Teradyne, Inc. (TER) Earnings Call Transcript & Summary

All right. Good morning and good afternoon, everyone. I'm Krish Sankar, the semi equipment analyst at Cowen. The next company presenting is Teradyne. And we are fortunate enough to have Andy Blanchard from Teradyne, the Head of IR, and most of you investors know him very well. And as you all know, Teradyne is one of the leaders in test equipment and becoming a formidable force in the world of robotics. So with that, Andy, thank you very much for your time, and I hope you have been doing well.

Yes, I'm doing very well. Thank you. All things considered. And thanks, everybody, for joining virtually. And we are ready to go, Krish. You tell me you got a long list of questions queued up and so does the audience. So I know we have a limited amount of time, let's just jump in.

All right. Thanks, Andy. So let me first start off with all these concerns and noise around this commerce department rule. On the civil military fusion on one side and then the ban on Huawei. And my understanding is that Huawei, either directly or through their OSAT, is a customer of Teradyne. So I just want to find out from your vantage point how do all these rules and politics impact Teradyne?

Well, see, if we take one at a time, the military piece of it is pretty straightforward. It's more of a bookkeeping and compliance activity than anything that we think we'll have a meaningful substance, at least to Teradyne. The more complex one is the trade with Huawei. As we noted in our financial filings, Huawei was an 11% customer in 2019. Just for context, they were a 4% customer in 2018, a 1% customer in '17. So it was quite a run up. And we also noted in our commentary in our earlier earnings calls that last year was a very strong year for infrastructure -- 5G infrastructure in China, in particular. And so there's a linkage there, of course. And we also noted that we expected that to decline in 2020 compared with 2019. Now we did see 5G infrastructure business in the first quarter. In fact, it was a little bit stronger than we had expected. But still, the earlier commentary that we expect that to decline full year remains in place. Now as far as the new trade rules, in particular, our internal legal team and compliance team, along with some consultants out of the D.C. area, are actually grinding through that. And this is going to be an unsatisfying answer to investors, but we -- there's still enough ambiguity in there that we don't know. And we're obviously going to make sure that we are 100% compliant with the letter of the law, but on the letter of the rule. But we're still trying to -- we're still [Technical Difficulty] exactly what the impact on us is. And as we go through that, we'll make folks aware.

All right. And then the 11% Huawei exposure, how much was it was direct to Huawei? How much was it was OSAT?

That's a very good point. And the way we handle those filings is we -- to try to provide context, Huawei by themselves, in particular, their subsidiary high silicon, was substantially less than 10% customer. However, like most fabless semiconductor companies, they develop test programs on our equipment or our competitors, and then they depend on their supply line to go fulfill the actual requirements -- the volume. And so they specify to their subcontract partners the equipment to use, in this case, Teradyne. And so in aggregate, not just from them, but in aggregate, they were an 11% customer. You can assume that a small portion of that was -- a very small portion of that was actually purchased by Huawei or, in this case, HiSilicon directly.

All right. That's helpful. And then, Andy, you mentioned about the big China 5G base station deployment last year, which clearly benefited you. How do you look at that going forward? There are some rumblings about China pulling in some of the 5G deployment in the back half of the year. Are you seeing that? And is there any way you can quantify in terms of base stations deployed? How many testers are being bought so far in 2019? And how do you envision that evolve in 2020 and beyond?

Yes. We don't -- we can't get to that level of granularity. I will point out, though, that obvious point. But at the end of the year, the annualized shipment rate was far, far greater than the number of base stations installed by the end of the year, right, as they built up capacity. The way we're looking at that last year, 2019, it was about $200 million or so were spent for semiconductor test equipment for 5G. And we think the vast majority of that was for infrastructure, and the vast majority of that was in China. We think in 2020, that shifts over to handset. And to date, the customer demand is overwhelmingly on the handset side. And we expect that will continue. As I mentioned earlier, the expectation was for that to fall off in -- the base station demand to fall off in 2020. We -- that continues to be our view. And the demand, especially what we're seeing in second quarter, is really, really strong for handsets. A portion of that, of course is 5G. And that's the natural topography of the mobility market where the second quarter is the biggest quarter of the year by far. And that it drops off as you get later into third quarter and fourth quarter. And then next year, we start it all over again.

Got it. And then on that handoff from base station to handsets, can you just talk a little bit about your market share position in base station? And as it moves into handsets and potentially down the road even to millimeter wave, how would it impact your market share or revenue profile?

Yes. The -- that's a great question. Now historically, the base station market, when we did 3G and 4G, the base station market wasn't a significant market, and it didn't -- it was kind of combined with ASCs and a whole bunch of other things. And it didn't really pop up as an end market segment. This time around, at least with one customer, it's been pretty significant. And we think of that 200 million or so last year, maybe 60% or 70% of it accrued to Teradyne. So that was a big year for us as that geography was building out. I think around the world, there's a variety of chip makers that provide devices to various base station makers, and our share varies as you go along. But in general, our market share is in the 40s. It kind of varies from low 40s to low 50s, and I would expect it would follow that path going forward. But what we're seeing right now, again, is the handset side taking off. And interestingly, you mentioned millimeter wave. We describe an expectation that there would be these waves of adoption, right? You have 6-gig base stations, and sub 6-gig handsets and then millimeter wave base station, millimeter wave handsets. And in fact, this year, even though the market forecasts are for a relatively small number of 5G handsets, maybe 200 million to 300 million out of 1.2 billion, 1.3 billion total, and a small portion of those 5G handsets would have millimeter wave capability, perhaps less than 10% of them. In spite of that, the millimeter wave demand for test equipment has been much stronger than we expected. And as we mentioned in the earnings call in April, it's likely to be north of 50 million, and it could well be high double-digit millions of dollars for Teradyne, right? And of course, we don't have all of the share. So that's a much larger market for a small number -- relatively small number of handsets. And part of that is because there's no install base of test equipment for handling 29 gigahertz and above. So any incremental demand has to get incremental test equipment replaced. But it still feels outsized. And so either the units are far bigger than the market forecast, as we're expecting. The test times are much longer, which is what we're hoping for. Or there's yield issues or some other manufacturing artifact that's bumping that up. But again, it's a far stronger market than we had expected.

Got it. Got it. And then when we compare and contrast the 5G base station from China last year to the handset and millimeter wave down the road, do you think that the 5G base station buildout was kind of a big enough [ piping ] scenario where you spend a lot? And going forward, the handset transition might be more spread out over a couple of years? So do you think you might see a surge for the next couple of years just as 5G gets adopted?

Yes. I think you're going to see -- there'll be steady buying for base stations with some -- thinking of python events along the way, right? So I think you'll see some surges because, by the way, in North America, we haven't seen a lot of spending for silicon going into base stations. Now maybe that's a competitive thing or maybe it's something else going on, right? But we haven't seen the level of focused investment for North America or for Europe, for that matter, compared to what we've seen going into China. I think the -- we're in the very early stages, though, of the handset buildout because, as I mentioned, again, maybe it's 200 million or 300 million handsets. In North America, you'd expect all of those eventually to have millimeter wave capability, and people are looking 20 million to 30 million -- or I think if people -- if there were 20 million, people would be surprised, 20 million millimeter wave handsets this year. So there's a lot of runway to go there. And we expect that 5G is going to roll out over a 3- or 4-year period, at least. And that's just on the handset side. You and others have done analysis on what it means for industrial IoT, what it means for automotive. Those are still quite a ways in the future. But the handset piece of it and infrastructure, we think, has got a lot of light.

Got it. Got it. I mean, if you look at all of these, like the 5G base station or handset, the mobility, kind of all of them rolled up under your SOC test business. And I know that you guys took away the full year view because of COVID. But if you look at SOC, it looks like this is probably the fifth year in a row that it could grow. And on top of it, you have a very strong first half. So how should we think about SOC on a go-forward basis? Do you think we are in a new structural regime? Or do you think the cyclicality will eventually come back and you're going to see some kind of a digestion?

No, I think we're in a different place. And that's -- and I'm not suggesting that volatility is gone forever because we will -- at some point, we'll have a down year. Maybe it's this year, maybe it's next year. But I mean we are still in the semiconductor capital equipment market, and it's not up and to the right forever, right? But it is structurally a much stronger market than it was pre-2014. And in our view, what's going on there is the drivers of our demand are complexity of devices and growth of devices. And those have been going on -- that's been happening since semiconductors were invented. But we had a period post tech bubble where it was tremendous innovation on the test equipment side and on the design tool side that dramatically improved the productivity and quality of test. And so in our case, it was parallel tested. The test is more efficient. The design tool makers came up with some very clever ideas for scan-based testing and being able to test much more productively with relatively lower volumes of data and still be able to get the quality levels required. That has slowed. That rapid increase in productivity has slowed. So now we and our competitors and the design tool guys continue to make innovations to make test more productive but not at the rate that we had in that kind of 14-, 15-year period after the chart of the century. And so now we're seeing much more connection between increasing complexity and increasing test SOC market size. And again, I think we model it going forward, that's somewhere in the 4% to 8% range. In fact, since -- if you average 14 and 15 through the end of last year, we've been growing at about 11% as a market, for the SOC market. So it's been very healthy. And another note that's important is as semiconductors get into more critical applications, automotive is one, but even in smartphones, the quality levels for smartphones are much higher. The requirements being put on the chip makers supplying into the smartphone space are much higher today than they were 10 years ago. Next year, they'll likely be higher than they are this year. And as a result, that's driving more test seconds compared with a device that's going into a jellybean consumer product. And so that's an added lift to the market. So I think it is structurally different. I'm not arguing that it's up into the right forever. I am sure we'll still have volatility, but it's not at the level that we saw in the mid [ odds ].

Got it. So I mean it makes a ton of sense, Andy, what you're saying. Things are more difficult and more challenging. But then if you try to like compare and contrast it to your largest customer, Apple, you publicly disclosed in the past. 2017 was a big customer, slowed in 2018 and 2019. It looks like it's coming back again. Do you think they will still be in this like hyper-cyclical cadence of like spending for 1 year, digesting it for 2 years? Or do you think with all the things that you said with the complexity, their behavior might change, too?

Yes. And I can't speak to any one customer. But I think in aggregate, the trends I was describing will continue, right? I think any one customer in any one period can implement a new design tool, right? Can implement a new manufacturing technique. They can -- any number of things that they could do that could make their specific demand be more volatile. But I think in aggregate, those trends remain.

Got it. Then the other thing I want to touch base on the semi test side is the memory business where you guys have clearly gotten a ton of traction. Now you already had a decent footprint in NAND. It looks like that has expanded. You also got in like some really good market share wins in DRAM. So can you just talk a little bit about the memory test landscape, you versus Advantest, and where you're seeing the major traction and where should we look for future share gains from here onwards?

Yes. That's a nice story because memory -- Teradyne entered the memory market in 2008 with the purchase of Nextest. And Nextest, we acquired them, they had about 4% share of the memory test market, all of it in flash. And we entered that because Advantest had 80%, 90% market share, and they were kind of able to take a monopoly rent out of that industry and redirected at us and SOC. So we acquired Nextest. They had an architecture again that's called Magnum that was low cost but very scalable. And as the speeds of flash devices increased driven by smartphone adoption and SSDs, the -- and in particular, in smart phones, the use of cameras and video, that really supercharged the speed requirements of flash. And their -- Nextest's product was far better suited to handle those increases in speed because those increases in speed came faster than even our direct customer. The test center managers expected. And so we had a product. Our competitor didn't. They were a bit behind. And we were able to get in the package test side of the market kind of 75%, 80% of the flash package test market because our product had the technology to deliver the economics that were superior. And so we've built on that product line, the Magnum, Magnum V, and now more recently, Magnum Epic, to expand it into the DRAM space. So with flash alone, we were able to get market share up into the high 20s. The 28, 29 -- in 2017, 2018 time period. In 2019, the market contract from 2018's 1 billion down to 2019's kind of 600 million. Most of that reduction came out of the DRAM side of the market. We had small exposure. So bam, we look like geniuses, and our market share is in the low 40s, right? Well, at the end of '19, we had a significant design win with a Magnum platform product called Magnum Epic and low power DDR5. And so that's our first entry into the DRAM package test part of the market. Again, it's done -- you don't win in semiconductor test on price. You win on innovation. And so when there's a technology inflection point, it gives us an opportunity to bring a new product to market that showcases our ability to solve the problem more proactively than a competitor and so on. So that was a great design win, and we're seeing the result of that here in the first half of 2020. We're aggressively working to expand beyond one foothold on the beach right now for DRAM package test, but that will unfold as the year goes along. But we had $80-plus million, $84 million or so in memory test sales in the first quarter. It will continue to be very strong in the second quarter. And the second half is a little bit of a question mark. But the things that drive demand right now, the things that are driving demand right now are the technology inflections. So customers -- even if volume is relatively modest for these new technologies, LPDDR5 or the new flash standards, the customers still have to put test capacity in place. And that's beneficial for us, and that will make that a pretty healthy market. We estimated -- in January, when we looked at the market size, we estimated that it will probably grow 15-ish percent or so this year. And then we kind of -- we haven't commented on the market since then because of COVID. But all things equal, I would expect it would grow year-on-year.

Got it. That's very helpful, Andy. And then just a question from an investor. In Q1, were there any 10% plus base station customers in your Q1?

I can't -- we had a 10% customer. I can't talk about what the end market was because that would be akin to commenting on the customer.

All right. No worries. Just one final question on Semi Test before I jump into the robotics side. Auto industry has been weak for a while. Are you seeing any recovery there? Or is it like the worst is behind? Or is the second derivative turning positive? Or it's going to stay like this for a while?

I don't know how long it stays like this, but we haven't seen that turning around yet. And both on the auto side -- auto and microcontroller and then just industrial analog has remained pretty muted. And that's -- in the case of microcontroller and automotive, that was through all of 2019 as well. So we're 6 quarters into that. And we don't see a turnaround in the near future.

Got it. Got it. So then onto robotics, clearly, like -- so robotics has been, over the last several years, has been a very successful acquisition, although the growth rate has kind of slowed. How do you look at the business today in a post-COVID world given the fact that there's still need for factory automation, but you still have your usual set of competitors in Asia and you're also trying to add more, for lack of a better term, bells and whistles with the ActiNav process kit, et cetera. So how should we think about the [ RealWear's ] robotics cobot business on a go-forward basis? And what would you say is the right growth profile to use for that business from here?

Yes. It's a good question. UR has been a great addition. We acquired them in 2015. They went from about $40 million in 2014 to just short of $250 million last year. I think 248 million for the full year. But you're right, the growth slowed in 2019. UR has high exposure to the automotive supply chain. Something on the order of about 40% of UR's business is related to automotive, most of that in the supply chain in, not in the brand names, although we sell to BMW and Volkswagen and Fiat Chrysler and all the rest. But it's majority suppliers. And that industry has been facing a very, very strong headwind. So I think even before COVID, we were looking for automotive to recover. At the same time, we're aggressively expanding the range of applications that we're serving here and in end market verticals as well. COVID comes along, and I think that's certainly going to have an impact given that between 2/3 and 3/4 of our business is North America and Europe, Europe being the bigger, and both of those economies are not back to work yet in -- especially on the industrial side. And we expect that when they do come back to work, first order of business is protecting employees and getting all of the infrastructure in place so that they can do that. And automation will be the next step, right? So I think in the near term, COVID is probably a negative because it's slowing down industrial activity. In the longer term, the same factors that drove the growth all along, the availability of labor, the cost of labor, the quality of end products, those all remain in place. And if anything, with some of the social safety that's in place because of COVID, it's going to be even tougher to attract employees because the safety net is higher, especially for dull, dirty, dangerous jobs. In addition, you now have this potential COVID effect of, boy, we want to put some distancing between employees. So maybe instead of laying out my whole work cell differently, I take every other employee, if it's a repetitive task, and I put a collaborative robot in there and I get the distancing in that way. In addition, there's a lot of talk -- we haven't seen hard data on this, but there's a lot of talk about rearchitecting supply lines in order to eliminate single point geographic failures. And then -- so you may see, whether it's in still in low-cost regions or stuff coming back closer to point of consumption, you may see that manufacturing work get reshuffled, if you will. And that would certainly be positive for us because as people are looking to bring on capability, the economics remain in place, and you still got to be -- you've still got to have a cost structure that's competitive. And our robot's ROI in less than a year in high-cost regions. And so that's a very attractive economic argument.

Got it.

The strategy there, by the way -- Krish, I don't mean to interrupt you, but the strategy there remains to take a piece of automation equipment and make it available to companies that in the past didn't have access to it. So you don't need a big IT department or an industrial engineering department to bring our UR collaborative robot in place. That's always been the case. But now we're building out this network of plug-and-play components, whether they're grippers, sanders, screw drivers, welders, we've been building those out through the program called UR+. We've got over 200 of those certified apps available in the market now. And just this past spring, we stepped that up with something called UR+ applications. We're now -- instead of a bunch of individual certified building blocks, now you can buy the whole house. So if you want to do -- ActiNav was the first of those for doing industrial bin picking. Now you buy the ActiNav solution, and it has the path planning software, the gripping technology, the 3D camera, all integrated together. We packaged that, again, that same model, a factory floor technician is able to program and 3D vision is typically very, very complex. We've simplified that. Path planning is a PhD kind of stuff, we've simplified that so that operators on a shop floor -- technicians on a shop for can train that and easily redeploy it. And that's, we think, is a very, very big deal, and we're going to continue to really invest in those UR+ applications to make it easier and easier and easier burn for companies that don't have the scope of an automaker to be able to deploy automation safely and economically.

Got it. Andy, let me ask you this question, right? So 5 years ago, when you got UR, there were 3 skills. Now it looks like you're going up the weight spectrum, going to larger, more weight capacity. You are adding more like ActiNav and all these things, you're adding more 3D sensing capabilities. Some of them arguably are -- you outsource it to other people on the vision side, et cetera. Are these effectively the natural evolution of the cobot business? Or is it more a reaction to some of your competitors like the Japanese robotic guys who always had a higher load capacity versus a traditional year?

No. I think it's the former. I mean it's a natural evolution because we generally don't run up against the traditional robot makers. And we're not trying to build a [ well blind ]. We're not trying to build a 100 [ geo ] payload. This is human scale automation, right? But one of the things that we discovered was for early adopters who are really excited about bringing the power of automation in, they would figure out how to do all the integration tasks. For the run of the mill guy who's just trying to get -- he's just try to get his product out the door -- the customer's product out the door, he's -- he may be less inclined to spend a couple of months figuring out how to integrate a camera from one guy and a gripper from one guy, I think, software from the third. He just wants to figure out how to pick things out of a bin and do it easily and get it -- and have it pay for itself in a year. And so that's the reason for the -- that's the reason for kind of going up the stack. It also, by the way, provides competitive moats because the more that we can integrate that and our competitors are still trying to figure out, first, trying to get a mechanical solution that is robust and reliable and then try to integrate end effectors to solve specific tasks, but that all adds time and money. And so the more that we can provide a single point solution that's going to add again a competitive moat on the ease of use and other cost.

Got it. And then my final question, Andy, is on capital allocation. I think you guys kind of scaled back the buyback assessment in Q2 because of COVID. Just kind of curious when will that return. And on the M&A front, I think ever since the LitePoint, you guys have completely pivoted away from Semi Test. So is the focus on M&A still going to be along the path of robotics, et cetera? Would you ever consider back to the basics on the Semi Test side?

Yes. We suspended the share repurchase program because of the uncertainty related to COVID. I think the Board will look at that quarterly and decide when to bring it in -- bring it back in. But I think that was just an -- what's the expression, an abundance of caution, right, in going into an unknown future. Regarding M&A, we look at industrial automation primarily. And within industrial automation, with the filter of [Technical Difficulty] that can make the workplace safer and more productive and deliver higher-quality product. That's kind of the lens that we have looked at with universal robots, with MIR, with AutoGuide, with Energid, they're all focused on how do you make the workplace more productive and safer. And then so we'll continue to look at that. And our kind of working hypothesis is that the COVID economic disruption is going to bring valuations in a little bit because there's a lot of very cool companies working with very clever technology, but they are a little stretch on the valuation side. So we're looking at making sure that we have the financial resources that if valuations come in, we can act on. With regard to test, we look at -- we still look at a lot of things in test, but it's a much more mature market. So there aren't as many interesting opportunities there. But occasionally, we had a small acquisition a little over a year ago called Lemsys out of Switzerland. And it was, again, a small acquisition, but they had a very focused product for doing very high current, very high power test semiconductors for EVs and other electric vehicles. So we're excited about that, but there's a smaller universe of things in test.

All right. Thank you very much, Andy, really appreciate your time, really. Thank you very much, and have a good rest of the day.

Yes. As always, good to see you, Krish. And nice to see you handling this COVID thing okay.