PDF Solutions, Inc.

be able to give us a forecast of more substantial growth okay great and then finally moving over to the model I was a little surprised to see some of the optics come down in the quarter as I thought you know you've been ramping on several programs lately curious it may be a little more color behind the sequential drop-off and then what your expectations are for the next couple quarters from an expense point of view yeah so R&D as you mentioned in the commentary during their prepared remarks

was primarily due to the merit increases. We do expect that to grow a little bit in line with what we're estimating for revenue or a little bit lower. In terms of SG&A, we also mentioned on the call, there were some savings from one-time items and also the timing of expenses during the year, so we expect that to grow. Overall, our goal is going to stay the same, that revenue growth should exceed the growth that we see in our total spend, both cost of sales and OPEX combined.

Okay, helpful. Well, thank you both for the questions today. Thank you, Tom.

And moving on to our next question.

Yes, good afternoon. This is Gus Richard with Northland.

A couple of questions on these partnerships with IBM and SAP. How is that impacting, you know, the work you guys need to do for, you know, integration to a customer system? And sort of how does it affect your you know, your cost and your ability to build standard products?

Yeah, that's a great question, Gus. You know, we have integrated Accenture with MES systems for forever, as long as, you know, I can remember. And we have something like, you know, 10 different scripts and we supported 10 different MES systems with different scripts and whatever. What we noticed once we started moving Accenture to the cloud was, wow, we could standardize this. So you don't have to go through and customers Yeah, they're going to always want to have a special report or a different way of cutting the data, but the piping, the connections between the systems, that should be, you know, established. So now with SAP's S4 HANA, which is a cloud solution, and Accentio also now, you know, greatly a cloud solution for most of our customers, we can more formally standardize, you know, do a one-time integration that works. And yes, with each customer, there'll be another, you know, customization and tuning, But that base pipe is already there, right, which greatly reduces the effort the customer needs to undertake in order to be able to connect, let's say, you know, test flow operational and yield data with, you know, shipment scheduling and planning information. So, you know, we think this is really very valuable to decrease the burden the customers needed to undertake in order to be able to get kind of more real-time integrated, information flows between the different silos within their organization. And it's enabled by the fact that, you know, more and more of these systems are on the cloud.

Okay. And Ian, any sense on, you know, sort of how much goodness this is for gross margin?

Yeah, not yet. We, yeah, ideally we could just make it where what the customer only buys from us is you know, cloud and time-based licenses and we don't do any deployments, that would be wonderful because that would, for sure you're right, the deployments are not as good margin as, you know, the ongoing recurring revenue from the system once it's up and going. So I don't know that we've quantified it yet, but that is, you know, a secondary benefit of why we're doing it. The primary benefit is to drive up total usage and hence the, you know, business impact we can have for the customer and the revenue they spend with us. The secondary impact was it will come at a benefit of gross margin.

Got it. And then just, you know, the CHIP Act is now passed. Clearly a boatload of that money is going to go for capacity. But even more important to realize what the government is trying to push for is advanced process technology available. I'm sort of in a US based boundary and I'm sort of wondering, you know, to what extent you can help your customers, you know, either create, you know, you know, EDA and design enablement, you know, internal IP, you know, is there a role for you to play with your characterization vehicles, you know, as, as The push for more advanced semiconductor manufacturing in the U.S. now comes to fruition.

Yeah, that's a great question. We were very excited about about the chipset passing, and we think that is a significant event for the industry overall, for PDA specifically. And there's a couple of things here. I think the impact is quite broad, and I'll touch on the one that you bring up, Gus. But I also want to point out when you think about moving manufacturing to high labor cost areas like the United States and Europe, What you first need to think about is how can we use software and analytics and machine learning to drive down what has been a very, very labor-intensive part of the industry, whether even the front-end fabs have an awful lot more of engineering and grunt work that goes on. We've been inside fabs all over the world for decades. If you look at how they're run, In East Asia, there's a lot of engineering work that goes in that we in the West had tended to not invest in. Now, with the advances of AI and analytics, you can replace a lot of what's wrote and detailed engineering analysis of equipment data, of operational information in the factory with machine learning. We think that's a very important opportunity for our systems, like Accentio, as we bring it out to the industry, because these fabs here are going to need a lot more automation. Secondly, the point you bring up is a really good one. What happened in Asia over the last two decades was factories served many, many product groups. That is the rise of the foundry model. And the foundry model says it's really hard to get a great return on investment as an integrated device manufacturer. So now as you see manufacturing coming back to the U.S., you see a number of U.S. companies and entities that are putting factories in the U.S. What's being built in the U.S., whether it's from Intel or TSMC or Samsung, are foundry capacity, which is different than what's gone on in the U.S. in the past, which is primarily IDM capacity. If you look at our history, as people try to open up to foundries, They need exactly the point you bring up. You need to have a way of characterizing the variability due to layout, due to design factors, and how that affects the transistor performance, and hence the PDK. In the first quarter of this year, we announced that follow-on contract for using PDF's capability, characterization, and systems for design for manufacturability. It really was the first step with people anticipating what the CHIPS Act is going to do. If you look at how our customers are using our design for inspection, it's going to greatly be used for helping bring up new designs and new varieties of IP because it's design aware as it does the inspection. And again, if you want to do that in a labor and cost sensitive market, you want to use a lot more machine learning and analytics, which is what's embedded in the design for inspection capability. So when we look at our characterization design for inspection, as you bring foundry to the rest of the world, where I think you're going to have to employ a lot more software and a lot less human capital, we think it's a phenomenal opportunity for PDF overall.

And that brings me to my last question. You know, DFI, you know, clearly a great lab tool and helps people get products into production. And I'm just wondering where you are in terms of increasing the throughput and capabilities so that it might leap from lab to SAB.

We've been monitoring usage and the applications. In the second quarter, there was just an explosion of applications for the system as we were able to demonstrate improved capability on the machine and were able to demonstrate with the software some very sophisticated automated analysis of design sensitivities. It went from running individual wafers a day to five to 10 wafers a day, and its utilization is really shot through the roof. We feel really good about where we are for this becoming a much more standard way people bring up new designs and control them in these advanced nodes. And as you look at the roadmaps with more 3D structures, more novel ways of, whether it's gate all around, power via, way for bonding, we feel the future yield issues are really well lined up for what a design for inspection can do.

okay in is that leading in the near term this is my last question i promise um any additional um sales of dfi tools we do expect it to impact our business second half of this year okay great great thank you thank you for your patience sure no problem

And once again, as a reminder, if you would like to ask a question, you may do so by pressing star 1 on your telephone keypad. Also, a voice prompt on your phone line will indicate when your line is open. Please state your name before posing your question. Moving on to our next caller.

Hi, guys. It's Blair Abernethy with Rosenblatt.

Hi, Blair. Hello.

Hi, guys. Adnan, one quick one for you. I wonder if you could talk about your backlog a little bit. 184 million this quarter. Down a little bit it looks like. I think Q1 was around 197. Was there any FX in that backlog? Any FX impact? And also, what's the duration of that backlog? What's your visibility into 2023 at this stage?

Yeah, sure. So no, not from an FX perspective. But look, I mean, reality is our business is starting to get engagements and customer attention for larger and larger deals. As these deals are getting larger, we're engaging not just with the engineering level, but also with the executive level at these organizations, and also broadening our business into the other partnerships like John mentioned. Net-net result of these is some of these larger deals take a little bit longer, and frankly, that's the only reason. It's a little bit of a function of timing where we are in the current quarter with respect to some of the deals that we're closing, which is why you saw that dip in the backlog number. Overall, like John said, we remain very confident and expect the second half from a booking perspective to be stronger than the first half of the year. And, you know, as far as looking forward to 2023, we continue to build on our momentum, and that's going to be the plan to end the year on a solid note that positions us well, even better from a recurring standpoint where we can have better predictability to our growth numbers as we go into next year than we did coming into the current year.

Okay, great. Just on the product side, John, the AdvanTest obviously is starting to do well with the first dynamic parametric testing, but the new applications, I wonder if you can just give some color around some of the new applications, and are they going to take a year or so to ramp, or do you think because you're already out there with the one product, it might go a little faster?

Yeah. What we announced at Advent Test Voice was applications that take advantage of their edge high performance computing box. So Advent Test innovated the idea of putting an AI computer right next to the tester, wired deeply into the way the tester collects information in real time. So now you have a lot more computing power. Historically, When you do screening at test, you establish some very simple rules. If, you know, the test, if this chip is tested well, but all the chips around it were bad, that's called good-die-bad neighborhood, you reject that anyway. These are very simple rules. And you do that in part because the amount of compute that is available to you on the tester is quite small, so you have to be very efficient. Now what Adventest is opening up is the ability to do much more deep machine learning. So these first three apps take advantage of PDF's more computationally intensive algorithms and bring them close to the tester so you can do more real-time capability. And we'll announce later on in the second half of this year additional applications that take advantage of PDF's ML pipeline and even more intensive computationally intensive algorithms. This will allow for a couple of things. smarter and more intelligent screening and binning, which is very important for customers in the system and package. We want to marry the right chips together that are similar in performance characteristics and also allow for more security because the entire approach is containerized. So, you know, a lot of our phallus customers that test at the OSATs, they want to make sure the information all the way from the cloud all the way down to the machine is encrypted and And that work goes on in a very encrypted way. So even within the OSAT, the OSAT has very little visibility in how things are screened. That's important, as you may know, in this world of chondrofit chips when you have a tight supply. So, you know, it's really kind of taking advantage of what the partnership intended, right? PDFs, more and more sophisticated algorithms. And Advent has the ability to build unique systems that take advantage of their detailed and deep knowledge of tests. How long it takes for that to drive revenue, I think we don't know, Blair. We'll find out. I hope we see an acceleration. We hope in part by announcing many of them at once, we're able to stimulate a lot more demand and do in parallel what we did serially with dynamic parametric tests.

Great, thank you for that. Just another question I made on the partner side. IBM was mentioned earlier in the call. Maybe just give us a bit of an update on how things have progressed there. I think it's been almost a year now. What sort of stage are you at with IBM?

Yes, that's great. So IBM, we have integrated Accenture with SciView. It drove business for us and continues to drive business for us where they sell. When they sell SciView, they also introduce and sometimes sell Accentio even as part of an IBM contract. So then the contract between the customer is with the customer and IBM, and then IBM turns back around and licenses Accentio from us. We have a number of ongoing selling opportunities. As you know, MES systems like SideView tend to sell when a new factory is built. So they're less frequent than tester installations or Accentio deployments. But we do have a number of them ongoing with them. And it also is now affecting how we work with SAP, too, because all three of us work together in some factories. And so we've now, in some companies, it's now started talking the three of us together with customers around MES analytics data, NERP data available to customers in real time. And so many of these partnerships actually interact with each other, right? So what we're doing with Siemens effects, what you do with tests, you know, K&S on the wirebonder with final tests. There's ways that we're bringing the partners together more and more tightly, not just one-on-one between us and each partner.

That's great. Thanks for the call, John. Thank you, Blair.

And once again, as a final reminder, if you would like to ask a question or pose a comment, you may do so at this time by pressing star 1. We'll pause for just one moment. And there are no further questions at this time. That will conclude today's conference. We thank you for your participation and you may now disconnect.