泰瑞達 (TER) 2017 Q3 法說會逐字稿

Good morning.

My name is Zetania, and I will be your conference operator today.

At this time, I would like to welcome everyone to the Teradyne Q3 2017 Earnings Conference Call.

(Operator Instructions) Thank you.

Mr. Andrew Blanchard, you may begin your conference.

Thank you, Zetania.

Good morning, everyone, and welcome to our discussion of Teradyne's most recent financial results.

I'm joined this morning by our CEO, Mark Jagiela; and our CFO, Greg Beecher.

Following our opening remarks, we'll provide details of our performance for 2017's third quarter and our outlook for the fourth quarter.

The press release containing our third quarter results was issued last evening, and we're providing slides on the investor page of the website that may be helpful to you in following the discussion.

And 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.

Additionally, those forward-looking statements are made as of today, and we take no obligation to update them as a result of developments occurring after this call.

During today's call, we'll make reference to non-GAAP financial measures.

We've posted additional information concerning these non-GAAP financial measures, including reconciliation to the most directly comparable GAAP measure where available, on the investor page of our website.

Also, between now and our next earnings call, Teradyne will be participating in investor conferences hosted by R.W. Baird, UBS, Goldman Sachs and (inaudible) America.

Now let's get on with the rest of the agenda.

First, Mark will comment on our recent results and the market conditions as we enter the fourth quarter.

Greg will then offer more details on our quarterly financial results, along with our guidance for the fourth quarter.

We'll then answer your questions.

And this call is scheduled for 1 hour.

Mark?

Good morning, everyone, and thanks for joining us today.

In my prepared remarks, I'll cover 4 topics: our 2017 highlights, key trends which we expect to drive long-term growth in our test businesses; an update on Universal Robots' expansion and strategy; and an early view on how we're thinking about 2018.

Greg will then provide additional color, along with the financial details.

As you saw in the release, our sales and earnings for the third quarter were above our guidance, and our fourth quarter outlook is substantially above our earlier view.

This performance is a result of a very healthy Semiconductor Test market, especially in memory test.

Seasonally, our third quarter Semiconductor Test orders were at the highest level since Q3 of 2000.

Additionally, Universal Robots continued to show strong growth, with sales up 70% from the same quarter of last year, well above our 50% baseline rate.

Through 9 months, we've generated over $1.65 billion in sales and non-GAAP earnings of $1.88 per share.

For the full year, at the midpoint of our fourth quarter guidance, we expect sales of above $2 billion and non-GAAP earnings of around $2.22 per share.

This puts us well ahead of our $2 per share EPS goal that we originally targeted for 2020.

2017 will be the second consecutive year of growth in the SOC test market with an estimated $2.6 billion market size, up 8% from last year.

We expect Teradyne's SOC test sales to grow above that rate at about 20%, and our market share will increase several points from 2016's 51%.

In memory test, we expect the 2017 market to be about $650 million, up roughly 55% from last year.

Q3 memory test orders of $65 million were at record levels, and our full year memory test sales will grow about 20% to about $180 million.

Our memory share will fall back a few points, as some of the segments we do not yet serve are growing faster than the overall market in 2017.

Combining memory and SOC test for the first 9 months, our sales are up about 23% from last year to $1.3 billion, while the combined market has grown about 15%, clearly a strong year of market growth and an even stronger year for Teradyne growth.

At Universal Robots, third quarter sales were about 70% above the same period last year, with all regions above our 50% growth baseline.

Through the first 9 months, we've grown 77%, a rate that will likely moderate a bit in Q4.

For the full year, we expect UR to grow about 65%.

In addition to top line growth, we are also showing improved gross margins, as the benefits of cost-down and selective price increases are making an impact.

In System Test, year-to-date revenues of $112 million are down 20% from revenues in the same 2016 period mainly due to a falloff in Storage Test demand.

However, we expect to recognize revenue in Q4 for a new application of our Storage Test product that will bring the group to about flat top line performance with 2016.

This new product test chips in a simulated system-level environment to capture elusive failures that are hard to detect on traditional ATE.

The initial application is in complex, high-volume, mobility-focused semiconductors.

At LitePoint, sales in the quarter were up 11% from Q3 last year and up nearly 20% for the first 9 months, off a weak 2016 market environment.

This growth has been the result of modestly expanding demand in the connectivity space.

New connectivity standards like 802.11ax WiFi are not expected to drive meaningful new demand until 2019.

In cellular, 5G is in the early preproduction stage and not yet driving volume.

However, we are seeing growth in LTE narrowband IoT space.

For the full year, we expect LitePoint to grow at about a 10% to 15% rate from last year.

As we first noted in 2014, after many years of trending down, the Semiconductor Test equipment market inflected and has been growing since 2015.

In 2016, we modeled the SOC test market to grow at about a 1% CAGR on a trend line basis from the average of the 2014/'15 market sizes of approximately $2.2 billion.

Through 2017, the actual SOC market growth has been about 8% from that level.

In light of these trends, we will update our model and targets in the January call.

While recent signs are encouraging, we expect volatility to be a part of the test markets in the future, as they have been in the past.

The key point is, even with volatility, the market is on a growth trajectory and the demand drivers we identified earlier remain in place.

These include the diminishing impact of parallel tests, increasing device complexity and increasing quality demands of semiconductor customers.

In memory, volume increases, higher bit density of 3D NAND and the proliferation of high-speed interfaces that are now commonplace for smartphone and SSD applications has driven growth of the memory test market this year.

For Teradyne, we've been successfully growing share in the NAND package test segment with our Magnum family of testers through the discontinuity created by the emergence of high-speed flash interfaces.

At Universal Robot, our strategy remains unchanged.

Beginning in 2016, to accelerate growth, we aggressively increased our quarterly OpEx investments by about $5 million per quarter to -- or nearly 70% up from the prior year.

We followed that up by increasing 2017 spending by a similar amount.

Those investments have enabled UR's growth rate to increase from 57% on a stand-alone basis in 2015 to the mid-60s for the full year 2017.

We've also improved gross margins along the way.

In the same way, in 2018, we'll continue this pattern of investment at UR to deliver further growth.

Greg will outline specifics of those investments, but one I'd like to highlight is the UR+ open architecture partner program.

Building out our ecosystem of partners that provide the hardware and software peripherals for specific end market applications is essential to drive our growth.

UR+ provides partners with the ability to adapt their products to natively operate on our UR platform through an API that links them to our ease-of-use software paradigm.

The customers get both applications breadth and a seamless ease-of-use experience, allowing faster, simpler cobot deployments.

At UR, we provide a universal cobot and a universal software platform that enables an ecosystem of innovators to provide plug-and-play add-ons that customize our products for specific end applications.

UR+ continues to enable an expanding breadth of applications.

For example, next month, at the FABTECH trade show in Chicago, there will be a demonstration of a UR+ gas metal arc welding package.

The product is for high-mix, low-volume applications that can be deployed in existing manual welding booths.

No floor plan changes or special facility changes are needed.

This reduces the costs and improves repeatability for metal fabrication shops.

From a programming perspective, the operator has only to identify the start and stop points for the welder, and the cobot does the rest.

A very different application is a small U.S. eyeglass maker that is using a UR+ partner gripper with our UR5 cobot to manufacture frames.

The ability to change tasks quickly gives this customer the flexibility to compete with much larger players.

The common thread between these applications is the tight software integration, allowing fast programming and short implementation cycles.

As we look ahead, we expect UR growth to remain at a 50%-plus clip for the foreseeable future.

We continue to explore additional opportunities, both organic and through M&A, to further accelerate UR's growth.

We expect labor availability, costs and product quality requirements will continue to drive demand for UR's unique human-scale collaborative automation solutions.

As is typical this time of year, visibility into our customers' demand for next year is limited.

That being said, our preliminary estimate is for an SOC test equipment market of about $2.3 billion to $2.7 billion.

In memory test, we forecast the market to be between $700 million and $800 million.

We expect the key end-demand drivers to be memory, the mobile market, automotive and the industrial markets.

While the next year is difficult to forecast, we are confident that expanding complexity, short semiconductor life cycles and increasing use of semiconductors across all segments of the economy will drive long-term growth.

This, combined with the continued market share gains, provide positive momentum for Teradyne's Semiconductor Test business.

At System Test and LitePoint, we expect similar market conditions next year as we saw in 2017.

Wireless Test growth will likely begin to turn up in 2019, followed by a more significant buy-in in 2020 as 802.11ax and early 5G cellular buying picks up.

In summary, for the full year 2017, we expect revenue of over $2 billion and to exceed our $2 EPS target well ahead of our 2020 plan.

While it's difficult to accurately forecast the size of the Semiconductor Test market in 2018, we're confident of the long-term drivers powering that market.

Combining the strong test business with the continued 50%-plus growth at Universal Robots positions us well for the road ahead.

With that, I'll turn it over to Greg for the financial details.

Thanks, Mark.

And good morning, everyone.

I'll start with a quick summary of 2017, as the finish line is in sight.

I'll also add some commentary on the trends and actions that are driving our growth, along with the third quarter results and fourth quarter guidance.

Starting with the 2017 financial summary.

Both our top line and non-GAAP EPS are projected to be up quite nicely over last year.

Factoring in our fourth quarter guidance at the midpoint, sales are tracking to be up 19% over last year, while non-GAAP EPS is expected to be up 47%.

For the full year, the projected non-GAAP EPS of $2.22 is 3 years ahead of our midterm $2 target outlined last year.

This favorable pull-in is principally due to a stronger semi test market, along with ongoing share gains.

We're also getting solid earnings lift and high growth from Universal Robots while managing our average share count.

We'll update our model and 2018 capital return plans on our next Investor Conference Call in January after we complete our midterm planning.

As Mark noted, we remain driven by rising complexity and unit growth rather than new nodes.

Semiconductors are at the heart of today's connected world and ubiquitous in many products or services that we use in our daily lives.

This central role fuels a constant flow of new designs,[strengths] and packaging advancements.

These changes bring added complexity, which often leads to longer test times and initial yield degradation, both of which trigger added test capacity.

The added complexity also drives the need for more robust test coverage to find the hard-to-detect faults earlier and in certain cases to fine-tune the electrical characteristics to hit optimal performance.

In a more granular level, ATE customers also have high switchover costs from one platform to another as they develop proficiency and tools around the tester programming and debugging environments.

This necessitates significant product advantages to move market share.

Consequently, targeting the right segments and winning customers remains critical so that share gains come largely from a rising tide in the waters we're in versus trying to get every point of share gain from battling the incumbent.

We also aren't directly driven, as the front end is, by new nodes.

Nonetheless, these performance advancements and strengths add complexity, big growth and often unit growth as well, so eventually we get our share.

We'll remain volatile simply due to the nature of being a derivative to a very large market, with small inflections having a pronounced impact on us.

Our volatility on a quarterly basis is tied to predictable consumer cycles and principally tied to new mobility launches in the fall, along with back-to-school and holiday electronics buying.

Our volatility on an annual basis is somewhat more opaque, but it's largely tied to the jump in device performance.

Compared to our past, we're far less volatile than in the prior decade due to a more efficient supply chain, shorter device life cycles and a broader portfolio.

The key actions in semi tests remain keeping a sharp focus on selective segment and customer targeting while maintaining strong gross margins and lean OpEx.

In the near term, we expect growth in China, memory and the continuation of the past demand trends.

This disciplined approach has allowed us to maintain an average non-GAAP PBIT rate of 22% at the company level since the start of this decade.

By comparison, this is 23 points of non-GAAP PBIT above the prior decade, when we gave profits and more back in downturns.

We're also pleased to report that we're on track to gain about 2 points of ATE share this year to about 50%, which would make this the sixth straight year of semi test share gains.

Shifting now to our high-growth automation business, Universal Robots.

The trends are clearly very favorable.

There are several third-party reports that have the cobot market at 1 billion or more in 2020, which fits with our assessment.

In short, there are tens of thousands of tedious and highly repetitive human-scale tasks that would benefit from UR's safe, low-cost and easy-to-program cobots.

We expect that it'll become increasingly critical to automate these repetitive, tedious tasks to maintain high quality and cost competitiveness.

In addition, we see some cobot subsidies being offered in China to ensure that their local companies take advantage of cobots.

Shifting now to the key actions to stay ahead and continue to grow at 50% or more.

First is about building a greater awareness of what is possible today with cobots.

Many potential buyers still do not understand how easy it is to automate repetitive tasks without cleansing or redesigning the workshop.

Much of our business is from client managers at large companies who buy locally without corporate involvement or small- and medium-size enterprises that move quickly from awareness to sell.

We do expect an inflection with larger buying in the future versus the small ordering of about 2 cobots on average today, as larger companies, which tend to move cautiously to new technology, embrace the power of UR cobots.

To accelerate this inflection, we're sponsoring many more trade shows, advertising programs, web educational content, co-funded distributor, cobot sales resources and so on.

This awareness campaign will continue as if a very big task with many possible cobot end-users.

We expect these end users will expand over time as easier-to-use accessories addressing more tasks are developed on the UR platform.

Critically important, though, is how we plan on staying ahead, as we expect new competitors will join the cobot field.

First is strengthening our sales channel of distributors and integrators both in increasing their sales velocity and getting the best channel partners aligned to our platform.

Through more UR sales and tech support, along with advanced training programs, we're growing the product sales velocity this year by about 50% for partners that were onboard with us last year.

We're also drawing up a global map with strategic partners for the spots we're not adequately covering.

Next is to get an apps ecosystem of turnkey solutions on our platform so that we'd lower the implementation costs and risks.

This allows us to have many third-party-developed plug-and-play solutions that can be accessed from our UR+ portal.

We haven't heard of anyone else with this type of third-party ecosystem backing them.

Third is, of course, to invest in R&D to make our cobots even easier to program with, for example, adding new wizards that shortcut programming even further.

Lastly, we'll continue to leverage Teradyne's strengths to improve UR's overall performance.

Shifting to System Test group.

Our mil-aero group is driven by avionic upgrades such as faster bus interfaces for advanced radar and advanced sensors, while production board test is driven by automotive, industrial and server-ed PCB demand.

In mil-aero and production board tests, we're growing sales at about 3% this year and tracking to modeled profitability.

Our third leg in the System Test group, Storage Test, is tied to sporadic HDD and SSD demand.

And the new system-level test application, which Mark outlined is launching in the market now.

We expect to be solidly profitable, starting in the fourth quarter.

In Wireless Test, LitePoint will remain in a lull before tooling for 802.11ax and, further out in time, the size of a 5G production cycle start.

Our actions have been to resize the business last year and stay focused on new growth.

So far this year, we're running at modeled profit percentages and getting good traction on the longer-term opportunities.

Now a reminder on our capital allocation plans.

We're buying a minimum of $200 million of our shares this year while returning about $56 million in dividends to shareholders.

So far this year, we've spent $152 million to acquire 4.6 million shares at an average price of $32.66.

Since 2015, we've bought back 27.1 million shares on aggregate at an average price of $22.06, totaling $598 million.

Our cash and marketable securities total $1,848,000,000, up $228 million from the end of the second quarter due to strong profits and strong accounts receivable collections in the quarter.

We have $742 million in the U.S., and the balance is offshore.

About 80% of our annual cash generation will be offshore this year.

Moving to the details of our third quarter.

Our sales were $503 million.

Gross margins of 59% was our highest quarterly rate in 4 years, driven by favorable product mix.

A clear bright spot in gross margin is the improvement of Universal Robots' margin to 58% from 54% in the third quarter last year.

Company non-GAAP operating profit rate was 26%.

And non-GAAP EPS was $0.54.

We had one 10% customer in the quarter.

You see our non-GAAP operating expenses were $163 million, down $10 million from the second quarter due to lower variable compensation accruals on decreased profit levels.

Total company OpEx in the third quarter this year at $163 million is up $13 million from the year-ago third quarter due to higher variable compensation accruals on higher profits and higher spending at Universal Robots.

We expect our full year 2017 OpEx, excluding Universal Robots and normal changes in variable compensation, to be essentially flat, while UR's full year OpEx will grow year-on-year to about $64 million from $43 million in 2016.

Looking ahead, we plan to keep aggregate spending flat in our test businesses, except of course for variable compensation which will move without profitability and growth.

OpEx at UR will step up in the first quarter and grow in the second half as well.

We've included a slide that shows all of our OpEx changes are due to Universal Robots' growth or swings in variable compensation.

Now moving to the segment-level detail.

Semi test bookings were $295 million, with broad-based strength in memory, microcontrollers, analog, image sensor and mobility.

SOC test orders were $230 million.

And memory test orders were $65 million, a quarterly record driven by flash applications.

Semi test service orders were $43 million of the total.

Semi test sales were $397 million in the third quarter, with SOC making up $350 million and memory test the balance.

Semi test service revenue totaled $70 million in the quarter.

Moving to systems test.

Orders were $42 million in the quarter, and sales were $35 million.

Shifting to Wireless Test.

We booked $33 million, and sales were $31 million in the third quarter.

At Universal Robots, orders in the third quarter were $40 million, and sales were also $40 million.

Regionally, UR's third quarter sales broke down 43% in Europe, 26% Asia, 23% North America and 8% rest of world.

Sales for the fourth quarter are expected to be between $420 million and $450 million, and the non-GAAP EPS range is $0.31 to $0.37 on 199 million diluted shares.

Q4 guidance excludes the amortization of acquired intangibles.

The fourth quarter gross margin should run about 55%, down from a very strong third quarter due to product mix.

In total, OpEx should run from 35% to 38%.

The operating profit rate at the midpoint of fourth quarter guidance is about 18%.

Shifting to taxes.

Our full year tax rate is expected to be about 17.25%, up 75 basis points from the July estimate due to strength at Nextest, our memory business, which is a U.S. business that carries a higher tax rate.

Please note that we expect our tax rate to step up to 19% for 2018.

On a quick housekeeping note, please be advised that we expect no material changes from the pending revenue recognition changes required under ASC 606, which takes effect from January of 2018.

Our free cash flow year-to-date totaled $406 million, driven by strong profits.

In summary, we're on track to hit our $2 non-GAAP EPS plan 3 years early.

We're gaining share in ATE for the sixth straight year.

We're growing Universal Robots above 50% again this year.

And we're maintaining steady financial discipline and returning capital.

With that, I'll turn call back to Andy.

Thanks, Greg.

Zetania, we'd now like to take some questions.

(Operator Instructions)

(Operator Instructions) Your first question comes from the line of Jagadish Iyer of Summit Redstone.

Two questions.

First, I was just wondering.

Why is there a resurgence in memory test?

Is there an inflection in testing that you can bring into light?

And how should we think about DRAM and NAND split here?

And I have a follow-up.

Yes.

So memory test really took off this year.

It surprised us.

If we go back to this time last year, we were estimating that the memory test market would be roughly $500 million.

It's going to be perhaps $650-ish million.

It's really driven not by a technology change per se.

The bit density growth is one thing that drives memory test demand, the unit volume.

And then changes in device interface speeds obsoletes older equipment.

So we do see obsolescence going on in flash and DRAM tests because of the higher-speed interfaces.

And then outside of that, most of it is just bit growth and unit growth.

There's also a lot of construction going on of fabs for memory expansion in China.

Those have actually not yet affected the memory test demand.

Those are still to come online.

It'll probably be a late '18 or '19 heavy test equipment tooling cycle.

So as we look out over the horizon, provided the bit demand remains robust, which it looks like it will, we should have several good years for memory demand -- memory test demand.

That's great.

And then as a follow-up, while you continue to invest in UR, when can we see an inflection in the operating margins going forward?

You have seen some uptick in the gross margins, but realistically how should we think about the operating margin as we look at, say, over the next 12 to 24 months?

Jagadish, this is Greg.

This year, for example, we expect to be -- meet our 15% operating profit target or be slightly above it.

So it's happening this year.

We think, long term, we'll get to 20%.

And the long term isn't 3 years away, so it could be sooner than what we earlier were modeling.

The gross margins have improved nicely, and we're growing sales at a very high rate.

So I would expect the 20% operating margin will become more in sight as we get further to the next year or two.

Your next question comes from the line of Krish Sankar of Bank of America.

I had two of them.

First one, Mark or Greg, this year, out of $2.6 billion in SOC test market, mobility was probably $1 billion or $1.2 billion; GPU, $100 million.

And auto test looks like $400 million.

Can you give us a similar composition for what you think SOC looks like in 2018?

Yes.

For 2018, again, it's a little bit of reading tea leaves at this point, but I do not expect that mix to change much.

The automotive and mobility spaces that came on strong this year should continue to be strong next year.

Memory again should be strong next year, so not a lot of change in the mix.

Got it.

Got it.

And then a follow-up on the robots business.

Some of your competitors have higher payloads, while you guys still have just 3 SKUs of URs.

Do you plan on developing a higher payload for the cobot business?

No, we have no plans for a higher-payload cobot.

Our cobots are exactly human scale, very flexible, pass without redesigning a cell.

So we don't see the higher payloads as a market that fits what we're trying to do.

Your next question comes from the line of Atif Malik of Citigroup.

Mark, if I look at the comments out of your peer in Japan, they're talking about a $3 billion test market for next year.

You guys are a little bit below that, $2.3 billion, $2.7 billion.

And you're generally more conservative.

What takes this to the high end of the SOC test market next year?

And then I have a follow-up.

First of all -- yes.

First of all, the peer -- Advantest forecasts for the market sizes exclude service.

And our sizes include service.

So actually, they're at $3 billion excluding service; and $800 million for memory; and I believe, $2.2 billion for SOC.

In the service business, you have to add another $500 million to $600 million to that, to their -- so in truth, we're pretty close to each other.

And for us, this year, $2.6 billion market is -- the tale of the tape isn't complete here.

We could end up with a slightly higher market than $2.6 billion by the end of the year.

But as we run into next year and the visibility we have at least through the early part of next year shows continued strong demand, we could easily be upside of the north end of the range we're talking about now, as we were last year.

This time last year for the SOC market, we said $2.1 billion to $2.5 billion.

It's coming in at $2.6 billion.

So it's -- the message I really want to give is it's very difficult for us to create a precise forecast at this point in the year.

We look at a lot of top-down factors.

Very few customers give us forecast because they're unable to forecast.

So we're looking at what we see in the pipeline around complexity trends for the devices that are coming out of design and into production to try to estimate next year's demand.

Got it.

Very helpful.

And then Greg, on the gross margins, you guys made improvement this year.

Can you just talk a little bit more about what drove the margins higher this year?

You mentioned product mix.

And as Mark commented, you're looking at a similar mix next year.

How should we think about gross margins broadly for next year?

Yes.

This was a very strong year for us, gross margins up.

SOC was very strong.

LitePoint was stronger than we expected.

On the flip side of the ledger, we had less Storage Test business, which pulls margins down.

And when you turn that around, we expect next year more Storage Test business, so that would be one thing by itself that would move the margins a bit down.

And there's probably a healthy amount of memory business coming our way next year, too, and some of that might have lower margins as well.

So margins are always a bit difficult to forecast.

The good news is we've been able to get material costs down year after year; and make significant improvements at Universal Robots, for example.

So we have a stronger starting spot, but we do see a couple of headwinds with Storage Test and some memory business.

But apart from that, those are the only 2 things I can think of that could have a quantifiable impact.

Your next question comes from the line of Richard Eastman of Robert W. Baird.

Just 2 quick questions on the semi test business.

Mark, when you look into the fourth quarter here, we've had 2 years in '15 and '16 where we've had this pull-forward of orders into the Q given one of the big mobility customers' plans for new products.

As we move here into this fourth quarter, does that order trend on the mobility side for semi test, does it look more like '13 and '14 where we have more of a normalized kind of order pattern?

Or is there still this expectation that we'll see the big mobility vendors, chip vendors, pull and make sure they get their test in the -- test orders in the queue earlier?

Yes.

That's always, for the last several years, been a big question.

And it's really hard to call because the order window is about a -- between these '13, '14, '15, '16, '17, it's a 8-week window that can move depending on their internal planning cycles and their order release timing.

So the timing isn't clear.

It could be late Q4.

It could be early Q1.

I think the ship-off schedule, independent of when the orders book, will be similar, meaning Q2, Q3 kind of concentration, as it's been in the prior years.

Okay.

And then just a question on the semi test.

The orders here in the third quarter, I presume, were kind of on the UltraFLEX-M side or Magnum test side given that you referenced memory.

Does that impact the gross margin that's in backlog on the semi test side?

A little bit.

The -- our memory test gross margins are a bit below the average in semi test.

It's maybe 4 to 5 points of swing there.

But given that relative size of memory at $65 million, let's say, to the total, it's not overly impactful.

Right.

And the fourth quarter gross margins, we've got it down because of Storage Test and memory.

I should add that both of those businesses generally have lower OpEx, but the lower OpEx comes with lower gross margins, so they all have good PBIT.

Okay.

And can I just sneak one -- sneak a question in on UR?

Could you just maybe speak a little bit to the 400 basis points of gross margin improvement year-over-year?

Is that coming from price?

Is it component cost-down, supply chain stuff?

Just trying to understand maybe how do you get that kind of gross margin improvement year-over-year.

Most of it is from price.

We announced a price increase early in the year, which has now rippled through.

All the orders were shipping now.

We've done material cost-down, too.

That's a smaller part of the improvement.

The good news is that material cost-down continues next year and next year.

So we do expect our supply line group from Teradyne should be able to help Universal Robots continue to lower material costs and get the best commercial terms and strategic sourcing in place.

So that will continue.

So we're very pleased with the performance overall in the Universal Robots gross margin.

And in truth, it's largely software is what we're delivering with.

It's very reliable, the chemical components.

So we do see over time that the -- there should be some opportunities to keep the margins quite healthy.

Your next question comes from the line of Patrick Ho of Stifel.

This is Brian Chin calling in for Patrick.

First question, just going back to the semi test business and the SOC business in particular, could you just characterize the utilization environment right now at your customers from a signal perspective now versus what you'd typically expect and how that again sets you up maybe for your typical seasonality in the business into early next year?

Yes.

This is always a strong utilization period after having installed a lot of equipment in the summer period.

Right now new product -- electronic products are peaking in production.

So utilization is very high.

It is -- compared to prior periods, it is running roughly close to where it's been in prior periods, a little bit stronger Q4 to Q4 of prior periods.

Now that doesn't necessarily -- we've not found that to be a prognosticator of what's coming next year, but that is where we are right now.

Okay.

That's helpful.

And switching over to the UR business.

Just curious, what is your installed base now for cobots?

Curious how much of that's still in Europe on a relative percent basis; and sort of, if you think those metrics, what -- from an installed base perspective where that could be exiting next year.

And the second part of that is just to take a different tact on your margin growth in the business.

Is it -- it's possible here that -- given the strength in the semi test side of the business, that you're even under-investing in UR and you could even pump up investments even more and push up maybe some of the margin targets just to continue you -- to set yourself up for a really strong trajectory in that business moving forward.

Right.

That's a good point.

And we do expect to invest more in OpEx growth in next year.

Obviously, as we continue to grow above 50%, the sales growth is higher on a bigger number.

So the OpEx is going to probably go up more than $20 million.

We've increased it last 2 quarters, but we'll create a plan or pencil to that as we get towards the end of the year.

So we've seen many opportunities to fan out Universal Robots in different regions, distributors, the ecosystem partners, so there's no shortage of opportunities.

In terms of where the cobots are, the mix that we described in the prepared remarks has been fairly consistent.

Europe is above 43% and moving around a little bit.

Asia is 26%.

North America is 23%.

That's this past quarter.

8% is rest of world.

And all the regions are growing at a very high rate.

Long term, we would expect China to be very significant.

But today, there's many applications in these higher-cost countries that are being deployed.

And I mentioned in my remarks that in China, you have some subsidies from government entities, which could accelerate the cobots faster in China as those policymakers see the advantage of bringing cobots to their workforce.

Next, we have Toshiya Hari of Goldman Sachs.

My first one is on semi test.

Mark, you guys have talked about strength in the automotive and industrial end markets for, I think, several quarters now.

And I think historically, these end markets, whether it'd be digital or analog, would be on for several quarters and then off for several quarters and kind of back on again.

But I think at this time, it seems like the cycle is extended in a positive way.

What did you see in these end markets, I guess, this Q3?

And what are your expectations going forward?

Yes, you're correct.

And typically, the pattern you mentioned has been true, and this one has extended longer.

Third quarter was strong again.

What's happening in fourth quarter looks to be pretty good.

I think there are several things that are new in the dynamic here.

One is that the electrification of the automobile is something that -- although in the past it has been a slow bleed, it's starting to become an avalanche of electronics moving toward model years, let's say, 1 to 3 years away from now.

Whether that's hybrid vehicles, EV or traditional vehicles with ADAS, all of that is fueling a lot of new designs and new complexity.

I mean, the complexity of the semiconductors, we're talking about next generation in automotive are much higher than prior generations, so you have this double effect for tests where it's a high-test-intensity environment, to start with.

Plus, a step-up in complexity and ubiquity has really changed that sort of fits and starts dynamic, I think.

So we're pretty positive on the next few years for automotive electronics.

Great.

And then I have a follow-up on memory test.

Can you remind us what percentage of the TAM you guys actively address today within memory test?

And I know you have new initiatives in place to potentially expand your SAM, but where could that percentage number be in 12 to 24 months?

Yes.

So the primary segment that we serve is flash final test, and we think we have a pretty high share of that segment.

It's roughly, a $200 million portion of that, let's say, $650 million TAM this year for memory tests.

So the concentration we have now is there.

We are moving the Magnum product line now into more wafer applications, which is closer to a $350 million new market opportunity for us that we should start seeing next year as an adder.

Your next question comes from the line of Mehdi Hosseini of SIG.

Yes.

I want to go back to your -- you briefly mentioned M&A.

And given the prospect of changes in taxation, would that accelerate the M&A strategy?

Or does it have no impact there, which will be pretty fair, using any change in taxation to strengthen capital return program?

Mehdi, I don't see that any possible tax changes would cause us do any different in our M&A approach.

So much of our M&A approach is, is it the right fit.

Does it accelerate Universal Robots growth with the differentiation obviously with the financial return?

And if there's any sort of tax play, that's more of a bonus thing that we think about, but we don't put that into or put that on as something that should drive us on the direction.

It's much more the fit and the advantage it can give us.

Okay.

And then you mentioned a new TAM opportunity for wafer application, wafer test.

Is that driven by increased adoption of wafer-level packaging?

Or is there any other driver that you can help us understand?

Are you referring to memory wafer test?

Yes.

Yes.

You mentioned that you're looking at an additional $350 million TAM, yes.

I'm just trying to figure out what the driver for that.

Yes.

Well, that segment of wafer test for memory has for many years been a relatively large segment.

We, in introducing the Magnum, chose to focus on flash final test because that is where the interface speed discontinuity first presented itself and gave us the opportunity to take market share.

Now that we're established there, we've been able to take some of the architectural benefits of Magnum and seek places in the wafer test -- pre-existing wafer test market where we can exploit that technology.

So I wouldn't say something is changing.

It's that now we're in a position to take that platform into a pre-existing, large wafer sort market.

Got it.

And if I may just ask one clarification to your comment about the demand trend.

In Q1 is when you typically have a strong backlog.

This year, your backlog had a historical high of almost $870 million.

Given the demand trend that you highlighted, the strength in the underlying trend for each business unit, should we assume that you can at the minimum hit similar backlog level by early next year?

No, I don't think you should assume that.

It gets back to the discussion we had a bit earlier because the timing of the orders in late Q4 versus early Q1 is one factor this year.

But sometimes, the orders even shift in from Q1 to Q2.

It's not a one lump quarter that typically drives the summer demand.

There's an initial baseline order that gets placed; and then follow-on incremental orders as the true demand for, let's say, the summer peak starts to unfold.

So those orders run from anywhere late Q4 all the way through, let's say, May, and they're spread across that period.

So it's hard to say that you snap a line at backlog at the end of Q1 or at the end of Q4 and make any meaningful judgments from that.

Your next question comes from C.J. Muse of Evercore.

I guess, first question, when I look at your sizing of the SOC market for '17, your expected revenues in semi test and what you talked about in memory, it looks like your market share is up about 600 bps, around 57%.

So the first question is, is that the right math?

And is that the kind of market share that you would expect to retain in SOC into 2018?

Well, yes, I think, first of all, the actual market share gains for '17 will depend on our shipments in Q4 and the market size.

We've said $2.6 billion.

It could be $2.65 billion.

It could be $2.7 billion when the -- everything is done, but we will likely be up in share anywhere from 2 to 6 points, let's say, in SOC.

We'll -- next year, I think what we'll end up doing is it'll be a year of consolidating that share.

Our plan typically is to try to pick up 1 to 2 points a year and be on that pace.

We've done much better this year.

And so I think, for next year, as we're looking at it now, we're going to try to maintain the share position we're going to exit this year with and look for on the SOC side more market momentum.

In the case of memory, we're going to see both market momentum and an expected share gain there to allow for growth.

Very helpful.

And as a follow-up question, as you think about gross margin specifically for your SOC business, obviously, it's very early predicting what the market size will be next year.

But if I take kind of the midpoint, it's down roughly 5% year-on-year.

As you think about kind of the mix shift going into next year, how should we think about it?

And again, I know it's early.

How we -- how should we think about gross margins for the SOC business year-over-year?

I think the SOC margins will be generally similar.

It's possible they're down 0.5 point or so.

We had a few credits that will reverse, meaning some inventory that was previously reserved that we sold.

So that comes in 100% in profit.

So we could have some charges we're not anticipating, retrofits or -- but we can't really forecast those, so there may be a little bit of movement there.

But if you just had the product itself, ignoring the credits or the offset-type charges, I would think SOC would largely be similar.

We get material cost down each year, but there's a little bit of price erosion each year.

They tend to stay in some equilibrium over time.

And I think the one thing that we can point out is that Storage Test and memory, certainly Storage Test will be much bigger next year.

And that will have a downward impact.

Other side of ledger, we have Universal Robots, which has improved their margin throughout the year.

So there's a bunch of things in the mix.

As we get to January, we'll have a better analytical sense as to what to guide for the year.

I guess I was trying to get a read on how you're thinking about higher-margin auto, industrial, as opposed to lower margin, all things being equal, digital.

Do you have an early read on that year-over-year?

Not really.

Like I said, I think the mix next year isn't too different from this year in terms of those segments.

Your next question comes from the line of Edwin Mok of Needham.

First question, though, just I guess stick with semi tests.

I think one of your customer or some of the foundries talk about high-performance computing potentially will become a big driver longer term.

How do you see that driving the test market?

Do you think that could be a bigger incremental or a big driver for test market?

Well, any high-complexity digital device is a driver.

And so to the extent more, let's say, AI, deep-learning applications require specialized, complex-array processing-type digital, that will be a balloon.

Absolutely.

And are you seeing that right now?

Or is that something kind of in the long term?

Well, I think it -- no.

Not right now.

It's -- there's a lot of discussion and buzz around that type of application, but even if you look at places like GPUs, which are those kind of processors, it's somewhere in the $100 million to $200 million test market per year.

So it's relatively small, anywhere from 5% to 10% of the market.

Okay.

Great.

That's helpful.

And then just jumping on to UR.

I think you guys talk about investing on OpEx and have been growing OpEx in the last few years as you kind of broaden out your -- a piece of your distribution, right?

I'm just curious.

As you look into '18, do you see needs to further invest in R&D, especially in software?

It seems like that's going -- almost likely to be the next step for you guys to kind of increase kind of the capability of UR robot.

Is that where we should expect increased spending in 2018?

And is that the way to kind of think about how you spend, well, on your OpEx; or how you could increase your OpEx [in your outlook] going to '18, between OpEx in R&D and SG&A?

I would expect, in '18, we're going to increase both the distribution and marketing, similar to how we've done in prior years but more higher dollars because, again, last 2 years was $20 million and $20 million.

Our plan to grow 50% or greater from our ending '17 is considerably higher sales growth in dollars, so we're going to need more OpEx to field all of those distribution initiatives and programs.

In terms of R&D specifically, we are going to continue to up R&D.

We see a number of opportunities to make the cobot extend into other spaces that it's not in today.

And this gets back to a little bit the question earlier, would we go for a different cobot size?

Frankly, we see so many opportunities with our 3 cobots that the challenge for us is getting the -- as many people that are talented onboard working in the right direction.

That's the bigger challenge versus there's any shortage of attractive opportunities.

Your next question comes from the line of Weston Twigg of KeyBanc Capital Markets.

First, at the risk of beating a dead horse a little bit, on the SOC market outlook for next year, you have it down just a little at the midpoint.

I'm wondering if you could be more specific on what you -- has you a little bit concerned, which segment has you a little bit concerned about the market maybe being a bit lower next year.

I don't think it's any particular segment.

It's maybe just a lot of experience over many years in the industry that, the visibility at this point being low, I would be reticent to be too aggressive.

I think, come January, we'll be able to have a better view of that, but maybe I'll just chalk it up to some sage conservatism at this point.

Okay.

That makes sense.

The other question I had was on the Universal Robots side.

I'm wondering if you could update us on what you think your current market share is and also just why the Q3 revenue really didn't grow much sequentially.

Was there anything that prevented a little bit faster growth quarterly -- or sequential quarterly growth?

So on the second point, the sequential quarterly compares are really tough in UR's case.

Q3 tends to be a slow quarter because of vacations in Europe.

Q4 tends to be a big quarter in the past because of people trying to meet year-end goals.

I think I mentioned that this year's Q4 year-over-year compare will probably be closer to the 50% growth number, meaning the year for us will finish at about 60, mid 60s.

That's because we've changed our incentive programs to try to smooth out the end-of-year rush to buy that we've seen in prior years.

So sequential ordering in UR is not something that's very meaningful, I would say.

Year-over-year comparisons are better.

Okay.

And market share?

So market share is a tough one because there's no reliable third-party reports on this.

There -- and there's a growing population of cobots.

If you go out and Google cobots, every quarter, you'll see a few more and a few more out there.

None of them are competitive with us in terms of the situations that we're selling into, so it's still a pretty greenfield environment.

But we've said in the past, roughly 60% share based on reports that at this point are over a year old on market size estimates.

I don't believe we've lost any share, but I think it's going to be hard for us to be too precise on that until we get some reliable third-party reporting.

Your final question comes from the line of Farhan Ahmad of Crédit Suisse.

Can you just talk about how you are forecasting the SOC test market for next year?

If I think about the semi revenue growth, that has been stronger this year, up since 2010; and almost tracking at about 10%, excluding memory.

Can you just give us a sense of, well, how you are going about forecasting the SOC test market?

And what kind of semi test growth are you assuming for next year?

So the things that drive our market.

Certainly, unit growth is important for us.

That generally correlates to semi revenue growth, not always, but unit growth is important.

Complexity growth is important.

So at this stage, what we're looking at is hard to forecast next year's unit growth, so we're looking at complexity growth.

We're looking at devices at our major customers that are in preproduction that will grow, that would ramp next year; trying to get a sense of do we see the same trends in jumps of complexity, which will mean test intensity and test time.

What do we see happening with parallel tests?

Next year's amount of parallel tests kind of gets set early, now through Q1 because the programs are in development.

So we look at trends around complexity, parallel tests, test time.

And the thing that we don't have a good read on right now is what does unit growth kind of look like.

I'll just add.

Inside the company, there isn't enormous amount of time trying to analytically answer that -- those questions because they're really not answerable.

It -- there are so many uncertainties.

And we're much more focused on are we executing against our market share goals, our gross margin goals and so forth.

And we know, every time we speak to you guys, we need to talk about the market, but it's not something that we put enormous energy in because it's something that -- it's not very knowable.

Got it.

But can you just give us a sense of what kind of unit growth are you assuming in the forecast?

Is it a deceleration from this year; or similar, at the high end maybe and maybe a deceleration at the low end?

I think at the high end, it would be similar unit growth to this year.

The low end would obviously be a significant fallback.

But again, as Greg said, there's not a lot of analytics that go into the ranges I'm giving you.

Got it.

And then in terms of your margins, this year has been pretty phenomenal.

The operating margins have been above 25%.

And can you just maybe talk about at a high level?

Do you think these margins are sustainable or -- and we can grow from here as your revenue grows?

Or do you -- should we expect somewhat of a moderation next year?

I think so much of it is tied to this prior conversation: Where exactly is the market size next year or any point in time?

What we feel good about is the long-term trends, but calling any 12-month window is more difficult for us with precision.

But that what's -- what really drives our profitability because there's such good drop-through on higher sales in semi test.

We don't need to add manufacturing people.

We don't need to add engineers or salespeople.

So that's the biggest swing factor in our P&L, is the market size.

The things that we can control, obviously, getting more market share.

We've been doing that.

Improving Universal Robots, we've been doing that.

And then improving our other System Test and LitePoint businesses, we're doing that as well.

But the wild card in all this is what is the semi test market.

That's the biggest single thing to our profitability.

Okay.

And operator, we're going to close this up.

Thanks, everybody, for joining us.

And for those in the queue, I'll get back to you immediately after this call ends.

This concludes today's conference call.

You may now disconnect.