Good morning. My name is Thea, and I will be the conference operator today. At this time, I would like to welcome everyone to the Teradyne Quarter Two 2018 Earnings Conference Call. All lines have been placed on mute to prevent any background noise. After the speakers' remarks, there will be a question and answer session. If you would like to ask a question during this time, simply press star and the number 1 on your telephone keypad. If you would like to withdraw the question, press the pound key. Thank you. At this time, I would like to turn the conference over to Andrew Blanchard. Please go ahead, sir.
Thank you, Thea. Good morning, everyone, and welcome to our discussion of Teradyne's most recent financial results. I'm joined by our CEO, Mark Jagiela, and our Chief Financial Officer, Greg Beecher. The press release containing our second quarter results was issued last evening, and we are providing slides on the investor page of the website that may be helpful to you in following the discussion. Replays of this call will be available via the same page after the call ends. The matters that we discuss today will include forward-looking statements that involve risk factors that could cause Teradyne'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.
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 will make reference to non-GAAP financial measures. We have posted additional information concerning these non-GAAP financial measures, including the reconciliation to the most directly comparable GAAP financial measures where available on the investor page of our website. Between now and our next earnings call, Teradyne will be participating in investor conferences hosted by Needham, KeyBanc, Davidson, Citi, and Deutsche Bank. We'll also be meeting with investors in September in Chicago at the International Manufacturing Technology Show, IMTS. Let's get on with the rest of the agenda. Mark will comment on our recent results, current market conditions, and outlook for the year.
Greg will offer more details on our quarterly financial results, followed by our guidance for the third quarter. We'll answer your questions, this call is scheduled for one hour. Mark?
Hello, everyone, and thanks for joining us this morning. In today's call, I'll cover our recent financial results and current business conditions, describe how we're looking at the rest of the year, and highlight some of the new products that we expect will power our growth in the future. First, the results. Our second quarter non-GAAP EPS of $0.59 on revenue of $527 million exceeded our guidance primarily due to our test businesses strengthening through the quarter. In semiconductor test, memory was a standout with sales of $67 million in the quarter, bringing our memory test revenue for the first six months to $139 million, up nearly 2x from the year-ago period. Memory test demand in the third quarter looks even stronger, setting us up for significant year-over-year growth.
This demand is being driven by both our traditional flash package test customers and a nice mix of new business for our recently introduced Magnum wafer test product. Elsewhere in SemiTest, SOC test demand remains strong. Other than the significant year-over-year drop in an isolated mobility segment that we outlined last quarter, demand in other parts of mobility and the overall SOC market has been stronger than we expected when we updated you in our April call. Our Eagle Test product line is a bright spot, with sales in the first six months up 17% compared to a year-ago period, as analog and automotive demand continues to grow. Our latest 2018 market size estimate for SOC is at the high end of our $2.2 billion-$2.4 billion range, and the memory test market is now estimated to be in the $900 million-$1 billion range.
At LitePoint, results were also above plan as Wi-Fi test demand for smartphone applications strengthened. Six-month sales at LitePoint are up 9% over the year-ago period. Despite being in a technology lull, pending new wireless standards like 802.11ax for Wi-Fi and 5G cellular, cell phone makers continue to add incremental wireless capabilities to their products, driving up test times and fueling our business. Our 802.11ax production test product is in the market testing access points and ready for the likely 2019 ramp as cell phones begin to adopt this technology. Likewise, both our millimeter wave and sub-6 GHz cellular production test products are being qualified now at top suppliers for likely production volume starting in 2020. Our system test business also is strong with the six-month sales up 47% from the same 2017 period.
Most of this is due to storage test demand, which has more than doubled from the first quarter on greater than expected HDD and system-level test demand. This will continue to be a lumpy business, and we've organized our operations accordingly, allowing us to deliver above-model profits in strong periods and model or better profits across the full year. In industrial automation, Universal Robots' robust growth continues with quarterly sales of $57 million, up 45% from the same quarter of 2017. Late in the quarter, we introduced our new e-Series of cobots. The e-Series extends the capabilities of our 3, 5, and 10-kilogram payload cobots by providing higher performance data processing, enhanced ease-of-use software, Category 3 safety certification, integrated force torque sensing, and a long list of other improvements to make the arm even easier to use.
The higher power data processing platform and software enhancements allow our UR+ ecosystem partners to design more powerful solutions and apps for our platform. The Category 3 certification further reduces the safety hurdles required to deploy our cobots, and the integrated force torque sensor gives our cobot a sense of touch, which is important for a wide range of tasks like gluing, polishing, surface inspection, and light assembly. With the acquisition of MiR in the quarter, Teradyne now has the leading platform in the rapidly expanding autonomous mobile robot market. In the second quarter, MiR delivered full quarter sales of nearly $6 million, up over 85% from the same period of 2017. In late June, MiR expanded their product line with the introduction of the 500-kilogram payload autonomous mobile robot, the MiR500. The MiR500 delivers the same low-cost, flexible, and easy-to-program characteristics of the MiR100 and 200-kilogram robots.
The MiR500 extends the range of industrial material movement applications that can be autonomously automated. Whereas the 100 and 200-kilogram robots are optimized for automating human-scale material movements in manufacturing, the MiR500 allows the additional automation of forklift and other pallet-moving systems. In the months ahead, we'll continue to strengthen MiR's distribution network, lineup of peripheral suppliers, and invest in their organizational capabilities to support their 50%-100% growth for the foreseeable future. Overall, we expect our Industrial Automation segment to grow at a 50%-55% clip this year, in line with our 2021 earnings model. Looking out at the full year from a financial perspective, we expect the second half to look very similar to the first half and the company revenues to be split about 50/50, with the Industrial Automation segment higher and Test lower compared with the first half.
Looking further ahead, I'm sure many of you are trying to forecast the size of the semiconductor test market next year. We won't get our first look at 2019 demand until late this year, and won't have a good sense of the market until sometime in the first quarter. While we know the test market will remain volatile in the future, we are confident that the long-term drivers of the market, device unit growth, and complexity are intact. Despite this volatility, the Semi Test market has been growing at about a 7% plus annual rate off the 2014-2015 market average. This supports our confidence in the $3.50-$4 earnings target in 2021, which assumes a model growth rate of 2%-4% trend line growth in Semi Test.
In addition to core market growth, we've been investing in new products and acquisitions to create a portfolio that can grow ahead of the curve. A few of those are just entering the market this year, and others will roll out in the quarters ahead. Some enhance our competitive position in served markets, while others expand our served markets. In semiconductor test, as noted earlier, the Magnum V's expansion beyond flash package test is underway as we've started delivering systems for both flash and DRAM wafer test. Recall that the flash and DRAM wafer test market is more than twice the size of the flash package test market. This market is expected to drive our memory market share from the high 20s to the high 30s over the next few years.
In semiconductor test, our R&D teams are working on products targeting new standards in the high-speed serial and RF markets. Our just-introduced UltraSerial60G instruments allow customers to fully characterize and test chips with interface speeds up to 60 Gbps. Targeted devices that use high-speed serial interfaces like PAM4, the instrument provides customers a cost-effective way to fully verify the performance of devices that deliver the immense bandwidth needed for 5G and cloud-based AI applications. In RF, our new instruments are testing silicon for a range of emerging standards, including 802.11ax, 802.11ad, and 5G cellular that are moving through our semiconductor customers' labs on their way to early production. The growing silicon content enabling AI at the edge is driving up device complexity and test time.
Whereas training algorithms usually get developed in the cloud, those honed algorithms are often deployed in the real world in edge-optimized silicon for AI inference and decision-making in real-time. Teradyne is well-positioned in this growing edge AI space. You can see this today in autonomous driving and automotive ADAS systems, where sensor data is interpreted and decisions are made in real-time to optimize safety and performance. You can also see in the increasing range of smart speakers interpreting voice commands with a combination of local and cloud-based AI. Smart cameras are another emerging trend where localized AI algorithms pre-process image data to make real-time decisions. In addition to organic investments, acquisitions will continue to play a role in our growth. Having recently completed the acquisitions of MiR and Energid, we continue to have an active M&A pipeline combined with a robust capital returns program.
In the first six months of 2018, we've repurchased $361 million of Teradyne shares and paid $35 million in dividends. We remain committed to this balanced capital allocation plan going forward. Now I'll turn it over to Greg.
Thanks, Mark, and good morning, everyone. I'll provide some brief comments on 2018 and then cover the second quarter results and third-quarter outlook. At the halfway point, our first-half sales total $1 billion, $14 million, and at the bottom line, we've achieved $1.04 in non-GAAP EPS. We're pleased with our first-half financial performance, particularly in an off-year for the mobility segment of our semiconductor test business. As described last quarter, our prior year's mobility tooling ramp will not repeat in 2018, which is largely why our first-half non-GAAP EPS is down $0.30 from a year ago. Even with this air pocket, our gross margins are running at 57% and our non-GAAP PBIT rate is 23% for the first six months. It's important to highlight that the same positive SemiTest structural dynamics, which we've described in the past, remain in place.
Starting in 2015, advancements in parallel testing hit their practical limits for complex mobility devices, which caused the ATE market to grow again. Previously, advancements in parallel tests had a compressive effect on the ATE market and masked the underlying growth in complexity end units. We expect mobility devices will continue to increase in complexity, with test times trending higher, offset at times by improvements in tester technology or customer optimization work. As a result, the mobility test segment will remain an important growing market for semiconductor tests, albeit volatile, as we've recently seen. The net of these trends is that even with lower mobility shipments in 2018, we should have solid profitability and free cash flow. Staying with SemiTest, ATE demand has been healthy this year across the automotive, industrial, and memory segments.
In automotive, ADAS, IVI, and hybrid vehicles are driving more silicon content along with shorter device life cycles. Automotive semiconductors also require testing under different temperature environments and take more test seconds than other devices, as escapes are costly. Longer term, AI should drive even more semiconductor content in automotive, where we have strong market share and expect above-average market growth. The ubiquitous industrial and microcontroller devices, where we also maintain strong share, are being driven by an overall healthy economy and greater end applications. Both EagleTest and the J750 have had strong demand two years running. These end markets are also finding new applications in energy management, factory monitoring, and IoT applications. Turning to memory test, as Mark noted, we had record first-half memory test sales of $139 million and are on track to grow our memory top line for the sixth year in a row.
Overall, our SemiTest business is tracking to its fifth consecutive year of non-GAAP PBIT solidly above 20%. Now moving to our high-growth industrial automation segment. We had sales of $62 million in the second quarter. Universal Robots was $57 million of this total and grew 45% from the year-ago quarter. Universal Robots set a new quarterly gross margin record and is on track to achieve a 60% gross margin for the full year as material cost savings from Teradyne supply line group continue. Recall that Universal Robots had a 51% gross margin when we acquired them in 2015. Universal Robots is also using its own cobot arms to assemble new UR cobots, improving quality and costs and freeing UR employees from many tedious and uncomfortable repetitive tasks.
MiR had second quarter sales of $4.5 million from the date of the acquisition and $5.7 million for the full quarter on a standalone basis, over 85% higher than the year-ago quarter. Energid made up the balance of industrial automation revenue. Staying with Universal Robots, we've been ramping our field distribution aggressively this year and expect those added resources to help us achieve higher growth in the second half. We expect that there will be waves of adoption that should cause UR sales to be above or below our 45%-50% model growth target for UR. For example, we've been selling mostly to SMEs and plant managers at the larger companies. At some point, we expect a wave of demand as larger companies adopt cobots in greater quantities as they become more aware of their fast payback and flexibility.
We also expect that new enabling technologies will open up much larger opportunities that to date have only just been scratched. For example, there are many tedious tasks across industrial manufacturing that increasingly can be automated with advanced path planning software and commercially available 3D vision modules. We believe that Energid's easy-to-use motion control software will provide the needed path planning capability to make automation of these tasks economically viable. There are also more and more third parties investing their R&D on our open platform to create plug-and-play solutions for an expanding array of tasks. For example, at Automatica, the large German automation trade show held in June, we counted over 40 third parties using UR cobots to demonstrate their products or end-of-arm accessories ranging from laser surface inspections to light assembly tasks. At MiR, we also expect a strong second half with mid to larger accounts driving demand.
Similar to Universal Robots, we have a significant market lead with our ease of use, fast payback, and distribution network. We'll also bring Teradyne synergies to MiR in supply line management, global expansion, and balance sheet muscle. Overall, in 2018, we continue to take strategic actions necessary to make solid progress towards our midterm $3.50-$4 non-GAAP EPS plan. Recall that the single largest growth contributor from the $2.34 non-GAAP EPS achieved last year is Universal Robots. With the new e-Series lineup, we've taken another big step forward. Moving on to the System Test business, defense and aerospace, production board tests, and storage tests are all expected to operate at model profitability or better for the year. Shifting quickly to wireless tests, we operated above model profitability in the first quarter and for the first half.
Turning to the balance sheet, our cash and marketable security balances stand at $1.3 billion, down $284 million from the first quarter. We returned $244 million of capital in the second quarter. Principally through $227 million in share repurchases, a bit more than planned due to opportunistic buying, and $17 million in dividends. Our share repurchases since 2015 total $1 billion at an average price of $27.12. Recall, we plan to buy back at least $750 million of our stock this year and return approximately $70 million in dividends. Looking ahead, we also expect to use our cash to make additional industrial automation moves to further our lead in next-generation industrial automation. That is why we plan to keep the balance sheet flexible. Moving to the details of the second quarter. Sales were $527 million, the non-GAAP operating profit rate was 25%, and non-GAAP EPS was $0.59.
We had no 10% customer in the quarter, and gross margins were 58%. You'll see our non-GAAP operating expenses were $175 million, up $10 million from the first quarter, primarily due to the expansion of UR's distribution programs, the inclusion of MiR and Energid for a full quarter, and higher variable compensation tied to higher profit levels. Our second quarter OpEx of $175 million versus a year-ago period is up by $3 million due to industrial automation acquisitions and distribution build-out, partially offset by lower variable compensation compared with the prior year. We plan to continue to invest aggressively to keep our high-growth industrial automation business on track against our 2021 model while keeping test OpEx flat, apart from variable compensation changes tied to profitability.
The detailed segment and unit-level sales for the second quarter, including the geographic breakdown for UR and MiR, are shown in a table in the presentation. Turning to the guidance for the third quarter. Sales are expected to be between $540 million-$570 million, and the non-GAAP EPS range is $0.59-$0.66 on 188 million diluted shares. Q3 guidance excludes the amortization of acquired intangibles, restructuring, and other non-cash convertible debt interest. The third quarter gross margin should run at 57%, and total OpEx should run from 31%-33%. The operating profit rate at the midpoint of our third quarter guidance is about 25%. Let me emphasize that we expect industrial automation quarterly OpEx to approach $35 million towards the end of the year, up from $16 million in the fourth quarter of 2017.
Notwithstanding our planned industrial automation OpEx investments, we expect this high-growth segment to achieve mid-teens operating profit rates for the full year. Shifting to taxes, our full-year tax rate is expected to be about 16.5%, up a half a point from our April estimate on stronger memory test sales at Nextest and analog test sales at Eagle. In summary, we've significantly extended our product lead at Universal Robots with the e-Series lineup. We added the higher payload MiR500 to MiR's product offerings. Outside of Teradyne, there's a growing pool of third parties developing solutions on our next-generation platforms. We see multiple waves of growth ahead in industrial automation.
We're also delivering solid profitability across our test and automation portfolio, taking the actions necessary to achieve our 2021 model that has our industrial automation segment at a $1 billion business, all while continuing to return significant capital to our investors. With that, I'll turn the call back to Andy. Thanks, Greg. Thea would now like to take some questions, and as a reminder, please limit yourself to one question and a follow-up.
At this time, I would like to remind everyone that if you would like to ask a question, to press star one on your telephone keypad now. We'll pause for just a moment to compile the Q&A roster. The first question will come from Richard Eastman with Baird.
My first question is around, Mark, I think you addressed this in your commentary. Is the expectation for the second half revenue for Ter still roughly equivalent to the first half revenue for Ter?
Yes, that's about what we expect. Of course, the industrial automation should be a bit stronger than the first half, and test will be a bit down, but 50/50.
Okay. Just a quick question around the systems test business as well as the wireless test businesses. I guess, if you do that math, was some of the system test business and wireless, was some of that pulled into the first half, some of that revenue? Again, just curious if the lumpiness around the quarterly revenue for those two businesses, or have we kind of stepped up to a new revenue level for those two businesses?
I'll take that one. It's principally in storage test within system test is inherently a lumpy business, and we get large orders concentrated. It's because we have a thin customer portfolio, and they tool up periodically, and when they tool up, the orders come all in one concentrated buying cycle.
Mm-hmm. Okay. All right.
I'd just add on wireless test, there's no evidence there of any sort of pull-in. I think the incremental demand we saw around cell phone Wi-Fi was just addressing the inherent complexity growth there, and our customers were a little caught short on some capacity. It's not a forward pull-in that we can see.
Okay, excellent. Thank you.
The next question will come from Vivek Arya with Bank of America.
Thanks for taking my question. First one on memory. When I look at what we hear from some of the front-end players and from some of the customers, and we look at the pricing trends, they seem to be more muted than the strength that you are seeing in memory. Why the disconnect? Are these share gains? Is this new product? What's the visibility around the utilization and orders in this segment?
Okay. First of all, visibility in memory is always pretty short, maybe it's three months. With that being said, there's always a disconnect in time between the front-end facilitizing fabs, which tend to be large multi-billion-dollar investments that customers make to bring a fab online, with a lot of tool orders rolling in and such. Then the incremental investment they make once the fab is running to facilitate test. As they start to turn on production in these large investments, they incrementally grow wafer volume and incrementally add test along the way. The lag time can be as much as six months to sometimes a year between when the tools go in on the front end and the fab's running at full volume. That's why you see a timing difference.
We don't see, in terms of test demand for memory, any slowdown at the moment, but again, I said the visibility is about three months on that. The other thing I will reinforce that I said before, the market this year is very strong, $900 million to $1 billion of test capacity. A more sustainable level, I believe, is $700 million to $800 million. Will we see that next year, for example? It could very well happen. A lot will depend on what the indigenous China customers do with their facilitation of test.
Got it. For my follow-up, it's clear that you're trying to move the company more and more towards this faster area of industrial automation. Do you need to change anything organizationally as you look at the transformation of the company, or do you think the way you have it set up right now is the right organizational structure?
Yeah. I think we're doing it exactly right at the moment. As industrial automation grows as part of the portfolio of Teradyne, there will be some adjustments we make structurally along the way. Right now, we have pretty autonomous and separate operating groups for each of those businesses. The one advantage we have here is leveraging our manufacturing capability, where we've achieved significant gross margin improvements. Leveraging where we need to and get some immediate benefit, but letting the IA business run with its own DNA and its own management team is the principle. Even now with MiR and Energid in the loop, we've got an umbrella strategy, but operating-wise, we're still pretty much running independently. At some point, a couple of years down the road, that could very well change, but not in the short term.
Okay. Thank you.
The next question will come from C.J. Muse with Evercore.
Yeah, good morning. Thank you for taking my question. I guess first question, trying to understand the sequential growth into the September quarter. Can you share with us how you're thinking about the storage side, whether that is sustainably strong or that downticks? As part of that, I would assume that you're seeing a meaningful acceleration in the UR IA part of things.
Yes, C.J. The storage test into the third quarter does downtick. That's not driving the growth. There's some good semiconductor test growth, particularly in memory in the third quarter. That's a piece of it, and that's some new opportunities that we are shipping to. It's a big plus. Certainly, UR is a big part of the second half. We target 50% for the year. For the first half, we're at 40%. We're targeting 60% growth over the corresponding period from a year ago. We've got a lot of actions underway, and we've hired a lot of people to make sure that we have the resources to pull that off.
Just to be clear, for total IA or UR, you're targeting 60% year-over-year growth for the second half of calendar 2018?
We're targeting the whole segment to get 50%. UR is 45 to 50. We want to get to 50. The first half we're at 40%. We hired a lot of people in the first quarter, the first part of this year, and they take about six months to become effective. We would expect a stronger second half, which you traditionally have in industrial automation.
Perfect. Makes sense. Just a quick follow-up on your target model for calendar 2018. Do you still expect, given the changes in mix here, that gross margins can be flat to up in 2018? Embedded in your OpEx guide in the slide deck, should we be thinking total OpEx, given what you know now of around $690-$695?
The gross margins, as you noted, have been improving over a period of time, and they're actually at the high end of our model for 2021 already. We're actually doing much better than we planned on that front, and that's coming across the portfolio. Universal Robots is a piece of that as well. We decided not to update that model quite yet because there'll be other things that could go the other way, or there's a lot of assumptions. We feel very good where we are at the gross margins. The OpEx, it's going to be the same strategy. We're going to keep test flat for the next several years, and industrial automation is going to continue to grow expenses.
If sales go up another 50%, I would expect OpEx would go up, but you'd end up with a similar profit rate, somewhere around 19%-20%.
Excellent. Thank you.
The next question will come from John Pitzer with Credit Suisse.
Yeah, good morning, guys. Congratulations on the strong results, and thanks for letting me ask the question. Mark, Greg, in your prepared comments on the SOC test side, you specifically talked about things your customers can do to optimize them. I'm just wondering if you could just help me understand and give me a little bit more detail. What levers do they pull for optimization? By how much do you think that can take down test time and sort of retard overall unit demand for testers? What's the real-world trade-off? Are they just willing to have things be good enough, and that's the optimization? Just help me better understand that.
Okay. Well, there's different levers customers can use to try to optimize. Historically, the one that we've talked about a lot in recent years that has reached its end is parallel test. That particular technique has run its course. There's other techniques. Another technique would be just trimming out tests in a long test program because you believe the probability of failure of that particular item is low. You do that with some risk that you might get some escapes that could affect either the end product manufacturing or the consumer experience. In environments where there's not a lot of technology change, if, for example, we're not moving to new silicon or we're not moving to a new lithography node, it's the same part that's now one year, two years old. That's a technique that people historically have used and continue to use.
It's not new. One of the things that works against that is a lot of the silicon rep rate here and the cycle time is quite short. Every year, as silicon gets reinvented, there's not enough time to learn what tests you can prune out and still be, let's say, safe that your quality level isn't going to dip. That's another one that as, in the mobility world, the cycle times have become short, that's sort of gone by and by. There's nothing really new here. Those are two techniques. Test time, let's say eliminating tests is still prevalent. Parallel test is less so. I'll just add to what Mark said. There are some techniques that aren't designed. They can impact test.
If yields all of a sudden improve substantially, you need to do less tests, because if yields are lower, you need to test more units to get the right number of good units out. The yield level between the prior generation and new generation does affect how many testers you need, and that's not really within great control of the customer. Also, when do they finish their tape out and when do they go to market? Is that three months? Is it four months? How much time is there for the peak build cycle? There's a bunch of other variables that get very complex and you have multiple moving variables. The ones that we see long-term drive the trend line steadily with these aberrations are the complexity and unit growth.
That's helpful. Then for my follow-up, guys, just on the SemiTest side, on the memory, clearly, we've seen some adjustments in front-end capacity. I'm wondering if you could just talk a little bit about some of the offsets that you might have around share gains that you're seeing in memory and/or just test times in memory going up as memory speeds are increasing and complexity's going up. How do I think about that?
Well, what I said earlier on the macro front is that slowing front-end investments in memory fabs probably impact test anywhere from six months to a year later. A lot of what is being bought today in test is for fabs that were facilitated six months to nine months ago. That's one thing. The new markets that we're entering in memory in wafer test are essentially two x the size of the flash final test market that we've been playing in up until this year. If you roughly, for rough numbers, say we've been participating up till this year in a $200 million flash final test market, we've now expanded our served market to $600 million by adding wafer test. We will incrementally gain share into that space over the next few years.
By the 2021 earnings target that we put out there, we expect we'll be in the high 30s to close to 40% overall share.
Perfect. Thank you.
The next question will come from Atif Malik with Citigroup.
Hi, thank you for taking my question, and good job on execution. First question on mobility. TSMC, on its earnings call last week, talked about a strong 7nm ramp, unlike 10-nanometer, which was a transitional node built for one key customer, 7nm is being widely adopted for high-performance computing, GPUs, server CPUs. TSMC is not going to ramp down 7nm capacity to convert to advanced nodes. What does it all mean for your semiconductor mobile test business for next year? I have a follow-up.
Yeah. The adoption of 7nm in mobility is happening now. It's not just a next year phenomenon, it's a this year phenomenon. The effect on test is really Two things. One is, it certainly means the devices have more transistors, so they have longer test times. That's a generic balloon for the test market. The other thing that Greg mentioned is the yields. What will yields be at seven nanometer versus what they were when people transitioned to 10? If yields are better, the test demand will be slightly lower. If yields are worse, the test demand will be slightly higher. The early indications on seven nanometer is that it's a pretty solid node, we don't see yields worse, therefore, there won't be another balloon from that.
In terms of what that means for next year, will people start early transitions to, or when will they start transitioning to five nanometer? I think if people stay on seven nanometer for two or three years, there'll be incremental yield improvement, which means there will be less test dedicated to testing bad parts. If we sit at a node for two to three years, that has a slight negative effect on the test market.
Any impact from China tariffs, as China is a fast-growing market for your industrial automation business. I understand you ship out of Europe for your UR and MiR businesses. Any impact from China tariffs that you are seeing?
The only impact to date is, the automatic test equipment code got caught up in the list of industries that tariffs would apply. When our contract manufacturer, Flex from Suzhou, China, ships into the U.S., there is a 25% tariff. As you can imagine, there are not a lot of shipments from China to the U.S. because most test equipment is going into Asia. For those shipments, say it is about $50 million, we are going to do the assembly and final configuration test in other locations, not China, and therefore, we should avoid the tariff, which would otherwise be about 25% of $50 million. Apart from that, there is no other impacts that we are seeing, and we have mitigated that. There is obviously uncertainty as to what might come next.
All we can do is sort of have contingency plans in place, which we do, in terms of if we have to move locations, how we would do that.
Okay.
The next question.
Thank you.
The next question will be from Toshiya Hari with Goldman Sachs.
Great. Thanks for taking my question. Mark, I had a question on the Magnum platform. How differentiated is it relative to solutions from your competitors, like Advantest and some of the local Korean companies? Is it differentiated enough for you to avoid any price competition? Kind of related to that, you talked about traction with the platform. Is it one large customer that you're involved with, or is it spread across multiple memory customers?
I think we're talking about the wafer test space. In wafer test, it's multiple customers. The competitiveness of the memory test space is actually reasonably high. It's a smaller set of customers, and it tends to be head-to-head competition. Other than the recent phenomena of new players emerging on the scene in China, it tends to be entrenched competitors any place they go for our share, or we go for theirs. I would say, in general, it's a more competitive space. You can see that reflected in both their gross margins and our gross margins. Ours are not that far out of line. They tend to be in the, I think they average in the low 50s, but it is more price competitive. We do have differentiation. We don't enter a market without differentiation.
The differentiation that we had on the flash final test side that allowed us to gain quite a bit of market share in a short time was tremendous. That happens once every 10 to 15 years. That was extraordinary and great. The differentiation we have in flash wafer test is not as great but still significant, which is why we've already had two customers convert to the platform and have others in flight. I don't expect that margins will be such far out of the norm of what we've seen in the past, and I do think there will be some disciplined share gains we'll see over time as we roll this product out.
Okay, great. I had a follow-up on the UR side of things. Mark, in your prepared remarks, you talked about the e-Series and some of the new tasks that it can address relative to your previous platform. I was just curious, to what extent does the introduction of the e-Series expand your SAM in cobots? I guess you guys have talked about waves of adoption in the UR business over the next couple of years. Does the introduction of the e-Series kind of allow you to grow above 50% over the next several quarters, or should we not think that way for the UR business? Thank you.
Okay. I think the e-Series won't specifically increase the SAM, but what it will do is increase our ability to have velocity into the SAM. Take some applications like polishing as an example. We've, up until before the e-Series was introduced, had polishing applications that relied on some third-party partner add-ons to the product and some relatively complex programming on the part of the customer. What the e-Series does for polishing is tremendously simplifies those applications. It embeds a lot of the touch and force measurement into the native architecture of the product, and by extension, simplifies the programming. We should get more velocity into polishing, as an example, from the e-Series. That's true across a broader set of applications. Separate from what the e-Series enables, there's other things happening in the ecosystem that will expand the SAM.
Those get back to, Greg mentioned 3D vision as a technology that is developing rapidly, and as it becomes practical to adopt that on the cobot, we will expect the SAM to increase to allow more, let's say, picking applications to be addressable with the product. That's coming. That might be something we see more start to pick up next year, but that's coming. To your other point or question, can we grow greater than 50% and the waves of adoption that Greg talked about? We do think there will be that opportunity as we get these two things solidly into the market. The effects of that are probably more next year than this year.
Thank you.
The next question is from Timothy Arcuri with UBS.
Hi. Thanks. My first question is on SemiTest. I know, Greg, you'd been talking about SemiTest revenue being, or rather SOC revenue was roughly going to be 50/50 for the year. I know that you just said that total company revenue will be 50/50, but previously you had said that SOC revenue would be 50/50. Now that you're taking the TAM sort of up to the high end of the market, or the high end of the range, do you think that the back half for SOC could be a little better than 50/50?
I still think it's in the 50/50 neighborhood. It could be, but I wouldn't say pencil that in. Again, we didn't have the big Q2 surge. Everything else is looking pretty solid. We have trouble predicting the fourth quarter this far away, but at the moment, it looks like it could be reasonable, and our strong quarter is the third quarter, so that's the quarter that's going to cover a little bit of whatever happens in the fourth quarter. It could be 50/50.
Got it. Okay, awesome. For Industrial Automation, you guys were saying 50% this year growing before MiR, and MiR was going to add, I think, $25 million to $30 million when you add MiR and Energid for the year. That was going to add about 10% for the year. Now, it sounds like you're still saying that the whole segment's going to grow 50%, maybe 55% for the year. It seems like a little bit of a downtick. It seems like maybe you've lost $20 million to $25 million for the year in that segment. Am I not reading that right? Is there something happening there?
No, MiR is on track. Their plan is to get to about $30 million of revenue. We could handicap that and say $25 million, but keep in mind, they were $12 million last year. Whether they get to $25 million or $30 million, it's going to be 100% type growth plus. We're quite confident we'll be mid-20s, but the stretch target is $30 million.
Yeah, I guess my question was more that you were saying that the segment would grow 50%. You bought a company that's doing $30 million during the year, and you're still saying that the segment grows 50%. Something seems like it got lost.
No. Last year it was $12 million in sales. Last year. This year, it's going to be $25 million, maybe upwards of $30 million. That's 100% growth. It may be the time period that you jotted down or what we said, there's some confusion. When we said 50% to 100% growth, we were talking about the prior year as a starting point. Even with that said, we would expect next year to be 50% plus growth. The thing about MiR is it is so much earlier than Universal Robots. Therefore, customers are even less aware of what's possible. When they do learn about it, we tend to have leads that can be 10 pieces, 15 pieces. It's not one or two.
It might be a bit lumpy, but it certainly has the ability to grow at a much faster rate because it's small, and the orders are larger quantities.
Okay. Yeah. I was talking about the segment as a whole, but yeah, great. Thanks.
Okay, I lost it.
Just to clarify on that one question again, for the total IA segment, what we've been saying is UR grows at about 50% plus a year. We add in MiR, which is growing in that 50%-100% range, and this is over the course of a couple of years. We would expect the combined segments to be growing in that long-term model we put out 50%-55%. There's obviously room for some upside on that, but as we've demonstrated with UR, we've had three years of growth beyond the 50% level. We're still sticking to that 50% level for this year as well. Adding MiR in does give us some upside potential.
The next question is from Tom Diffely with D.A. Davidson.
Yes, good morning. When you look at the new e-Series, it sounds like a lot of that is software driven. Does that give you an opportunity to retrofit your 25,000 or so robots in the field?
No, we would not be retrofitting those cobots. This product is different enough that it would not make sense, and those cobots are all being used very productively. No.
Okay. Just to follow up on the question about your facilities in China, how long would it take you to move final assembly from a Chinese facility to another region?
It would probably take us, to be up in full production, probably six weeks. If the rest of the world is doing the same thing, then we might run into a problem.
Mm-hmm. Okay. Thank you.
The next question will come from Mehdi Hosseini with SIG.
Yes, thanks for squeezing me in. Just going back to the industrial automation, appreciate the breakdown between UR and MiR. You mentioned the longer-term growth of 50%-55% with a bias to the upside. Is that upside going to be dependent on additional acquisition? I have a follow-up.
No, it's not dependent upon additional acquisitions. It's dependent solely on new enabling technologies that expand what these products can do, and these enabling technologies are things like Energid path planning or these vision modules. I'm speaking about the arm now. There's new technologies that we see that should hit the market and be usable with an easy-to-use framework, and that expands the market. We don't need to buy anything or acquire anything.
Got it. Shifting gear to SemiTest, specifically SOC. I like the way you framed the opportunities as 7nm, but I want to dig in. To me, it seems like the incremental growth drivers are driven by GPU and new ASIC chips, two areas where historically Teradyne has had lower market share. Is there any specific program or what are your thoughts about getting share? Because in order to hit that 2%-4% CAGR growth for SemiTest, you need to see your SemiTest market share rebound. In my opinion, the GPU and ASIC will become critical. Any thoughts will be appreciated.
Yeah. In my prepared comments, Mehdi, I was talking about how edge AI, and ASICs in particular, are areas we are participating in and have good share. The GPU area itself is not something that we have good share in. We obviously have some targets there, but it will take technology discontinuities to enable us to get into the specific GPU market. The growth rate, if you talk about the growth rate of these class of parts, today, as a baseline, roughly speaking, we think that the GPU and ASIC AI-related chips are $100 million-ish of ATE tests, and they're probably going to grow at 10%+ maybe 20% clip over the next few years. That's a combination of both GPU and ASIC, which incorporates these AI at the edge elements.
For that piece of the market that we think is also going to be growing in this 20% clip rate, we will and are participating. For the GPU segment, we're not counting on it, but we are positioned for certain technology discontinuities that could give us a shot.
Great. Do you mind if I just have a follow-up? On the ASIC side, do you think your customers are going to be subcoms? Is that going to be more like foundry, or is there a new category of a customer that is emerging? I'm not sure who are the customers looking forward for this specific area.
Well, it's like other parts of the fabless world. The actual test equipment buyers will be foundries and subcoms, and that's true today. The actual selling of the capability of the tester will be to the design house. In some cases, you'll find an automotive manufacturer designing an ASIC for autonomous driving for their vehicle, and it will be that design team that will choose the test equipment. In other cases, an automotive manufacturer might outsource the design to a third-party silicon house. That third-party silicon house will be the house that decides on the test supplier. It's usually almost always the design team.
Got it. Thank you. Very helpful.
The next question will come from Krish Sankar with Cowen and Company.
Hi. Thanks for taking my question. I have two of them. One is, on the memory market, looks like your market share is probably in the mid to high 20s, I guess you've spoken about getting to a 40%-plus market share over the next few years. Is it primarily coming from moving into flash wafer test, or is there a way to quantify how much is flash wafer test and how much would DRAM be as part of that market share mix? I had a follow-up.
It moves around every year, but a way to think about it is the market segment's 25%, 25%, 25%, 25%. Final test, 25% DRAM, 25% flash, wafer test, 25% DRAM, 25% flash. By entering the wafer test portion of memory test, we've increased our served market from 25% of the overall market to 75% of the overall market. Now, in any given year, that can move a little bit, but average it over three or four years, that's about how it breaks down.
Got it. All right. That's very helpful. I had a follow-up on the wireless test. I understand 5G is going to be mass deployment still like two, three years out. If you look at some of your comps, like Keysight, they're already beginning to get traction on 5G testers. My understanding was, from a test standpoint, this could be probably more of an early cyclical, but I'm just trying to wonder why you guys are not seeing the traction in 5G while some of your peers are.
Our differentiation is much more in the production environment when 5G gets rolled out in high volume. We excel in that environment. We have the easiest-to-use testers, great throughput, great support. The company you mentioned and others tend to be more lab-based, they're there earlier, the features that they have often isn't what you need in production. It's probably too costly, but it's very useful in a lab.
Thanks, [Mate].
Okay. Operator, we have time for just one more question, please.
Yes, sir. The final question is from Patrick Ho with Stifel.
Thank you very much. Mark, maybe if you could give a little bit of color in terms of the SOC trends that you saw in the June quarter. Was it a steady kind of buildup that led to the upside, or was it a kind of a sudden surge at the end of the quarter with some of the marketplaces you talked about, power management, analog, automotive?
It was pretty steady. Analog, for example, throughout the quarter, continued the order rate at a very steady clip. It usually can be lumpy, but our principal customers in automotive and industrial that are analog suppliers have just been, for a couple of years now, but increasingly, buying at a steady rate. Automotive, similar. If anything, we saw one isolated area that I would say might have been a bit of a surge toward the end of the quarter, and that was around image sensors, which is usually a mobility-related issue, image sensor testing. Everything else was pretty steady.
Okay, great. This is my follow-up question, following up on the SOC test trend. These trends, obviously, as you mentioned, have been actually pretty consistent over the last few years on these type of devices. I know it's a little bit early, but as you look at 2019, given some of the new marketplaces, new applications, and new products that are coming out, do you see another healthy year in these type of, I guess, the low-end IC device test market for 2019?
Yeah, I think that's really hard to call. I don't see anything that is an anchor at the moment. I do think the real next uptick coming, and it's probably late 2019 into 2020, is 5G. That will produce a lot of demand because the technologies tend to obsolete the semiconductor test equipment that's out there. That will be a very discernible and additional wave that comes. The other markets and the other buying that's going on now, I would say, feels and has been for a couple of years and feels right now to be just steady as she goes.
Great. Thank you.
Great. Okay, everybody, thanks for joining us today. This concludes the call and look forward to talking to you in the days and weeks ahead.
Ladies and gentlemen, thank you for participating in today's conference call. You may now disconnect.