PDF Solutions Inc (PDFS) 2017 Q3 法說會逐字稿

Good day, ladies and gentlemen, and welcome to the PDF Solutions conference call to discuss its financial results for the third quarter ended Friday, September 30, 2017. (Operator Instructions) As a reminder, this conference is being recorded. If you have not yet received a copy of the corresponding press release, it has been posted to the PDF website at www.pdf.com.

Some of the statements that will be made in the course of this conference are forward-looking, including statements regarding PDF's future financial results and performance, growth rates and demand for its solutions. PDF's actual results could differ materially. You should refer to the section entitled Risk Factors on pages 10 through 17 of PDF's annual report on Form 10-K for the fiscal year ended December 31, 2016, and similar disclosures in subsequent SEC filings.

The forward-looking statements and risks stated in this conference call are based on information available to PDF today. PDF assumes no obligations to update them.

Now I'd like to introduce John Kibarian, PDF's President and Chief Executive Officer; and Greg Walker, PDF's Chief Financial Officer. Mr. Kibarian, please go ahead.

Thank you, and welcome, everyone. If you have not already seen our earnings press release and financial results presentation for the quarter, please go to the Investors section of our website where they are posted.

The third quarter was a busy one for PDF Solutions, with significant progress on all 4 of our 2017 strategic objectives, moving from a company whose technology is primarily used by foundries when they bring up new leading edge nodes to a company whose products and services are broadly used across the semiconductor industry throughout the life of the products and processes.

First, we have had a goal of being a critical component to the entire Chinese IC industry from manufacturing to system companies. In this quarter, we signed 2 contracts, taking a big step towards this objective. A fabless entity, that is part of a larger system company, selected our Exensio platform for its yield and test management solution.

We believe we were selected because: first, Exensio is unique in the marketplace with a strong position in test, fab-yield analysis, fab-process control, packaging and control and analysis. Second, this customer has a long -- has a history of using our Characterization Vehicle data from foundries at multiple nodes around the world. Using Exensio, the customer can leverage the unique proprietary CV data to increase insight and learning from their product test data. And finally, our strong engineering team in China provides them world-class support locally.

Besides a strong win for Exensio, we also signed a Yield Ramp contract for a 28-nanometer foundry in China that is deploying both polysilicon and high-k versions of the process. We have been working with this customer in early R&D. And after making strong progress, they committed to a new factory in production ramp. This gainshare contract will align PDF and the client's interest for over a decade to come.

Our second strategic objective is to evolve our Characterization Vehicle infrastructure to be valuable for customers' usage over the life of a node. PDF's Characterization Vehicle infrastructure and the yield ramps they enable are known to be very effective for the development and ramp of leading edge processes, such as 14-nanometer and 7-.

We have believed that as foundries create more derivatives of 28-nanometer and 14-nanometer processes, offering embedded memories, ultralow power, RF and high-voltage, PDF's Characterization Vehicle infrastructure can become an important link for the fabs and fabless to more tightly couple capabilities of the technology and the designs they enable.

To this end, in the second quarter, we signed a contract with a foundry customer to deploy CV infrastructure through aligning process control with fabless designs for a mature 28-nanometer process. This contract demonstrates that customers see the value in using our CV infrastructure to help them meet their customer satisfaction objectives.

Our third strategic objective in 2017 has been development and initial market penetration of our Design-for-Inspection solution. Design-for-Inspection, or DFI, brings electrical characterization, the lingua franca between design and fab to in-line process control. This allows fabs to control 3D structures and provide a link between fab layout -- fabless layout patterns and process control.

As NVIDIA's John Chen said at our users' conference this summer, the industry needs a solution to detect the undetectable, particularly for complex, large AI chips that are going into mission-critical systems like autonomous driving.

In 2016, we installed our first DFI system as part of an overall yield ramp engagement. The first use usage was included in the initial contract and concluded this summer. Based on the value provided, the customer signed a separate contract for the continued use of the DFI system. This contract is consistent with our other DFI contracts as we are providing subscriptions to the yield information the system provides rather than selling the tool. Our DFI business model, therefore, is in-line with our Exensio business model where the majority of our sales around a ratable basis.

At the end of the third quarter, we now have 4 DFI systems in the field, driving value for our customers and establishing a new business for PDF. From January this year until now, the first 2 machines have inspected collectively well over 1,000 wafers across numerous on-chip instruments, demonstrating that they can be used in production fab.

As to our model with the Characterization Vehicle infrastructure, our system serves the data and information to the customer. We have the ability to then update the design of the on-chip instrument, IT, software and eProbe systems based on our learning. With now over 200 patents filed, many cycles of learning on our systems, we believe we're making significant progress in using e-beam as an electrical test instrument.

We had planned on demoing a customer wafer on our second generation e-beam tool, the eProbe 250, in the third quarter. Due in part to the incremental learning we have made by analyzing the data we have generated with the eProbe 150, we decided to move this demonstration out 1 quarter. We believe this gives us additional time to strengthen our capabilities and tune the demonstration to the yield problems that customers are experiencing today.

Further toward our DFI objective, we continue to have numerous tape-outs that include DFI content. Over the past 12 months, we have almost doubled the number of known tape-outs with DFI. We know of well over 90 chips that have DFI content on the radical, up from about 50 in total 1 year ago. We anticipate further deployments at fabless and fabs as we are able to demonstrate the unique capabilities of DFI and build additional tools.

Finally, the last key element to our strategy is Exensio, our big data real-time platform for electronic supply chain. Over the past quarter, we achieved 3 substantial developments specific to Exensio. First, we had a strong bookings quarter, including, as I mentioned earlier, a fabless company that's part of a larger Chinese systems company.

Also in the quarter, our existing fab customers purchased and deployed more Exensio modules for process control and yield analysis. We also had strong bookings of our products for back-end assembly and test. Overall, we had approximately 200 contracts close in the quarter, a record for us.

Second, we closed the acquisition of KINESYS ALPS product. As we talked about on our last quarter's call, ALPS allows customers to ID and track all components that go into a package or system. This is critical for systems and package applications like sensors and IOT.

In the short time we have owned this capability, our customers have told us that systems companies are requiring them to have the traceability information that they found ALPS is uniquely able to provide. These customers are excited about having the data ALPS provide integrated into Exensio, so they can close the loop between test, fab and packaging to improve their product quality, yield and reliability.

Third, we had a successful Exensio users' conference with engineers from 34 companies attending. Today, we have over a 100 customers of the Exensio platform. At our users' conference, and later as I attended executives' meetings around the world, we consistently heard customers tell us they are excited about Exensio's solution across the entire supply chain and the collaborative model we are bringing to the industry. We are encouraged by the significant opportunity we have with Exensio.

We believe our innovations in Exensio and DFI will also change the way we bring CV infrastructure to our customers. In the past, the majority of our cost in our solutions revenue was related to human capital cost. As DFI and Exensio contribute more and more to this revenue, our costs migrate to depreciation of our DFI systems, PD fast testers and computing hardware to support our customers' use of the software and infrastructure.

Moreover, while solutions revenue was historically tied to project revenue, it is more now a blend of license fees frequently tied to customers' test and equipment capacity and usage fees. Because many of our contracts include CV test chips, DFI content or systems and Exensio modules, it is more and more difficult to break out the individual contributions. However, Exensio and DFI represent approximately half the solutions revenue today.

As we look at the overall market, fabless chip companies and system companies are doing well. We are seeing a mixed bag at the foundries. Multiple foundries have demonstrated they can succeed on bringing up a new node. However, they cannot maintain filling that node throughout the life as their largest competitor have defined derivative nodes that provide new value to the end customers with foundry capacity. These follower foundries will struggle to maintain utilization rates if they don't evolve.

We are engaging with customers now to demonstrate how we can help them maintain and improve volume by moving more rapidly in developing derivative nodes using our infrastructure. Our gainshare results for the quarter represents what we are seeing in the general market. Some 28-nanometer customers' volumes were quite weak while others had stronger performance. On top of the 28-nanometer gainshare, we saw reasonable performance on 14-nanometer volumes.

In the short term, we expect our customers to have variable volumes. However, as we look out to 2018 and beyond, we believe volumes will improve at our customers. For Q4 and 2018, we believe the market will continue to validate our long-term strategy. While 2017 has not been a strong year for us from a revenue standpoint, we believe it has been a strong year in terms of the progress on our strategic objectives, setting up a stronger 2018 and beyond.

Overall, the third quarter demonstrated our moving from a company whose technology is primarily used by the foundries when they bring up new leading edge nodes to a company whose products and services are broadly used across the semiconductor supply chain throughout the life of the products and processes.

I will now turn the call over to Greg, who will review our results in detail and discuss our Q4 outlook. Greg?

Thank you, John. As you may have seen in our earnings release, we have posted in the Investors Relations section of our website a management report with detailed comments regarding the financial results of PDF for the quarter. Given that, I'm going to focus my verbal comments for the quarter on a few key highlights reflected in those results.

First with revenue, total revenues of $26.5 million for the quarter were up $2.2 million as compared to Q2 2017. Solutions revenues, at $19.2 million, increased by $2.7 million from Q2 while gainshare revenues, at $7.3 million, decreased by $500,000.

As John said, gainshare volumes increased during the quarter, primarily driven by a continuing 14-nanometer ramp at one of our major customers and 28-nanometer volume increases across multiple customers. The Q3-over-Q2 improvement in volumes and related revenue, however, was offset by the one-time performance bonus recognized in Q2 related to achievement of R&D milestones at 14-nanometer -- at a 14-nanometer customer that was not repeated in Q3.

During the quarter, the increase in solutions revenue was primarily related to closing 2 contracts that we originally expected to close in the prior quarter and higher activity levels related to a large memory project booked in Q2.

Additionally, DFI revenues increased during the quarter due to the conversion to a standalone agreement of an eProbe 150 system previously deployed as part of a yield ramp project. This is the same project that John was referring to.

Finally, Exensio revenues increased during the quarter based on strong bookings, including several repeat customers increasing the user counts or tool license counts in their applications.

GAAP net income for the quarter was approximately $600,000, up by $400,000 over Q2. Non-GAAP net income for the quarter was $3.6 million, up $900,000 from Q2.

Looking at expenses, on a GAAP basis, total expenses for the quarter were $26.1 million, approximately $1.2 million higher than the previous quarter. This increase in expense was primarily due to recognition of previously deferred project cost on an engagement that was signed in the quarter, also an increase in development activity on the next-generation eProbe tool and nonrecurring expenses related to cost-reduction actions implemented during the quarter. The increase was partially offset by lower SG&A expenses due primarily to lower legal expenses related to the KINESYS assets acquisition, ALPS, which was completed early in the quarter.

On a non-GAAP basis, total expenses for the quarter were $22.5 million, approximately $700,000 higher than the previous quarter. The increase was primarily due to the items previously mentioned for the GAAP increase, with the exception of the impact of nonrecurring severance payments related to our cost-reduction actions. Those have been excluded from non-GAAP expenses.

On a GAAP basis, cost of sales was $12.4 million, approximately $1.1 million [higher] than the previous quarter. This was primarily due to the recognition of deferred cost mentioned earlier, the nonrecurring cost-reduction actions mentioned earlier and a higher stock comp expense. On a non-GAAP basis, cost of sales was $11 million, approximately $700,000 higher than Q2, once again primarily due to the recognition of deferred project costs.

On a GAAP basis, R&D expenses were $7.9 million, approximately $600,000 higher than the previous quarter, primarily due to the increase in development activity on the next generation of the eProbe tool. On a non-GAAP basis, R&D expenses were $6.8 million or $400,000 higher than Q2, again due to the eProbe tool development activity.

On a GAAP and non-GAAP basis, SG&A expenses were $5.7 million and $4.7 million, respectively, with each decreasing approximately $500,000 from the previous quarter, primarily due to lower acquisition-related legal expenses and a variety of other miscellaneous expenses.

Looking at the balance sheet, total cash, at $100.8 million, declined by $8.3 million during the quarter. This reduction was primarily the result of stock repurchases totaling $8.6 million, the cash impact of $3.8 million in acquiring the ALPS assets from KINESYS; property, plant and equipment purchases of $2 million, most of which were related to the development of our DFI solution; and $600,000 for the purchase of company stock related to the settlement of employee tax obligations and RSU grants, the total of which was offset by cash generated from operations of $5.3 million and stock option exercises and employee stock purchase plan purchases of $1.5 million.

Accounts receivable, including total unbilled receivables, was $62 million in the quarter, which represented a decrease of approximately $1.7 million from the prior quarter. DSO for total accounts receivables decreased from 239 days in Q2 to 213 days or an 11% decrease over the prior quarter.

The total current and long-term unbilled AR balance of $29.8 million decreased during the quarter by approximately $900,000. Of the $29.8 million unbilled AR balance, we expect $20.7 million to be billed over the next 12 months, of which $10.9 million will be billed during Q4. Since the end of the quarter, we have collected approximately $2.5 million of the $32.2 million trade accounts receivable outstanding.

Our GAAP tax benefit for the quarter was $270,000. Cash taxes incurred for the quarter were $320,000 or a cash tax rate as a percentage of GAAP pretax income of 100%. For the full year, we expect our GAAP tax to be a benefit in the range of $1.5 million to $1.9 million. Cash taxes incurred for the remainder of the year are expected to be in the range of $300,000 to $400,000 for the quarter.

Looking at the remainder of 2017 from a total revenue perspective, we expect Q4 to be approximately in line with Q3. Similar to total revenues, we also expect gainshare revenues to be approximately flat quarter-to-quarter. Continued growth in our Exensio solution should offset the non-reoccurrence of previously deferred project revenues in solutions recognized in Q3. Total spending should also be approximately in line with Q3.

In summary, we saw improving results in all aspects of our solutions business during the quarter. In particular, our Exensio business demonstrated continuing growth at a rate higher than the overall industry. Additionally, during the quarter, we continue to make excellent progress on our development and marketing of our DFI solutions, including the conversion of the system I mentioned previously.

Finally, though on a dollar basis gainshare did decline quarter-over-quarter, after adjusting for the previously mentioned one-time performance bonus in Q2, we saw sequential improvements in both gainshare volumes and revenues.

From a spending standpoint, aggressive development of the eProbe 250 system will continue to drive high levels of R&D expenditures for the foreseeable future. However, during the quarter, the company undertook limited cost-reduction actions to partially offset this high level of R&D expense.

With that, I will return the call over to the operator for Q&A.

(Operator Instructions) And the first question comes from Jon Tanwanteng from CJS Securities.

John, can you clarify the e-beam or the 250 series demo push out that you mentioned earlier in your prepared remarks, is it adding more capabilities? Or are you needing more time to do what you originally thought you could? And kind of second, are you still on track to generate revenue from that 250 series machine in the first half next year?

Yes, we certainly were on track to be able to ship machines first half in 2018. We are adding slightly more capabilities, different capabilities as we've looked at the data we've seen on the -- off the 150. And also finishing up some capabilities that we wanted to have buttoned up a little bit tighter for the demo itself. So it's a little bit of both, Jon.

Okay, got it. And then, Greg, you mentioned expenses are going to stay elevated for the R&D portion on the 250. How long is that going to stay elevated for? Is there a point where that drops off after you start shipping them? Or are you going to keep adding capabilities? Does that mean R&D stays high?

I think by the time we get to the middle of next year, we should probably see a decline in R&D expense, but we'll see a pickup in cost of sales as we start to ship. But there'll still be an elevated level of R&D expense, but just probably not at the peaks we're seeing now.

Okay, got it. And then just from a customer basis, what has been the response so far to what you've been telling them about the 250. So are they excited? Are they -- how should we think about the pace of orders in the next year once you start delivering the first one?

Yes, so what we -- what we first started showing customers was the data they could get off the 150. In the spring, when I mentioned the capability of the 250 to one of the customers, the guy said, "Well, gee, we have got great data off the 150, what do we need additional measurement data for?"

Then as they started designing all of their vehicles for 7-nanometer, and they saw how much area they could consume up, they were very intrigued with the idea of the 250 and the ability to be able to look inside the die where the area is free, which we can't do on the 150.

And even be faster -- substantially faster on the scribe line as we started finding customers were using up the scribe line area in ways that we hadn't anticipated and some of the incremental capability we're putting in the 250s to make it even superior in the scribe line. Far enough ahead of the scribe line capability on the 150 to be valuable for that application, too.

So I think we've had very strong response from the fabs and fabless customers about the 250 and the 250 demos. And then, on the 150, incrementally, additional information on measurement types that they're very intrigued with, and some of the changes they're making to the 250 are to be able to meet the incremental data types and information they want off the systems. I mean the customers are really seeing this as an electrical test data, and so they're asking for more and more subtle electrical test information off these systems.

Okay, great. And then, finally, you mentioned that your foundry customers are having a hard time filling their factories, especially over the life of your contracts. Have you considered any other changes to the proportion of what you charge upfront versus the back end gainshare royalties to kind of minimize that risk?

Yes, we have, Jon. From our perspective, it will be a good thing to do. That doesn't specifically help the customers though. And over the longer term, right, they won't be viable customers at some point. And I think, particularly with the investments that the industry is making in China, those factories have to fill up at some point and because of just the geopolitical situation out there.

So we are looking at ways on how to become important to them for that aspect of their business. We just -- we'll adjust it both ways, but we particularly see finding ways to make it a win-win in a longer term, more viable way.

And your next question comes from Tom Diffely with D.A. Davidson.

So following up on the DFI tools, what are the prospects for additional sales on the 150? I know earlier this year, you thought that there could be a few more going out the door by the end of the year. Is that still the plan? Or are customers now waiting for the 250 to be ready?

We are still in the position to be able to ship additional 150s. We've planned for that, it'll be the end of this year or early next year. And we still have pilots and demos going on with customers on 150 capability. We believe they will have a life and value in the manufacturing facilities for a time to come, particularly for test vehicles.

Over the longer term, the 250, we believe, as far as some of the things we added in R&D this past quarter, could even do that even more cost-effectively than the 150. So over time we may shift to simplify -- both the customers and us, a balance of machines in the world.

But we do believe, at least, through our -- and our plan has been for quite a while that for a good chunk of 2018, even though we'd be bringing out 250s and selling them in the marketplace, a sizable fraction of the revenue we generate on DFI in 2018 has always been expected to be of the 150 platform.

Okay. And then when you look at the next few sales potentially of the 150, is it more likely to be to new customers or repeat orders to existing customers?

We have both discussions going on with customers right now, Tom. It's hard for me to forecast what would happen first.

Okay. And then when you look at the spending going forward, you said it was going to be roughly flat quarter-over-quarter, what kind of increase in spending or OpEx is there from the acquisition?

It's very minor. We primarily did an asset acquisition, and we brought a few people onboard, but it's a very small number. And we expect that acquisition to be accretive almost instantly.

Okay. And then when you look at the gainshare part of your business coming up a little bit on a core basis then it looks like it's going to be flat in the fourth quarter. Is it still a situation where the 28-nanometer is going to continue to be weak? And I thought there would have been a little bit of ramp in the 14-nanometer. Has that business essentially maxed out at this point?

We don't believe it's maxed out, Tom. We do believe the 14-nanometer business will improve in 2018 as well as it improves -- as it improved in the second half of this year and the first half of this year. We thought the second half of the year was roughly flat Q3 to Q4 due to different customer seasonality, some were up and some were down.

As regarding to 2018, we think there will be more 14-nanometer gainshare revenue than there was in 2017. Overall, in the 28-nanometer, customers, particularly the newer customers, the customers that are newer in the node, they are still struggling to get meaningful and sustained volumes. The ones that have been in the node for a while, I would say that they're -- their volumes are a mixed bag, some are having decent quarters and some are not.

Okay. And then just to clarify, of the 4 150 tools that are in the field, are 3 of them now on a subscription model?

2 are on a subscription model, 1 is part of an overall IYR engagement. And so it's -- revenue was part of the overall purchase the customer made. And the fourth one is used primarily for R&D on a JDP related to a specific electrical measurement.

Okay. And finally when you look out to 2018, do you think there will be more growth coming from kind of core business in Exensio or from the new DFI tools?

Exensio is a ratable bookings business, and so it tends to build out over time, so is DFI. But it's had a couple of years of, 2016 and '17, of very strong bookings. So even with relatively modest incremental bookings in 2018, it will still build revenue. And DFI is coming off a smaller base. So on a percentage basis, it's going to grow faster than Exensio will. In terms of absolute dollars, we still Exensio to be a bigger contributor in 2018.

The next question comes from Tom Sepenzis, Northland.

Just wondering if you could talk about some of the capabilities that you want to add to the eProbe 250 that's causing this push out?

Yes, so the customers, in particular, one of our foundry customers, is very interested in being able to relate what the eProbe [piece], what we call the Electric Response Index and the electrical value that, that actually represents.

So for our structures that measure leakage between fins, they want to be able to understand for a given, what we call, ERI value -- I know that sounds really technical -- what would be the current level -- well, how much leakage current would be there roughly in certain buckets if you were to actually be able to probe that electrically. And the same for resistances, they want to be able to understand specifically not just that there is a highly resistive contact as opposed to a short, but what is that resistance value in a bucket.

So the customers want us to be able to get more fidelity around the resistance buckets. We're not able to do that on the 150. We think there's some things that we're adding to the 250 to improve its ability -- its sensitivities that we're trying to be able to get to.

Great, that's helpful. And then just in terms of modeling purposes, I know you don't give specific guidance, but directionally in 2018, should we be thinking about some of the newer products adding growth to the topline? Or how should we be looking at that?

Yes, I think that you can expect in '18 that we would see growth in the Exensio business, the DFI business, once again, off a small base. So the absolute dollars aren't of the same scale. And then also, we are expecting some level of recovery in gainshare.

All right. And then lastly, if you could just provide any kind of update on 3D NAND progress. I know that you've got one customer, but any additional customers on the horizon that we should be thinking about? Or how's that first customer going?

Yes, it's going quite well. We have been able to demonstrate a lot of interesting information that will help them accelerate their program. We've also been able to talk or have discussions with other customers in 3D NAND and go over the capability, both that we're able to see on DFI as well as what we're able to see on electrical test.

And we will release, starting in the first quarter of this year, an updated pdFasTester system specifically for nonvolatile and 3D memory and XPoint memories to allow for even better characterization of that -- of those processes and technologies.

And your next question comes from Andrew Wiener with Samjo Capital.

With respect to the new Chinese 28-nanometer customer, can you give us a sense of when you'd expect gainshare from that customer -- to see our first gainshare?

Yes. I believe, Andrew, that would be late 2019.

Okay. And do we have any -- I mean, what's our sense of what the initial wafer capacity they're planning to build?

Yes, they're building a facility, I think it's a 40,000 wafer start facility. It's relatively large. What the [shell] can handle for a capacity and what they tool up, we will wait and see. They've almost completed the shell now.

Okay. Then secondly, when you think about the guidance for Q4 for gainshare, does that include -- is there an expectation of any milestone payments similar to what you achieved in Q2?

No. There's not.

Okay. With respect to the arrangement with GLOBALFOUNDRIES, one of their large customers disclosed that under the terms of the wafer supply agreement that they purchased certain wafers from a foundry other than GLOBAL. Do we benefit from that in any way?

I didn't get a chance to read that customer's WSA agreement or any disclosures, so I'm sure I can comment. I know that GLOBALFOUNDRIES has on their 40-nanometer technology, technology alignment with Samsung. And of course, both those foundries we benefit from. I don't know if that disclosure would have been their customer buying wafers from a foundry other than Samsung. And so like -- if that were the case, then we would not benefit.

Okay. And with respect to the -- sort of the tweaks or whatever, however you want to call it, on the 250, given that we're sort of responding in large part to certain things customers are asking for, how do we think that will delay getting sort of a first tool in a customer? And do we think it potentially shrinks the time to commercialization given the sort of level of interaction at this point?

Yes, this has been -- that's a good question, Andrew. This is part of the calculus that we made in the third quarter. We have a really tight ongoing working relationship, particularly with one of the customers on exactly what things they want to be able to characterize. This is part of the reason why we have a JDP going on with them on specific electrical mechanisms they think are critical to advance the products.

And we have had a dialogue with them about what we are able to see with the 150, what we can see with electrical characterization itself, and what we think we could do for a 250 capability. And they are very vocal about what things they're interested in.

So we are trying to adjust to what we think they say, what they think they need. And we believe that, that would help us on the commercialization and time to revenue. Hence, part of the reason why we took this additional pause.

Okay. Does that put then getting the first tool into -- first beta tool into a customer into Q1 of next year? Is that the thinking currently?

It would be Q -- yes, the first half of the year, for sure. I think it depends on how successful we are with the data. We've actually designed with them a relatively detailed beta. They are sending multiple wafers to us across different conditions to see what we're able to do. So I think it depends on how long it takes us to complete all of those wafers they want to look at.

At this time, there are no more questions. Ladies and gentlemen, this concludes the program. Thank you for joining us today.