PDF Solutions Inc (PDFS) 2018 Q4 法說會逐字稿

Good day, ladies and gentlemen, and welcome to the PDF Solutions, Inc. conference call to discuss its financial results for the fourth quarter and full year ended Monday, December 31, 2018. (Operator Instructions) As a reminder, this conference is being recorded.

At this time, I'd like to turn the call over to our host, Gary Dvorchak, you may go ahead.

Thank you, Jerica, and good afternoon, everyone. We appreciate you joining us for PDF Solutions' Fourth Quarter and Full Year 2018 Earnings Conference Call. We distributed the press release earlier today after the close. It's available on the company's website and from Newswire services.

Joining me today is John Kibarian, PDF's President and Chief Executive Officer; and Christine Russell, PDF's Chief Financial Officer. First, John will review the operating highlights, and then Christine will review the financials for the quarter, and then we'll take questions.

Before I turn the call over to John, I'd like to make a brief statement regarding forward-looking remarks. This call contains forward-looking information regarding future events and the future financial performance of the company. We caution you that such statements are predictions based on management's current expectations or beliefs. Actual results may differ materially as a result of risks and uncertainties that pertain to our business.

We refer you to the company's press release issued this morning -- this afternoon, excuse me, under our SEC filings. These documents discuss important factors that could cause actual results to differ materially from those contained in the company's projections or forward-looking statements. We assume no obligation to revise any forward-looking statements made on today's call.

In addition, we'll be referring to non-GAAP financial measures during the call. These non-GAAP financial measures are not prepared in accordance with generally accepted accounting principles. A GAAP to non-GAAP consideration -- reconciliation as well as an explanation behind the use of non-GAAP financial measures is available on our press release, prepared remarks and in the appendix of slides.

With that, I'll now turn the call over to John. John?

Thank you, and welcome, everyone. If you've not already seen our earnings press release and management report presentation for the quarter and full year, please go to the Investors section of our website, where each has been posted. Today, we will discuss our results in the fourth quarter, our performance for the full year. The factors that we believe position PDF well in the marketplace and the implications for 2019.

Several years ago, we embarked on a multiyear initiative to evolve PDF into a semiconductor analytics platform company. The key component to our platform is differentiated data sources, such as the Design for Inspection and Characterization Vehicle data.

Our analytics enable customers to have foresight. We give them the ability to see what is likely to happen in the future manufacturing test or operations and prevent or correct for an impending problem. This is a significant improvement for the industry. The conventional approach involved using data for historical insights in events that already happened. Our approach now is proactive, not reactive.

The Exensio platform is being deployed across the supply chain. Customers and suppliers can work collaboratively to squeeze more value out of their manufacturing technologies at a time when Moore's Law is diminishing.

We are now moving our revenue model to a ratable subscriptions. We believe this model is more predictable and less variable than the old approach, which was a combination of solutions deployment fees and Gainshare incentive. The foundation of our new business model is nearly complete. We have made substantial progress in 2018, including important achievements in the fourth quarter. Let me elaborate on those achievements and then review the full year.

Importantly, we achieved a key milestone for the eProbe 250. As we announced in the last -- as we announced last month, we shipped our first eProbe 250 as part of an order booked in the fourth quarter. This 1 year initial contract is broad as it also includes electrical test. Most importantly, the contract prices the eProbe 250 in a production fab, which is a watershed event in the history of PDF Solutions.

We believe we won this business because the eProbe 250 has the unique capability to measure in line during wafer processing. As such, it can predict the probability of downstream yield or reliability issues. This is particularly important to our fabless customers in computing applications for server, cloud and automotive, where reliability is paramount.

The second major achievement in the quarter was a contract that we believe demonstrates the value of enterprise-wide data integration enabled by our Exensio platform. We closed a paid pilot with a long-time PDF customer that until now used only elements of Exensio in their manufacturing.

The customer is now evaluating Exensio as an enterprise-wide platform. They need a platform that can drive higher quality, improved operations efficiencies and maintain high yields. This pilot is intended to demonstrate that Exensio can meet their price needs. The contract is proof-point that customers recognize the merits of our platform vision.

Our third achievement was seeing initial market acceptance of our cloud-based Exensio platform. We have hosted Exensio for years, but that offering was more limited in its capabilities. Our new public cloud-based product not only offers complete Exensio analytics functionality, but we also add a series of value-added services.

In Q4, we closed our new -- our first cloud Exensio contract for a fabless customer. In general, the analytics business is accelerating, which underpins our confidence in the new time-based licensing model. We closed a number of other significant contracts, including a renewal for a customer that is using a combination of Exensio and our electrical CV infrastructure to better control their fab.

In 2015, this was our first account to implement the time-based model. This renewal demonstrates the strength of our time-based model. With the renewal, this contract will equal or exceed the longevity of our old Gainshare contracts, will capture equal or greater value and will have less variability.

Overall, during the quarter, we made significant progress in leveraging our investments in R&D to execute on our long-term objective of becoming the leading provider of semiconductor analytics.

Now let me step back and review our accomplishments during 2018. The new business model is starting to deliver. Our real-time analytics business represented well over 60% of our solutions revenue. This is up meaningfully from 2017 when analytics was under 50% of our solutions revenue.

Last year, what we heard consistently from the market is a general need to use data coming from the fab and other sources, including electrical test, to predict quality and reliability. Customers don't want to wait until they have reams of test data to apply rules to achieve better quality. In today's fast-paced world, this conventional approach delays product and manufacturing efficiency and increases costs.

Let me offer some highlights from 2018 that support our confidence. First, at IABM, a prominent silicon innovation conference, STMicro discussed how they were able to quickly characterize yield and quality metrics and their non-volatile-embedded PCM technology by using PDF's CV infrastructure. Importantly, they were able to do this without having to run an entire process file.

Another highlight was that one of our fabless customers who used an eProbe 250 to identify failures during manufacturing, the customer noted that our detection capability was consistent with their quality and burn-in failures. Our tool has the potential to detect problems in real-time that otherwise are not screened for 2 or 3 months. Importantly, this capability was resilient. We detected problems across 2 technology nodes, 3 products and 3 foundries.

Finally, in 2018, we showed our Exensio's networking capability is able to combine data, detailed fab data with assembly test and test information to optimize product quality and manufacturing efficiencies. Exensio's network of test facilities now covers most of the major OSATs and lead foundries.

This is important because customers' production flows are run at a mix of internal and external facilities. Exensio is the only platform running across the supply chain from tools in the foundries to the assembly fabs to the test floor. Throughout the entire manufacturing flow, internal and external, our customers can react in real time to improve their manufacturing.

Going into 2019, we are confident that we will, again, make meaningful progress in the evolution of our business model. We believe that the foundation is built, and we can now go-to-market more aggressively and demonstrate compelling unrivaled value proposition. With the completion of the eProbe 250 hardware development, R&D costs will transition to driving applications and revenue opportunities. Future capital expenses will be aligned with future business.

In other words, with few exceptions, we have the hardware design we need to drive the growth of DFI.

Second, this coming year, we will see our analytics business to grow well beyond 70% of solutions revenue. Analytics will become the driver of PDF's presence in the market place. We believe the Exensio platform is sufficiently broad and engineering effort has moved to solutions development. These solutions, in many cases, imply AI machine learning to squeeze more information from CV infrastructure and other in-fab data to provide foresight to customers and additional business opportunities for PDF.

As Q4 demonstrated, the investments in Exensio and DFI are driving the strength in business activity we are experiencing. We believe 2019 will be a strong year in bookings compared to 2018. Given the ratable nature of our model, the revenue will follow from the business activity. Because of our capital expense management and expected uplift in gross margins as software becomes a larger component of our product sales, we expect improvements in our 2019 non-GAAP income over 2018.

The evolution of our business model entails risks in investments. We are grateful for the patience and latitude our shareholders have given us during this period. We're also thankful for the dedication and support of PDF employees. As we steward the company through this important inflection point, we remain committed to reignite profitable growth by delivering even greater functionality and value to our customers through the sustained leadership in the market.

Now I will turn the call over to Christine to review the financial details.

Thank you, John. Most of you will have seen our financials in our earnings release. In addition, we posted in the Investor Relations section of our website a management report with financials and comments regarding the results of PDF for the quarter. So I'll focus my comments on a few key areas.

All of the financial results that we provide on this call are on a non-GAAP basis, which excludes stock-based compensation, amortization of intangibles, onetime and restructuring charges. Please refer to our press release for our GAAP results and GAAP to non-GAAP reconciliation.

Revenue for the fourth quarter was $19.7 million, a 2% decrease from the prior quarter. Solutions revenue decreased quarter-over-quarter by $1.3 million and Gainshare revenue increased by $839,000. The decrease in solutions revenue was primarily a result of lower hours spent on IYR projects, partially offset by increased sales of Exensio DFI solutions.

Exensio DFI revenues benefited from the extension of an existing Exensio contract with a large Asia-based company and the renewal and purchase of additional product with a major Taiwan fab. The increase in Gainshare revenue was primarily at the 14-nanometer node.

Cost of sales for the fourth quarter was lower by almost $1 million at $9.7 million compared to the third quarter. The lower costs are a result of our restructuring activity initiated during the period.

Total operating expenses for fourth quarter were $11 million, which is $248,000 higher than the third quarter, primarily due to a onetime payment to our former auditors for their final fees.

Compensation was lower as we continue to manage down our employee base through voluntary attrition, slower hiring and our fourth quarter [risk]. Travel expense and the cost related to subcontractors were also lower than the prior quarter. We continue to analyze all the costs of the business and manage our resources wisely. We reduced headcount year-over-year by 52 from 417 to 365 at the end of Q4 due to lower resource requirements for IYR. And with DFI development largely completed, we're able to shift our resources to Exensio.

The company posted a loss of $366,000 for the fourth quarter of 2018 compared to a profit of $205,000 in the prior quarter. The company was profitable on a pretax basis at $113,000 with a tax provision of $479,000 driving the bottom line loss of $366,000. Non-GAAP per share loss for Q4 was $0.01. Shares outstanding for Q4 were $32.3 million.

Looking at the balance sheet, cash at the end of Q4 was $96 million compared to $97 million in the prior quarter. We generated $4.3 million cash from operations, and we used cash for PP&E for new eProbe builds and also the earn-out payment for the acquisition of Kinesys.

Accounts receivable decreased slightly in the fourth quarter to $29.3 million compared to the prior quarter of $30.1 million. DSOs were 135 days compared to prior quarter of 135 days.

Turning to the year 2018, it came with a number of challenges. 28-nanometer loading at the smaller foundries remain challenged and GLOBALFOUNDRIES exited 7-nanometer. These headwinds put pressure on our revenues. As a result, in 2018, we reduced costs.

For the year 2018, solutions revenue was $60.1 million compared to prior year of $74.4 million. The reduction in revenue was in our classic IYR business. The analytics part of the business grew year-over-year. We anticipate that our analytics product suite will continue to be the growth engine for our company. Gainshare revenue declined by 6% year-over-year. Gainshare revenue, which is a volume-based royalty, is tied directly to our Classic IYR business.

Cost of sales was reduced year-over-year by 9%. Cost declined from $42.3 million in 2017 to $38.4 million in 2018. This was a result of our cost streamlining initiative. Looking ahead, there are a couple of factors which can put upward pressure on our gross margin. Our fixed costs are mainly compensation, which is expected to be stable going forward. Therefore, incremental revenue falls mostly to gross profits. Second, our mix is improving as we generate more analytics revenue.

Operating expenses also were lower in 2018 compared to 2017. OpEx in 2018 was $44.8 million, down from $46.3 million in 2017. Our cost-containment focus resulted in reductions in headcount as well as travel expense, use of subcontractors of financial services.

Since compensation comprises almost 60% of our spending, the headcount reduction materially contributed to lower expenses. Our total spending, including both cost to sales and OpEx, decreased by $5.4 million year-over-year from 2017 annual spending of $88.6 million to 2018 spending of $83.2 million. We posted a profit of $3.1 million or $0.09 per share in 2018. Shares outstanding for 2018 were $32.2 million.

Turning our attention to our balance sheet for the year. We generated $13.3 million of operating cash and used $13.1 million for PP&E, which included leasehold improvements for our company headquarter move and expansion of our clean room and labs and the initial build of eProbe 250s. We used $5.2 million cash for stock buybacks during the year.

That concludes our prepared remarks. We'll turn the call over to the operator for any questions. Operator?

(Operator Instructions) Your first question comes from the line of Jon Tanwanteng with CJS Securities.

Can I ask, the eProbe 250, is that now installed on the floor of your customer? And has there been any initial feedback? How are they experiencing it?

It arrived before the New Year's holiday in Asia. It was secured for earthquake, they went away for a week, and I think now it's being installed as we speak.

Okay. Great. And is it generating...

It won't be generating data till effectively Q2. And they continue to use the eProbe 150 until we have the eProbe 250 up.

Okay. Great. And are you taking that 150 back? Or is it staying there?

Normally, it will come back, still to be seen. Contracts, it will come.

Got it. Jumping to the Gainshare, the $7.1 million in the quarter, which was nice to see, is that all from GLOBALFOUNDRIES 14-nanometer? Was there any China in there at all? And just any color on how that played out? And then 28-nanometer was in the next...

That is more than -- of course, it's much more than just GLOBALFOUNDRIES. That includes foundries in Taiwan as well as foundries in other parts of the world, including China.

Okay. Was there any other 14 besides GLOBAL? I guess it's a...

There was [almost] 14, yes. Actually, a couple of 14s, now that I think about it, 2 other 14s.

Okay. Great. And any update on the situation with GLOBALFOUNDRIES now that they have terminated their 7 investment? Are you going to see any compensation for the contract you had with them? Or are you still working on that? Any update on timing and kind of size of what the resolution may be?

Yes. We -- As we said, our contract is noncancelable. We -- We'll make -- we will collect all the money from that contract either from that direct contract or additional business, and we are in discussions with them on that as we speak. We have a very long history with them, and we don't see any issues there.

Okay. And just to be clear, there wasn't any in Q4, right?

No, we took -- there was -- because we had no hours in Q4, we had no revenue rec for that contract in Q4.

Okay. Got it. Christine, can we expect operating expense to be lower as we go forward from the restructuring and the cost-saving efforts you're doing? Or are you going to reinvest that somewhere?

Well, I would call it flattish except for one area. Since our revenues are going to be increasing slightly, as John mentioned, we do have some variable expense associated with that, that includes sales commissions and bonuses. So to the extent revenue increases, those commissions and bonuses will increase.

Okay. Got it. And then just on an overall revenue basis, John, as you look into '19, you talked about the analytics being the growing piece of your business. Can you just clarify what analytics means? Does that mean DFI plus Exensio? Or is there other stuff in there? And kind of what was that base of business in 2018 and kind of what are the expectations for growth in '19 overall?

Yes, so our solutions revenue in 2018 was around $60 million for Christine's words correctly, and we said north of 60% of that was analytics. And what that includes is Exensio as well as contracts like the contract I described on the renewal, where it's a combination of Exensio and CVi, where a customer will use that Characterization Vehicle infrastructure in a fab on a ratable basis, and it does include DFI. Anything that's on a time-based license or a license model, primarily basically analytics as the core, Exensio as the core.

Okay. Great. And you've previously talked about the growth for Exensio as on a stand-alone basis, and we can kind of model what DFI may do depending on the number of machines you sell. What do you think analytics does year-over-year on an absolute growth basis?

Yes. I don't think we've given out that specific number, right? We said now we believe the overall solutions revenue will grow, and it will be north of 70% of the solutions business.

Okay. Got it. But I assume that the other side would be the traditional [highway or] would be shrinking (multiple speakers).

Yes. We anticipate that decreasing, but overall, solutions will grow.

Your next question comes from the line of Tom Diffely with D.A. Davidson.

Yes, maybe a similar comment on what you think of Gainshare this year?

Yes. We've modeled it to be slightly down over this past year. Although we do get -- I mean, we do find it tends to be sometimes more resilient than we model, but we expect it to be slightly down.

Yes. And we do expect it to continue to be lumpy quarter-over-quarter.

Okay. And would you expect, at some point, to either get a lump sum from GLOBAL? Or I guess, maybe it doesn't have to be if it's built into new contracts, but is that kind of a risk to the upside at some point, you might get a slug of -- or slug of payment?

Yes. Of course because we haven't finalized what we're doing with any, it's hard for us to forecast or envision it, Tom. But we expect that it will -- if we reach resolution, that will have an impact -- a positive impact on this year.

Yes. It would. And what you have to remember is since we've been recognizing revenue on a percent of completion basis, hours invested, that any dollar agreement we come to with GLOBALFOUNDRIES will not necessarily reflect the revenue that we post. However, there should be a gain in the revenue for that because we've earned it.

Okay. And Christine, maybe you could just go over the OpEx one more time. It look like went up during the quarter, but it sounded like from your commentary, it should have gone down with the cost-cutting programs.

Well, actually, we had one area where it did go up, and we transitioned auditors. We let go our former auditors. We started with new auditors, and when that happens, you can have -- we did have several hundreds of thousands of dollars of transition fees in which the former auditors are charging for access to work papers and transition and things like that. So that's a onetime expense.

Okay. So would you expect SG&A going forward to look more like third quarter numbers?

I would expect SG&A, because of the variable nature of commissions and bonuses, to increase during 2019 as the revenue increases.

Okay. All right. And then, John, when you look at the eProbe 250, it sounds like it gets up and running over the next quarter or 2 and then gets into production. Once it gets into production, are they going to need a second tool right away for redundancy? Or is this more of a beta tool in production? Or how would you describe it for the rest of this year?

Yes, I think we've modeled the rest of this year that they're using one tool primarily as -- prove out as a manufacturing application, not in an R&D mode. We would expect if they decide that this is important for manufacturing, one tool would not be safe even from a redundancy standpoint let alone from a volume standpoint. But frankly, we built the model assuming it's one for the year, although we don't -- we obviously don't work towards that, we work towards doing better than that.

Okay. And then based on what you know about the number of data points looked at and evaluated. What -- is there a number of tools needed for a typical line? Or is there some way to describe what the opportunity is on a either per fab or per node basis?

Yes. We still believe what we said before, which is for modest sampling of a fab -- of a meaningful fab, say, a 30,000 or 40,000 wafer start fab, you would need a handful of machines to be able to even keep up with that. And we are -- frankly, we are just chipping away, Tom, trying to get it to be okay, let's get to more than one for a production application. I believe it will expand from there. But ultimately, yes, the -- we need a handful of them at least.

And then when the ramp hits, what is your lead time to build the tools?

Lead time is quite substantial, but we have ways of managing that down by preordering some critical lead item parts. But we're no different than any other company building a complex machine. It can be somewhere between 9-plus months.

Okay. But have you taken steps just in case there's a few more required for this year to be able to (multiple speakers)?

Yes, we've done that. Yes, you're right, Tom, we've done that. We've preordered some of the critical parts to bring that down.

Okay. And then moving over to the Exensio business, obviously very nice growth over the last year. I'm just curious, do you feel like you're saturating the semiconductor market with the tool, and you need to have more apps -- or kind of a land and expand strategy for the growth there? Or is there still plenty of greenfield growth ahead?

So as we said on other calls, we have something over 130 customers that use our analytics Exensio primarily. And when we look at those customers, almost none of them -- including that one that signed that pilot for the enterprise pilot, none of them really actually have the entire Exensio because Exensio was born from a series of point tools that in 2009, we architected into an integrated -- started architecting into an integrated system.

And we believe actually -- and we've had discussions with customers, that there is a lot more potential for them to use analytics inside their organizations, and we could grow greatly our annual revenue in those accounts before we would need to start looking at things outside of those 130. And of course, we are looking and talking with customers that are outside of those 130. There's a lot of system companies that are becoming chip companies these days, but there's tremendous amount of growth from where we are right now inside that customer base.

Okay. Is that multiples of current revenue levels that substantially...

Yes. Yes.

Okay. And then just finally when you look at the cuts you made to the personnel, is that going to hinder your ability to potentially ramp up with some of the traditional solutions business with some of the new customers in China? Or is that business that you are just trying to push over to more of an Extensio -- Exensio model?

Yes, that's a great question, Tom. So we believe with customers that are using Characterization Vehicles, like we did this in this contract, we signed -- the renewal we signed in the last quarter, that we can provide the -- for someone that's, let's say, running a factory at 65-nanometer, we can provide Exensio and the vehicles and not put consultants in there to help them get to targets and charge them a ratable subscription.

So for those classes of customers, we believe there's a tremendous opportunity for the Characterization Vehicle technology that doesn't require a team to help them get value out of it. When you're ramping up or developing, let's say, 7-nanometer, you're spending billions of dollars, and you've got tremendous risk and there's a lot of new analytics that we are bringing to that problem. So we provided people to help the customer get value out of it fast.

For mature technologies, there really isn't the need for us to provide people to help them get value out of it fast. There isn't a fast really for them, and we can monetize it that way.

For places in China, where people are trying to bring up even things like 28-nanometer and 14-nanometer very rapidly, we will still provide consulting teams as part of that, as part of an overall engagement. But we will make sure we get paid for that on good margin, and we can source that out of our overall locations in Asia. We have a number of engineers in Taiwan and in China that are more -- the salaries are a different level than our U.S. and European salaries.

Your next question comes from the line of Tyler Burmeister with Craig-Hallum.

Most of my questions been answered, but if I can just tie a bow on the guidance here. You said orders should pick up, and revenue should follow this year, Gainshare being down modestly, solutions up. So in total, do we think that we can grow revenue year-over-year in 2019?

We expect modest growth in revenue. That's why I said, revenues will follow. We expect substantial growth in bookings and modest growth in revenue.

Okay. Perfect. Just to clarify that. And then on the DFI side, maybe you could talk a little bit about longer-term the opportunity and what the timeline of that might look like? Are we -- do we expect -- is there any vision to start to see more material orders later this year? Is this a 2020, even beyond that, kind of thing still? Just maybe an update on the more material timeline of DFI.

Yes. We expect -- we have a number of pilots going on, if you kind of go back and listen to our conference calls, Tyler, over the last few quarters, we were very heads-down focused on one specific customer. We were very happy to get that order. We believe there's tremendous more opportunity in that customer itself, and we do remain very focused on delivering value on that customer.

But starting in very late Q3 and into Q4, we started doing demonstrations of the capability for other customers. In my prepared remarks today, I talked about a fab customer who had 3 products from 3 foundries across 2 process nodes, right?

So with DFI, the fabless are our best marketing tool because they put the content on their designs, and then they tell the foundry that it's there, and they'd like to see a wafer come to our lab in Milpitas. So we began that effort, let's say, the second half of 2018, and we anticipate that bearing fruit in 2019.

And as we -- because of the nature of our ratable business model, it will have a more modest impact on revenue in '19 compared to the bookings level impact it would have. And that would then give us a more substantial run rate in 2020 should we be successful with the pilots. So I mean, the short answer is, booking will be up, and it will have a modest impact on revenue and a more meaningful impact on revenue the following year.

Your next question comes from the line of Gus Richard with Northland.

I have some questions. Thinking about the fixed -- the IYR business going forward, is there going to continue to be some legacy revenue in that product line? Or is that just going to dwindle to nothing over the next year or 2?

We have a number of Gainshare contracts, including Gainshare contracts in China that have quite a long tail on them. So we believe there's still a substantial number of years left on Gainshare. There is a big question when those factories go up and how -- what the ASPs are on the wafers, et cetera. So we've -- in our own models, we've been very careful about handicapping that to be small.

We do believe there is need for customers, as I explained on a question that Tom had asked, we do believe there's need for customers when they want to get up quickly in a node where the team helping them get up quickly in the node is of value to them. We are migrating the tail of that from primarily Gainshare, which the customers perceive as a royalty to a license for ongoing use of technology, much like the TBL that we have, time-based license that we -- I discussed on that renewal.

So we are looking for ways because we want to take down our risk on their achieving volume. That's to be [carried], but we still see value in that way of deploying for customers and customers, by the way, still see value in us deploying in that way. It's just been very difficult for us to get a return on some of the more riskier investments in Asia for us.

And we will continue to have IYR revenue above and beyond the Gainshare revenue. We will continue to have IYR revenue during 2019. It's just going to be a smaller fraction of our total solutions revenue.

No, I got it. It makes complete sense to me. Essentially, you're going to be licensing people or Characterization Vehicles, and you'll get paid for that work and helping them ramp a process.

And then in terms of revenue recognition on the DFI system and sort of how that impacts the cost of goods, do you have to turn on a depreciation when it ships to the customer? And then sort of how do you recognize revenue, particularly in the first quarter, when you're shipping but not necessarily -- ship the product to install but not necessarily running product?

Yes, the DFI arrangement, first of all, is revenue recognized out as ratable revenue. And I think you already know, we continue to own the piece of equipment, and we begin the depreciation when it's deployed and the...

May I take the second part of that question? So it's -- the machine is not up and running, but there is a 150 there running. And so they are actually running wafers today on the 150 machine. So the contract is set up where, although it's a little over a year, they only get a partial year use of the 250 and a partial year use of the 150.

I see. And so you'll get a ratable amount of revenue each quarter over the period of the contract?

Exactly. Right now (multiple speakers).

And then the depreciation will turn on on the 250 when that is starting to run production?

That's correct.

When it's put in service, yes.

Yes.

Got it. And then in terms of your number of customers that have DFI structures on their products, has that expanded beyond the customers you've had over the last year?

It -- actually, have a look at the statistics recently, Gus, I know of a number of tape-out activities going on right now that are our new customers. So if it hasn't extended, it will be expanding relatively soon. I think the shipment on the 250 was a pretty big deal for many of us in the industry.

Got it. And then finally, just on the sales pipeline on additional 250s, can you handicap getting a second customer, say, by the end of the year?

That would be our anticipation. We have also ways of deploying 150s in applications where they make sense as well, and we see some opportunities there as well. So it will be a mix [in terms of that] that we generate -- that generates revenue this year.

And your next question comes from the line of Jeff Bernstein with Cowen.

You touched on having Exensio now pretty ubiquitously across the supply chain, and I hope you could just flesh that out a little bit more in terms of what it means for customers. What it means for the potential to gather GLOBAL kinds of data, et cetera.

Sure, Jeff. So Exensio runs at fabs, fabless, OSATs, test houses, system companies and is -- actually, OEM did include it on tools like wire bonders, pick-and-play systems and assembly flow, you can get the ops module from Besi, from Kulicke & Soffa and others, for example.

So there's a couple of things that that's done, when we've deployed at the OSATs, we deploy in a way where they work collaboratively with their end customers. So we generate revenue both from the OSAT and the fabless customer, and they can therefore share data across the Exensio dex node to help improve operational elements, which really is the OSAT's responsibility as well as quality elements, which is really the fabless -- or system company's responsibility.

We also have situations -- for example, we have large fabless companies that are used to certain kinds of Exensio reports. So in the past year, we had a sale at a smaller foundry where they were winning that fabless company, and they wanted to deploy Exensio so they could generate the same reports that the fabless company expected.

So -- and that's all part of the -- and if the fabless company has a license for Exensio, they can load that same web report directly into their Exensio. All Exensio web reports are interactive. So it's not a static report. That means they can go and say, "Well, wait a minute, they should have screened out those couple of chips," and it automatically updates the report.

So all the reports are interactive and also shareable in that way. So when one of your suppliers on Exensio, when you get a report from them, you can make a tweak to that report without having to go back and wait the 24 hours for them to come back and work and fix the report for you when it's something kind of trivial. That's a big deal for an engine that's trying to generate something for a 7 a.m. meeting.

Do you guys have the potential at some point to create a data business that has benchmarking kinds of stuff across different players, et cetera? Are customers interested in that? Is the data security there to be able to do that in a way that people wouldn't mind?

That's a great question, Jeff, yes, and it's true that we do have that capability. We maintain a lot of the systems for our customers on our host of business as well as all the nodes at the OSATs. And we maintain use of that data, what are called secondary data rights, for the purposes of benchmarking, developing and optimizing better algorithms and helping customers be more efficient with their manufacturing.

So over time, that would be a more and more significant business. And if you look at our -- my prepared remarks, when I said that we moved to the public cloud and that we can offer a series of value-added services on top of that, that's exactly in the direction that a lot of our customers want us to go: heightened security, heightened traceability of their material as well as heightened security about who is looking at my data, which we have ways of providing security in that and different 2- and 3-factor identifications.

And ultimately, then benchmarking and analysis and alerts about what's going over -- on in the overall supply chain. But that's something that will come -- will become more available in the future than it is today.

Okay. And then lastly, just the challenges of advanced packaging, what do those sort of mean to all of this?

Yes. So that's -- when we started out with Exensio, and we started building out all of the advance modeling, we focused it on the factory, the fab itself. And customers in 2014 educated us and me, in particular, that we were missing the point and there was a tremendous amount of complexity going into the assembly flow.

That's why we purchased ALPS in 2017 because the traceability across that assembly process is quite important. There are many, many components going into a package these days, 10 or 15 components. I mean, if you were to go look at the construction analysis inside your iPhone, there's 4 packages that have -- at least 4 packages that have many components in them. And many of them need that traceability to know exactly which part of the supply chain it came from.

That responsibility falls on the laps of the fabless company or the system company and hence, manufacturing for them is becoming more complex. When you go back to when Moore's Law was driving the industry, and it was just a new node, all of the complexity of manufacturing and all the risk resided at the foundry. With advanced packaging, that is a shared responsibility between the foundry, the assembly house and the system company itself.

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

John, maybe you could update us on a macro perspective how things are going in the opportunity in China? I know the ecosystem for a domestic capability there has developed at a slower rate than maybe the public believe, you guys were always a little more cautious on it, but what's going on now and how are you looking at that?

Yes, that's a great point. And I've said in a number of meetings with folks that I believe China had its sputnik event this past year. When the U.S. government shutdown ZTE and then subsequently what's gone on between the U.S. government and Huawei, I think it was a wake-up call for the overall semiconductor industry in China.

And I think that creates a tremendous opportunity for PDF because they need to be able to ramp those technologies. I would say, in the early part of 2017 -- '18, my perspective, when I would visit with fabless and system companies, they were not so excited about using the capacity inside China. They were very happy with their worldwide suppliers.

And I'm sure they're still very happy with those worldwide suppliers, but they see tremendous risk that they could be cut off from them. And so I think there's a higher degree of urgency. We felt that in the fourth quarter, in our dialogues with customers we anticipate as we go through this year an increased business activity there as a result of it. And I jokingly -- when I talk with my friends over there, I said, "I think you experienced sputnik." I think it's the same kind of thing.

With respect to the pipeline or development activities related to the eProbe 250, given the change in the business model associated with electrical characterization and inspection we're doing and the nature of it being in production versus in development. How has that changed the type of customer we're talking to, in particular, has it opened the doors at some of the largest logic players where we either historically haven't been able to get in or one where we were kicked out?

Yes. Of course, Andrew, we have to be super careful about talking about any specific company, but customers have always liked our technology up and down the industry, and the business model has -- had been a factor. The fact that DFI provides a way that we can create value in manufacturing means we can charge for continuous usage rather than charging a royalty upfront.

And that is much more amendable to many of our customers, including the largest companies that have not been our customers in the past. And that's increased our dialogue with the entire industry over the last few months.

Okay. Christine, do you have what the bookings growth rate was on the analytics business in 2018 versus 2017?

Bookings is not a number that we disclose. So certainly, bookings rose. I can give you that kind of qualitative direction, but we don't report on bookings.

Do you have metrics around either renewal rates on Exensio, either on a customer basis or on a revenue basis?

It's a great question. Andrew, we're pulling all that together. And as we go out through this year, we'll be disclosing more about what we see our renewal rates being. This example I gave last quarter was kind of the first really major one that we could talk about, and there'll be more that we'll be able to talk about as we get through this year.

Yes. I mean, John, as we've had this conversation, I think it's incredibly important as we transition the business that we provide metrics that will help people model the business and that are sort of more akin to other subscription or software or analytics-based businesses.

Yes, so you're preaching to the choir on that one, Andrew. We...

John and I violently agree with you.

Yes, we'll be doing that as we get through the year.

(Operator Instructions) Your next question comes from the line of Rob Ammann with RB -- I'm sorry, RK Capital.

Yes, my question was also primarily around some sort of bookings metric and the plan going forward, but I would imagine deferred revenue is not a very good proxy for calculating an implied billings sort of number given different contract terms or billing terms. Is that fair?

That's correct.

It's fair.

Okay. And then maybe the second question, but thanks for your help with the model at least baking in one 250. Can you tell us if the model contemplates additional 150 placements over the course of the year?

In this year, very little, just a little bit at the end of the year.

Okay. And how many are out there now? Is it 3?

There's 3 out there now.

(Operator Instructions) And there are no further questions at this time. And now I'd like to turn the call back over to the presenters for closing remarks.

Thank you, everyone, for attending. We look forward to talking with you again soon. Good day.

This concludes today's conference call, and you may now disconnect.