Teradyne, Inc. — 2024 Q1
Transcript
Each turn shows the speaker, their inferred role, the section, and that turn's net sentiment (×1000).
Greetings, and welcome to Q1 2024 Teradyne, Inc. Earnings Conference Call. [Operator Instructions] As a reminder, this conference is being recorded.
Thank you, operator. Good morning, everyone, and welcome to our discussion of Teradyne'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 the first quarter of 2024 and our outlook for the second quarter of 2024.
Thanks, Traci. Hello, everyone, and thanks for joining us this morning. Today, I will summarize our first quarter results and discuss some of the trends in the semiconductor industry and robotics that we believe position Teradyne for long-term growth. It is useful to comprehend these longer-term growth drivers because the industries in which we operate are inherently cyclical. Sanjay will then provide greater financial detail on our first quarter results, our current outlook and additional financial information.
first is SAM expansion through new offerings. Second is growing our OEM solution provider in large account channels. Third is building increased recurring revenue through service and software offerings.
Thank you, Greg. Good morning, everyone. Today, I'll cover the financial summary of Q1, provide our Q2 outlook and full year planning assumptions. Now to Q1. First quarter sales were $600 million, which was $10 million above the high end of our guidance with non-GAAP EPS of $0.51, which was above the high end of our guidance of $0.38. Non-GAAP gross margins were 56.6%, above our guidance due to favorable product mix, higher volumes and improved operational efficiencies. Non-GAAP operating expenses were $251 million, both flat year-over-year and up slightly compared to our fourth quarter. Non-GAAP operating profit was approximately 15%.
Thanks. We will go ahead with questions.
[Operator Instructions] The first question comes from the line of Mehdi Hosseini with SIG.
Yes. Two follow-up. It seems to me that despite the fact that you highlighted mobile as an area of weakness, it had already been dialed into your January guide or commentary for the full year. Would that be fair?
Yes, it would, Mehdi.
Okay. And this brings up what I want to really ask you. When I look at TSMC, they just announced their annual report, and obviously, Apple's mix of revenue went up. And I'm using that as a kind of trying to better understand the dynamics. So if TSMC is running or generating more revenues, that means more chips for that particular customer is coming out. And at some point, utilization rate for those testers are going to hit that 85%, 90% threshold. And I'm trying to look beyond the next couple of quarters.
Mehdi, this is Greg. I think you've got it right that utilization is continuing to increase. The equipment that is in place is being used quite efficiently through the year. So the loading is relatively constant. And we definitely believe that it's a matter of when, not if there's going to be a need for significant additional capacity to support that kind of product line.
Next question comes from the line of Vivek Arya with Bank of America Securities.
On the compute TAM, the $1.5 billion, I think that's up roughly 15% from last year. When I look at the size of the accelerator market this year, it's supposed to double. So is that one kind of high-level way to look at how much the accelerator market is growing versus how much your tester compute TAM is growing? Or is there a different way? So for example if, let's say, the accelerated market grows 30% next year, what does that imply for your compute TAM?
Interesting question. Thinking about the TAM, that is the overall size of the market, so what we would get, what our competitor would get. And the size of the accelerator market has gone up a lot this year. Most of the testers to support this year's revenue in accelerators was put in place last year. So if the TAM that was supported last year supporting that sort of big growth that we've seen in the end market this year, what's going to happen this year, the $1.5 billion that we talked about is going to support a significant increase again next year.
And if I could add just a quick comment. In my prepared remarks, we noted that the -- our estimate was revised for 2023 for the compute market. When I do the math of our midpoint of $1.5 billion in '24 versus our old, you do get a 15% increase, but we revised that in my prepared remarks, I noted to $1.4 billion. So I think the growth is 7% versus 15%. I just wanted to point that out.
And just a quick follow-up. A number of your WFE peers have started to see the benefits of 2-nanometer and gate-all-around. I'm wondering what does that mean for Teradyne? When will you start to see those benefits? And what does that mean for tester intensity? I assume that's more '25 or '26. Any color around that would be very useful.
Sure. So we are expecting to see some testers being used to support the engineering work in early parts in 2-nanometer probably towards the end of 2025. We don't expect any volume associated with 2-nanometer gate-all-around until 2026. And the way that we're looking at that right now is that there's nothing particularly idiosyncratic about that node in terms of tester demand.
Next question comes from the line of Toshiya Hari with Goldman Sachs.
I had one question on memory and another one on your VIP business. On the memory side, you guys talked about a $1.25 billion TAM for '24. I was hoping you could provide a rough split between DRAM and NAND, how you're viewing the market, what the contribution could be from HBM, and if you can speak to your competitive position, your market share position within HBM, that would be helpful.
Sure. Why don't I take the numbers, and maybe the competitive position, Greg, if you want to take? So the range that we provided was $1.2 billion to $1.3 billion, with the midpoint of $1.25 billion. And we think from a DRAM perspective, it's roughly 80% of that versus flash of 20%. Historically, I would say DRAM has been in the 40% and 50% of the overall memory market. So that's roughly the split. And then HBM we think is -- yes, we think that number is about $500 million. It's going to work upwards, but we think that number is above $500 million out of the $1.25 billion.
So in terms of competitive position, the share patterns that have existed for a few years are persisting. We have very high share in flash final test, we have strong share position in DRAM final test and we have a lower share position in both flat wafer sort and DRAM wafer sort. So what's happening in the memory market? Like last year, there was a lot of technology-driven buys that were pushing final test purchases. This year, the spark is coming back to the memory market, and that is driving the need for more capacity, and that is driving wafer sort purchases.
Great. And then as my follow-up on the VIP side, we've all seen quite a few announcements from the Googles, the Amazons, the Metas of the world in terms of what they're working on. I'm curious when those announcements kind of translate into revenue for you guys. I know it takes time for those projects to ramp. But I think you have pretty good visibility as a test provider, you're probably pulled into those projects early. So is it a '25 dynamic, is it '26? And how should we think about that contribution as some of those projects ramp?
Well, from a vertically integrated producer perspective, it really kind of started back in like 2022, 2023. In 2024, like right now, there are hundreds of testers that are being used to test VIP source parts for us. And probably a similar number from our competitor. So it has happened. The thing that hasn't happened because of the -- the low utilization driven by the mobile slowdown, that hasn't translated into as much new tester purchases as we would have expected in a stronger market.
Next question comes from the line of C.J. Muse with Cantor Fitzgerald.
Yes. I guess first question, Greg, in your prepared remarks, you talked about cautiousness on the second half, though seeing better upside opportunity than downside risk. So I was hoping to dig deeper into that. If you could share what would be the end markets that could be upward [indiscernible] as we progress through the year.
So we have limited visibility, so just bear in mind that I'm going into the realm of speculation here. The things that could drive a higher second half is the leading edge of some sort of a recovery in mobile. Right now, we have a pretty low baseline baked into our plan. So essentially any capacity shortfalls would immediately turn into business.
Very helpful. As a follow-up, you talked briefly on Edge AI. And so I was curious to hear your thoughts on how you see that play out in the smartphone arena. It sounds like this is a year where everyone is trying to see what sticks and what use cases and then maybe next year is the year. Would love to hear whether you agree with that assessment? When you think the earliest we could see increased content to support Edge AI and smartphones? And how you see that playing out and impact Teradyne's test business?
Yes. So that's a great question. We've been having very rich discussions internally about this to try and figure out when you consider a smartphone to be AI-enabled. And we're thinking about it in terms of the complexity of the processor that goes into the phone. There have been neural processing units and AI features in phones for years. But now, the AI opportunity, the edge AI opportunity is driving things towards pushing the amount of silicon area used for AI up towards like 50% or 1/3 of the silicon area going into that.
Next question comes from the line of Samik Chatterjee with JPMorgan Chase.
Maybe just to stick with the mobile ecosystem a bit. Any sort of change in your thinking about the primary smartphone customer there? I know you've said previously, it's less than 10% is what your expectation for this year is. Just checking in terms of how much of the sort of change in 2Q or sort of thinking about the back half, any changes in how you think about the cadence of purchases from the primary customer or any change in your overall taking for the year for that customer.
I hope I caught all of your question. If I didn't, then please correct me. We don't comment about specific customers. I will share that we don't expect that our historically largest customer will be a 10% customer this year. I'll also share that we think that there is more of -- that we've gotten sort of the bottom dialed into our plan, that there's definitely more upside to downside in terms of the plan associated with mobile in general.
Got it. And you got that question. So for my follow-up, if I can ask on robotics. Can you give us some sense of how you're thinking about first half versus second half there? And what's the -- right now, any updated thinking on profitability for the year for the business? How do you track in 1Q and sort of how you're thinking about the full year for robotics.
So I'll start off with sort of a qualitative answer in terms of the drivers, but then I'll pass it off to Sanjay to give you more of the precise split first half, second half and the profitability. So what's going on right now is we have a plan for the year that's going to be essentially sequentially growing quarter by quarter. And the reason that the plan is laid out that way is because there are new growth drivers that are coming online throughout the year. There are drivers that are in place from last year in Q1, that's specifically the UR20 and UR30, the heavy payload cobots.
So roughly the first half of the year -- well, first, as Greg noted, we expect to grow sequentially in Q2, and we expect to grow throughout the year for the reasons Greg just noted. But in the first half, we have 42%-ish plus or minus, and then 58% in the back half of the year. We do expect to be profitable this year. I would say, over the year, and I would say that, that would be single-digit profitability. I will comment, I think you asked about Q1. We were not profitable just given the seasonality of the quarter. That should give you the context, I think that you asked.
Next question comes from the line of Krish Sankar with TD Cohen.
I had a couple of them. First one, on the memory side, how much of your memory revenues were from HBM in the March quarter. And I think Sanjay, you mentioned HBM revenues could grow 5x this year. I'm kind of curious, are you over-indexed to 1 customer, because my understanding is of the 3 HBM customers, 1 of them in sources, and within other 2, can you just talk a little bit about your share position there? Then I have a follow-up.
I'll take the first one on the memory in the quarter. Yes. So I want to say roughly about 45% of the memory revenue that we had was tied in the first quarter to HBM.
Yes. So Krish, one quick comment. The 5x comment about HBM was referring to the market size for HBM memory. So last year was about $100 million of TAM for HBM. This year, we think it's about $500 million worth of TAM for HBM. Our revenue is not going to go up by a factor of 5. It's going to go up. It's going to go up significantly. But we're not going to be able to hold sort of the 50-ish percent share of HBM that we had last year.
Very helpful, very helpful. And then just a quick follow-up. You spoke about your revenue trajectory for the year. You said September should be similar to June and then growth in 4Q. I was just curious because that seems to be against normal seasonality. So what's the deviation this year?
Yes. Interesting question. I'd say that I remember several years back when we talked about we were going into a downturn and that downturn was 4 to 6 quarters, and the downturn has gone a lot longer in the way of mobility. And as Greg noted, we believe that our forecast considers mobility at a point where we feel comfortable that it's going to be achieved. And so I think as things pick back up, the historical seasonality, I put it in the context of some segments are going to be recovering and we have some new segments that are -- or some segments that are growing fairly well. So I think the seasonality comment with regards to the test perspective, or test portfolio of businesses we have. It's a little bit off this year.
This is Greg. Just one additional bit of color. If you think about the seasonality pattern that we developed really for a decade, 2010 through 2021, that seasonality, if you looked at all of our segments with the exception of mobile, there wasn't a really marked seasonality. Like those automotive, compute, industrial, they kind of chugged along. The seasonality was really driven by the mobile TAM, and that was driven by consumer buying behavior, right, that you needed to have things for the holiday season and for Lunar New Year. All of that inventory needed to be built up. And so there was a concentration of capacity that would go in, in Q2 and Q3, with mobile not driving right now, the seasonality is very, very muted. So I would expect to see that the seasonality will return once the mobile market is stronger, but I don't think you can use normal patterns to predict the way things will look in '24.
Next question comes from the line of Joe Moore with Morgan Stanley.
I wonder if you could talk to the smartphone issues that you just mentioned. Away from your biggest customer, you sort of saw the smartphone suppliers reduce balance sheet inventory by 15, 20 days in Q4. And they talked about doing that again in Q1. So I would expect you to have low visibility, but shouldn't you get some recovery as they stop depleting that much inventory? Is there any kind of nuance in terms of how much of that inventory has been tested already in die bank, things like that. But shouldn't we have some optimism that this is going to get better.
Well, I think we certainly have optimism that it can get better. And that's basically because we've dialed in the bottom. And the way we do this is we typically will look at tester utilization numbers. And frankly, those numbers are all over the map. Some indicators show that capacity is tightening. Other indicators are showing that it's relatively -- utilization is tightening. Others show that it's relatively flat. Our qualitative checks, when we are talking to people in the ecosystem, they are telling us that capacity -- that utilization is getting tighter and it's forecast to go up from there.
Thank you. Next question comes from the line of Brian Chin with Stifel.
Just to clarify, maybe firstly, was the third quarter pull-in, was that for AI, for compute, or for memory, or for both? And just to be clear, your share of the $200 million memory TAM increase, should we think about that being closer to, say, 40% if the emphasis is on wafer sort. And kind of sorry one more part to this. But I didn't catch this earlier, but if HBM demand is biased to a single customer, do you expect memory sales to be lumpy this year kind of from a quarterly perspective?
Yes, I'll take the kind of the Q3 to Q2. It was, I'd say, mainly in compute and some ADAS that was accelerated and -- just looking up something.
Yes. So in terms of the TAM increase, we moved the TAM up by $200 million. And I think -- I think your guess of $40 million is a bit low. I think we'll be up a bit more than -- we'll get a larger chunk of that up than that, probably in line with our historical share level, kind of 35% to 40% of that TAM increase should probably go to us. In terms of lumpy sales, no, I think actually the memory business is driving -- we have deliveries that will be stretching out through the year. And I think the demand is relatively steady and potentially increasing. So I think on a quarterly basis, we don't expect a lot of memory variation.
Okay. That's helpful. And then switching gears to robotics. It seems like your omnichannel and new product focus is driving a lot of this improved momentum in that business. But can you also just touch on the broader environment you're seeing in robotics, and where, if any, you see sort of macro like headwinds or tailwinds? And last part of that, what vertical or application was that customer that placed the record MiR order? And did that include any of the MiR1200s?
Okay. So in terms of the macro environment, it's pretty weak. I mean it was weak throughout 2023, and we don't see a significant improvement right now in terms of the end market conditions. If you look at PMIs, some regions are clicking up slightly, but we haven't really seen that turned into like a hot market. Having said that, we're in a part of the robotics market that is very, very low penetration. And so we are not -- we have seen our business results vary with the macro conditions, and we think that, that is indicative of the -- that was one of the things that drove us to go and make the changes in terms of adding new products and driving channel development, because we think we're 5% of the way into something that could be huge.
Next question comes from the line of Gus Richard with Northland Capital.
Congratulations on the results. On the Industrial Automation, you've got a partnership with NVIDIA on AI. And sort of one on the limiters to penetration is the lack of a VAR channel, having to have to work with people to implement this stuff. Does multimodal AI or AI in general sort of simplify the implementation? Can you simplify the programming and implementation of robotics with that infusion of AI?
Yes. So for sure, the great thing about AI, there are two primary benefits to AI. One is that it's far easier to design a solution that is able to deal with variation, whether it's part variation or location variation or other uncertainties that exist in normal manufacturing environments, the solutions end up being far more resilient. So that means that more people would be willing to adopt them.
Got it. And then on the semi test side, you've got sort of -- it's going to be a while before we get to 2 nanometers, yields at 3. I don't think we're that great. And the AI accelerators are reticle limited. How are you seeing the expansion of use of chiplets sort of beyond AI accelerators and servers and FPGAs. Is that beginning to broaden out?
Not really. Well, one thing is that most of the advanced packaging capacity in the world has been consumed by the people who are doing cloud AI, high-performance computing. So there's more demand than there is supply for it. So I think that's limiting it to the markets that are willing to pay the most for the ability to do it. I think that it's possible that chiplet technology will migrate out of high-performance computing potentially -- I mean it could potentially migrate into some mobile applications. I think that's going to be a relatively slow process because the price points are very, very different.
We have time for one last question. The next question comes from the line of Steve Barger with KeyBanc Capital Markets.
I'll be quick. Greg, you said share gain opportunities in HBM are good. Is that primarily due to you having capacity to support a fast-growing addressable market, or is there something unique in your current or future testers that will make you a supplier of choice for some variations of HBM?
I think it's definitely the latter. So we believe we have a differentiated solution, both in terms of ability to support data rates out through HBM 4 and also in terms of being able to use same platform across multiple insertions. So we think that we have an ability to deliver more cost-effective performance test of HBM, and we believe that we're going to make some progress there.
Understood. And for the large account channel in robotics, you said there's a gestation time before revenue. Was this largest ever MiR order a function of work prior done to the pivot? Or did it come after? And then to your point about being more immune to cycles, is that just because large accounts are more likely to invest through cycles and are less swayed by near-term conditions?
Yes. So let me take the large account question first. And I need to apologize to Brian because I didn't answer his question before. So the largest order that we've ever gotten from MiR, it came from an automotive customer. And it is also our historically largest customer for MiR. By no coincidence, it's also a significantly large customer for UR. So we have been selling them robots for a long time. The one thing that has happened since we established our large account effort is that we've been applying many of the strategic account management techniques that we've been using for decades in semiconductor test towards addressing those accounts and really organizing ourselves around the way that those large accounts acquire new equipment and also how to take care of the equipment that they have. So I think it's certainly an account that predated that effort, but I think our ability to serve that account has improved with this effort.
And then just about being more immune to cycles, is that because large accounts will invest through cycles? And so that pivot will make you less cyclical yourself?
I think actually, the large accounts may be the segment that has -- that will continue to be sensitive to end market conditions, because large accounts will work off of budgets. And if they don't provide budgets for automation, then those purchases won't happen. The thing that we're really trying to do is make sure that we are adding enough new opportunities to drive growth even if end market conditions are weak.
Ladies and gentlemen, we have reached the end of question-and-answer session. I would now like to turn the floor over to Traci Tsuchiguchi for closing comments.
Great. Thank you all for joining us this morning for our first quarter earnings call. We look forward to being in touch with you all through the course of the quarter. Thank you. Good day.
This concludes today's teleconference. You may disconnect your lines at this time.