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.
spk00: Greetings. Welcome to Taradine's fourth quarter 2023 earnings column webcast. At this time, all participants are in listen-only mode. A question and answer session will follow the formal presentation. If anyone should require operator assistance during the conference, please press star zero from your telephone keypad. Please note this conference is being recorded. At this time, I'll now turn the conference over to Andy Blanchard, Vice President, Corporate Communications. Mr. Blanchard, you may begin.
spk14: Good morning, everyone, and welcome to our discussion of Terranite's most recent financial results. I'm joined this morning by our CEO, Greg Smith, and our CFO, Sanjay Mehta. Following our opening remarks, we'll provide details of our performance for 2023's fourth quarter and full year, along with our outlook for the first quarter of 2024. The press release containing our fourth quarter results was issued last evening. We're providing slides on the investor page of the website that may be helpful to you in following the discussions. Replays of this call will be available via the same page after the call ends. The matters that we discussed today will include forward-looking statements that involve risk factors that could cause TerraDyn's results to differ materially from management's current expectations. We encourage you to review the safe harbor statement contained in the earnings release, as well as our most recent SEC filings. And additionally, those forward-looking statements are made as of today, and we take no obligation to update them as a result of developments occurring after this call. During today's call, we'll make reference to non-GAAP financial measures. We've posted additional information concerning these non-GAAP financial measures, including reconciliation to the most directly comparable GAAP measure, where available on the investor page of our website. Looking ahead between now and our next earnings call, Ternine expects to participate in technology or industrial-focused investor conferences hosted by Wolf Research, Citi, Susquehanna, and Morgan Stanley. Now let's get on with the rest of the agenda. First, Greg will comment on our recent results and the market conditions as we enter the new year. Sanjay will then offer more details on our quarterly results, along with our guidance for the first quarter. We'll then answer your questions, and this call is scheduled for one hour. Greg?
spk03: Hello, everyone, and thanks for joining us this morning. Today, I will summarize our fourth quarter and full year 2023 results, comment on our early view of 2024, and describe the market assumptions underpinning our updated midterm earnings model. Sanjay will then provide financial details on all of these topics. We delivered Q4 financial results in line with our guidance. The clear highlights for the quarter were memory test, where DRAM tester revenue more than doubled from Q4 2022 on HBM demand, and our robotics team's execution in growing sales 17% from Q4 2022, and 50% sequentially as we ramped shipments to address the record backlog of our new UR20 cobot at Universal Robots. Balancing these strengths in memory and robotics was continued softness in our other test markets. Shifting now to the full year, 2023 was the second consecutive year that the size of the semiconductor test market declined as the industry digested record shipments in 2020 and 2021 driven by COVID related demand. The SOC test market has declined 21% from 2021 and the mobility portion has contracted 55% over the same period. Most of that decline occurred in 2023 with SOC demand down 17% year on year. Mobility has historically been the largest sub-segment of the SOC market and an area of strength for us in semiconductor test and in wireless test. Despite the weakness in demand, the trend towards vertically integrated producers is continuing. This new class of customer provides an opportunity for Teradyne to gain share in high-performance computing, a segment where we have historically had low share. These customers will take time to ramp, and so we are focusing on capturing sockets. On that front, 2023 was a great year for us, capturing the majority of key VIP sockets where we targeted at customers in the United States and in China. Although automotive was strong for most of 2023, at the end of the year, elevated inventories for our automotive customers began to slow down the rate at which they add test capacity. Our latest estimate of the semi-test market for 2023 is about $4.8 billion, with $3.9 billion in SOC and $900 million in memory tests. Our 2023 financial results reflected the test market weakness, and when combined with a tepid robotics market, company sales were $2.676 billion, down about 15%, and non-GAAP earnings of $2.93 were down 31% from 2022. As we enter 2024, let me describe how we're looking at the current market. Visibility is limited, but we are cautiously optimistic. In SOC test, chip inventories remain high, and sub-con tester utilization remains in the 70s. These combine to create a headwind to tester demand in the first half of 2024. The inventory overhang now extends to parts of the legacy automotive market, which began to soften late in Q4. With this softening of demand, lead times for our SOC testers are now back to pre-COVID levels, generally less than 16 weeks. This enables customers to wait to place new orders until their demand is certain, limiting our visibility. Our current view is that these headwinds will abate mid-year, and we are planning for a stronger second half in SOC test. Several leading indicators point to a second half improvement. For example, We've seen no slowdown in chip development activity at our customers, so the product pipeline remains healthy. While still below normal seasonality, utilization levels are beginning to inch upwards, especially in the OSAPs, which points towards improving mobility and compute demand. Mobile phones are expected to adopt AI capabilities in the premium tier in 2024, and more broadly in 2025, which should accelerate complexity and be a positive for test demand. Despite the slowdown we are seeing now in automotive demand, the key driver for this market is the increased electronics content per vehicle, not end vehicle sales. Automotive semi-devices are forecast to grow an 11% CAGR through the midterm, and this and device complexity is increasing. Therefore, we expect this lull will be short-lived. In memory test, the story for 2023 is that despite high-end market inventories, demand for new testers was driven by the retooling required to test higher speed flash and DRAM devices, especially HBM DRAM. This drove our memory share to a record 43% in 2023. In 2024, We expect continued memory revenue growth driven by the volume production ramps of the technology introductions we saw in 2023 and continued R&D investments for even faster devices. As business conditions improve for memory makers, this may drive increased demand to support capacity expansion. Shifting now to other test markets. Wireless test demand is expected to remain at current levels in 2024 due to weakness in the networking equipment demand and excess capacity in smartphone test. In system test, we expect continued strength in defense and aerospace markets and expect modest growth in demand in production board test. We expect storage test will remain weak in 2024 due to excess test capacity in the HDD and SLT test markets. Having said that, in 2023, we added important new SLT customers in mobility and high performance computing that are setting us up for midterm growth. Shifting to robotics, we had a very strong Q4 as our UR20 ramp continued and we launched the new UR30 late in the quarter. Looking at robotics for the full year 2023, in addition to the good progress on the new product front, our channel transformation work continued nicely with our OEM sales growing nearly 10% as we added more than 50 new OEM partners. We've also expanded the number of large accounts we manage directly from approximately 100 to over 250. We expect Q1 of 2024 to be down more than the normal seasonal dip as the extraordinary Q4 shipments are digested. Thereafter, We expect quarterly year-on-year growth through the rest of 2024. Combining all these points and with the proviso that our view into the second half is limited, we're modeling 2024 company revenue to be roughly flat year-on-year with 44% in the first half and 56% in the second half. Within that, we expect lower test revenue in 2024, which reflects the sale of our DIS business, reducing our full-year semiconductor test revenue by approximately $100 million. Excluding the effects of that sale, expected test revenue would have been about flat in a roughly flat market. In semi-test, our early estimate of the SOC market size for 2024 is $3.6 to $4.2 billion. And our estimate for memory is $1 to $1.1 billion for a combined semiconductor test market at the midpoint of $4.95 billion. We expect robotics will grow in the 10% to 20% range in 2024. Turning now to our midterm model. Despite the longer than expected downturn in the mobile test market and the softness in robotics and demand, we remain optimistic about the long-term potential of our test and robotics businesses. This is shown in the update to our mid-term earnings model. At the midpoint in 2026, we expect to deliver earnings per share at over two times the 2023 level and a revenue growth of nearly 60%. As we've noted in past calls, the key drivers of that growth include process node advances to three nanometer, two nanometer, gate all around, and backside power delivery. These are all on track or even accelerating, and they enable higher transistor counts, higher complexity, and that drives longer test times. Good progress in the emerging VIP space with key wins at design and targets and high share at two of the three leading ASIC design houses, all of which drive future revenue. Advanced packaging, including chiplet technology, which requires higher test intensity at the wafer level, driving longer test times. Compelling applications of Edge AI for ADAS and smartphone co-pilots that are driving demand for more processing power, more memory, and wider bandwidth communications. All of these factors accelerate test demand. There's an aggressive roadmap for increases in memory interface speeds in DRAM, hbm and flash that will continue to drive technology-driven retooling in the memory market and finally in robotics we are still at less than five percent market penetration and we are confident that our channel strategy will unlock the long-term growth potential of this market in addition the application of ai is expanding the range of tasks that our robots can serve while our new products will expand our served market and decrease the effort required for customers to automate. At the company level, compared to last year, our growth outlook has shifted to the right, but the slope of expected growth is largely unchanged. The duration of the downturn in mobile demand has been longer than we expected, and the softness in the industrial automation market that we and our peers have seen has really impacted our expectations of growth in robotics for 2023. Sanjay will provide more financial details of the model. As we look at our results for 2023 and the outlook for 2024, we are focused on improving gross margins and maintaining tight financial discipline while making the necessary R&D and customer-facing investments required to capture the long-term growth potential in both the test and robotics markets. To maintain that financial discipline, we will be looking to see signs of top-line growth before allowing OPEX to materially increase. We operate our business with midterm plans that track long-term historical trends and the future demand drivers in each of our businesses rather than trying to predict short-term cycles. In any given year, results will land above or below that trend but that trend line has provided a reliable baseline for planning. As expected, 2023 was below trend line, but the underlying demand drivers remain in place, and we're executing our plans to capture that future demand. We're excited about the opportunities ahead, and we have deep confidence in our team's ability to capture those opportunities. With that, I'll turn things over to Sanjay. Sanjay?
spk09: Thank you, Greg. Good morning, everyone. Today I'll cover the financial summary of Q4, the full year 2023, provide our Q1 outlook, some planning guidance for full year 2024, review our updated earnings model, and outline our capital allocation plan. Now to Q4. Fourth quarter sales were $671 million, with non-GAAP EPS of 79 cents, both in line with our guidance. Semi-test revenue of $431 million had SOC revenue of $327 million with high memory shipment of $104 million, as Greg noted in his highlights. System test group had revenue of $86 million, down 14% from Q4-22. Growth in defense and aerospace was offset by weakness in storage test and production board test. White Point revenue of $25 million was weaker year over year due to continued weakness in cellular and PC markets. Robotics revenue of $129 million was up 17% from fourth quarter of last year and up 50% sequentially on seasonally high demand and the impact of UR20 and UR30 new product shipments. Non-GAAP gross margin was 56.6% in the middle of our guidance range, and non-GAAP operating expenses were $245 million in Q4, up $2 million from Q3 23. Non-GAAP operating profit rate was 20.1%. Some other financial facts. We had one 10% customer in the quarter. Tax rate, excluding discrete items for the quarter, was 12.6% on a non-GAAP basis, and lower than planned because of geographic mix. Gap tax rate was 14% in Q4, excluding discrete items. We repurchased $51 million of shares in the quarter. Turning to the full year results, we had revenue of $2.676 billion. Texas Instruments was the only customer greater than 10% of our revenue for the year. Gross margin for the year was 57.4%. OpEx was $990 million, and operating profit was 20.4%. Non-GAAP EPS was $2.93. We generated $426 million in free cash flow in 2023. We returned $468 million, or 110% of free cash flow to our shareholders through share repurchases and dividends. We ended the year with $937 million of cash and marketable securities. Our tax rate for the full year, excluding discrete items, was 15.5% on a non-GAAP basis and 15.25% on a GAAP basis. Semi-test revenue for the year was approximately $1.8 billion, with SOC revenue contributing $1.43 billion and memory $386 million. Our SOC sales contracted 16% in 2023, in line with a continued mobility market weakness partially offset by strength in auto and industrial. In memory, our sales grew about 4% driven by new technologies like UFS4 and Flash and LPDDR5 and HBM and DRAM. System Test Group had revenue of $338 million in 2023. Combined defense and aerospace and production board test sales were both flat, while HDD and system level tests in our storage business were down in total nearly 50% from 2022. In wireless tests, revenue of $144 million in 2023 was lower year on year, given the decline in handset units and continued weakness in the networking and PC markets. Now to robotics. Robotics revenue in 2023 was $375 million, with UR contributing $304 million and MIR $71 million. The group had mid-teens profitability in Q4-23, but for the full year had an 8% non-gap operating loss. While UR was profitable in 2023, we continued to lean into R&D and channel enhancement investments in MIR. Now to our outlook for Q1. Since our October call, SOC test demand has softened, which impacts our Q1 outlook. Q1 sales are expected to be between $540 and $590 million, with non-GAAP EPS in a range of $0.22 to $0.38 on 162 million diluted shares. The first quarter guidance excludes the amortization of acquired intangibles. First quarter gross margins are expected to be the low point for the year and are estimated at 53.5% to 54.5% due to lower volume and unfavorable product mix. Oplex is expected to be roughly flat with Q423 and run at 42% to 46% of first quarter sales. Non-GAAP operating profit rate at the midpoint of our first quarter guidance is 10%. A few points to assist you in modeling 2024. First, the gross margin profile. In 2024, we expect gross margins to increase through the year. Gross margins are forecasted to improve to 58% to 59% for the full year, driven by improved product mix, including the sale of the device interface solutions or DIS business, lower business resiliency spending, and operational improvement. Moving to the strategic partnership we announced with Technoprobe in November, 2023. The agreement has several key components as follows. First, Technoprobe will purchase Teradyne's DIS business. This business provides advanced interfaces that connect our testers to customers' chips for test. Second, Teradyne will make an equity investment in Technoprobe, acquiring 10% of their outstanding shares. Third, Teradyne and Technoprobe will work together on a series of projects to expand the performance of semiconductor device interfaces to enable customers to realize the full performance of our test systems. We expect the transactions to close in the second quarter. DIS had 2023 revenue of $103 million, and the sale will reduce Teradyne's expected 2024 revenue by approximately $100 million, assuming the Q2 close date. It will not have a material impact on earnings. We plan to account for our investment in TechnoPro using the equity accounting method. Regarding OpEx for the full year, we expect full year 2024 OpEx to increase 5% to 7% driven by strategic projects to grow share in our test businesses. The growth will be back half loaded. Robotics is expected to grow revenue 10% to 20% in 2024, enabling robotics to be profitable. The group will have a similar revenue profile as the company with second half expected at 56% of full year revenue. Our gap and non-gap tax rate is forecasted to be 16% in 2024, excluding discrete items. Turning to capital allocations. Our strategy remains consistent as we take a balanced approach to maintain a minimum cash level of $800 million, which enables us to run the business, have cash reserves set aside in the event of a significant downturn, and have dry powder for M&A. For reference, from 2015 to 2023, we've returned over $4.3 billion to shareholders through share repurchases and dividends, which is 97% of our free cash flow. Earlier this month, we also announced a 9% increase in our dividend to $0.12 per quarter. We expect to close our investment in Technoprobe and our divestiture of DIS in Q2, which will consume an estimated $440 million of net cash. We will limit our share buybacks in 2024 to an amount necessary to offset dilution from equity compensation and our employee share purchase program in order to build cash backup to a minimum goal of $800 million. Therefore, we do not expect to materially reduce the share count in 2024. Moving to our midterm earnings model. As we do each January, we've updated our model. We share this with investors to provide insight into how we look at the markets we serve, our competitive positioning, and ultimately, the growth and earnings power of the company. A few points for context. First, we've kept 2026 as the year end so you can make an easy comparison with last year's update. Second, financial ranges are down to reflect our revised view of the markets in which we participate. Greg outlined the key drivers for the test portfolio, including device technology trends, complexity, and unit growth, which we anticipate will drive ATE growth over the three-year horizon. Our test revenues are expected to grow at a CAGR of 12% to 18% from 2023's muted level to 2026, driven primarily by a recovery in mobility test demand. For robotics, we're looking at a strong market opportunity. The drivers over the middle include labor shortages, new products and applications, including a growing range of AI-driven products and our channel transformation work. Our robotics revenues are expected to grow at 20 to 30 percent CAGR from 2023 to 2026. The updated earnings model will drive 2026 revenue to approximately $4.3 billion and non-GAAP EPS to $6.50 at the midpoint of the model range. Our model at the midpoint has a revenue CAGR from 2023 to 2026 of 17% and on EPS at 30%, highlighting the operating leverage of both robotics and test portfolios. Gross margin is estimated at 59% to 60%. OpEx as a percentage of sales will be 28% to 31%, yielding a non-GAAP operating margin of 28% to 32%. Summing up, we had a challenging year in 2023 and still delivered 20% operating profit on a non-GAAP basis, generated $426 million in free cash flow, and returned 110% of that cash to shareholders. We did this while we strengthened our competitive position in test and robotics with new products, invested in channel transformation in robotics, and announced a strategic partnership in test interfaces to drive semi-test share growth. Looking ahead, we expect a return to growth later this year and beyond, driven by favorable trends in both test and robotics, and the financial impact of that growth is reflected in our updated midterm earnings model. With that, I'll turn the call back to Andy. Andy?
spk02: Thanks, Isaac. We'd now like to take some questions, and as a reminder, please limit yourself to one question and a follow-up.
spk00: Thank you. If you'd like to ask a question today, please press star 1 on your telephone keypad and a confirmation tone will indicate your line is in the question queue. You may press star 2 if you'd like to remove your question from the queue. For participants using speaker equipment, it may be necessary to pick up your handset before pressing the star keys. One moment please while we poll for questions. Thank you. Thank you. Our first question will be coming from the line of Mehdi Hussein with SIG. Please proceed with your questions.
spk10: Yes, thanks for taking my question. The first one has to do with the midterm guide. It seems like there's about a 13% decline in your SOC TAM for 2026. What I want to better understand is how do you think about the mobile processor and especially the transistor density versus a slow start in industrial, how should we think about those two end markets impacting the 13% reduction to your 2026 TAM for SOC? And I have a follow-up.
spk04: Hi, Mehdi. This is Greg. So the composition of the TAM in the midterm model, if you compare this year to last year, I think the biggest change is that We think that the compute part of the market is going to be at a higher level than we did last year. And we think that the mobile portion of the market is going to be up from where it is now, but will not be at the level that we were modeling last year. So we believe that the pace of transistor density is proceeding basically at the same rate that we thought it was proceeding last year. The real difference is that the end market, you know, sort of end unit market predictions going out through the midterm are a bit softer in this model than they were last time. And, you know, certainly the impact of cloud and edge AI in the market is a relatively new factor and is strengthening the compute part of it.
spk10: Okay, great. And the follow-up has to do with your prepared remarks, I believe you said you have a new compute customer. And if that's true, if I heard you correctly, how does the ramp of that new customer would look like? Is it more like a 25, 26 impact? And if you could just provide any additional color, it would be great.
spk04: So there are sort of two comments in my prepared remarks. One talked about socket wins for VIPs. And those socket wins are in both cloud compute and in edge devices. And the right way to model that is perhaps a little bit starting towards the end of 2024, but definitely hitting much more seriously in 2025. The other reference that I made was for an SLT win for a compute customer. And that's definitely something that's going to take, um, until 2025 to build to a material volume.
spk10: Okay, great. Perhaps I'll follow up with Andy about what VIP is.
spk04: Oh, that's sorry. We, um, uh, because most of the, um, there are a number of players that are designing their own Silicon. Um, and, You know, typically a big portion of those are the hyperscalers, but there are players in that space that don't fit the traditional hyperscaler model. So, you know, like a large integrated automaker or people that are doing custom silicon development for their own handsets or customer devices. So we prefer the term vertically integrated producer, but you can read that as hyperscaler.
spk00: Okay, thank you for coming. I appreciate it. Our next questions are from the line of Timothy Akuri with UBS. Please proceed with your questions.
spk13: Thanks a lot. I'm wondering if you can go through the TAM numbers, how the different segments came in in 23. I know the breakout before was Compute 1.3. Mobility was 900. Autos were, I think, a little more than 600. If you can just, did the market come in about as you thought it would for 23? Can you shape 24 for those segments? I know it's flat at a high level, but I would imagine autos are probably down a couple hundred million and maybe computes up a couple hundred million. That's the delta. Can you just run through that?
spk09: Sure. Hi, Tim. It's Sanjay. So we estimate the SOC TAM in 2023. You were going down the correct numbers that haven't really changed, but just for specifics, we view the compute market at $1.3 billion. mobility at 900 million, auto at 600 million, industrial at 400 million, and service at 700 million. That's how you get to the 3.9 billion. And then the midpoint of 2024, compute is 1.4 billion, mobility at 0.9 billion, auto and MCU at 0.5 billion, industrial at 0.4 billion, and service, SLC service at 0.7 billion to get to 3.9 billion at our midpoint.
spk13: Got it. Okay. So, so it's, it is compute. It's just a little bit of a mix shift. Okay. Um, and then can you talk about your largest customer, um, your big customer on the mobility side, which, which was actually not your largest customer in 2023? Um, I know you have a range of outcomes this year. Um, I'm not asking for you know numbers on that range but can you talk about sort of what you think at the midpoint you know how big they could be as a person is there a case where they're back at percent of your you know revenue range in 2024 or is that just off the table so um hi this is Greg you know our historically largest customer
spk04: We've said that like in 2023, they are going to come in at under 10%. Looking forward into 2024, our base case again has them at less than 10% of total revenue. Right now, our visibility into their second half demand is quite limited. And we usually have a better picture of that in the April timeframe. So if we get any better information about it, we'll share it at that time. But the thing that you should think about is that there isn't much downside to our base case around anything that happens with our historically largest customers.
spk13: Got it? Okay. Thank you, Greg.
spk00: Our next question is from the line of CJ Muse with Canterford Sherald. Please proceed with your questions.
spk12: Yeah, good morning. Thank you for taking the question. You know, in your prepared remarks, you talked about confidence in the second half recovery for SSC tests. So I was hoping you could walk through, you know, what's giving you the confidence and how you see that recovery emerging. And perhaps you could speak to, you know, OSAT versus end customers, content, edge AI, you know, whatever you think is really going to move the needle for you guys and drive that second half recovery over the first half.
spk04: Sure. So the leading indicator that we tend to look at most is really around utilization rates. And right now, those utilization rates are low, but we did start to see OSAT utilization start to pick up. So they're still below normal seasonality, but they're definitely improving. And we believe that they could reach sort of the point the level that would trigger buying by the third quarter of this year if the trend that we're seeing continues. The key end market trend that we're looking for is unit growth in smartphones and PCs. PCs have been in a real pit. And so just from a refresh cycle, we're expecting to see that recovering a bit, and that would be a tailwind for us. The thing that we also saw in the fourth quarter was that utilization came down a bit in IDMs. And so primarily, that's where most of our businesses concentrated for industrial and automotive. So we saw the industrial part of that start to weaken in the second half of 2023. And then late in Q4 of 2023, we also started to see automotive begin to weaken. Our model right now has that, that's a relatively short lull in demand. And we think that that's mainly driven by the fact that new model year introductions every year for the automakers has a significantly higher attach rate for electronics. And that's driving a pretty good tagger in the automotive semiconductor market. So we think the fundamentals are good to drive growth in automotive. So we think that this demand lull is going to be relatively short.
spk12: Very helpful. And if I could follow up on the SLP side. You talked about wins across both mobility and HPC. I know you have one very large customer. Would be curious, you know, the timing of the ramp of revenues here and, you know, could they in aggregate get to kind of the scale of your very large customer or or that will take, you know, years to do that.
spk04: So in the prepared marks, I was referring to system level test wins for mobility and for compute. And if you put that into context, the customer that we won for mobility is not as large as our primary SLP customer. but they have the potential to drive significant volume. In compute, that's something that's going to take a while to take off. The, you know, the device plans, basically for SLP, it was a hyperscaler customer in compute, and the volumes for those devices is going to start at a relatively small level and then increase over time. So I don't think that any of those in isolation would rise to the level of our historically largest customer, but it's more additional incremental tailwinds for us. Great. Thank you.
spk00: Our next questions are from the line of Vivek Arya with Bank of America Securities. Please proceed with your questions.
spk05: Thanks for taking my question, and I appreciate you giving the compute time, which is now the largest market. Could you help us break that between more client compute versus data center, and then the growth rate that you're mentioning, the 1.3 to 1.4, that seems very modest compared to all the very high growth rates that we see in AI and data center type products. If you could just give us a little more color, you know, how much is client versus data center, and then why only such modest growth in testing? Why don't we see more of a correlation with the growth in data-centric products?
spk04: Hi, Vivek. This is Greg. So first of all, let me talk a little bit about client versus data center. As far as we can tell, in 2023, Most of the test capacity ads in compute were really in support of data center. The client market has been very, very quiet. Looking forward into the future, we think that the capacity ads required for cloud are going to moderate a bit. They've added a ton of productive capacity in 2023, and that is going to be felt in higher end volumes in 2024. So they're at sort of a new high watermark in terms of that part of the market. The client side, there are two things that I think are going to accelerate things. The first is that The unit volume in PCs is at a very low point right now. There's really nowhere to go but up. And the other is that we've been seeing the push in large language models and AI primarily on the training side right now. And looking at the roadmaps for our customers, they're putting a huge push on custom silicon for edge AI inference versus the training part. So I think we're going to see a shift over the next three years to a much more balanced market, whereas 2023 was dominated by data center. As we get out through this midterm, it's going to be much more balanced between data center and client. Now, in terms of the modest size, the thing that I want to just, I want to try and provide a little bit of context that if you look back to the high watermark for the total SOC TAM back in 2021, it was about a $4.9 billion market. And in that $4.9 billion market, less than $1.2 billion of it was in compute. Now, fast-forwarding to this year, the overall TAM is down to 3.9. And Compute is representing 1.3 billion of that total. Now, if you look into 2024, that's a very high level that it's at right now. And that level supports significantly increased production for these devices. The last thing that I'll point out is these devices are big and complex, and they have a high test intensity, but the unit volumes are relatively small, and the margin in the end market for these devices is very high. So the revenue that compute makers are getting on these devices is pretty great. But it means that the number of testers required to fulfill that revenue demand is a little bit lower than it would be in the mobile space.
spk05: And for my follow-up, I actually had two small ones. I don't know whether they're related or not. But there was some media reports about, you know, Teradyne kind of pulling out of the China market from just a manufacturing footprint. I just wanted to see if there was any clarification. around that. But the bigger question, Greg, that I have is when I look at this midterm model and, you know, the forecast of growing, right, kind of low mid-20s and 25 and 26, if I look at the last decade, the only year Teradyne grew above that rate was just like one of the years and that too was 2020. So why keep such a high forecast and that too for two two years, you know, what is the kind of the visibility and confidence in hitting those kind of growth rates?
spk04: Yeah. So first, just a quick comment on China. The news report is really nothing new. We've had a multi-year effort going on to increase the resilience of our supply chain. that involved moving production locations for some of our products um we still have an expense an extensive supply chain that is in china we have over 600 people in china we're competing for business and winning in china um and you know i i think that uh uh it's the it's a good headline, but there's nothing new in that story or in our commitment to that region as a place where we expect to grow. So, on the midterm model, if you look at the, you know, if you look with a longer lens than you're talking about, you know, we've been in this market for 60 years, and we've seen the dynamics of the market as it comes out of downturns. You know, so in, you know, coming out of 2001, coming out of 2009, and then, again, the COVID-related demand increase in 2020, it's not that unprecedented to see significant growth in the years following a downturn. And looking at the state of the end market for mobile phones, and the end market for PCs, we really expect to see a pretty good snapback. The other X factor that I think we are becoming increasingly confident in is how generative AI is going to impact complexity of devices. So we were really kind of wondering what was going to be the next thing that drove complexity at the enhanced sets in cars at the edge and it appears that the technologies incorporated in these large language models and the amount of compute that they need to execute is something that is a very positive tailwind both in the mobile and the compute part of the market one final reminder is as you look at this when you think about edge ai A lot of that is going to accrue to growth in the mobile part of the market and the automotive part of the market, not in the compute part of the market. Thank you, Greg.
spk00: Thank you. Our next questions are from the line of Krish Sankar with TD Canwin. Pleased to see you with your questions.
spk11: Thanks for taking my question. The first one is for Greg or Sanjay. When I look at your Q1 guidance compared to the last time we were at that revenue run rate, Your earnings power is almost half of that. I'm just wondering, is there a function of maybe lower gross margin due to product mix, like less auto-analog testers? Or is it because of higher investments in IA, a combination? I'm just kind of curious. And then a follow-up to that.
spk09: Sure. So I think it's really fundamentally two things in Q1. One is much lower volume, but then really product mix. say product mix is a little bit of a bigger driver. Nothing more than that. It's product mix and volume.
spk11: Got it. Got it. Kind of like a more philosophical question for Greg. You know, Greg, clearly, obviously, on the mobile test side, you have a good position. And then the last several years, it seems like your focus was on IA. And then now you've got this big investment in Technoprobe, which kind of came out of the blue. And historically like vertical integration between test and probe cards has never worked. I'm just kind of curious, is it because of the last few years you took your eye off the test market that now you're beginning to see a shift away from mobile to compute and you need to do certain things or is there something else going on?
spk04: So I think there's no real change in our strategic priorities. And so our strategic priorities has always been to look for accretive investments and opportunities to put free cash flow to work for our investors in the most effective way. We didn't take our eye off the ball on tests when we were talking about our industrial automation business. And as a matter of fact, over the period of time that you're talking about, the industrial automation business was accretive, that we were making money and that IA business had a rough year in 2023, mainly because of end market conditions. So we haven't changed our strategy. The thing that's different and the thing that really motivated the investment in Technoprobe is that there is a friend in the test market that is drawing the tester and the interface closer together. And that's really driven by the complexity and performance in the end market. So if you look at the data rates that are required for SOC and memory devices, if you look at the bandwidths required for RF devices, if you look at the number of devices that customers want to test in parallel all of those are driving a tighter integration between tester and interface. And by establishing a partnership with TPI, we believed that we were going to be able to achieve an advantage in terms of unlocking value for our customers that exists in the technology that Technoprobe has in those interfaces and our testers have in their architecture. So I wouldn't say that the Technoprobe investment reflects a change in strategy, more it reflects the fact that we have been monitoring the trends in the test market and we discerned an opportunity for us to do something great for our customers and our investors.
spk10: Thank you, Vic.
spk00: Our next question is from the line of Brian Chin with Stifel. Pleased to see you with your questions.
spk01: Hi there. Good morning. Thanks for letting us ask a few questions. Maybe Greg in past discussion, maybe my numbers are a little bit off, but I believe you've anticipated the hyperscaler ASIC companies maybe driving or representing 400 million or so of incremental growth in the compute TAM in coming years. I guess how large do you see this market by 2026 and based on your customer wins and ongoing traction, what do you think your market share of this incremental could be in that timeframe?
spk04: So, um, I think I'll end up phoning a friend with Andy here in terms of what we've said in terms of projected TAM for that market.
spk02: Yeah, we've said it's in the 400 to 600 .
spk04: Right. And we've said that there's modest growth in the overall compute segment over that period of time. So we're going to see a slight decline in sort of the traditional compute part of the market and this 400 to 600 million of new hyperscaler or VIP. The way that you can think about this is we're always putting up a good fight to try and win share in the traditional compute space, but we really think that our opportunity is to gain share in that 400 to 600 million dollar chunk. And what we've seen is, you know, say we're at a below 20% share in the traditional market, in terms of socket wins, the sockets that we see in the market and we're competing for, we're definitely winning more than our fair share. So we're winning the majority of sockets that we're targeting. And so we're very hopeful that we'll end up with a pretty good split inside of that 400 to 600.
spk01: okay yeah that's that's helpful and then maybe uh focusing on the the memory test part of the uh test tam um i think your main competitor as of last night is forecasting a much stronger year of your growth in memory tests um than you this year though they're largely tied off to dram i guess how can we reconcile your up 10 with with their sort of maybe more optimistic forecast and does that do you think potentially represent upside to your view this year
spk04: I think it could represent upside to our view. But the thing that I'll remind you is that if you look at the memory test market, we typically break it up into four chunks. So two types of memory, DRAM and NAND flash, and then the wafer sort, and then the final test. Our highest share is in the final test for both DRAM and for flash memory. What we've seen in 2023 is technology-related buying in those spaces, and that has been a great tailwind to our share in that market. What we don't have clear visibility into or real forecasts from our customers is how much capacity they're going to need to add at wafer sort. So that's not a technology-driven retooling space. That's something where they can use the same testers for new generations of parts. and they need to have production volumes that drive additional acquisitions. It's entirely possible that our competitor has better view into the long term needs in the wafer sort for some of those customers than we do. So and it's also possible that we have not seen the benefit in that part of the market. So I'm pretty confident in our in that we have a good view of our memory business for 2024, I think there is an upside potential depending on how much capacity add is required as the memory inventories come down.
spk06: Okay, very helpful. Thanks, Greg.
spk00: Our next question comes from the line of Semek Chatterjee with JP Morgan. Please proceed with your question.
spk07: Hi, good morning, and thanks for taking my questions. Maybe for the first one, if I can ask you, in relation to the long-term model of the 2026 model, you are outlining strong growth in the robotics segment, as you see a lot of opportunities for growth. How should we think about, as you sort of invest towards that growth, what profitability can you drive to in that 2026 model for robotics? And I'll follow. Thank you.
spk04: Hi, Semec. So in terms of the long-term model, our goals remain kind of consistent for our robotics group, that our target performance is 20 to 30 percent growth and our target profit is 5 to 15 percent. What we've said previously and what we are still operating to is as we are watching the growth develop in that market, If it appears that incremental investment is not yielding higher growth rates, then we will feather those back to try to increase the profit range. But as long as we believe that we're at this low penetration and there are fruitful investments that we can make, we would prefer to make those investments and continue to drive growth. The key thing that we're doing in operating that group is we're really focusing on maintaining high gross margins. So we're in excess of 60% gross margins for the group now, and we are intent on keeping those gross margins at that level so that we have the option to sort of dial the profit that is appropriate to the growth rate we're achieving.
spk07: Sure. That's helpful. And for my follow-up, just a question on gross margin for the year. I think for the full year 2024, you're guiding to 58 to 59. With the starting point you have, that does imply, I think, if I do the math, that you would be at some point during the year crossing 60. Am I sort of calculating that right? Is that sort of what you're implying? And What are the drivers to get to that 60% level during 2024 itself with the volume challenges that you're seeing right now in the test market? Thank you.
spk09: Yeah, so you're right. In some of the quarters, we do anticipate to be higher than our model, really coming back with stronger product mix. In various quarters, we have obviously higher volume. And then some of our higher product lines or some of our higher margin product lines do come back. So it's volume in the quarters, but mainly product mix and product line mix that are driving the improvement. I should add that there's also operational efficiencies that we're working on that we anticipate that will help margins as well in the back end.
spk07: Okay, great. Thank you. Thanks for taking my questions.
spk00: The next questions are from the line of Tushaya Hari with Goldman Sachs. Please proceed with your questions.
spk08: Hi, good morning. I had two as well. The first one on HBM, Greg, can you characterize your competitive position in this market? I guess number one. Number two, how big was HBM as a percentage of your memory business, either in calendar 23 or Q4? And with your customers sort of expecting, you know, 50, 60% growth on an annual basis over the next couple of years or several years, should we think about, should we assume, you know, the growth rate in your business to be in that zip code or could there be, you know, retooling or parallel tests that sort of deflates that number?
spk04: Okay, so... you packed a lot of parts into that question. So let me start unpacking it. So first, in terms of competitive position, right now we are roughly splitting the HBM test market with our competitors. So if you look at overall share, that's a positive to our share, that there are multiple competitors in the overall memory space. the HBM part of the market is pretty much a clean split between us and our competitor. In Q4, we think HBM represented more than 50% of our memory shipments. So it was a huge factor in that quarter. And in terms of growth, we think that there is the potential for growth, that there are new HBM competitors that are coming on the scene. So there's both a unit volume growth. I think that a lot of that, the capacity for a lot of that is in place now. But there are also standards changes, HBM3E and HBM4, that are coming. And those are driving retooling for performance tests. And we think that that is going to be a potential driver for us in the back half of the year in memories.
spk08: Got it. Thank you. And then as my follow-up on the robotics side, just wanted to get your thoughts on 24. You're guiding the business up 10 to 20%. You talked about, obviously, the long-term value proposition, which makes sense. You're ramping the UR30. You talked about some of the initiatives you're doing from a channel distribution perspective. So I'm just curious, you know, if you're guiding the long-term up 20 to 30, why up 10 to 20 this year? particularly given the fact that 23 was down, you know, close to 10%. What's sort of weighing on growth this year? Thank you.
spk04: Well, I think the key thing that is limiting our optimism is that even though we had a really great Q4, we had a really great Q4 because we introduced basically a blockbuster product. 24% or so of our revenue in Q4 came from that product. Underlying that, there's still some fundamental weakness in the industrial end market. And there are predictions that that is going to ease, that the demand is going to come back relatively strongly. But we are entering Q1 of 2024 with PMIs at a relatively low level and a fair amount of some regions that are quite quiet. So we have, you know, Optimism for continued growth through the year, you know, both quarter on quarter through the year and each quarter in comparison to the year prior. But we're coming in with a relatively low Q1. So we wanted to be careful in terms of where we set the bar for growth for the full year.
spk12: Thanks, Ray.
spk02: Okay, and operator, we are out of time. So folks that are still in the queue, I'll get back to you later. morning but thanks everyone for joining and we look forward to talking to you in the weeks ahead.
Disclaimer