PDF Solutions, Inc. (PDFS) Q1 2019 Earnings Report, Transcript and Summary

Good day, ladies and gentlemen, and welcome to the PDF Solutions, Incorporated conference call to discuss its financial results for the first quarter ended Sunday, March 31, 2019. [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 PDF's 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 demands for its solutions. PDF's actual results could differ materially. You should refer to the section entitled Risk Factors on Pages 12 through 21 of PDF's annual report on Form 10-K for the fiscal year ended December 31, 2018, 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 obligation to update them. Now I'd like to introduce John Kibarian, PDF's President and Chief Executive Officer and Christine Russell, 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 management report presentation for the first fiscal quarter, please go to the Investors section of our website where each has been posted. Our objective is to be the end-to-end analytics and manufacturing control provider of choice to the semiconductor industry. PDF is evolving from being important to our customers for ramping new nodes, to being even more important for controlling quality, reliability, yield, and operational effectiveness across the lifetime of semiconductor products. This evolution is facilitated by PDF's unique silicon foresight, our ability to see what will happen in production. Foresight starts from our position across the entire supply chain. Wafer manufacturers, foundries, OSATs, IDMs, and fabless and system companies all use Exensio. Foresight is possible, even for new materials and processes, because of the unique measurement capability enabled by DFI and PDF Characterization Vehicle technology. And foresight is automated because of the large datasets the industry has, but often does not leverage. By applying machine learning and AI to the petabytes that PDF's Exensio's system processes for the industry, new levels of productivity can be achieved. During the quarter, we made progress in four areas. The first was deploying state-of-the-art measurement systems to aid our customers in generating insights into their new materials, processes, and products. The second was the signing of a number of new contracts for Exensio, our end-to-end manufacturing 4.0 software platform. The third was accelerating our customers' move to Exensio's cloud offering, which is designed to provide them with unprecedented scalability and performance, while enabling us to deliver more software as a service. And the fourth was through acquiring additional AI capabilities proven in the semiconductor market with our recent acquisition of Stream Mosaic, announced earlier today. For my prepared remarks, I will talk about each of these four areas, as well as discuss the implications for our business model and expectations for the remainder of 2019 and beyond. Along the way, I will touch on the general environment and the resulting opportunities for PDF. In January, we announced the first shipment of the eProbe 250 to a leading high-volume production foundry. The eProbe 250 was designed to be a production control system for monitoring electrical characteristics in line. We believe this will enable our customers to see factors which limit product quality and yield in line that were previously only seen at the end of line. Discovery of performance and quality defects at the end of production process is expensive. Moreover, the most expensive defects discovery is once the chip is sold and in the market. DFI is designed to help our customers avoid these problems. In the first quarter, we installed the machine in the factory. It is now being qualified for production. Testing is mainly focused on reproducing our lab measurements in their production facility so that the customer can rely on them to make control decisions. Beyond this first installation, demonstration activity for DFI is accelerating. Interest is high at multiple leading-edge logic providers, as well as [indiscernible]. We're getting positive feedback from the early technical pilots that we believe bodes well for the future. We are also getting our first customer feedback on potential applications outside of leading-edge logic, and those are also promising. Our goal for DFI for the remainder of 2019 include achieving a qualification at our first production installation, as well as converting pilots to additional business as we get into the second half of 2019. In Q1, we also shipped our newest electrical test system, the pdFasTest 380+. This system, while very valuable for leading-edge logic, also has capabilities critical for characterizing next-generation non-volatile memories, such as PCM, MRAM, reRAM, and vertical man. As I've mentioned before at IEDM, STMicroelectronics, who is a leader in embedded PCM for automotive, acknowledged the value of PDF's characterization capability in their IEDM paper. With 256 channels, the 380+ was able to demonstrate the ability to do in parallel what customers often can only achieve on engineering bench tests. This is important for customers optimizing technologies where they need to achieve fast cycles of learning. Moving on to business activity in the quarter, we had a strong start to the year in our analytics business, which includes Exensio, DFI, and Characterization Vehicle technology. Analytics revenue is recognized on a ratable or subscription basis and was again over 60% of our solutions revenue. As encouraging, it was the vast majority of our bookings in the quarter as well. Contracts signed in the quarter demonstrate the growing adoption of Exensio. Here are several examples of significant contracts. An enterprise deployment of Exensio-Yield on the cloud for an [indiscernible] customer; additional licenses and other services for an IDM deploying Exensio-Control to additional process tools; additional Characterization Vehicle capability, and additional big data analytics pilot deployments; renewals for a large analog company deploying Exensio-Yield and other Exensio components; advanced analytic pilots at both U.S. and Japanese IDMs, in one case including pilot cloud deployment; and finally, an extension for 7-nanometer engagement for a system company characterizing a new foundry. These bookings demonstrate that customers value the integration of our Characterization Vehicles and end-to-end analytics that are possible with Exensio. This leads to my next topic, Exensio in the cloud. From an historic perspective, recall that we have had a hosted offering for some time through the acquisition of Syntricity. This ran primarily through our servers at a colo facility. In Q4, we started signing customer engagements for a more complete version of Exensio delivered through conventional cloud providers. This provides us the opportunity to deliver a better, more integrated SaaS capability for our customers. Interest grew significantly in Q1. The interest in cloud is driven by the ability to leverage our big data capability. As customers with on-premise deployments of Exensio look to take advantage of our big data version of Exensio, they realize it may be time to leverage the cloud for scalability, and eventually, AI and data interoperability. This gives us the opportunity to show them the advantage of our cloud offerings, which include incremental services for added value and the ability to reduce the customers' overall cost of deployment. Cloud also enables us to accelerate our development of new machine learning and AI capabilities in Exensio. AI techniques are well known now, and the issue is data. PDF has large amounts of characterization data. As we move to the cloud, we can make the benefits of algorithm improvement available to our customers, without the customers necessarily needing to waste effort in data wrangling. This brings us to our acquisition of Stream Mosaic. Stream was an AI company focused on the semiconductor industry. Nvidia was featured as a customer on their website. Semiconductor engineers are quoted describing how Stream's capability helped bring yield prediction to operations, enabling customers to improve costs and product quality. Through our review, we found Stream shared the same foresight vision that we have. Early data in manufacturing can be used so customers can achieve higher quality and reliability at a lower cost. We are thrilled to have the Stream team as part of PDF. The Exensio platform has been architected to take advantage of new, innovative AI algorithms, such as those from Stream. Our evolution to monetizing our technology in ratable subscription model enables us to more efficiently capitalize on new software, such as that from Stream. As we look to the remainder of 2019, our original thesis for the year remains intact. We expect continued customer adoption of our Exensio analytic software, Characterization Vehicles, and DFI. These will all be provided to our customers on a ratable basis. Customer feedback indicates a high level of interest in utilizing our capabilities through a SaaS model, which will further enhance our opportunities. We are excited about the progress that we've made in Q1. Ratable business models change over years, not quarters, and we will remain focused on building on this foundation. With that, I will turn the call over to Christine, who will review our financial performance. Christine?

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 comp, amortization of intangibles, and one-time and restructuring charges. Please refer to our press release for our GAAP results and GAAP to non-GAAP reconciliation. Starting with revenue, first quarter revenue was $20.5 million, a 4% increase from the prior quarter. Solutions revenue was $16.7 million, a sequential increase of $4 million. Gainshare revenue was $3.9 million, lower by $3.2 million from the prior quarter. As John mentioned, analytics was again over 60% of solutions revenue. This demonstrates the traction we're getting in shifting our revenue base to higher-margin software. Notable deployments include revenue from the DFI eProbe 250 installed for production at a leading foundry. We also booked significant revenue from an Exensio deployment at a major European semiconductor company. The lower gainshare revenue was due to lower shipment volumes at one foundry. While our customers balance their load relatively uniformly, they time shipments to when their customers request ships. This can make gainshare lumpy. Now, turning to cost of sales and gross margin, non-GAAP cost of sales was $6.9 million, which was down $1.9 million from the prior quarter. The lowered cost and resulting gross margin improvement were attributable to several factors. First, expanding and growing our software business yields a higher margin compared to our former business model, which was consulting with a gainshare royalty expected at a future time to offset the initial investment. We also reduced our headcount during the first quarter, and that reduction came primarily from the IYR consulting business, which is cost of sales. We invested in sales and marketing and R&D for the software business and made headcount increases there. DFI gross margins improved as well, as we received revenue from the major foundry that we announced at the beginning of this year. The lower revenue in gainshare for the quarter, which is 100% margin and thus reduces gross profit 1 for 1, was offset by revenue in the quarter in conjunction with an amendment to resolve the early cessation of services under a customer contract. This amendment revenue is 100% margin, as there are no associated costs. We'll continue to receive additional revenue in 2019 in connection with this amendment. As such, the net improvement in gross margin was really driven by software content and resource realignment. Those accounted for the difference and will continue to do so going forward. This is the outcome we expected from our strategic evolution to focus on time-based software licensing. Over time, our gross margin should drift up to levels associated with other software companies, in the 70% to 80% range. Now, operating expenses. Non-GAAP operating expenses were $12.6 million, which is $1.6 million higher than the fourth quarter. As I mentioned, we added resources to R&D and sales and marketing, but our net total spending still declined. Combining cost of sales and operating expenses, the total spending is lower by $300,000 sequentially. In the first quarter, we focused on resourcing R&D and sales and marketing to support the software business. In particular, we have invested in bringing in business development and marketing skills. In addition, to put our OpEx in perspective, during 2018 we had a very small number of sales representatives who were paid on traditional base and commissions linked to bookings. In 2019, we have substantially increased the number of individuals compensated with a combination of base and commission tied to bookings. This aligns better with a traditional software staffing and comp model. The total net spending decline was due to lower headcount. We're finishing up the reduction in force initiative that we started in the fourth quarter of 2018. This is because locations outside the U.S. are typically more time-consuming and complex to enact reductions. As John mentioned and we announced earlier today, we acquired Stream Mosaic. Although we're not disclosing the purchase price, we brought on the small team from Stream and plan to add headcount during 2019 to this group, and to support our cloud initiative. Now, turning to the bottom line, we posted a non-GAAP profit of $842,000 in Q1 compared to a loss of $366,000 in Q4. Non-GAAP earnings per share was $0.03. Shares outstanding for Q1 were 32.5 million. And now we'll turn our attention to our balance sheet. Cash at the end of Q1 was $90.4 million, compared to $96.1 million in the prior quarter. We used $5.7 million of cash, of which $3.9 million was used for stock buyback. We used $2.4 million in cash for material and labor to continue to build the new eProbe 250 EB machines. The cash use was partially offset by proceeds from our ESPP. As I noted, we recently deployed one of the new EB models to a major foundry. That deployment resulted in an important booking, which converts to revenue ratably during this year, showing that our investment in new eProbes is starting to deliver a return on investment. Accounts receivable of $53.2 million represented an increase of $1.6 million from the prior quarter. DSOs were 150 days in Q1, compared to 135 days in the prior quarter. Two customer accounts are the primary reason for extensive DSO. We hope to have both those accounts caught up by the time we report second quarter results. That concludes my prepared remarks. Before we go to Q&A, I want to mention some investor relations activities this month. We'll be presenting and holding one-on-one meetings at three investor conferences in May. Those are the B. Riley Conference in Beverly Hills on May 22nd, the Callan Technology Conference in New York on May 29th, and the Craig-Hallum conference in Minneapolis on May 29th. I hope to see some of you there. We'll now turn the call over to the operator for any questions. Operator?

[Operator Instructions]. Our first question comes from John Tanwanteng from CJS Securities.

Christine, can you give us a little bit more detail on the resolution and settlement with the foundry that, you know, ceased the 7-nanometer activity? How much was that of a benefit in the quarter, and how much do you expect the rest of the year?

Well, we don't disclose the terms of our amendment with the customer; $3.3 million was related to that contract, that amendment to the contract, and we will be receiving additional revenues the rest of the year.

Okay, does it make sense to annualize that, or does it tail off?

It's all going to be in 2019, and it will not necessarily be ratable. It will be a little bit lumpy, but it will all be concluded by the end of 2019.

Okay, thank you. And then, just a little bit more color on the gainshare in the quarter. You mentioned that it was lumpy based on delivery schedules to customers. Do you expect that to normalize going forward and the match up with semiconductor demand from that client?

We don't necessarily count on our customers who pay gainshare to deliver the gainshare royalties ratably, and they're very seldom even. I think you can look back on our history and see that gainshare has always been very variable. So we would like to think that our customers would have linear production, but as I noted in my remarks, sometimes they are subject to the demands of their customers who don't want to take the shipment at a certain time, so we're just going to have to take our gainshare as it's reported to us.

Okay, fair enough. Any more color on the ramp-up of the series 250 at the Asian foundry? Is it measured performance standard so far? How soon are you going to be able to reach the lab performance?

Yes, I'll take that, John. This is John. Yes, so we're on schedule to basically meet our targets that we have set for ourselves for Q2, meaning that it will be scanning the same types of wafers that it scanned in our lab in Q4 of last year in their manufacturing facility and generating the same reports. And then, we expect as we get through the quarter, it will be released. They will then do a series of additional things, not from a qualification standpoint, but more to understand usage models and how to best use the information. So we expect it to be this work to be done throughout Q2, which was our original target.

Our next question comes from Gus Richard from Northern Capital Markets.

Just a point of clarification. That amendment to the contract, you said it was $3.3 million that was an impact in the current quarter?

Yes.

Got it. And then, John, I think you mentioned you were seeing some interest in DFI outside of leading-edge logic. Would that be the memory market, or is there another market that is showing interest?

There are a couple of markets. We're further along in the memory market than we are in some of the sensing technologies. What DFI has demonstrated to be really good at is measuring very small leakage currents and being able -- you know, what the foundry in Asia really pushed us on was being able to see leakage currents down from a picoamp up into the microamps with a precision that's never been seen before with an EB machine, and then do that at huge scale. Right? You know, hundreds of millions to billions of those in a reasonable time. And we're finding that, you know, you can take leakage and then understand lots about structures for other markets besides leading-edge logic, memory being one of them where that tends to be very important. You can use it as a model for structure shape and structure [indiscernible] and also sensing technologies, where leakage currents are important. And so we're finding -- you know, it's like any time we've built one of these things, like the original pdFasTest 180, once you start showing it to customers, they start telling you ways it could be used that you hadn't really imagined. And we're experiencing some really positive feedback right now in the memory market.

Our next question comes from Christian Schwab from Craig-Hallum.

This is Tyler on for Christian. Thanks for letting me ask a couple questions. So, first, I was wondering if you could give any more color on this acquisition, the scale of it, if there's any revenue or expenses associated with it that you could provide us.

Sure. You know, they're on the order of four folks associated with the acquisition, you know, so you can kind of do the math on U.S. headcount and what costs would be for that. From a revenue standpoint, you know, they really had some pilot deployments and a real production deployment with one leading [indiscernible] customer, but revenues were small, you know, really from our standpoint, not consequential. We really bought it -- you know, if you go back and look, before 2015, right, the Exensio business was a very small part of total PDF. We made a series of acquisitions, the first one being assets from [indiscernible] subsequently Syntricity, and then subsequently, KINESYS. Each one of those companies, while all of them were larger than Stream Mosaic, actually, from a revenue standpoint, they were all, you know, single digit, $1 million to $2 million, $3 million a year revenue when we acquired them. Yet, by bolting them onto the Exensio platform, we were able to generate a lot more opportunity than they represented on their own. Stream, we feel, that way will be, and even multiplier-wise, a more significant opportunity. Because all these AI systems, it comes down to being able to have the data wrangled in a way that you can make use of it, right? I think Forbes quoted something like 80% of the cost and effort is getting to data loaded and aligned. That's what Exensio does for a lot of our customers. You know, at the 2016 or '17 User Conference, the Qualcomm keynote speaker pointed out that Exensio was the single biggest database at Qualcomm, right, to our users. And we have a number of customers that tell us that. So our customers have huge datasets that are aligned, and therefore analyzable. And something like Stream really has a hard time getting to scale without being able to sit on top of a perch like that. So by putting Stream on top of that perch, we believe we can actually grow it quite substantially, even though today it's a relatively insignificant amount of revenue.

That's great, thanks. And then, I understand the gainshare revenue piece is going to be lumpy and it was down this quarter. But assuming the rest of the year it's kind of somewhere between this quarter and Q4, the segment for the year is still going to be down pretty meaningfully. Last quarter, you said solutions would more than offset gainshare declines this year. Do you still believe that's going to be true?

Yes, in my prepared remarks, I said that we -- our thesis for the year remains intact. We still believe that, more so after the first quarter than we did after Q4 last year.

Our next question comes from Tom Diffely from D.A. Davidson.

Maybe first, a question on the transformation that you're doing with your employee base. Are you fairly well through the transaction from IMIR to the sales service and support, or is there more to do on that front?

Yes, Tom, this is John. Greatly, we are through that. As Christine said in her prepared remarks that we did let some folks go, we did hire some new folks that had skills that we needed, and we did re-target folks that were working on one thing that we felt had the skills to transition and work on Exensio and other things. By and large, we're done with that. We do have -- we are seeing a lot of activity in certain areas, like cloud for example, where we will make some incremental hires. And we do see some geographies where we will work on what we do in the field and improve our field operations. But, you know, that transition of getting people targeted to the new areas is by and large done.

Okay, and I guess in the same vein, what kind of capacity do you have right now to entertain new clients for the EB 250 tool? You know, is it limited by the tools you have to provide clients with, or with the people to support them?

Yes, it's a great question, Tom. You know, engineering at this stage is always super important because, you know, we're trying to prove things out, so we're trying to make sure we don't spread our engineering resources too thinly. You know, if you look at this journey we've been on, PDF has invested very significantly. As well, our lead customer has invested very significantly. And now, I think the new companies, we can help them use DFI quite quickly. So there is value in the methodology outsized of the initial tool. So as we look at business models for -- as we acquire new customers, there is value in licensing the IP on top of the machine. So, you know, we are working to make it -- I can't say that we've done this yet -- so that our business is not limited by our rate at which we build machines.

We have a follow-up question from John Tanwanteng from CJS Securities.

Given the activity on DFI you mentioned in your prepared remarks, do you feel better or worse on the expectation of shipping more machines in 2019 and into 2020?

We feel better about our ability to ship this year.

And also, you mentioned an extension at a foundry for 7-nanometer characterization.

It's for a system company.

Excuse me?

It was for a system company to characterize the new foundry for 7-nanometer.

Got it, okay. I'm just trying to figure out where that new 7-nanometer deployment would be. Is that an American company, an Asian one?

Yes, well, you know, unfortunately -- fortunately, unfortunately -- there's not much activity going on in the U.S. on leading-edge logic these days, so if I were a betting man, I wouldn't be betting on this continent, though we don't disclose these things, as you know, John.

Okay, got it. And do you have a ballpark for Exensio growth, either year-over-year or quarter-over-quarter, or how much absolute dollars it did in the quarter?

Yes, we have not broken that out yet. As we get through the year, as we collect a lot more data, we're going to be communicating more and more information about how the overall analytics business is going. You know, at this stage, we have our own -- you know, we're starting to build a track record of this, to be candid with you, John. You know, two years ago, we didn't really track it very closely, so we didn't have a lot of historical data that we could really work on. But we will be more communicative as we get out of the next few quarters.

Okay, fair enough. And then, Christine, just on the operating expenses that you did in the quarter, should we expect that level to hold steady or just increase a bit with Stream Mosaic under the hood, and maybe some additional salesforce, or are there other factors to think about?

I think you should expect those to increase a small amount. As I mentioned, we are investing in some business development skills, in some marketing skills. And we have also put in place a more robust sales commission to incent our salespeople to bring in bookings. You can also expect to see some additional expenses around the cloud deployment, and a few more additional expenses around kind of adding to the Stream Mosaic team.

Okay, got it. Is that outlook for OpEx any different from what you expected at the end of Q4, ex Stream Mosaic?

No, it isn't.

We have a follow-up question from Gus Richard, Northern Capital Markets.

Yes, just a two more follow-ups on DFI. First of all, how many established customers now have your structures in product [indiscernible] or product?

You know, Gus, I don't have the numbers off the top of my head. I know well over 100 tape-outs have happened, and I believe there are over 5 or so [indiscernible] maybe 6 or 7 that have DFI structures in them on one node or another.

Okay, so 100 tape-outs and 5 or 6 customers.

Yes.

And then, in terms of the memory market and sensor market, could they be serviced with a 150 in order to get them up and running? Is there a potential for you shipping some of those this year?

Yes, there is, both for some logic applications as well as those. Yes, it's something we're exploring now, Gus. I don't think we have a definitive answer yet. So far, the only work we've done in the memory market has been actually on the 150.

Okay, so there is customer interest in the 150?

Oh, yes.

[Operator Instructions]. At this time, there are no more questions.

Thank you everyone. We look forward to talking you again next quarter. Goodbye.

Ladies and gentlemen, this concludes the program. Thank you for joining us today.