Good afternoon, and welcome to the first day of Needham's twenty-sixth Annual Growth Conference. My name is Quinn Bolton. I'm the semiconductor analyst for Needham & Company. It's my pleasure to host this presentation with Atomera. Atomera is a semiconductor materials and technology licensing company focused on deploying its proprietary silicon-proven technology into the semiconductor industry. The company has developed Mears Silicon Technology, which increases performance and power efficiency in semiconductor transistors. Joining me from the company today are Scott Bibaud, President and CEO, and Frank Laurencio, CFO. Scott, Frank, welcome to the Needham Growth Conference. Thank you for joining us.
Thanks a lot, Quinn.
Before handing the presentation over to management, I just want to remind folks listening to the webcast that if you would like to ask a question, you can submit that through the dialogue box on your screen. I will come back after the presentation to moderate the Q&A session. So, Scott, over to you.
Okay. Thanks a lot, Quinn. Today I'd like to go over our, you know, I'd like to do a quick introduction to our company for new investors and also dig in a little bit more on some of the latest news, a little bit deeper in it. So, let's get into it. So first of all, Atomera is a company that, whose primary product is something called Mears Silicon Technology, which is a thin film that we can apply when people are manufacturing semiconductor wafers, and it results in... it improves the transistors so that they get higher performance, lower power, and lower cost. We're not a manufacturer of those wafers, though.
We're a capital-light technology licensing business, and since we've gone public, we've grown our engagements with customers to the point where we're working with more than half of the world's largest semiconductor makers. We've got licenses with five companies and JDAs with another two, and we have a very strong patent portfolio and, and management and engineering team to, to deliver this product, and I'll talk more about those later. So Mears Silicon Technology is actually a quantum engineered silicon technology, and what do we mean by that? This is not a material that's naturally occurring in the world. It was designed on a computer at the atomic level, and then gradually, built up to the point where we have not only created the material, but figured out the best way to possibly build it in production.
You can see over here on the left, we're showing a diagram, which is a cross-section of one of our, a silicon wafer, and you can see these, these little lines in here. That's actually Mears Silicon Technology, is that sandwich of silicon and partial monolayers of oxygen, which does all the magic that our technology delivers. You build this technology in a fab with normal semiconductor tools that are made by some of the biggest companies, including ASM, Applied Materials and Kokusai, and we've been working with these companies for more than a decade on their tools, so we know how to install their tools. There's not a lot of mystery involved with getting our tool up and running inside someone's fab.
It takes just a small amount of effort to change some gas connections into it, and then our team brings our know-how and technology, and we can start building a film like the one we're showing over here on the left. Now, when we deposit that film, we put it in transistors. Like in this case, we're showing this yellow and blue area inside a transistor between the source and drain. So our MST goes there, it forms a superlattice, and that allows oxygens, sorry, electrons to move faster. It blocks leakage down from the gate, and it does a number of other things to improve performance of the transistor, so you get that higher performance and lower power that we talked about earlier. And this is not a theoretical thing.
We've built thousands of wafers of silicon and tested it out to prove that our technology actually delivers on what we say it does. One of the things about MST is that it has to be properly integrated into the customer. So if you think about Atomera, part of our company is the part that's developing materials and testing them to see how good they perform. The other part of our company is an engineering team whose expertise is in helping our customers integrate our technology into their manufacturing flow. Because when we put our technology in, it causes changes inside some of the other surrounding steps of the technology. So in order to get the best performance, they need to work very closely with us, and we have a lot of know-how to help them out with that.
So, how does it work once we do integrate it properly? Number one, it increases carrier mobility, which means electrons and holes flow much faster when MST is in place. It enables dopant diffusion engineering, which means when you make a transistor, you can see in these pictures on the right, these different colors are different areas of the transistor that have been implanted with dopants. And as you manufacture and subject it to high heat, those dopants move around uncontrollably. But with our technology, we can help to hold them in place, and that brings big advantages. It minimizes gate leakage and variability across transistors, and it enhances reliability. And by enhancing reliability, it allows people to overdrive their chips and still get the same type of reliability, which means they can make them higher performance.
On the right, you can see this transistor architectural evolution. When we invented MST, the entire industry was using planar transistors, and we brought benefits in these areas between source and drain. In about 2014, we moved to FinFETs in the industry, and now they're doing work on gate-all-around. You can see that the transistor architecture gets more and more complicated, and wherever you can see one color of this transistor touching another color, is a place where MST would be very valuable. So as the transistors get more complicated and the die size, I mean, and the transistor gate length gets smaller, like today, we're talking about 2 or 3 nanometers, MST becomes more and more valuable and more and more needed to solve the problems of the industry.
That's why today, MST works in many different process areas and application areas, including the ones shown here: BCD, RF, advanced nodes, and memories. We work with most of the very large players in the industry that have their own factories, so integrated device manufacturers, like those shown here on the chart, and foundries, like those shown here. We also do some work with the fabless manufacturers, but fabless guys are not traditionally our customers. But if we can talk with a fabless customer about using our technology and convince them that that's the right thing for them, they will go to their foundry partner and encourage their foundry partner to adopt our technology. So they're more influencers than direct customers.
I mentioned earlier our big suppliers are. We're also a very close partner with Synopsys, because Synopsys makes some of the simulation tools called TCAD that the whole industry uses to simulate what a manufacturing process will look like, and therefore what level of performance you're gonna get from your transistors. And so we do a lot of work with them and write our own TCAD software for our customers to use. You know, one of the challenges in our company is that our customers tend to be very secretive. They don't wanna say that they're working with us. They don't want us to talk about what we're doing with them or where we are in our efforts.
So we've come up with this phased chart showing from when we first start planning to work with a customer until we get into volume production, and we regularly report on that during our earnings calls to give you an idea about where we are with customers. The most critical phase being this third phase, where they do multiple iterations of runs, of wafers, and may do, may do many runs, which can take a long time to get done, or it may be a quick set of runs where it takes less time before they move into the installation phase, which is really where the customer can deposit their own MST, and then they can be on the path to production. The way we get revenue during these is three ways.
First, we get some engineering services fees, but it's really not the focus of our company. The real focus is on our licensing, which is structured into upfront license fees, which we break into three different categories so that customers can buy low cost, low risk integration licenses in the beginning, and as they learn more about it and see more details, they can buy more expensive manufacturing licenses. And finally, a distribution license, which gives them the rights to sell technology using MST in it. Finally, if they get into production, they will pay us royalties on the shipment of their products, and that's really the main driver for what we're trying to accomplish here at Atomera. Sometimes we enter into joint developments with customers, which span multiple of these phases, and they're all customized, and so we report on those somewhat separately.
You can see here our history of growing our engagements with customers, and I'm happy to report that in Q4, we entered into our second installation with STMicro, and even during the quarter, we entered into that installation and completed that installation, which is why we reported an increase in our projected revenue for Q4 of this year. But let me talk a lot more about STMicro here, because I know that that's something that people are very interested in hearing about. We've been working with STMicro for a number of years. They saw big improvements on their technology using our MST when we first did experiments with them back in 2019 and 2020.
But they, for various reasons, they held on to the technology, kept talking with us, and then in 2022, they reached out and said, "Okay, now we're ready to take this to production." So we negotiate a license with them, and in late April of last year, we announced a full commercial license with ST that gives them from the manufacturing license all the way through the licenses for going to production and defines what the royalty rates will be.
So where we are today with them, in that typical chart that we show you, everything in gray has been really completed, and we, we just completed installation, which means that our manufacturing license is not only in place, it's also fully paid for at ST, and now they are in the process of productization, which they've been working on since we signed the license, and will enter into qualification and then be in production. And so in front of us, we still have upfront license fees for the distribution license, which will happen during the qualification period, and then the royalties that will come during production. So let me just talk a little bit more about what this productization effort looks like.
Today, and even since the beginning of May, we've been under process development with TCAD, so in using modeling software with ST. We could have done at the MST installation very quickly after we announced our license, but because of some logistics issues on their part, it took them till November to get started on that. I don't believe that it slowed down the schedule for this overall process with ST, because there was a lot of work to be done under process development anyway. So now that the MST installation is done, and the next step, which will be commencing now, is that they'll be manufacturing wafers using MST and then taking those wafers and testing them.
To the extent they're getting the type of performance improvements that they've been able to simulate in TCAD, then they'll be ready to move forward to the PDK. They may decide to do multiple iterations of this manufacturing and silicon validation, but because MST is installed in their fab, they can do that very quickly. Just to give you an idea, when they have to send wafers to Atomera to deposit MST, and then we send it back, that can add a month or two months to the process every time they want to run wafers. In this case, they can run wafers in 20 minutes or so, and it takes no time at all for them to add MST to their wafers.
So once they reach the level that they've got the performance levels they're looking for, and we know that they are already available... I mean, that we've been able to prove that we've gotten to those levels in the past, they will freeze the process design kit or the PDK, and enter into process qualification. So now the PDK is really like a database of how the transistor will work when someone designs with them at their company. So all of the chip designers will access the PDK when they're designing new chips. So as soon as the PDK is frozen, all chip designers will start using it.
As a matter of fact, in my experience, when chip designers know a new PDK is coming sometime in the future, they slow down their development of chips on the older PDK, because they want to get access to all the new features they're gonna get on the new PDK. So generally, there's a pent-up demand of new chip designs when you release a new PDK, and we believe that will be the case, so they'll be starting a lot of chip designs. Those will be happening in parallel with process qualification, and when the chip designs are done, they can actually start in production, even before process qual is complete. That's a process that's called risk production. When the process qual is complete, they'll release it to production, and then chip designs will be under normal manufacturing rules. In any case, the...
As soon as those chips ship into customers, then we'll be eligible for the royalties that we expect will be coming. Now, we don't have any definitive schedule from ST on when this will happen. We hope that we would expect to see this happen by the middle of next year. And yeah. And so we don't have much insight into the volume ramp or what markets these chip designs will go into yet, but we hope to get that in the near future. Anyway, it's very exciting. Today, I would say, well, not today, but since we started, there was some skepticism about when and whether we would be able to get a customer into production with our technology.
I can tell you that I am, I am very confident that ST and where they are in this process, will be getting into production with this technology, and it's only a matter of us turning the crank now with them to the point where they'll ramp, and we'll be, we'll be generating our first royalty revenue. Now, that being said, we exited last year with a lot of momentum in many different areas. This MST-SP and SPX is the work that we've done in the power, 5-volt and higher voltages power area, which is similar to what ST is doing, although they're not using our technology, they're using their own.
We're also doing a lot of work on MST for RF-SOI, and it's a space that we have a lot of customers and are engaged, and we, you know, a lot of confidence in the technology that we have there can bring us some big gains in the marketplace. MST for advanced nodes, we've been working with some customers on their advanced nodes technologies, and, again, last year we wrote some white papers about all the advantages that MST can bring for advanced nodes. You know, every time the industry moves to a smaller node, they need help. It requires an ecosystem of users to get to the point where they can take one of these new nodes to production, and our MST technology, we believe, can provide a critical element in getting there.
And then finally, we wrote some white papers and showed some information on MST for DRAM this fall. We've known that MST was good in DRAM for many years, but only in the last 12 months have we started to get traction with some good memory makers. And so we're hopeful that this will lead to licenses and production with customers in the DRAM area, because of our ability to improve the variability of the sense amps that are used when people make DRAM technology. So one other area that I just wanted to talk about briefly today, I haven't talked about before, but it really is something that we're starting to see demand for.
So in artificial intelligence, you know, the performance requirements in artificial intelligence with algorithms that are being run are driving really unprecedented compute workloads, and it's very hard to put all of the different computing requirements for a high-end, AI chip onto a single die. So, you know, when you're making single die, you don't want them to be infinitely large, because at a certain point, the yield starts to fall off when the die area gets too big. So we see that the way the industry is going after solving this problem is called heterogeneous chiplet architectures. There's a good picture of it over here on the right.
You use one of these complicated interposers and develop a number of different chips that interconnect on the interposer, and therefore, you can optimize your chiplets for different process nodes. So for example, on a typical interposer design, there might be a power management chip, which really, in most cases, should be developed in some of the older process nodes, like 130 or 90 or 65 nanometers. But then you have your high-end CPU that's gonna be developed on a 10-nanometer process or a large FPGA that's gonna be on the most bleeding-edge node that they can get to. By using chiplets, they can really optimize the performance of each of the chips and interconnect them together. But what this also means is that some of the mainstream nodes and leading analog nodes are gonna require additional work.
Additional work, meaning performance enhancements to help them achieve the levels that they need to get to, to deliver on the computing demands of AI. And so you can see that because MST provides benefits across all of these different process nodes, it really is a fertile ground for Atomera to continue to get new designs in the mature nodes. Finally, let me just say a word on our patent portfolio. So today, we have 322 patents, and our patent portfolio continues to grow very strongly. During the course of this year, we had some of our earliest patents sunset, but the rate at which we're growing patents is far out, outstretching the small number that we have sunsetting. So I continue to believe that our patent portfolio will grow strongly in the years ahead.
But not only do we have a lot of patents, these patents are covering, you know, most of the areas that you would need to do MST, including the core method and device architectures, and also, where we discover next-generation technologies that can be developed using MST, we patent that as well. And we patent them both domestically and in appropriate foreign jurisdictions where we think they are most compelling. I showed earlier this TEM image showing a silicon chip cut in half with our distinctive, lines of oxygen and silicon. This is a very important thing for patents because this means that it's discoverable. If we believe some company is trying to pirate our technology, steal our patents, then we would be able to cut their chip in half and prove that they are using MST when they didn't have a license. So that makes...
When a patent is discoverable, it makes it much more valuable. And also, I would say, being able to build this, that's shown up here, is very difficult without our know-how that we've developed over many years. So we do license our know-how to customers. Licensing know-how has no expiration date, and so that makes our patent portfolio even more valuable. One way you can judge by how valuable and useful our patents are, is by how other companies are coming up with ideas about how they'd use MST and filing their own patents on it. Those patents are great. It means that they've studied our technology and see a lot of value in it.
It also means that if they really want to implement it and take it to production, that they'll have to execute licenses with us to build the MST that's part of their patents. And finally, this is our last public financial statement, which is from September thirtieth. I will say that our cash equivalent at that point in time was just over $20 million, which was a comfortable cushion for us. And our, as you can see, our operating expenses are, you know, modest to the point where we can continue to operate the company quite safely with the cash level that we have. So let me just wind this up.
You know, today we, we've demonstrated that we can have a very high margin, recurring revenue financial model, which we think is really the strength of the company. We've a very strong technology patent portfolio and balance sheet, and good traction with top customers, and ramping commercial licenses. I think the, the STMicro license is important for many ways. Not only does it signal to investors that we will get to the commercial phase and, and be a revenue-bearing, I mean, a, a royalty-bearing company. But it also signals to a number of our potential customers out in the marketplace, that other large reputable companies are licensing from us, so it's safe for them to do so as well. So with that, I think I'll turn it over to Quinn for Q&A.
Perfect. Well, thank you, Scott. Thank you, Frank. I guess, you know, maybe if you could, I don't know if you disclosed this or not, but you talked about sort of the three different types of licenses, the integration, the manufacturing, and the distribution. You've got a number of customer engagements, and so, you know, just kind of wondering, do you size? You know, what would a typical integration license be? You know, are these hundreds of thousands? Could it reach a million? Same for manufacturing or distribution licenses. And then do you also, you know, to the extent ST sounds like they'll be the first to potentially go to production. You know, is it a royalty based on the wafer ASP, the chip ASP?
Or how do you, how do you sort of collect the royalties, or what is the royalty rate based on?
Yeah. So, Quinn, yeah, first I'll address the three upfront licenses. So our list price for those is, you know, the integration license, we try to keep it fairly low to encourage people to try out our technology, and it's $ a few hundred thousand. Then, when we get to a manufacturing license, then we're asking people for about $1 million. And then when we get to the, the distribution license, at that point, you know, it's very safe. They know exactly what they're gonna be getting, and so we ask for about a $3 million upfront license fee for that. So, overall, about $4 million-$4.5 million that we would like to get from customers.
Then once we get into volume production, we've always talked about a 1%-3% of selling price royalty rate, so, and that's based on what the customer sells. So if it's a foundry, it's based on the selling price of wafers, and if it's a IDM, then it would be on the selling price of the chips. There are exceptions to this. You know, when we came up with those guidelines, we were really focused on some of the mature nodes, and it really works well for those mature nodes. For some of the more advanced nodes in memory, it may not work as well, because in memory, they ship so many wafers. The production of wafers in the memory industry is just massive.
And so probably the royalty rate I talked about might be a little bit high for them, because we're just gonna be getting so much revenue, even with a lower royalty rate. And the same as can be said for the most advanced nodes. In the most advanced nodes, they're selling the wafers. I think TSMC announced that their 2-nanometer gate-all-around wafers would be selling on the order of $30,000 per wafer. And so we probably will not get 1%-3% of revenue on those wafers, but still there's a huge revenue opportunity, even at a lower level with them.
Got it. It and I may have missed the exact number, but it looks like you had about 23 or 24 customers in your engagement pipeline. Do you provide any sort of sense as to the split? You know, what of that... You know, how much of that is, you know, kind of analog, RF? Is any of it today in the more advanced nodes, FinFET? You know, probably, I would imagine, early for gate-all-around. And you mentioned the DRAM work, which was a little bit more recent, so maybe early for DRAM as well. But do you give any color sort of by device type, what's in that pipeline?
Yeah. We have not broken it down in the past by device type. Unfortunately, these days, the semiconductor industry is so consolidated that if I announce that I'm working with, you know, one customer in a certain area, the other customers will know who it is. They can, for some reason, pretty easily figure it out. We run into that problem all the time, and customers are very, very sensitive about us disclosing. But I can tell you everything you asked about, analog, RF, advanced node, memory; they're all in that pipeline, yes.
Got it. Okay. Another question I had is, you, you sort of highlighted RF as, you know, one of the applications. In the presentation, you talked about the Mears technology increases electron mobility. So is the RF, is that more sort of a, you know, a high-performance, frequency-driven process? Does it tend to be more things like power amplifiers? Because you also talked about being able to have very low leakage current. So, you know, what, what are some of the applications you're seeing in the RF segment?
Yeah, there's a number of applications in the RF segment, but I would characterize them more as in the RF front end, and in particular, for products that are using RF-SOI substrates today, which are a substrate that was innovated by Soitec. A lot of the RF manufacturers are using them today for power switches and for LNAs and a few other components. The RF-SOI substrate brings a lot of advantages, but it has a few flaws that the industry has known about for years and just couldn't do anything about. We believe that MST helps to solve one of those flaws, that is, is preventing them from getting to much higher levels of performance, and we've proven that on silicon on a number of occasions.
So we're hopeful that that will be incorporated in some of the next generation RF-SOI processes that are being developed in the market.
Okay. Is it something specific to the Soitec process, or is it more broadly, just general SOI? I think ST and others have worked on a fully depleted SOI process that they use in some of their fabs. So, wondering... You know, maybe I'm trying to connect too many dots here, but,
No. Yeah, it's not specific to Soitec. It's more. And by the way, this is not fully depleted SOI. This is something called RF-SOI, which is a trap-rich substrate with an oxide layer on it, that Soitec made it, but there are other companies that still manufacture similar RF-SOI wafers, and they would have the same type of problem. It's not a company-specific problem. It's just one of the limitations to that architecture, that has to do with the way dopants move in an uncontrolled fashion. And one of the great things about MST is that we can engineer where MST is installed and integrated to control the dopant flow, and so help to solve problems like this exact thing.
Got it. You'd also mentioned, you know, sort of integration of the MST, you know, does require some perhaps process changes to other process steps. And so, you know, I don't know if this is things like you've got to be very careful once you put an MST film down, about hitting certain temperatures and, you know, post the MST. But, you know, what types of process steps might you have to work with customers as you look to integrate the technology?
Yeah, a classic example, I just talked about dopant diffusion. So, you know, when you make a semiconductor, and the more advanced it is, the more implants it has, and implants put dopants into certain areas, and then when you go to the next step and you heat it up, they move around kind of uncontrollably. If you could make an ideal semiconductor, you would put the dopants where they're supposed to be, and they would never move, but that's not how it works in real life. With MST, we can deposit MST, and it helps to control those dopants and hold them in place.
So what that means is, maybe, before they had MST, they would put far too much dopant in a spot because they knew a lot of it would diffuse away, and they'd be left with the amount they wanted, roughly. Once we put MST in, they have to put less dopant because it's gonna stay there, right? So that's how they might have to change it. That's an example of what they might change for the different processes. It's not rocket science, but it is small changes, and we can help customers through that because that's what we're expert at.
Yeah, so that almost sounds additive rather than, you know, "Geez, I've got to do some corrective actions and other process steps." It sounds like the MST could actually help you in some of the other wafer processing steps as you just mentioned.
Exactly. Exactly.
Yeah.
Yeah, and I think what we find interacting with customers is, we're not reinventing the wheel. Sometimes we're presenting them with changes to other steps that they may have considered, but the materials they were using made those impossible. So sometimes we can enable things, and this is true with, like MST-SP and, which is really sort of an asymmetric design for a power device. These are things that are already well known to people who are much more expert on the transistor than we are, but we're delivering them a material that enables them to do certain things.
Got it. I believe in the past, you guys had some CapEx associated with the epi chambers that deposited the MST film. I assume at this point, do you have the epi chambers that you need? Is there a significant CapEx requirement going forward, or, you know, it would, I imagine as you go into production or the customers begin to manufacture the wafers, that CapEx would fall on their balance sheets, not your balance sheet.
So it's not CapEx for us because, you know, owning and operating epitaxial reactors is very expensive. It's really the operating cost, so because you have to have a full, you know, clean room to house them. So, what I would say is it's a very big component of our operating expense. So, a big part of our R&D expense every month is the lease payments that we're making to, you know, two separate tools that we have. One of those tools has two chambers in it, so it can run both 200- and 300-millimeter wafers. And so it's not really— We have talked about two, you know, tool expenses, but really, it's leasing those.
You know, that's where we put the film on the wafer that's being evaluated by the customer. When we transfer the recipe to the customer, as we did last quarter with STMicro, they're now doing that themselves, and so that removes some of the workload from us, but it just frees up that capacity for us to run internal experiments and experiments for customers.
Got it. Got it. I know we're getting close to the end of the session, so just a couple more questions from me. You know, what are your biggest priorities for 2024? I know one of the nice things about the Needham Growth Conference is it's here in early January, and so we get to ask you the big question, you know, like what's... You know, what's in store for 2024? What's on your highest priority list?
Yeah, so, it's a great question because I'm working on company goals as we speak as well. So, you know, our number one goal is to make sure that STMicroelectronics gets into production, gets in production as fast as possible and with as high a benefit as possible, so it gets adopted across the most possible products. That's number one. So our team has really been focusing on that. Second thing is to really try to bring more customers out into licenses to the point where we can expand our customer base and have a more diversified revenue path as we go forward. We absolutely are focused on that, so I would say that's our second biggest goal, driving across the company.
And then third, we have a number of R&D initiatives that are underway that will provide us with not only additional application segments to go after, but even possibly some entirely new applications of MST that we would love to get to market, and especially ones that can get to market and start generating revenue faster than the traditional method of integrating into semiconductors. So those are our key care abouts for 2024 for the company. And we think there's really good potential out there.
Excellent. And the last question just, you know, to the extent you, you're meeting with investors, do you think that there's anything that you find commonly misunderstood that you'd like to try to clarify, or, or do you think that the investors have a pretty good handle on, on the story and the opportunity?
I would say the one thing I really want investors to understand is that this... It seems like there's still some skepticism about this STMicroelectronics, design, and I can tell you that within the industry, there's not skepticism, and people understand what's happening. It's moving forward. It's gonna go to production, and people need to be focused on that as our really, the proof of our business model, and so just want to be sure people know that, that that's gonna happen, and we're gonna make it happen in a big way.
Excellent. Well, I'm sure the first customer probably always takes the longest, but hopefully it opens the floodgates or at least the doors for additional customers, and so it'll be exciting to kind of watch the company's progress this year. With that, it looks like we're at the end of the session. So Scott, Frank, thank you so much for joining us at the Needham Growth Conference. We really appreciate it.
All right. Absolutely. Thanks for having us.
Thanks.
Thanks, everybody.