Good day, everyone, and welcome to the PDF Solutions, Inc conference call to discuss its financial results for the fourth quarter and year-end 2022 conference call ending Saturday, December 31, 2022. At this time, all participants are in listen only. After the speaker's presentation, there will be a question and answer session. To ask a question during the session, you will need to press star one one on your telephone. 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 demand for its solutions. PDF's actual results could differ materially.
You should refer to the section entitled Risk Factors on pages 17 through 30 of PDF's annual report on Form 10-K for the fiscal year ended December 31st, 2021, and similar disclosures in subsequent SEC filings. The forward-looking statements and risks stated in this conference call are based upon 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 Adnan Raza, PDF's Chief Financial Officer. Mr. Kibarian, please go ahead.
Thank you for joining us on today's call. If you've not already seen our earnings press release and management report for the fourth quarter and the full year, please go to the investor section of our website where each has been posted. As Adnan will discuss in more detail, 2022 was another record year for total revenue in a year where our analytics business achieved tremendous growth, growing 40% year-over-year. That on top of 63% growth in 2021, which benefited from the inorganic growth associated with our Cimetrix acquisition. Beyond the strong growth, we laid the foundation for continued success in 2023 and beyond. Today, we'll provide a summary for the fourth quarter in 2022, our perspective on the market in 2023, and then conclude with remarks about our growth for 2023.
During 2022 overall, in Q4 specifically, PDF was increasingly recognized by our customers as critical to their success in developing, ramping, and controlling new products and processes. During the year, more than 20 companies booked over $1 million in business with PDF, representing just under 90% of total bookings. These companies span the IC supply chain, from equipment makers to foundries to IDMs to fabless companies. They span most end markets also, from advanced computing to RF, to microcontrollers, to memory, and to analog chips. Our customers serve virtually all major markets that consume integrated circuits such as automotive, industrial, consumer, and computing. Beyond customers committing to PDF, partner activity in 2022 was at an all-time high as leading equipment, ERP, analytics, and cloud partners appreciate the strategic relevance of PDF and the opportunities that we can jointly provide our mutual customers.
Building on our success with Dynamic Parametric Test, in 2022, we announced numerous applications available on Advantest ACS Edge that leverage Exensio's analytics, AI ML, and real-time control. In the summer, we announced our first products available on the SAP application store. Our collaboration has been well-received, and evaluations picked up pace throughout 2022. In each case, whether it's Advantest, SAP, Siemens, K&S, IBM or AWS, we commit not just marketing effort, but R&D to fully integrate the partner's data and capabilities so it is aligned with the Exensio semantic model. This allows for our joint customers to have benefits in their manufacturing that are not possible before. We anticipate in 2023 and beyond, we will see increased contributions of these collaborations to our revenues. 2022 was a strong year in our product development, the result of continued and persistent R&D investments.
For our eProbe DFI system. This enables our customers to use design information for direct electrical inspection on their most critical parts of a chip to find defects and as a result, ramp faster. We also released our Guide Analytics module for Exensio, which leverages AI ML to direct engineers to focus on the part of their manufacturing data that is important to their success. With our customers storing hundreds of terabytes of data and moving to petabytes of data in Exensio, they need to leverage ML to make sure they find the needles in the haystack and control production to achieve the highest yield and reliability with the highest throughput. Speed to detect and control for our customers will increasingly benefit from our ML solution offerings such as Guide Analytics. There'll be more to come.
We made big advances in tool communication and data collection with our Cimetrix offering, supporting and leading new standards for single device traceability, tool data collection, and efficient and secure transfer of the data to the cloud. As a result of customer partner commitments to PDF and our strong product development, analytics revenue grew 40% year-over-year in 2022 to over $130 million. This is over double the $57 million we reported in 2020. With that rapid growth in analytics, which is obviously more than the 20% per year that we committed, came improvements in gross margin and operating margins. We exited the year with over $270 million in backlog.
I would add that we also track shadow backlog, which represents our expectations of future gain share, shipments of Cimetrix runtime licenses, and overage charges for Exensio Cloud usage beyond committed levels. When factoring reasonable expectations for these, we believe that the company is in a very strong financial position, further bolstered by the positive impact our products and services are having on our customers' businesses. Q4 in 2022 demonstrate the success of PDF's products and solutions in the market. I wanna thank our customers, partners, employees, and contractors for their commitment to the company. We begin 2023, the semiconductor environment is very different than it was when we started 2022. Today, our customers are slowing capacity expansions, and in some cases, suffering from oversupply. Offsetting those headwinds, we find that customers are continuing to invest in process and product development.
The significant shift in supply chain is increasing investments in new capabilities around the world. Geopolitical factors are increasing government support for reshoring capacity. The Economist magazine article pointed out recently, over $380 billion in government support, including the US CHIPS Act, has been committed around the world. Underpinning this increased activity in development and government support for R&D and capacity, we believe are two growth factors that are beyond the usual increased use of silicon in conventional markets such as phones, computing, and Internet of Things. We believe that technologies such as Silicon Carbide, high-voltage transistors, and semiconductor-like processing for EV battery manufacturing will benefit from the shift of the energy economy from carbon-based sources to electrification over the next 10 years. Production of these products will increasingly need advanced analytics and process control, such as what we provide.
Second, the increased use of AI ML comes at the cost of exponentially increasing amounts of computing. Again, PDF's capabilities to enable our customers to build these ever-increasingly complex systems in high volumes will be critical. As we look to 2023, our outlook factors in both the caution associated with the oversupply and the opportunities that these longer-term trends provide. With respect to caution, we know that our runtime licenses for equipment connectivity track with our customers' equipment sales. Our gain share tracks with our foundry customers' wafer shipments, and our overage charges for Exensio Cloud depend on the volume of chips processed. We do expect that some of our revenues could be negatively impacted by business corrections. With respect to opportunities, we anticipate continued increased demand in 2023 for our analytics platform, particularly to spur product and process development and reshoring production capacity.
This demand will increasingly benefit from our collaboration with partners, which we believe will contribute more significantly in 2023 than in the past. As a result, on balance, even against the backdrop of over 30% growth in 2022, we expect to grow revenue in 2023. Now I will turn the call over to Adnan, who will provide more detailed comments on our results. Adnan?
Thank you, John. Good afternoon, everyone. Good to speak with you again today. I hope all of you and your families are keeping safe. We are pleased to review the financial results of the full year and the fourth quarter of 2022. We posted our earnings release and a management report in the investor relations section of our website. Our Form 10-K with final results will be filed with the SEC in early March after review by our auditors. Please note that all the financial results we discussed in today's call will be on a non-GAAP basis. A reconciliation to GAAP financials is provided in the materials on our website. As John indicated in his comments, 2022 was a year of solid performance for PDF Solutions.
We generated record revenues of $148.5 million, versus $111.1 million in 2021, a 34% year-over-year increase. While 34% year-over-year revenue growth is remarkable in of itself, it is worth noting a few highlights. First, the growth in 2022 was entirely organic. After we completed the Cimetrix acquisition in 2020 and delivered 26% revenue growth for 2021. Second, the analytics business grew in 2022 at a 40% year-over-year rate. Third, we saw positive customer momentum in all our analytic product families of Exensio, Cimetrix, and LeadingEdge. Our booking for the year were strong at over $200 million and grew faster than our revenue growth rate.
While we are pleased with the strength of bookings, we are even more pleased with the strong and healthy backlog at the end of the year, exceeding a quarter billion dollars. Our backlog of $278 million grew 55% year-over-year and gives us a solid foundation for predictability and longevity of the business. The strong bookings and backlog are a testament to the value customers see in the PDF platform, including data collection, connectivity, and powerful analytics. For the full year 2023, we expect to grow total revenues at rates approaching mid-teens% on a year-over-year basis. Over the longer term, we continue to remain committed to our 20% or better growth rate target for the analytics business. For the fourth quarter of 2022, our total revenue was $40.5 million, up 36% versus Q4 of 2021.
Our analytics revenue this quarter also grew 32% versus the same quarter of prior year. We also saw strength in our IYR business, with 69% growth versus the comparable quarter of prior year, primarily driven by an increase in gainshare. Our gross margin was 74%, and we reported EPS of $0.19 per share for the fourth quarter. Turning back to the full year 2022 results, I will now provide detailed comments. For the full year, analytics revenue increased 40% to $130.5 million versus 2021. Our progress to become the leading analytics software provider for the global semiconductor supply chain continues to be strong. Analytics is now the dominant component of our overall business at 88% of total revenues.
Our Exensio product revenues are now starting to benefit from renewal uplifts of cloud deals we originally signed a few years ago, software sales for new fab site installations, and multiple module utilizations by our customers. Our Leading Edge solution has established once again its unique value proposition for customers with advanced nodes, where we provide the unique data collection and fast, powerful analysis of manufacturing-related data to improve yield for their next generation advanced products. For our Cimetrix products, we are emboldened by the number of new equipment design wins, certifying PDF Cimetrix products as the connectivity software for multiple new equipment platforms, giving us further confidence in our shadow backlog as these equipment products go to market and become more successful. Another data point worth noting, our full year analytics revenue for 2022 was more than the total company revenue for the prior year, 2021.
This is the second year in a row we have achieved this milestone. For 2022, IYR revenue comprised 12% of total revenues at $18.1 million, and was essentially similar to our IYR revenue last year, even though this year we did not have the benefit of a legacy gainshare contract from a long-tenured customer. We are also pleased that gainshare from some new customers grew this year, which we will continue to watch carefully. Gross margin for 2022 increased to 71%, up from 64% in 2021, or an expansion of approximately 700 basis points. We were able to grow the gross margin because we carefully managed costs to not grow faster than our revenue growth. In particular, we managed our cloud spend and focused the efforts of our people and resources.
It feels good to have now achieved for the full year 2022 the gross margin target of 70%, which we had set as our long-term target. Turning to the operating expenses, we also controlled the growth of our R&D and SG&A expenses to stay below our revenue growth, thereby expanding our operating margins for the year. Our main expenses in the R&D and SG&A category were related to headcount, as we invest in highly talented and geographically distributed workforce and ensure we build a sales, marketing, and G&A infrastructure to support our growing business, all while benefiting from economies of scale. Our R&D and SG&A expenses as a percentage of revenue came in at 31% and 24% respectively, both less than last year's comparable levels.
For the year 2022, we reported EPS of $0.60 a share, demonstrating a meaningful growth compared to the $0.08 per share we reported for the year 2021. CapEx for the year totaled $8.4 million, versus $4.1 million in full year 2021, as we invested primarily in data collection systems for our Leading Edge business. For the year 2022, we generated positive cash flow of $32.3 million, compared to $4.2 million for the prior year, 2021. We are pleased with another year of positive operating cash flow generation consistent with our history.
During the year, we also bought back $22.5 million of PDF stock at an average price of just over $24 per share, which we believe was an effective way of returning cash to our stockholders and which allowed us to keep our weighted average diluted shares essentially flat for 2022 compared to 2021. Turning to the balance sheet, we ended the year 2022 with cash and equivalents of $139.2 million, compared to $140.2 million at the end of prior year, and we carry no debt. Our cash was essentially flat year-over-year, even after $8.4 million of CapEx and $22.5 million of stock buyback due to strong operating cash flow generation.
We are proud of the performance of 2022, remain committed to our long-term business fundamentals, and are pleased with the fortunate position we are in with our backlog of over a quarter billion dollars, all coupled with a healthy balance sheet. For 2023, we look forward to another growth year on top of the strong growth delivered in prior year, with year-over-year growth rates approaching mid-teens%. With that, I'll turn the call over to the operator to commence the Q&A session. Operator?
Thank you, Mr. Raza. Ladies and gentlemen, if you have a question at this time, please press star one one on your telephone. If you're using a speakerphone, please lift the handset before asking a question. Please wait one moment for our first question. Our first question comes from the line of Blair Abernethy from Rosenblatt Securities.
Hi, good afternoon, gentlemen.
Good afternoon.
Hi. Great quarter and great way to finish the year. I just wanna ask you a little bit about some more about the partners. You sort of touched on SAP in the quarter or in the prepared remarks. Maybe just expand upon how you're progressing with some of the major partner opportunities you've been working on the last two years.
Sure. Thanks, Blair. I'll kind of run through them all. Well, no, let's say three of them. Advantest, I said in my prepared remarks, we released, I think, four or five apps on the App Store. We expect over the next couple of months to release a number more, taking advantage of Exensio's feedforward and feedback and ML capabilities. Customers at different test insertions can use data from previous test points to make better control screening choices. Very important for our customers, particularly that are doing system and package, and they wanna make sure they match the right chiplet with the right chiplets in the right package. You'll see those more of those coming out this year and more announcements about customer adoption of that capability. Secondly, Siemens.
Really, Siemens software is used to generate test vectors, and then at the end, when the tests are complete, to do fault diagnostics to localize where in the chip the fault might be. You do that after you get the chip. It's tested at the final metal layer, maybe metal 15 or metal 17 these days, run that fault diagnostics. You would like that fault diagnostics to look at every other piece of information about that chip, its parametric test data, in-line factory data from other test insertion points. Exensio stores all of that. By integrating Exensio, our layout software for systematic models and tests, customers can do a better job on fault diagnostics and a better job localizing faults. We are now working with early customers on that capability in an early access program.
The extensions to that will also come out over the next couple of years that take more advantage of the capabilities from both companies. The third one is SAP that I also talked about in my prepared remarks. There, if you just look, you know, I think the way they've talked about it, and they put out a very nice white paper on it. You really wanna attach the top floor to the shop floor, as they call it. That, you know, shop floor being PDF, where we store an awful lot of engineering data that's inside a factory or for a fabless company from the factories where they operate. We communicate with most of the major Manufacturing Execution Systems, so our systems know where the material is in line.
They know how much consumables cost. They know how much parts sell for. They know where parts end up going in the end market. By combining the data in Exensio with the data that's in hana S/4HANA, you're able to, you know, be able to answer questions for customers on field return analysis, accurate costing by product, monthly changes in deviations in costing due to product variances, better insights on yield prediction and supply. Those are applications we're building with SAP are really how do you use engineering data with financial data to be able to have a more agile manufacturing environment. That's the third category of applications that we're rolling out. We'll expect to see all three of these make significant progress in 2023.
That's great, John. Just on the SAP, you know, the go-to-market there, I assume they're taking it to their install base as an add-on product. Is that how they're approaching it? What do you think sort of a trial period or ramp period here is on the customer taking up this solution?
We're in discussions with customers on this capability right now, and the way it's worked is it's a joint sale. We sell our capability, and of course they sell their capability. They, you know, obviously their reach, and the part of the whole reason why we did this partnership program is our partners reach into accounts is far, far deeper and broader than ours, right? Our sales capacity is really small relative to our product portfolio. We need their help to find the relevant customers, even sometimes customers that we know well, but they know the right part of the organization, the part of the organization that we don't have access to today. They are really the lead. They bring us in. They, of course, benefit financially from that.
We right now, in these early stages, we are the a contracting capability for each of the customer opportunities. It's a little bit different with Advantest, where they are the front door, and they are selling in their case, and we're secondary backup support. In the case of SAP, right now we're going in jointly.
Okay, great. Thanks a lot. Jump back in the queue.
Thank you. One moment for our next question. Our next question comes from the line of Tom Diffely from D.A. Davidson.
Yes, good afternoon. Thank you for taking my questions. First one to kinda jump in on the revenue outlook. It sounded like you said low teens is what the expectations were for this year. I'm curious, you know, when you look at the IYR business, it's, you know, averaged about $4.5 million a quarter for the last year or so. Is that our new baseline there? Or, where there's some one-offs this year that could cause it to be a little higher, meaning that most of the growth this year will come from analytics?
Yeah, that's a great point, Tom. In my prepared remarks, I said that we are watching the utilizations at our customer foundry factories because the baseline did seem to settle at around four and a half million dollars on gainshare. We believe that that may be in the first half of this year, there may be some pressure on that number as volumes will be lower than they were in the second half of 2022. 2022 also benefited a little bit from gainshare, catch-ups for some of our customers. We felt that the COVID impact in China maybe increased a fair amount of capacity in the second half of 2022, that we're questioning whether we'll sustain catch-up capacity that, you know, was not possible in 2021 due to COVID restrictions.
You know, we're being a little mindful on the IYR number, thinking it's a little bit, you know, last year may be a high water market, may be a little bit lower this year than it was last year. I think in terms of overall growth, as Adnan said, we said approaching mid-teens, so we do believe it's better than lower teens. We don't think it's on the higher end of mid-teens, you know, you know. We're trying to kind of directionally give some perspective there. Again, not just because of the gainshare piece, but as I said in my prepared remarks, you know, our customers that ship equipment, we expect runtime licenses to be down this year. Capital equipment companies are reporting about a 20% decrease in units shipped this year over last year.
You know, we do expect that we'll do a little better than that because the attach rate with the software is a little higher as they use more of the communication standards on the newer equipment than on the older equipment. We're being a little bit cautious on what we think those shipments will be. Similarly, you know, some of our customers, you know, when they go into overages on data storage and data volumes on Exensio, then pay for additional storage as they're churning through more silicon. We think some of our customers will be a little bit down in 2023 volumes over 2022 volumes. That may put a little bit of consumption charges in Exensio a little bit lower.
Okay.
As those offset the kind of core growth in analytics, which we still believe is quite strong and will continue on, very well.
Okay. I know several years ago, there used to be a little bit of seasonality in the IYR business where you had the occasional, I'm not sure if it was a budget flush, but you had fourth quarters that were sometimes packed pretty high.
Yeah
... you know, you had a little bit of a, you know, muted first quarter off of that. It sounds like the customer base, because it's different today, your view is much more of just, you know, the run rate of chip production as opposed to a fourth quarter budget flush.
Yes, that's correct, Tom.
Okay. Then that leads me to the analytics side. You know, obviously a nice sequential jump in analytics. Was there any kind of a one-time issue or budget flush in that as well?
I think part of it also, hey, Tom. Part of it, just to connect to prior comments as well on the gainshare, why to your point about Q4 expecting some seasonality, if you were looking at just Q4 over Q3, you'll see a little bit of a dip. Part of that was also because Q3 in of itself had some catch up.
Right.
If you go back and discussion from that point. No, I think some of the effects that you're seeing in Q4, we talked about the large bookings we were able to garner during Q4. Some of it is also early, 606 related impacts during the quarter that we generally experience. Nothing is difficult to remain confident about the business and our long-term growth targets for analytics of 20%.
Okay. Maybe one last question on the IYR business. Is there anything you can do during the quarter to track that, or is this really you just get a report and a payment from your customer at the end of the quarter saying this is what they produced?
Yeah, we do track it quite closely because, first, a lot of our customers, it runs through our scribe test structure, so we can measure unit volumes as they go through. It's not representative of all the wafers they run. It is a representative of some of the wafers they run, so we can look at those trends, gives us some indication. Of course, we have, you know, our engineers have daily dialogues with some of the customers that they're in there quite frequently. You know, to be clear with you, Tom, sometimes we feel like our customers don't always have the best visibility themselves, so they'll be more bullish, and then you talk to them a month later and the sky is falling.
Yeah.
You know, we, you know, we have a pretty broad perspective across the customer base these days. We look at things up and down because, you know, Exensio Cloud data, we can look at what storage loading looks like there. You know, what are the end customers doing, et cetera. You know, we apply our own little judgment on these things. Similarly with the capital equipment folks, I think, November, December, some of our smaller capital equipment customers were quite bullish when we met with them, and we were looking at signals that were less robust than they were.
Yeah.
you know, we try to take a balanced look at it, overall, you know, but we do watch quite a bit of data throughout the quarter.
Okay. Thank you. last question then. I was hoping just to get a quick update on DFI, maybe the accelerated ramp you're working on.
Yeah, it's a great question. And I did put something in my prepared remarks. I think, you know, this past year, the application we released and brought up for our customer was this DirectScan application, really, and it's very related to what we're doing with Siemens. You're taking that same software that, you know, digests all of the key critical elements of a product and identifying what is inspectable at each layer and what should you see. You can find a unique systematic defects and other problems. Scanning, you know, billions of features per hour, many billions of features per hour at this point, on a wafer. You scan a meaningful fraction of the wafer in the typical 2-hour budget that our customers give us.
We saw tremendous uptake when we got that installed in the first quarter of 2020, Q2 and beyond, we saw the customers use it quite meaningfully. We started doing pilots with other customers by them sending us wafers, really starting in the fourth quarter. I think I kind of alluded to that on my prepared remarks in November that we were anticipating that. We've been seeing positive results from that customer activity as well, and we expect to see a pickup of the business as we go throughout this year. Our perspective on the business, you know, we've been very cautious about the revenue impact this year, but we do feel it's gonna impact our bookings this year, incremental bookings this year.
Okay. Maybe just a quick follow-up on that. You know, some chips these days have 500 steps in producing them. When you look at your tools, how many insertion points or between how many different steps are they going into your system to evaluate?
That's a great question. When we first came up with this, right, the with the filler cells, we were really only, you only have to use the system at metal one. Now, you can use it earlier with this DirectScan approach as early as the contact layer. Once you started making the contacts in the middle of line through all of the metal layers in the back, because the software figures out what the patterns are.
I just saw the results from a customer where, you know, a scan somewhere in metal seven had a, you know, yield issue, and the software was able to tell them which of the defects at metal seven are really caused by a metal seven or contact or via below it, and which ones are caused by layers below that based on the pattern of the failures. You know, it's quite sophisticated in the way it post-processes the information. As a result, you can really look at any layer after you've built the first contacts. You typically start with the isolation and the thins and the gate and the polysilicon or no, metal gate. Those first three layers, you can't really use this approach, but after that, on every layer above, you can use it.
Wow. Okay. Great. All right. Well, I appreciate your time today, and I'll jump back in the queue.
Thank you.
Thank you. As a reminder, to ask a question, please press star one one on your telephone. If you're using a speakerphone, please lift a handset before asking a question. Our next question comes from the line of Gus Richard from Northland Capital Markets.
Yes, thanks for taking a question. By the way, great quarter, guys. I was wondering if you could just talk about, you touched on the $380 billion of government support, you know, likely benefiting from that spend in the U.S. because it's a little bit ahead of other countries. You know, India, Japan, and Europe are three regions that, you know, have big buckets of money. I was wondering, you know, are you engaged with programs in those regions? You know, what, you know, what might, you know, what stage is that in? Is it a ways away or near end?
Yeah, I think that's a good question, Gus. You're right. You picked other geographies where we do see significant activity. You know, PDF is pretty well known in our business for, you know, our leading-edge capability. Generally speaking, you know, when someone's going off and doing some of this, we do get contacted, and we've been contacted by a number of new entities. Of course, I can't say like, which ones and when, but around the world. We suspect that maybe it impacts bookings a little bit in 2023. At the max, it's really more of a 2024 and 2025 business perspective.
We do see, you know, those very high visibility ones, like the ones you mentioned, and a number of partnerships and joint ventures going on where folks are coming together for high voltage transistors, not just leading edge, you know, 28 nanometer and other kind of automotive applications. You know, some of the leading edge applications like you mentioned in Japan, the United States, Taiwan and Korea. It's very broad in terms of the capabilities.
While we call that our capability leading edge because the majority of our customers are doing things like two nanometer and gate-all-around, we do find that we get contacted by customers doing 28 nanometer and high voltage transistors because, you know, you need the same capability that our customers need on the leading edge. Maybe it's a different mix of vehicles and machines and software, but you need that same capability typically.
Got it. Got it. That, that's helpful framing that up. This is a little bit off the wall. I apologize. You know, clearly there has always been a market for black, you know, market chips. There's been some concern that, you know, some of the chips that shouldn't, from places are going there making, you know, weapons and doing AI. Of all the software that tracks the semiconductor industry, you track it from one end to the other. I've heard of some efforts to try to rein in the black market. I was wondering again, is that an opportunity for you or is that something, you know, you've been, you know, anybody reached out to you on?
Yeah, we've had conversations with government entities around that. We also are very active in the SEMI organization on standards. We put the standard for single device traceability, which is the SEMI E142 standard that our team, a company we acquired, Kinesis, really innovated in that area and put that capability as a standard document. We support it in our Cimetrix software in Exensio, but it's freely available. We've also extended that in the way that we've written standards with SEMI to propose using Hyperledger technology like blockchain technologies and have looked for partners in that area too, around making sure that whenever Exensio logs information about a chip as it goes through a test point or a wafer, you know, processing tool, you put in a ledger in a, you know, immutable way information about that chip.
That's very valuable, not just for the counterfeit issue that you bring up, but also because, you know, whenever you have an RMA situation and the chip comes back from the field, there's always a tension between the buyer of the chip and the seller of the chip about, you know, what was the root cause, and did you recall enough of the chips? Having clearly non-faked data is valuable for all parties in that situation. You know, in a no trust environment, this kind of information is very important. Firsthand, we had experiences with this as we were trying to build some of our fast testers for customers. In 2022, we purchased some critical chips, you know, mixed signal analog chips. 30% of them were failures.
When we went back, turns out the supplier is also an Exensio user. I sat back with the leadership team there and said, "Hey, you know, we just bought a bunch of your stuff and it was bad." They said, "Oh, yeah, we have leaks in our supply chain. We know about that." We know it typically is, you know, in the assembly area and system board manufacturer where there's risks that chips that were bad are labeled good and get out. This is an area that we do see as an opportunity for us. You know, it's hard to make it move economically for us quickly because you need adoption by the industry. Today, I think people are intrigued by it.
The engineers that chair those standards bodies are very passionate about how important it is, but we get modest interest from the customer so far.
Okay. Got it. Got it. Thanks so much. I think that covers it for me.
Thank you. One moment for our next question. Our next question comes from the line of Blair Abernethy from Rosenblatt Securities.
Hi again, guys. Just two things. I just wanna ask a little more about the DFI. You know, in the past, you've given sort of DFI tapeouts per year. It's on your website. I don't know if you've got any up-to-date stats on that to sort of show us how volumes have been progressing with the system. You know, another year has passed here, and you're getting more traction in the market. How are you looking at the, or how do you currently view the, you know, market size, market opportunity for DFI over the next five years?
Thanks. I'll answer that. First of all, Blair, it's a really good point. When you went back a few years ago, we tracked tapeouts because you needed to put our structures inside the design in order for the whole machine to work. What the team innovated it in the, you know, end of 2021, and then we installed for the customer in early 2022, was the ability to look at a product, even if it didn't have our structures in it. That's what we call the DirectScan ability. The ability. The software looks at the design and figures out at that point of the process, whether it's metal two or metal four or metal zero or whatever, where are all the features that look like a test structure at that point?
That we stopped tracking filler cell insertion because we didn't need it anymore for the application to take off. We've been checking, you know, recipes set up and, you know, usage at the early customers, that's actually picked up and been quite strong. Number of recipes set up. I think we're setting up a handful of recipes every week. In, you know, the last, let's say, six or eight months, you know, you can say probably 40 or so recipes set up just off of, you know, early adoption. In terms of the market opportunity for this, you know, we feel very bullish about it, but particularly more bullish as we went through 2022. The reason for that is twofold.
You know, this first ability to be able to look at any product, even if it doesn't have our filler cells in it, really greatly opened up the market opportunity. and the ability to look at any layer as opposed to just the metal one layer, as Tom alluded to in his question, means that, you know, instead of there being one pass for inspection, you know, you literally, you would have, you know, 10 or 15 or 18 potential layers you could use the machine. It's number of insertion points is much larger. Secondly, as we've met with customers, there seems to be an increasing growing understanding that you need to use voltage contrast in electrical inspection as customers are trying to bring up things like gate-all-around and other complex 3D structures.
You know, that pilot we did in Q4, the early work that we've done with another customer in Q1, a lot of it is around sending us wafers specifically for these 3D problems, where really the voltage contrast approach is the only way you can see it. The reason why people haven't done it in the past is the throughput typically on an E-beam machine is so poor you can't see a change, you know, a deviation in yield once the yield gets to be, you know, relatively good. Once you say at, you know, a one square centimeter chip is yielding 70%. You're not gonna see anything on a conventional E-beam machine yet. With a two hour inspection on an eProbe, you can still see yield detractors. Customers have given us that feedback.
I think that's really the, you know, kind of a inflection point for E-beam as an inspection tool. I think we'll see more proof points as we come through 2023 on that.
That's great. Thanks for the color, John. Just on the backlog, $278 million is. Adnan, do you have a current amount of that backlog, i.e., what you expect to recognize in the next 12 months and the duration of the overall backlog?
Yeah, great question. You know, I know it's a data point of interest to yourself and me. Right now we're not disclosing it. We're still gonna put the color that we do in normally in our filings, that the majority of that is going to be within the next two years. I can give you this, you know, perspective that every year as we do the annual operating plan, just like we finished and got done this year, same time as last year, our confidence and the demonstrated number, from what percentage of the next year's revenue is going to get contributed into the forecast that we have continues to be higher. That was a trend that we saw this year.
Larger portion than last year of the revenue coming into the following year is spoken for as a function of the strength in the backlog that we have. I think as we get ready for future sharing of the metrics, you know, we'll be thinking about when is the right time to share this. I can assure you we're certainly tracking it inside, and this is one we find very comforting as in whenever we do the annual operating plan.
Great. Understood. Thanks very much, guys.
Thank you. As a reminder, to ask a question, please press star one one on your telephone. At this time, there are no more questions. Ladies and gentlemen, this concludes the program. Thank you for joining us on today's call.