泰瑞達 (TER) 2017 Q2 法說會逐字稿

Good morning.

My name is [Britannia] and I will be your conference operator today.

At this time, I would like to welcome everyone to the Teradyne Q2 2017 Earnings Conference Call.

(Operator Instructions)

Mr. Andy Blanchard, you may begin your conference.

Thank you, [Britannia]

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 CFO, Greg Beecher.

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

The press release containing our second quarter results was issued last evening.

We're providing slides on the Investor Page of the website that may be helpful to you in following today's 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.

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 posted additional information concerning those non-GAAP financial measures, including reconciliation to the most directly comparable GAAP measure, where available, on the Investor Page of the website.

Also between now and our next earnings call, Teradyne will be participating in Investor Conferences hosted by Needham, KeyBanc, Citi and Deutsche Bank.

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 third quarter.

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

We'll then answer your questions, and this call is scheduled for one hour.

Mark?

Thanks, Andy, and hello, everyone.

Today I'll provide a summary of our second quarter and first half results, describe how we're thinking about the full year and provide some additional insight on our strategy at Universal Robots.

As expected, we had a very strong second quarter.

Our sales of $697 million were up 31% from the year ago quarter and non-GAAP EPS of $0.90 was up 64% from that same period.

This reflects strong demand across the Semiconductor Test space and at Universal Robots.

SemiTest was our biggest driver with sales of $593 million in the quarter and $949 million for the first half of the year.

This is the second consecutive year of high demand in the mobility test segment, as device complexity growth continues to fuel demand.

We are also benefiting from the capacity additions to support a steady increase in both the volume and complexity of automotive electronics.

In the automotive market, there's a healthy activity in new advanced technologies for hybrids, electrics and advanced driver-assisted automation systems.

At the same time, the demand has been increasing for more traditional technologies, such as microcontroller and analog device families, supporting functions like cabin controls and LED lighting, as these functions move from high-end to mid-range to low price cars.

Additionally, automotive camera growth continues at double-digit rates fueling image sensor test expansion.

All these factors point to continued strong demand for tech capacity for automotive electronics.

In Memory Test, our first half has also been strong, driven primarily by demand for flash package test systems.

Memory Test sales in the second quarter were $50 million, more than double Q1 sales, and up 22% from the year ago period.

We've noted in past calls that the combination of Flash bit growth and the move to higher speed interfaces, like UFS, are responsible for the growth of our Memory Test business.

These devices have interface speeds now reaching over 5 gigabits per second, which can't be tested on older legacy test systems.

As smartphone video capture and display performance improves, the memory performance must likewise improve.

Similarly, SSDs require interface speeds substantially higher than the operating speeds of older installed base testers.

We expect these higher performance requirements will be a long-term tailwind for the memory market, and a high-speed, high throughput architecture of our Magnum product line is resonating with customers.

For the full year, we expect the SOC test market to be towards the high end of the $2.4 billion to $2.6 billion range we described in our April call.

And we expect to gain 1 point or 2 points of market share.

For Memory Test, we've increased the top of our market size range by $50 million, and we now forecast a market this year of $550 million to $600 million.

We expect our Memory Test sales to grow in 2017, but below the overall market rate, as segments outside our Core Flash final test segment are growing faster this year.

In 2018, we plan to broaden our Magnum Memory Test footprint by expanding into new Memory Test market segments.

In the combined total ATE market, we expect to gain about 1 point of share this year.

At Universal Robots, we are tracking nicely to our 50% or greater annual growth target, with second quarter and first half sales up 57% and 81%, respectively, compared with 2016.

Bookings in the second quarter were down sequentially, as pricing changes that took effect in Q2 pulled some orders forward into the first quarter.

Greg will provide more color on UR shortly, but I will point out that we are very pleased with the overall performance of the business.

In the first half, about 60% of our growth came from distributors that were on board at the same time last year, while 40% of the growth came from new distributors.

China remains a bright spot this year, with year-to-date sales running more than double the same period in 2016.

We are also continuing to expand the range of programs designed to help our distribution partners improve their capabilities and their productivity.

These range from product and sales training to trade show support, to expanding the range of peripherals available so they can address an even broader range of end markets.

For example, our Universal Robots+ Open Architecture Partner Program added an application that provides a high-resolution camera and supporting software for surface metrology and 3D scanning.

Another recent addition allows remote cobot monitoring using a smartphone.

Collectively, we now have about 40 UR+ products available and we'll continue to expand the number to provide an increasing range of proven capabilities available to customers around the world.

These apps broaden the market reach, shorten the development time and improve the ease of employing the power of UR cobots.

In Wireless Test at LitePoint, we had another good quarter.

Sales were up 23% from Q2 of '16 to $28 million and up 24% for the first half when compared with the first half of 2016.

The business was profitable for the fourth consecutive quarter, as we've aligned our cost structure to the new market sites.

Meaningful market and revenue growth at LitePoint awaits the next round of WiFi standards, which we expect to begin deployment in smartphones in late 2018 and into '19.

We expect to see initial 802.11ax chipsets in terminal equipment deployed later this year.

Cellular test expansion will follow, with another wave of growth as 5G technologies begin to roll out.

Other than low-volume demonstration sites, the timing of significant 5G technology rollouts is likely to be in the 2020 timeframe and beyond.

In System Test, we saw quarterly sequential sales growth in the combined defense and aerospace and production board test businesses, of about 10%, but storage test remains quite slow.

We expect that business to pick up in the second half, as a new product in that group currently installed at a lead customer begins to generate revenue for the fourth quarter.

For the full year, we expect System Test sales to be similar to last year's $190 million, so we'll see sales this year a bit more back half loaded than last year.

Overall company sales consistent with last years will be down in the second half, as this year's wave of mobility tooling is largely behind us.

As a result, we expect first cap company sales as a percentage of the full year to be in the high '50s.

The strength in the first half makes this a few percentage points higher than in past years.

On the capital allocation front, as noted in past calls, we will continue our balanced approach with dividends and share repurchases providing direct return to shareholders and M&A providing the growth factor.

In the M&A area, we continue to look at a range of emerging technologies and businesses that have the potential to accelerate our growth in the automation space.

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

Thanks, Mark, and good morning, everyone.

I'll provide some brief comments on our first half financial performance and key annual goals, along with added commentary on Universal Robots.

And then I'll cover the second quarter results and third quarter outlook.

Starting with our financial performance.

We're pleased to be tracking ahead of our midterm $2 non-GAAP EPS plan, having achieved $1.34 of non-GAAP EPS in the first half of 2017.

The second half will, of course, be seasonally down in SemiTest, but we expect, we'll approach this $2 annual target this year.

We'll be revealing this target as part of our midterm planning process later this year, and we'll update you in the January call.

At the 2017 halfway mark, our stronger EPS performance has been driven principally by SOC test.

We're seeing broad demand across numerous segments including Mobility, automotive, Analog and Microcontrollers.

As Mark noted, the SOC test market is estimated at the high end of the $2.4 billion to $2.6 billion range this year, up as much as $200 million from last year.

We get very good earnings leverage from SOC market growth, with our market share position and our approximately 50% or better operating profit drop through on higher SOC test sales.

At the company level, we also achieved our highest first half sales this decade at $1,154,000,000, the drive of this very strong top line performance were SOC test and Universal Robots.

In SemiTest, first half SOC test sales of $876 million were up over 20% versus the first half of a year ago.

Mobility remained quite strong as expected, and automotive grew nicely.

More broadly, in Analog and Microcontroller Test, there were many end applications, and our customers are seeing growth driving new capacity needs.

So all told, a healthy but not overheated SOC test market.

As Mark also noted, Universal Robots had first half sales of $76 million, which was up 81% from the first half of 2016, as we're benefiting from our aggressive distribution investments.

We see many years of strong growth ahead, as we believe we're still in the early adoption phase of bringing flexible, human scale automation to a dull, dirty and dangerous task.

On the 2017 annual goals that I outlined last quarter, we're performing on plan or better for most goals.

Starting with our SemiTest market share goal, we're on a multiyear trend line to pick up at least 1 point of share.

While market share gain goals are important, we'll continue to be highly targeted and disciplined in a mature SemiTest market facilities to maintain healthy gross and operating margins.

Next, our balance capital allocation goal remains on track, with our buyback and dividend programs and highly selective M&A.

We continue to look at industrial automation opportunities, particularly for added software capabilities that can increase the sizable ease-of-use and flexibility advantage of UR's collaborative robots.

Next, UR is tracking to our 50% or greater annual growth goal at 81% at the halfway mark.

I should add that quarter-to-quarter, or other short term comparisons will have some noise.

In SemiTest, our goal to operate at model profits of 50% or better is on plan and the combine MilAero production for the test businesses, but in Storage Test, we've incurred new product development costs in bringing in a new product to a new market application, and expect to get back to our 50% or better operating profit in the fourth quarter.

At LitePoint in Wireless Test, our goal to operate in the black during the low and new standards is on track with our first half operating profit percentage in double digits.

So we're tracking quite nicely against the key annual goals.

Now, before I move to the second and third quarter details, let me quickly discuss UR's fit within Teradyne and UR's growth plan, as these are the areas where we get the most questions from investors and analysts.

First, let's start with some context.

We acquired UR in mid-2015, as we were looking for markets with long-term secular growth that had a connection to our core capabilities.

We had earlier been a leader in the consolidation of automated test equipment, or ATE, with the acquisition of Eagle Test and Nextest, and saw that there wasn't anything else closely adjacent to SemiTest that had long-term earnings power.

We did, however, have wireless test and production board test customers looking to automate the feeding of their testers without reconfiguring their lines.

And it didn't take a lot of imagination to see that this need for human-scale flexible automation was a tip of a very large ice berg.

If you scan their production lines, you'd see many dirty, dull and dangerous task that would benefit from human scale automation.

And we realized that electronics was but one of many end markets that could benefit from today's cobots.

The A or automate in ATE became fertile ground for us if we could find an attractive path to enter the cobot market.

(technical difficulty)

We acquired Universal Robots, the clear market share leader, in the emergent collaborative robot module, innovation was to encapsulate the complexity and provide the software user with high-level software, making it easy to deploy cobots.

Adding to their value, our cobots can be quickly repurposed as the pinch points in a production operation change.

These features combined with UR's attractive price points can yield a payback period as little as six months.

Since acquiring UR, we've grown it's OpEx from about $7 million a quarter in late-2015 to about $16 million a quarter in the second quarter of 2017.

This added investment is targeted at 3 key areas and positions us for continued growth in market leadership.

First, we have strengthened UR's distribution capabilities, including increasing the velocity through existing partners by feeding them many more leads, providing more and better training programs, adding more commercial and technical support and cosponsoring about twice the number of products exhibitions.

We're also continuing to open up new offices and add new partners to extend UR's geographic coverage.

Our distributors are just scratching the surface of a very large potential market.

Today, the average order size from our distributors is quite small at just 2 cobots, as most end-users are in the early adopter stage of deployment.

This broad distribution reach and effectiveness at selling in low qualities -- quantities creates a hurdle for others to replicate.

Even in our larger deployments of 100 or more cobots, we believe we're in the early stages of growth, based upon the customers longer-term plans and a number of untapped opportunities that we can see.

Secondly, we're investing and supporting third-party developers with our UR+ open architecture partnership program.

We've opened our software platform to an array of third-party developers, we're creating applications to run on UR cobots.

We then assist them by promoting many of their products on the UR+ online showcase.

Third, the UR R&D team continues to extend their performance of our industry-leading software to improve ease of use and shorten deployment times.

So we're adding to our first mover product advantage with a global distribution channel that can serve small and large deployments, while making it attractive for third-parties to develop applications on our cobot, bringing us additional new opportunities, both in manufacturing and service, and we're organically investing in the R&D necessary to maintain our product leadership.

Lastly, on the fifth, we've been able to bring larger company capabilities to UR in supply line management, global recruiting and expansion, product marketing, customer service and business development.

Of course, our electronics customer relationships are also tapped.

Shifting now back to the company level.

We paid $14 million in dividends and used $57 million to buy back 1.7 million shares at an average price of $33.60 in the second quarter.

This leaves us with $406 million remaining under our current $500 million stock repurchase authorization, and we intend to spend at least $200 million in total this year.

Since the start of our buybacks in 2015, we've bought back 25.4 million shares at an average price of $21.26.

Our U.S. cash and marketable securities totaled $751 million, down $27 million due to our share buybacks and dividend payment.

Our total cash and marketable securities balances of $1.62 billion were up $138 million due to strong profits and a reduction in inventory due to seasonally strong shipment levels in the quarter.

Moving to the details of the second quarter.

Our sales were $697 million.

The non-GAAP operating profit rate was 31%, and non-GAAP EPS was $0.90.

We had one greater than 10% customer in the quarter.

Gross margins were 56%, down 2 points from the prior quarter due to the mix of mobility business in the quarter.

You'll see our non-GAAP operating expenses were $173 million, up $12 million from the first quarter, primarily due to higher variable compensation accruals on increased profit levels.

Comparing our second quarter OpEx of $173 million versus the year-ago period, it's up by $15 million due to higher variable compensation accruals on higher profits, and an expansion of UR's distribution programs.

Moving to the segment level detail.

SemiTest bookings were $369 million, down 23% sequentially from seasonally strong first quarter bookings of $476 million, SOC test orders were $314 million and our Memory Test orders of $55 million was the highest in over 2 years.

SemiTest service orders were $79 million of the total.

SemiTest sales were $593 million in the second quarter, driven by strong mobility in automotive demand with SOC making up $543 million and Memory Test, the balance.

SemiTest service revenue totaled $68 million in the quarter.

At UR, orders in the second quarter were $33 million and sales were $39 million, a quarterly sales record.

UR's backlog was higher than normal at the end of Q1, as channel partners pulled in orders to the first quarter to avoid pricing changes.

Regionally, UR's second quarter sales broke down, 39% in Europe, 28% in North America, 24% in Asia and 9% rest of world.

Moving to System Test, orders were $29 million in the quarter, and sales were at $37 million.

In Wireless Test, we booked $30 million and had sales of $28 million in the second quarter.

Turning to guidance for the third quarter.

Sales are expected to be between $455 million and $485 million, and the non-GAAP EPS range is $0.39 to $0.46, and 200 million diluted shares.

This outlook reflects the expected normal seasonality in SemiTest orders.

The guidance excludes the imposition of acquired intangibles and the noncash convertible debt interest.

The third quarter gross market margins should run at 57%, up 1 point from the second quarter due to favorable product mix.

And the total OpEx should run from 34% to 37%.

The operating profit rate at the midpoint of our third quarter guidance is about 22%.

I should add that we plan to continue building UR's distribution capabilities to capture even more of the growth available in the cobot market, which should add a few million to our quarterly OpEx run rate, with quarterly UR OpEx increasing to around $20 million in late 2017.

Elsewhere, as noted, we have increased investment in our Storage Test group, to bring in market -- a new product to market and a new application.

Nonetheless, in our test businesses, we expect full year OpEx spending to be essentially flat, year-over-year, apart from normal changes in variable compensation tied to profitability levels.

Shifting to taxes.

Our full year non-GAAP tax rate is expected to be 16.5%, up 0.5 point from our April estimate due to strength at LitePoint and Nextest, both of which are U.S. businesses that carry a higher tax rate.

We're operating in 2017 above the prior 3 sequentially strong years, driven by SOC test strength in Universal Robots.

Our test businesses, in aggregate, have consistently generated strong profitability and free cash flow since 2010, and UR continues growing in excess of 50% a year and expanding its first mover advantage.

This consistent strong performance provides the confidence to stay the course and continue to buyback our shares.

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

Thanks, Greg.

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

(Operator Instructions)

(Operator Instructions) Your first question comes from the line of Mehdi Hosseini with SIG.

Going back to your commentary regarding building out the channels for the cobot business.

I'm just trying to understand how this buildout is going to impact booking?

Whether the new channel partners that you're engaging are going to have adverse impact on booking?

Or how should we think about the impact on booking as the channel is buildout?

And I have a follow-up.

Mehdi, it should have a positive impact.

Where we're building out, we're building out in regions and locations that we don't have good reach or contact with various end verticals or customers.

So we very much expect the trend to continue that Mark mentioned, that a fair amount of our growth comes from these new distributors, as they're getting us to places we had not gotten to before.

At the same time, we're improving the velocity of existing channel partners.

So the 2 of them together is what gets us this greater than 50% growth.

They're both critical to the ongoing high growth game plan we have.

Got it.

So the turnover by these new channel partners is not -- shouldn't have an impact on what they book?

That's what you're saying, right?

The idea is that they don't cannibalize the existing.

I can't say in every single case, they might not call on the same customer.

But generally speaking, we're bringing on other distributors or integrators that are getting us to places that we had very little or inadequate coverage.

Okay.

Very clear.

And one question on the SemiTest.

I want to understand the time opportunities offered by MCU or automotive, which is more secular versus the SOC, which is more concentrated among a few players.

SOC meaning the out processor and GPU vendors.

How do you see these 2 different sub-segments evolving and impacting the overall SOC test into the second half and into 2018?

The second half in 2018 is a pretty short time frame.

I will say that the variety of electronics going into automotive is growing.

So the traditional microcontroller and analog growth is still there.

It's probably running in the 7% to 8% growth range.

But when you add on the growth happening in memory, more sophisticated controllers, image sensors, these new devices combined with the traditional devices, bring the aggregate electronic growth rate for automotive closer to that 10% level, in terms of semiconductor content.

For test, because the test intensity is that much higher for a device going into an automobile, that segment will probably grow in excess of 10%.

And -- on average over the next 3 to 5 years.

Quarter-to-quarter is hard to call, but that's how we look at it over the 3 to 5 years.

I guess, if I were to rephrase my question, what percentage of SOC test is currently driven by auto and MCU?

And how is it going to change forward?

Yes, I think it's hard to quantify in any precise way, because of the flow of content from segment.

Even things like modems for cellular communication are now showing up in cars.

But as a rough feel, and I would say that the test market for automotive, electronics has roughly been in the $400 million range.

And that the combination of increasing content in test complexity, we'd expect that to grow in excess of 10%, probably in the 10% to 15% range on average.

Next, we have a question from Farhan Ahmad of Credit Suisse.

Just in terms of the pulling of the UR demand into Q1, can you just talk about what were the pricing changes and how did it impact -- how did it result in pulling into Q1?

And why shouldn't we be worried about the deceleration in growth in UR?

Yes, let me just give you a few data points on that.

So, we had some geographic price adjustments in the first quarter were increases.

That resulted in some of our distribution partners buying inventory ahead of that price increase that came in Q2.

Depending on the geography, it could have been as much as 15%, 20% increase.

But overall, it's a handful of percent price increase on the product.

If you looked at what we booked in Q1, we booked $44 million at UR, and we shipped $36 million.

So we built $8 million in backlog, which, traditionally at UR, would be quite high.

Most of that would be attributed to this buy ahead.

In the second quarter, we gave back $6 million of that $8 million that we built in the first quarter, and that essentially nets out the effect of that buy in ahead for the price increase.

So going forward, we'd expect a book-to-bill at UR to trend back close to that 1% ratio.

Got it.

And can you just tell us why did you increase the price?

Because this is a new emerging market and competition is coming as well.

So -- and I believe the margins, the gross margins are at least okay.

So why increase the price?

There were some regional differences in price that we wanted to bring the regions closer together.

So that there was less of an incentive to buy in one region and ship to another region by sophisticated customers.

We also had a UR3 that has higher cost gears when they're smaller.

And that product, we thought at the outset probably wasn't priced where it should have been, given its payback and the value that it brings.

So there were a couple of changes made, and they seem to have gone well.

Got it.

And then switching to the wireless side, can you talk about the 5G and how should we think about the growth from 5G as it relates to the wireless business that you have?

Well, as I mentioned in my remarks, 5G is -- there will be technology demonstrations, and there is certainly R&D development occurring now.

Very low-level buying around design validation for those.

From what we see, in terms of production rollout, reasonable infrastructure tooling and terminal equipment and handset deployment, that's 2020-plus.

Now when that occurs, we expect that the market for test, for cellular, will have a significant growth behind it.

And that could be -- the market could be a couple of hundred million dollars up during those early years of deployment.

So this year, if the market is closer to the $200 million to $250 million for Wireless Test that could grow to $450 million to $500 million.

And we see that running probably for 3 to 5 years during that 5G rollout.

Your next question comes from the line of Toshiya Hari of Goldman Sachs.

Mark, I wanted to follow up on the mobile SOC test question and kind of your outlook into 2018.

Clearly you had 2 consecutive years of very strong growth year.

I appreciate it's kind of early, but if you can give us your preliminary thoughts on how you think about the business going into '18, that would be great?

Any view of 2018 at this point is not based on any bottoms up view.

Typically that visibility is quite short, in a sort of a 3-month window.

But the trends, the trends that we've been seeing in the past few years that we've talked about.

Number one, test time growth due to complexity growth.

Number 2, the sort of capping of parallel test, certainly is evident again this year, and we expect and all indications are that, that will be true next year.

There are some aggressive moves towards new lithography nodes that substantiate that, and the advanced packaging technologies that are being employed, substantiate that.

So at a trend level, macro level, even without a lot of unit growth, which has been the case now for a few years in mobility.

The complexity growth has been and we expect it to continue to fuel demand for our test equipment.

And then, as a follow-up, I think in your prepared remarks, Mark, you talked about potentially expanding your product portfolio on the memory side in SemiTtest.

Can you perhaps elaborate on that?

And how would your SAM in Memory Test grow as a result of that?

I think historically, you've been strong in Flash package test, but elsewhere, I think you've been more of a follower relative to your nearest competitor.

So if you can talk about your SAM expansion opportunities in Memory Test, that will be helpful.

Okay.

Yes.

So you're right.

We have been quite strong in flash package or final test.

We've been able to move our market share in memory almost at a 30%, roughly the 30% level of the total market through that strength.

Where we haven't really participated much is in probe test or wafer test.

That represents -- it moves around year-to-year, but that represents roughly half of the Memory Test market in aggregate, which is a part of the SAM we haven't participated in.

So if you look at this year being close to $600 million, $300 million or so is in probe test.

That's for both DRAM and for flash combined.

So our plans are to move the product in that direction, move the Magnum platform toward more probe test applications, and open up the other half, let's say, of the Memory Test market where we have not participated.

And next, we have a question from the line of Atif Malik with Citi.

First question for Greg.

Greg, I'm looking at the transcript from the April quarter, and you guys well telegraphed that your order rates in Q1 are going to accelerate because of the pricing change, it's not a surprise that your orders are down in Q2.

My question to you is why report orders, your U.S. [front] and peers have stopped reporting orders, as they could be lumpy and there is volatility in them.

And sale side does not know how to model them, so why continue reporting segment orders?

And then, I have a follow-up.

Well, Atif, ironically we were just talking about that before the call.

So we'll revisit that.

And the reason we continue doing it is if we stop doing it, we don't want to create some uncertainty and some worry or some greater speculation that people try so hard to figure out what they are, and that's all we talk about.

But it is something we should look at.

And we've actually said before the call, let's look at what others do.

Obviously, when one has very short lead times, booking -- I mean reporting orders doesn't do anything.

It's misleading.

It's really shipments.

So it's something we will look at and come to a conclusion on.

Okay.

And a follow-up for Mark.

Mark, (inaudible) is talking about expanding use of cohost for AI chips for Nvidia and others.

Can you just talk about how would that impact your opportunity in the future?

Yes.

So I won't comment specifically about any one customer or technology.

But in general, the advanced packaging trend towards integration is -- as we've talked about, it's a positive impact for Test.

It increases the premium on known good die, because the assembly of that advanced package puts a premium on low-low failure rates, high, high yield.

So known good die goes up, which means test time goes up at prob.

And then the testing of the integrated package itself, once the various components have been assembled and mounted, the complexity factor goes up.

It scales more than just proportional to the transistor count.

When you have that collection of parts integrated on a substrate, it tends to drive up test seconds.

And it's another depressant on increasing parallelism.

So overall, for Test, it's a good effect.

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

I guess first question wanted to revisit an earlier question regarding SOC and you talked about complexity.

So typically, you guys have had an on year and then an off year and now we've had two on years in a row.

And I guess I'm trying to understand what has truly changed here?

Historically, semiconductors have been complex and will continue to be complex.

But is it has reduced parallelism that is now driving this uplift that is sustainable, is a confluence of perhaps emerging growth drivers related to AI and other factors like that?

How are you thinking about that from a very big picture perspective?

Yes.

So you're right.

Complexity of semiconductors has been increasing every year since the dawn of the semiconductor era.

So what's changed?

Well, the semiconductor complexity growth I would say is on a [silica] level not changed very much in terms of its progression.

The parallelism plateauing certainly has changed.

So that counterbalance to complexity growth that we've had going back a decade plus, is something that has changed.

The economics of trying to turn the knobs up higher on parallelism are just not there.

The expense of the probe cards and the time to market hit of trying to do that does not justify continuing to try to mitigate test time through higher parallelism.

So that's absolutely one factor.

And then the other factor that we just talked about that's changing its advanced packaging.

Advanced packaging is another new element in the mix, that incrementally drive higher test intensity.

So removing parallelism as a depressant and we still have complexity growth and a new element of advanced packaging.

Very helpful.

And I guess as my follow-up.

It looks like embedded in your September OpEx guidance, perhaps around $5 million, $6 million higher OpEx.

And I get -- you had that chart on the slide deck where you're spending more on UR and keeping SemiTest flat and then you have incremental competition.

But is that kind of up $5 million to $6 million run rate, what we should be expecting going forward through '18 or how should we be thinking about OpEx?

If you want to look at OpEx, I would take this full year '17 compared to '16.

And I'd expect $25 million more for variable compensation, which you wouldn't necessary carried forward to '18 unless you had a similar looking year.

And what's driving the variable compensation is we're hitting all of our goals, our growth goals, our market share goals, our profitability goals so the variable compensation is at the high end this year in '17.

The other place where spending is going up is in Universal Robots.

That's also about $25 million in OpEx for the full year.

Apart from that, our Test businesses are flat.

So if you want to look at '18 its vary -- it depends on what's you model for '18, your variable compensation will vary depending upon what you put in but Universal Robots will likely still increase in '18 but Test will remain flat.

Your next question comes from the line of Jagadish Iyer of Summit Redstone.

Two questions.

First on UR.

I was wondering how many distributors do you plan to add in calendar '17?

And how does this compare to calendar '16 and what is your take for next year for distributors?

Okay.

From about a year ago we are up about 1/3 in distributors, but it's I should start it's not a numbers game.

It's really are we getting the right partners in the right regions.

So we try not to get caught into just looking at the number.

It's more the quality.

But we would expect to be growing distributors for -- probably a couple of more years to spread out where our cobots are connected to different verticals.

And as the UR+ applications grow, the cobots will be taking into new tasks, new applications, new customers and that will also provide opportunities to call on different new accounts or new task to be replaced with automation because third-party developers have developed a nifty solution on our cobot.

Okay.

Fair enough.

Then on a bigger picture, I think you alluded to an earlier question here.

The front end and the equipment guys have indicated a new baseline in spending.

I was wondering, as units continue to grow and die sizes become bigger and bigger, is there a new baseline for SemiTest that we can expect us we look at it over the next 2 to 3 years?

Or do you still view it as a deeply cyclical business?

Yes.

I don't view it as deeply cyclical.

I do think there will be swings of 10%, 15% in market size in our future.

We had a pretty good consistent run but I don't think the underlying volatility -- although [dually] going forward will disappear.

The thing about test is that for 10 years up -- through 2013, what we called the capital intensity of Test relative to the revenue of semiconductors had been declining.

What's happened since then is that, has plateaued stabilized and may in fact be slightly growing.

So I would say, the baseline for Test has shifted from a compressing market relative to the growth of semiconductors to one that is now perhaps at least growing as fast, if not slightly ahead of the semiconductor market.

There's only 3 years, 4 years here where we have seen this trend so rather than prognosticate a long term vector, we are still looking at it, we have said we expect the market to grow at least on average at 1% per year.

We've been ahead of that, we look at that every year and we may change that model.

But that's how we look at it.

Next question comes from Edwin Mok of Needham.

First on the ATE side, I understand you guys have very strong position in image sensor.

We think the demand both in volume as well as continuity requirement relate to kind of these [AR DR]3D centric, just curious how does that affect the test requirement for image sensor?

Does that really help you on the ATE image sensor market?

Absolutely.

There's been a series of trends image sensors that have been very good for Teradyne.

The original -- obviously smartphone incorporation, then the front sight camera, now dual cameras on the backside, moving to what maybe even additional types of image sensors in cameras beyond the 3 and the high-end phones today, all of that has driven up unit volume, typically the pixel density has also been increasing.

So that is a combination effect on test demand.

Now we are seeing automotive and in the case of automotive it's multiple, multiple sensors, upwards of six image sensors or more in automobiles is the trend line for the high-end models, which will then move down to the midrange and low-end.

So all of that has been great.

All that being said, image sensor testing as a business is roughly in the $100 million to $150 million a year range out of the total SOC market of about 2.6.

So that gives you a rough sense of the size.

Great.

That's actually helpful color.

And then just moving on to -- on your prepared remark you mentioned that your China business is doubled.

I'm just curious -- and I know China is still a small part of your business, maybe around 10% of the business, something like that.

But just kind of curious, maybe talk about geographic roles opportunity.

Is China now the best growth opportunity for your UR business or is that the way to kind of think about where you -- we should see greatest geographic expansion for you all?

I think for the next couple of years, you should see high growth in all regions.

Obviously Europe and North America has a faster payback because there is higher labor cost.

But what we're seeing in China, even Singapore is you have government subsidies to get cobots in faster so they are more competitive globally.

So we see growth in all regions for period of time but long-term, through a long-term, I certainly would expect China to grow more than others long-term just given there's many more opportunities to automate in China.

Up next, we have Krish Sankar of Bank of America Merrill Lynch.

I have a couple of them.

One is, as you try to increase your Universal Robots sales, besides the quality of distributors, would you consider doing a direct sales force like what (inaudible) do or some kind of like a better mouse trap like having all the robots come with vision like what [reefing] does?

Yes.

So, a couple of points there.

Right now we don't have any view that a direct sales force is the right approach to the market.

The market tends to be quite diffused.

There may be some house accounts that develop over time if we talk about some very large thousands of cobot type deployments that could occur and those are things that we are looking at.

But by and large, 90-plus percent of the market over time, we view as being served through the distribution channel and partner program we have.

As it relates to incorporating other features in the cobot at standards, if you look at vision as an example, vision is a critical element in some cobot deployments but by far not all, in fact today, in Universal Robot's case, roughly 15% to 20% of our deployments do incorporate some form of camera and vision system but not all, so to burden every customer with the expense of a standard camera wouldn't be economical or prudent for the market.

It's not a universal requirement.

Also in the case of deploying vision, if you look at the variety of camera types that have been deployed depending on the application, that's quite diverse.

One camera doesn't suit all applications.

So if you are doing an inspection step, you certainly need a different camera type than if you're doing a product ID step.

Or if you're doing something like a pick and place where you need to visualize an object, locate it guide a robot arm to pick it up.

That is yet again a different camera and vision systems than the other applications.

So we don't see that there is a universal camera and therefore, it's not part of our universal product.

Got it.

Helpful.

And then a follow-up on the SemiTest side, you guys have done a great job in gaining share in mobility test and also on the analog and microcontroller side but from right, your penetration on the GPU side is like almost 0. Do you have any plans to gain market share on the GPU front?

Yes.

We're certainly active in the GPU space.

I would say certainly -- you're correct, our share there is below our average share in the industry.

So it's an opportunity.

It's not 0. But it is low.

So that's an opportunity for us.

We've demonstrated the ability to have a differentiated platform for these quick time-to-market high complexity parts.

And back -- I would say from about 2000 through 2013 or so , we've had a conscious strategy of focusing on mobility, analog, automotive segments and less focus on what was then computer PC type devices.

Now that we have moved to north of 50% share of the market, we've obviously got our sights on underperforming segments and that's one.

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

Mark, first off in terms of I think some comments made about the expansion and broadening of market application for cobot.

Can you give some specific examples of what marketplaces you've seen cobots go into that you may not have thought of say 1 year or 1.5-year ago.

Sure.

So the applications that typically we've targeted have been a big part of our growth.

Our automating the operating of industrial machinery like CNC machines, injection molding machines, conveyor operations pick and place.

Those are sweet spot, huge potential still but right up the middle.

Beyond that, we have seen applications for example, where the cobot starts to do things like polishing applications.

Where the repetitive and precise motions and forces needed to polish a piece of consumer equipment or automotive equipment has been an emerging application.

Another one would be inspection.

We -- putting a camera we talked earlier about vision, some of the applications we have are inspecting parts on an assembly line using specific camera types to look at features and compare them to CAD models.

That is something that's also growing, emerging and we hadn't expected.

Another one that's may be right at the cusp, that's really interesting is service.

Service applications.

So we've talked in the past about -- for example, in Singapore, there is an application where the cobot is equipped with special endo factors that gives sports massage of all things.

There's another application in Singapore that has been developed around manning a buffet station to make omelettes and eggs, that's another example.

There's an application in some geographies related to agricultural, where the cobot is used in dairy production to disinfect cows that come through for milking.

So it's really -- the nice thing about the approach and the universal aspect of the product is we've got an incredibly creative echo system of the distributors and customers out there innovating on how to use the product.

So we've really taken this approach to create the most applicable, general purpose universal product we can and encourage and incentivize and reward distributors who find these interesting and unique applications.

Right.

That's really helpful.

And maybe as a follow-up question on the UR side still.

Given your market share leadership to date and the distribution agreements that you put into place, the competition appears to not have transpired also as probably fast as you thought.

I guess maybe from your perspective, why is not that happen?

Or are these distribution agreements really putting the wall in front for competition not to come in?

My opinion is that it's more difficult than what others thought in getting the ease of programming and the ability to make the cobots very flexible when tasks changed that you could repurpose the cobot.

I believe copying the gears or the hardware can be done.

Now that's not always done properly, sometimes it's not as repeatable or is reliable but generally speaking that can be done.

But there's a lot of very clever software programming know-how that greatly accelerate virtually any task that could be done at Universal Robots and that has proven hard.

Other thing that we believe will be hard to do and people will eventually forget how to program it properly is going to be the distribution.

We're calling on and selling to many accounts that did not use automation before.

So if it's existing automation companies, they don't have a reach into these small or medium-sized companies or these new verticals.

So I think we've created a very good distribution system outside of the traditional automotive players and yes, you will also call it automotive.

And we've got this third-party ecosystem that are developing apps on our platform.

So we think we've got a number of advantages that should keep us in front and we still are surprised that there's a lot of companies you see at trade shows but we are surprised that we really don't see them in the field or they don't seem they have gotten much traction.

Up next we have a question from the line of Richard Eastman of Robert Baird.

Post the first quarter Conference Call, I think Greg, had made the comment that the SemiTest business I think is -- generally, the second and third quarter revenue is about -- just under 60% of the full year revenue.

58% might be the right number if you combine second and third quarter revenue.

And I'm curious, is that kind of weighting your distribution this year?

Is that still intact?

I think what we've said more recently this year is that the revenue in the first half will be a higher percent than in prior years because there was more demand early in the year for a number of reasons.

So the fall off in the second half, we think we will be sharper than prior years.

We didn't make any comparisons with second and third, we made the comparisons with the first and second quarter versus third and fourth quarter Rick.

Well, that was I think on the total revenue basis, you said, first half would be high 50s a percentage of total revenue I believe for tear up for the full year?

I think that's kind of -- I guess what I'm maybe getting on a little bit is the fourth quarter order number for the SemiTest business, obviously, very significant -- you know driven by some of the pull forward on the mobility side.

But I'm curious how the fourth quarter order number for SemiTest might look, given some of the other growth dynamics here that we've spoken to whether it would be automotive or microcontrollers or image sensors?

I mean how do you think what would be your best expectation for a SemiTest order number for the fourth quarter?

Let me just that for a minute.

What I tried to say in the past, and it's very true is that a few weeks worth of changes in order timing by some of our major competitors consuming hundreds of millions bills of dollars of bookings across the Q4 and Q1 boundary condition.

Even in the October call, that's hard to call, as to the exact timing but since there is generally speaking a ramp of shipments in Q1 and Q2 of the subsequent year to facilitate for the smartphone launches in the back half of the year, there is a slug of orders that typically come in.

But we can see $100 million, $150 million of swing in SemiTest orders that are not meaningful.

it's simply a few weeks of timing.

Is there a -- but I think more in terms of kind of book and ship.

I mean is there a basic baseline SemiTest order number for Q4.

If I look historically, it might be $185 million or something like that.

But what would be like a baseline number or is that also not the ...

I can give you just some recent history on that if you looked at Q4's core SemiTest.

And if you go back to Q4 '14, it was $225 million.

Yes.

And then in Q4 of '15, it was roughly $405 million, right?

And then Q4 of '16, it was $523 million.

I think that all makes the point that someone asked us earlier, do booking help you or do they confuse things, so we'll revisit.

Well, I actually, I think, I will throw in the bucket where I think bookings by segment do help you.

I mean, obviously, there's a great deal of lumpiness in the SemiTest side.

I think we all understand why but it's a big enough number to swing things.

I mean, I would suggest, I think the visibility does help but that would be my vote.

Could I ask you one more question?

You talked for a minute.

You had mentioned the SemiTest business potentially or the System Test business, excuse me, eventually, you're looking like maybe $190 million for the year.

And driven by an uptick in the Storage Test for the second half.

Do you have some visibility on that?

Because second quarter orders here in the System Test business don't maybe provide that visibility to us?

Yes.

We do have a little bit of visibility because there was an order that we had in backlog for quite some time, that's part of an R&D program that we're in the midst of with our major customer.

And so we do have some reasonable indication that, that should convert to revenue before the end of the year and that's what gives us that insight.

Okay.

And Operator, we are out of time.

So I am going to say thank you, to everybody for joining.

I know there's still a couple of folks in the queue.

I'll call you as soon as this conference call concludes.

Thank you so much.

This concludes today's conference call.

You may now disconnect.