quick question for Mike. In the September quarter, you said foundry and logic was strong. Is it fair to assume there's more logic or NPU BIOS? I'm kind of curious with the largest foundry, what is your market share today, and is there a potential to increase market share as we get some of the AI applications?
Mike Slessor, CEO
I think in the September quarter, we highlighted two pieces in the foundry and logic business. Certainly, the ramping of a chiplet-based or tile-based client microprocessor was a key highlight adding to the strength. But we also commented on some high-performance compute designs in the foundry space that drove some strong results. As we look at our market share across the industry, whether it be in foundry, foundry and logic, RF, memory, one of the key strategic pillars of the company is market leadership. And I think in the foundry and logic market, there is some opportunity to gain some share based on the investments we're making and the very strong customer relationships we have with those key customers.
Krish Sankar, Analyst
Mike, is there a way to quantify that market share number or…
Mike Slessor, CEO
Yes, I think it moves around quarter-to-quarter. Again, because probe cards are design specific consumables, a key design win in any quarter can move it around. I would say for most major customers they have us and our primary competitor in foundry and logic in about a 60-40 window. They need to have two suppliers for supply chain continuity, a lesson that was hard learned through the pandemic and beyond. And so I think we're competing in this window. And I think in most of the major customers, there are a couple of exceptions that offer upside opportunity for us, and most of these customers were in that window. I think the Q3 performance because of these two designs was pretty strong based on the data we currently have.
Operator, Operator
And our next question comes from the line of Brian Chin from Stifel.
Brian Chin, Analyst
Mike, maybe first, just thinking about the fourth quarter guidance, and in terms of the order activity that you might normally see on the foundry logic side of the business versus kind of what's implied in the guidance. Is the breadth of the design activity pretty normal at this stage but maybe the release of volume against these designs still leaving something to be desired, is that a fair way to characterize?
Mike Slessor, CEO
Yes, Brian, you’ve captured the essence of it. Throughout 2023, our design teams have played a crucial and distinctive role in the probe card business by customizing our architectures for each individual customer chip design. These teams remain fully engaged. However, the peak revenues we've experienced in the past, such as in the first half of 2022, are not translating now, except for a few exceptions like HBM and the tile-based client microprocessor. Generally, customers are seeing reorder volumes for these new designs that are lower than usual during a cyclical downturn. This behavior makes sense as inventories remain high across various sectors of the industry. While customers are eager to innovate and push out new designs, they also aim to reduce the costs associated with obsolescence when introducing them into a market with elevated inventory levels.
Brian Chin, Analyst
And then I guess kind of related, but when you think about augmenting that kind of environment for however long it sort of persists until things get better. HBM and AI-related revenue opportunities are certainly larger than you probably thought a couple of years ago when you set that $850 million revenue target. And so when you look at sort of co-packaged and the individual HBM and GPU components under test there, do you have a sense of what the size of that served addressable market is for advanced probe cards now and what it could look like in two or three years?
Mike Slessor, CEO
I think I'll get to that in a second. But I think it's an important element of the different puts and takes, referencing our target model or anybody's target model, we're still committed to the $850 million target model. But you can imagine that when we put that in place, there was pretty open trade with China for US suppliers. But on the other hand, we didn't have AI and high bandwidth memory and GPUs as front and center as we do now. So we're always going to have puts and takes on the way to a target financial model. Specifically with HBM and AI, for us and I think most other equipment and consumable suppliers in the industry, although the GPUs kind of get all the glory, HBM drives a tremendous increase in overall wafer volume. Again, because you're stacking individual DRAM die, each of those comes from its own wafer. But also just the memory, you look at the silicon area associated with the memory, it's much larger than the GPUs. So HBM looks like, as I said, for this quarter, the just completed third quarter, was more than half our DRAM probe card revenue at one of our largest DRAM customers, that's a contribution that we hadn't contemplated. And if you look at industry forecasts, there's still quite a bit of bullishness associated with HBM and AI in general. Maybe we can get to it in a follow-up but we're working pretty hard on penetrating the GPU market as well.
Operator, Operator
And our next question comes from Dr. Charles Shi from Needham & Company.
Charles Shi, Analyst
I would like to ask about Q4 in terms of seasonality. Mike, if I compare the Q4 FRT revenue or even the FRT revenue from Q3, it appears that Q4 is somewhat below seasonal expectations. I believe you may have addressed this in your previous responses, but I would like your thoughts on why Q4 is still below seasonality. Is this primarily due to lower spending on the PC or smartphone side? Any additional insights would be appreciated.
Mike Slessor, CEO
So Charles, as we've discussed before, seasonality has become somewhat unpredictable in this business, and I think that's the case for 2023 as well. When you look at how Shai analyzed the difference from Q3 to Q4, the midpoint of our guidance shows that the absence of FRT accounts for about half of that $7 million change. We are also observing weaker demand in foundry and logic, which we believe is temporary, considering the significant contributions from these areas in Q3 due to several major designs. There is currently a bit of adjustment happening among our customers as they utilize those probe cards for those significant designs and products. On the flip side, we are seeing some additional strength in DRAM, which we noted is largely driven by HBM and some DDR5. Those are the various factors at play. Once again, while there are elements of calendar seasonality, the strength in HBM seems to overlay what might be the typical seasonal and cyclic patterns in DRAM. As I mentioned, we are witnessing some level of adjustment from the strong results achieved in the third quarter for foundry and logic.
Charles Shi, Analyst
I was hoping you could clarify how much HBM revenue you generated in Q3 and what your expectations are for Q4. You've mentioned that Q3 and Q4 typically account for over half of the DRAM revenue from your largest DRAM customer, but we don't have specific figures regarding that customer within the $28 million to $30 million projection for Q4. Could you help us understand the progression of HBM revenue over the last two quarters and what to expect for the next quarter? In June, you mentioned it was around $10 million. We're trying to understand how variable this revenue is and whether there has been any sequential growth over the past couple of quarters.
Mike Slessor, CEO
Yes, there is going to be volatility quarter-to-quarter. I ballpark you that the absolute magnitude of it in the third quarter of HBM inside DRAM was about the same in round numbers, $10 million. We do expect that contribution to strengthen in the fourth quarter. But again, with probe cards being specific to each individual chip design and things like DDR5 also showing some significant activity and different customers ramping different designs at different times, we want to look at this on a multi-quarter basis, similar to the comments I made around market share. So on an absolute level, it's about stable. It was heavily concentrated in one customer. As I said, surprising perhaps that half of our DRAM revenue at this customer was associated with HBM. But we're looking at this as a longer-term trend where we continue to want to ride this tailwind, but also build differentiation and share not just the one customer who's leading in HBM DRAM right now but each of the three major DRAM customers, and we've got activity with all of them.
Operator, Operator
And our next question comes from the line of Vedvati Shrotre from Jefferies.
Vedvati Shrotre, Analyst
Can you help me understand if there's a difference in test intensity for HBM devices compared to chiplet microprocessors now that you have HBM and your chiplet-based microprocessors ramping?
Mike Slessor, CEO
No, I think we are seeing consistent increases in testing intensity across the board, which we previously estimated to be between 20% to 30%. This is a reliable guideline for both HBM and various logic and foundry designs. While some designs have significantly higher testing requirements and others lower, this variation is influenced by the customer yields for individual chiplets and the testing complexity needed for each one. HBM is particularly interesting because it allows for multiple test insertions. Each chiplet, similar to the MPU case, undergoes testing. Notably, in HBM, as the stack is assembled, there are several intermediate test insertions, some occurring at very high speeds to ensure the die will function correctly. This contributes to the overall 20% to 30% increase in testing, although there is considerable variability around that figure.
Vedvati Shrotre, Analyst
For my second question, building on your earlier responses, in foundry logic, you're competing with another company for each design. How do you determine if you are the market share leader for a specific design? Do you gain this understanding before the design begins to ramp up, or do you only realize it afterward?
Mike Slessor, CEO
No, you understand it before the design starts to ramp, because our customers don't really want to surprise us. Our lead times are relatively short. So you do have visibility into what designs you won and what designs that you have gone to your competitor. But that is not entirely predictive of the revenue because as we saw in the third quarter, customers can accelerate or decelerate their wafer start plans on that particular design essentially in real-time based on their customer demand. And so you can have won a majority of designs. But to put it simply, if you don't get lucky with the ones that are ramping, that might not result in a majority of the revenue-based market share but that swings both ways. Our teams are very focused on working with our customers to make sure we've got the maximum number of design wins. But as I said, picking the winners is something that's extremely difficult to do and probably impossible to do, to be honest, given the dynamics in the industry.
Operator, Operator
Our next question comes from the line of Tom Diffely from D.A. Davidson.
Tom Diffely, Analyst
Mike, just another question on the photonics business. So obviously, it's been good for your systems business over the last several quarters. But what does that look like in a couple of years when it goes from the lab to the fab, what kind of opportunity is that for you?
Mike Slessor, CEO
Yes. I think we've mentioned on the past several calls and highlighted again today the progress in silicon photonics and co-packaged optics. It has been a strong driver of our system segment revenues, somewhere around 10% to 20% of those system segment revenues as customers invest in early R&D. One of the key points we made in the prepared remarks is we're shipping really the first system for low volume production into one of the major foundries in the world so that they can serve their fabless customers with some of these co-packaged optics applications. When we look at that transition, it sort of depends on whose market forecast you believe in. If silicon photonics really is going to take off as a data center application in the next several years and chip-to-chip communication because of the power efficiency it offers, the power reduction it offers, we could see a sizable increase there. Now we've got some work to do, right? Our systems business is primarily focused on the lab. And as we bring this technology to the fab, together with the consumable probe cards that perform the electrical test together with the photonics test, there's a significant chunk of our R&D spend going forward to enable this. But I think it's one of the key areas where we see great engagement in early R&D, the transition to low volume production. And as we go from the 10% to 20% in R&D revenues, you could see that scale substantially, multiples of that if we get this right.
Tom Diffely, Analyst
And then, Mike, if you look at just the revenue guidance for the fourth quarter, it's somewhat similar to what we had a year ago. Just wondering if you could step back from a big picture point of view and just talk about the health of the industry today versus a year ago when revenues were projected to be the same?
Mike Slessor, CEO
Although the revenue levels are the same, the rate of change and the state of the industry are very different. Our third quarter revenue declined, and our fourth quarter guidance is similar to what we see this year, but that was off a very strong first half of 2022. We have been operating at those levels throughout 2023. There are some fluctuations, with HBM strengthening and some ups and downs in the foundry and logic business typical of different designs ramping at various times, but we see the industry in a much more stable condition overall. We believe we are in a bottoming phase. Listening to some key customers in the foundry and PC spaces, and with mobile earnings coming in the next couple of days, it appears that things are bottoming out and possibly strengthening a bit. Overall, the industry seems to be in much better shape. There are diverse opinions on when things will improve and when the next upturn will occur, but we feel good about our position given the capacity investments we have made to take advantage of that.
Operator, Operator
Our next question comes from the line of Gus Richard from Northland.
Gus Richard, Analyst
A lot of them had been answered but I did want to ask you about the expansion of the system and package, and is that starting to reach to mobile customers or are they starting to look to put multiple tiles in a package and increase test intensity there?
Mike Slessor, CEO
So we gave you a couple of examples of where chiplets or tiles are really driving our business. One, the client microprocessor, the other HBM, which has been a theme for a while. I think we're probably in very early innings associated with those. There are some R&D projects going on, but it's going to be at least a few quarters, if not the better part of the year before we start to see chiplets and tiles really infused into the mainstream mobile roadmaps. Having said that, there are a variety of places in high-end 5G mobile handsets where you do see advanced packaging and system in a package kind of characteristics. One is the RF front end where now we have multiple filters packaged together with an antenna and package. And those are areas that, at least when mobile was ramping and 5G adoption was strong, we saw a great uptick in the RF business. I think thematically, it all goes to us wanting to continue to broaden our exposure to advanced packaging. We view this as a secular shift in where value is created in the industry and then the supply chain and it's somewhere we're continuing to place bets, both from our R&D spending but also as we look to an M&A funnel.
Gus Richard, Analyst
And then just touching again on silicon photonics. You're shipping both system. Is that fully automated at this point? And can you talk a little bit about the mix going forward between systems and consumables as you ramp in volume?
Mike Slessor, CEO
We're still pretty early in the implementation into volume production. And so the system we're shipping in the quarter that goes into low volume production is still really an optical-only test. It's testing the optical part of the chip. Now we do have separate engagements on this combined when you co-package the photonic chip together with the electrical chip, at least to do functional tests and ensure that die works, it's going to need to be a combined electrical optical test. But that's in fairly early co-development with some of our key customers. So I'd expect, let's say, through 2024, the mix to continue to be pretty heavily systems-based.
Operator, Operator
Our next question comes from the line of Christian Schwab from Craig-Hallum.
Christian Schwab, Analyst
I just have a couple of quick questions. Is there any update on the large new customer opportunity that you could share?
Mike Slessor, CEO
I have a few updates to share. To provide some context, our market share in the logic segment, particularly regarding microprocessor applications, is significantly dependent on a single customer. Lately, the trends in the industry have not been very favorable for that customer. As we've mentioned before, our main strategy is to be a leading supplier to every major customer in the industry, and this is an area where we have identified a gap. We're focused on it and have made some changes in our technology and the technical team that works with this customer. The main issues we've faced in qualifying have been related to adapting our technology, which are technical challenges. This customer requires an additional supplier. Currently, there are only two companies that can effectively provide high-end foundry and logic probe cards, particularly given the demands of advanced packaging. We've been making changes for longer than we would have liked, but we remain committed to increasing our share with this customer.
Christian Schwab, Analyst
The second question is about the high bandwidth memory industry, which is expected to more than double its capacity. Samsung is finally producing it, and Micron is likely to improve soon. Is it correct to say, based on industry discussions, that your high bandwidth memory business could increase from $10 million a quarter to over $20 million a quarter at some point next year?
Mike Slessor, CEO
Typically, probe card intensity scales with wafer starts and output. Over time, there has been a solid relationship between overall semiconductor revenue and probe card spending, although intensity has increased. One potential concern is that as any technology matures, such as TSV die stacking in HBM, yields improve, which has been a counterbalance for us. However, as all three DRAM manufacturers ramp up production to meet the native demand for HBM in supporting AI, it's reasonable to expect our business could grow from approximately $10 million a quarter to around $20 million a quarter. This growth will depend on when these other customers increase their production and the market demand from AI. If those assumptions hold true, it seems feasible that we could nearly double our HBM business sometime in 2024.
Operator, Operator
And our next question comes from the line of Craig Ellis from B. Riley Securities.
Craig Ellis, Analyst
Mike, I wanted to start just by following up on some of the conversation around foundry logic. It seemed like a lot of the strength in the quarter was compute related either more on the client side or HPC side and some similar things going on in the outlook, but I didn't hear too much about mobile and things related to RF and foundry logic. So how do you think that shapes up as we exit your heading into next year and any notable changes on the mobile side in foundry logic as we're looking at the business?
Mike Slessor, CEO
I think our RF business, which we report inside of foundry and logic really is indexed pretty heavily towards mobile handset growth. And I think anybody who follows that industry knows that that's been a pretty ugly 2023 for handsets and the mobile ecosystem in general. We do see things bottoming in the mobile space. And there were some nice application processor wins and RF wins in the third quarter that did contribute to the results. They just weren't as major as the compute designs. I still feel like if you look at inventories, whether they be filters or even some of the app processors across the overall channel, there's some digestion, some usage for those chips before we really see the new chips that are driving our business to ramp in volume. But like most of the rest of our businesses, we've seen mobile and RF stabilize it, as I said, maybe even get a little bit better. Now one of the things to know about our RF business is some of the shortest lead times we operate in. So where our average lead times are better part of a quarter, RF lead times are actually shorter than that. And so we can see some very significant changes in the demand profile as customers change their wafer start plans. So not a lot of visibility, I guess, is one of the things I'm trying to say.
Craig Ellis, Analyst
And then the second question earlier in the Q&A, you noted that there have been pluses and minuses vis-a-vis the target model. One of the pluses would be AI, one of the minuses would be China. And I was just looking at the China country trends in the supplement that's released and noticed that China had ticked a few million in lower every quarter over the last four or five quarters. How much of that is cyclical versus anything that might be related to more structural things with shipment restrictions, et cetera? And more importantly, from here, what should we expect going forward in that geography?
Mike Slessor, CEO
So it's a good point, right? Our China business, I think, was again down sequentially, and probably at half the levels it was at the peak back in late '21 or early '22. To answer your question directly, it's almost entirely structural. As a US supplier operating at advanced nodes that are subject of export controls from the Department of Commerce, we've been limited in our ability to supply customers. And I think we've been quite clear in our expectations that we expect the domestic China business to continue to increase and at some point here probably go to zero as they cultivate probably not local China alternatives to us, but with our primary competitor in memory being based out of Japan, our primary foundry and logic competitor being based in the EU, they do have geographic alternatives that aren't restricted as strongly as we are. The part holding that business up is still the multinationals that operate in the region. And we've seen a recent relaxation of the export controls around them, they were granted extensions to the license, and those businesses are holding up fairly well.
Operator, Operator
Our next question comes from the line of David Duley from Steelhead Securities.
David Duley, Analyst
My question has to do with advanced packaging. Clearly, on the conference calls over the last few quarters, we've seen a whole bunch more chatter from both the test and assembly guys, Intel and AMD and whatnot. I'm just curious as far as more probe card business goes, what percentage would you classify as advanced packaging and what do you think the growth rate of that segment is? And I'm assuming that includes both the tile, the high bandwidth memory stuff?
Mike Slessor, CEO
And that's the way we classify it, too, David, for clarity, right? Regular flip chip, although, some people would call that advanced packaging, we're really targeting that classification at chiplet multi-die architectures assembled using techniques like hybrid bonding and TSV, because that's really where the increase in performance, the decrease in cost, the innovation acceleration comes from. It can sometimes be a little bit difficult to break out. But given what we highlighted in the third quarter, my guess is that somewhere around 20%, 25%, maybe even 30% of our business at present is associated with advanced packaging. The combination of the client-based microprocessor, HBM at the various customers and a smattering of other designs. So we're still clearly in very early innings associated with this. But if you look at the increase in test intensity associated with it, it's also one of the things that is underlying the strength in the third quarter results. The fact that some of the revenue concentration was associated with things in advanced packaging really does help out the overall top line. And that's why we're so optimistic about our exposure and the eventual growth advanced packaging will drive for us.
David Duley, Analyst
And when you look at both the unit volume growth that you see in advanced packaging plus this increased intensity and increased complexity, or more advanced packaging business over the next two years, what would you guess the growth rate is? Clearly, it's probably more than 2x the industry growth. But I'm just kind of curious if you have an idea of what it might be?
Mike Slessor, CEO
So we're going through annual planning and strategic planning right now, and it's one of the key things we're looking into. So over a two-year time frame, we do expect it to outgrow the industry significantly. I think 2x may be a little bit aggressive. But again, it's one of the reasons, given our competitive position, given the value we create for our customers in advanced packaging applications with things like our MEMS probe technology and co-packaged optics, that outsized growth is somewhere where we've got good exposure, and we're going to continue to prioritize and capitalize on.
David Silver, Analyst
And just one more clarification. As far as sub 7-nanometer and below foundry logic business, what percentage of those wafer starts are MEMS based probes versus other technology? Is it a vast majority at this point?
Mike Slessor, CEO
Yes, it's a vast majority of MEMS probes. There are counter examples where MEMS probes have not been adopted in areas like very large footprint, single die GPUs, for example, because the interconnect density doesn't require a MEMS-based probe, but it's a pretty reasonable rule of thumb that the vast majority of sub 7-nanometer foundry and logic devices do need to be probed with the MEMS probe cards.
Operator, Operator
And our next question comes from the line of David Duley from Steelhead Securities.
David Duley, Analyst
I have a couple of questions regarding planning for next year, as this is usually the time for budgeting. Specifically, I would like to focus on your CapEx budget. There has been the Livermore capacity expansion, but when considering next year's CapEx spending, could you identify where the largest portion of discretionary CapEx will be allocated? In other words, what initiatives or projects will you be directing discretionary resources towards?
Shai Shahar, CFO
So as you said, we are going through the planning process. And given our big footprint in Livermore in California, I expect most of the additional CapEx to happen in that facility. In terms of magnitude, I did talk about completing the majority of our capacity expansions that are required for achieving the target model, and that's where we expect the CapEx in 2024 to decrease certainly compared to the $60 million midpoint CapEx we expect in 2023. I don't expect it to go down significantly like to the $20 million a year levels that we had before starting this capacity expansion, but somewhere in that range is what we expect CapEx to be in 2024, and most of it in the probe cards business and most of it in Livermore.
David Silver, Analyst
And then I'd kind of like to ask a similar question on your R&D spend. So I think the third quarter R&D spend was an all-time high. And again, when you look out to next year, what are the new or significantly increased efforts in your R&D budget, if you could call out maybe the top one or two, that would be very helpful?
Mike Slessor, CEO
Really go back to the themes we've talked about today, advanced packaging, chiplets, making sure our product roadmap is aligned with the key customers who are adopting advanced packaging and chiplets in volume and making sure we're focusing our R&D spend to have differentiated products in those applications are really the focus areas. So in DRAM, making sure we're keeping our competitive lead in HBM. But in foundry and logic, there are a variety of new requirements and new products that we're releasing associated with higher speed, higher power, smaller pads, and higher pin counts that are really the focus of our envisioned R&D spend, I won't call it planned yet in 2024.
Operator, Operator
Thank you. This does conclude the question-and-answer session of today's program. I'd like to hand the program back to Mike Slessor for any further remarks.
Mike Slessor, CEO
Thanks, everybody, for joining us again today. We have a couple of investor conferences on our docket to end the year. I believe they've been posted on our website. So we'll hope to see you there to wrap up an eventful 2023 and look forward to 2024. Take care.
Operator, Operator
Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.