Weebit Nano Limited (WBT) Earnings Call Transcript & Summary

All right, everyone. We'll make a start of it. I'm really thrilled to be welcoming back Coby Hanoch from Weebit Nano. WBT is the ticker code here. We've spoken to Coby 3 times previously. But as I was just saying, it's actually been quite a while since we caught up, in fact, almost 2 years or so. And a lot -- let me tell you, a lot has happened in that time. So trying to bring myself back up to speed here. The first thing I looked at was the share price as you do. And it's up 50% since we last spoke to Coby. I was like maybe I should have paid more attention. Of course, it round tripped to a new all-time record low before it went up. And I think it shows you 2 things. It shows you one that the market is often wrong and it shows you two, that patience is a virtue. And I think it's particularly true when you're talking about a company that is doing hard things, and Weebit Nano -- it's hard to think of too much that's more difficult than this. So let me back up a little bit here. These guys are in the business of designing computer chips essentially, what's called next-generation resistive ReRAM technology. So this is a technology that's used in a host of different applications, but particularly things like Edge AI, autonomous vehicles, these kinds of things. The kind of thing where you need a bit of computing power away from the heavy gruntwork of the servers. And look, if you've been keeping up to date with developments in the tech space, you know that the demand for this stuff is only growing. And not only can it do things in a better way, but importantly, in a more cost-effective kind of way and a more energy-efficient kind of way and a more robust kind of way. So the technology has always had this promise. And I think what's really, really interesting about Weebit is that when we first spoke, they were still very much in the development phase, sort of coming out of it, and we've seen some really cool developments and milestones are being hit. But in the last year, in particular, things have really got super interesting, and we are very much in the commercialization phase now. And frankly, I think a lot of the market is probably oblivious to, and I'll admit that to my embarrassment, I have been in that camp. I'll put some numbers on this for you. Last full financial year, FY '25, it was about $4.5 million in revenue. They're on track to do at least $10 million in revenue in the current financial year. Why is that? Well, a whole bunch of developments, as I said, perhaps one of the most standout ones being this landmark licensing agreement with a giant like Texas Instruments. And we've also had successful qualifications at foundries like DB HiTek, et cetera. So things are really, really getting interesting now. So I'll welcome Coby in just a moment. Before I do, remember, as always, none of this is advice. And please, if you do have any questions, use that slido link, and I'll put them to Coby when we get the chance. But with that out of the way, Coby, thanks for your time today.

Thank you.

What a journey, man.

Well, it's something that I personally knew was going to be like this, obviously. I know that many people in Australia have less understanding of semiconductors and how it works and all of that. So for the longest time, I had so many people doubting and so many people not really understanding what is going on. Now I think we're starting to actually show people what is going on. It is a very exciting journey. It's -- I mean, in semiconductors, memory is a critical part. You can see it today, anyone who reads the news, even the nontechnical news, et cetera. Memory has different uses, and I want to make it clear. There's actually 2 main usages of memory. One is memory chips, just these big gigantic chips that all they have on them is memory. That's called a discrete chip. And that's something that Weebit is planning to address. But right now, the priority of that has actually gone lower with all the AI stuff, and I'll talk about that in a minute. The other place where it's really critical is in what's called a system-on-a-chip. That's the heart of any electronic system. You basically today have a situation where -- and many people might have heard of Moore's Law and everything is shrinking all the time. So today, an electronic company, the heart of their system is actually a chip where they design the core of that chip. And then they can put a whole system there. Now they don't need to reinvent the wheel. If they want a WiFi, how many companies have already done WiFi. They take a WiFi module from someone, they plug it in. When they need a processor, there's ARM or other companies that have already developed processors, they just take it and plug it in. When they need a memory, they just look for a memory module and they come to a company like Weebit and take it and plug it in. So we call this market the embedded market. You embed our memory into the chip, the customer's chip. So -- and the business model in this market is licensing. So we don't actually manufacture a chip. Weebit doesn't have a chip that it manufactures. It enables manufacturing a whole lot of chips. Now let's take a step back. What is nonvolatile memory? What is it that we actually do? Nonvolatile memory, as its name kind of implies is the memory that doesn't lose its data even when you unplug it from power. Now if you think about it, any electronic device in the world practically has the magical moment, I call it. You turn it on for the first time and you expect there to be some code, some data, key security keys or whatever there. So that's nonvolatile memory. And today, the technology that is prevalent, that's what everyone is using is called Flash. It's been around for decades, literally decades. And you'll understand why the potential of ReRAM is so great, again, because of that, once we sign an agreement with a company like the Texas Instruments, it's an agreement for decades. This is something in this industry, such a very basic core technology, you don't bring it into your system for just a few years. It just doesn't work that way. And I'll talk about the applications. So Flash for over 20 years already, people knew that it's going to hit the wall. It was so obvious, so clear and people have been looking for the longest time for something that can replace Flash. And there were so many different technologies out there and so many people tried so many things. But you basically have a situation where all of these things were too expensive, too difficult to manufacture. It came down to a couple of technologies finally about 5 years ago. There was a technology called -- or there is a technology called MRAM. Everyone thought that was the winner. It was already in mass production 5 years ago, everyone said, "Hey, great, wow. But MRAM is really expensive to manufacture. But the biggest issue is products with MRAM, once they hit the market, they started malfunctioning sometimes because, guess what, M is magnetic. So -- and we're surrounded by magnetic fields. And that's when ReRAM already got to the point where it's ready for mass production and people said, wow, something which is so cheap to manufacture, so easy. You just go into a standard manufacturing line, and it doesn't have all these magnetic issues. So yes, so we are now at the point where the market basically -- I think I can already say the market accepts ReRAM as the replacement of Flash. You see it in analyst reports that they're expecting the ReRAM market share to grow significantly in the next few years and basically become more than half of the market already. So it's really exciting times for us. And I mean, even the biggest consumer companies came out and said, "Hey, we cannot use MRAM. MRAM is just too dangerous for us. Our products go everywhere on this planet and stuff. So we're really feeling very bullish. And you can see now that the market is accepting us. The market is -- we've signed several key agreements. You mentioned Texas Instrument. We also signed Onsemi, which in Australia, less people know, but it's a very, very big company. It's a USD 30-plus billion market cap. And we've been moving forward with the technology. And yes, it's -- I mean, TI -- I'm sorry, TSMC, the biggest foundry in the world, foundry is a company that manufactures semiconductor. They develop their own ReRAM. And it's in mass production and used in products already, so people can actually see ReRAM in mass production. Our ReRAM is very close to mass production now. And it's a challenge. Everything in semiconductors. When you're talking about the core technology is very, very difficult, takes a lot of time. That's what you've been seeing as well, right? And many people lost their patience along the way, maybe. Actually, I have to say our shareholder base is very loyal. They might be the most knowledgeable people in semiconductor in Australia. It's really -- they learn the material, they understand what's going on and so many of them are holding for so long.

Yes. I mean it's such a virtue. It's a theme that you see a lot actually is that -- and it's very hard when you sort of see the future and you're very bullish on it and you think, well, this is the way the world is going, not recognizing that -- or perhaps recognizing and just having to make peace with the fact that even under very bullish and optimistic scenarios, these things take years. Now in the grand scheme of things, what's a few years, right? It's a blink of the eye, but in human time and share market investor time frame, we have trouble looking across the next week, let alone 3 years from now. But as I said at the start, I think that's done well for us as a group. And I think it's a message I like to really just underline and reiterate here, too, is that being too early is probably indistinguishable for being wrong. But if you want to wait until every single thing is absolutely green lit and clear, it's like, well, great, but most of the upside is gone. So there is a balancing act there, and it does require to look beyond a 4C or to look beyond the half year report and like what does a company -- what does Weebit look like in the year 2028, 2032. And that's where I think the opportunity comes from. One of the things I want to pick up on there actually that was very interesting, and I know we said this in some of our previous discussions is that there are long lead times with sort of getting licensing agreements and getting the fabs to sort of manufacture it. But once embedded, it is very, very sticky. And as you said, once companies start using it, I mean, they've got to design all this kind of stuff around it. It's not something that you're going to replace willy-nilly. Can you just elaborate a little bit more on that because I know that when I saw the Texas Instrument announcement, you sort of said, well, there's no commercials that we're releasing. We might not even see any revenue in the short term. It will be material in the medium to longer term, but it's sort of, again, just to make sure people have very both bullish but reasonable expectations. It's not tomorrow that it's going to all happen.

Yes. So there are a few things here that I need to explain. And I'll start off with the basics that, yes, these are agreements for decades. And the key reason is if you think about it, a company that designed -- let's just talk an automotive chip, okay? In the automotive industry, in many countries, once you have a chip out there, you're supposed to be able to supply replacement parts for 17 years, okay? That chip needs to be in the market 17 years, okay? So -- and there are so many markets like that. I mean we always think of things like, oh, we buy this product after 2, 3 years, we throw it away. We have something new. We have some -- there are a lot of chips that actually live a long time. That's one thing. And even if not them, then their derivatives, I mean, once you already design something and everything, you don't just go and totally replace everything all the time. You improve your core technology, you improve a lot of other things around, but you don't really go and change everything. And the memory technology is something that bringing it in takes a lot of time. When we sign an agreement with a TI or with an Onsemi, et cetera, we go into something like a 2-year plus or minus. It can be 2.5 years of transferring the technology to them, teaching them how to manufacture this technology. We're talking about a very, very -- I don't know how -- what word to actually use. I don't like to use the word complex, but it is a very deep technology. Weebit has 19 PhDs in chemistry and physics developing this technology. Do you understand what that means? 1/3 of my R&D are PhDs in chemistry and physics developing this technology. It's not an easy technology to develop. And that's why, by the way, I'll talk about the competitive landscape in a minute, but the competitors have basically practically all of them disappeared already. It's not easy to develop this technology and to avoid all the pitfalls and find ways to bypass all these obstacles, et cetera. You have to have a very, very strong... [Technical Difficulty]

Let's see. Is that me or you, Coby? I can hear you, Andrew, but not Coby. Okay. Well, not good. There you are.

Yes, I can hear you.

It's okay. We just lost you for 5 seconds.

Sorry. Okay. I'll repeat the last part. So our Chairman, Dadi Perlmutter, his team invented the Pentum processor at Intel. He was #2 at Intel for so many years. We have Atiq Raza, who basically brought AMD made it from a smaller company to a big substantial player. We see where AMD is today. Of course, I have to give credit also to Lisa Su. But we have Nissan-Cohen, the guy who is the founder of Tower Semiconductor, one of the top 10 foundries in the world, who, by the way, has a PhD in nonvolatile memory. So we have a very, very strong technical side on our Board. Obviously, we have a strong Australian side as well. The management team consists of people who have been in this industry for more than 30 years with a lot of experience, well-known worldwide because you have so many disciplines here, and I won't go into all the details, but there's the PhDs, what we call the device guys who basically imagine this potential memory bit. We have the process guys who take that theory and map it into the tools that exist in the fab. You have the analog guys who take that bit that you finally know how to make and then they make an array out of it. And then they put all the control logic around it. And there's so much intelligence that goes into that. And then you have the digital guys, which obviously have to wrap everything with an interface to the digital world, and you have the test and characterization and so many different disciplines. And all of these guys have to work very closely together. That's the secret of success of Weebit. So really I think we have an amazing team. And that's why people always ask about the competitive landscape and how is that -- what does that look like? I mentioned TSMC as a manufacturer. And again, let me -- one more term that I want to make sure everyone understands. Semiconductors are really difficult to manufacture. They're very expensive to manufacture. And I have yet to find someone in Australia who gives me a good answer to the question. How much does it cost to build one fab, one manufacturing site for semiconductors?

I know because you told me last time, but it's in the...

Last time it has gone up. Now the latest fab from TSMC is USD 50 billion. okay, USD 50 billion for one building. Just understand that. So you can understand now why even the NVIDIA, Apple, Google, Facebook, Qualcomm of the world do not have their own fabs. It just -- even for them, it just doesn't make any sense. And so you have companies like TSMC, which are called foundries that offer manufacturing-as-a-service. There are very few companies that are big and strong and historically already had their manufacturing sites like TI or Onsemi. And they actually -- we refer to them as IDMs, which manufacture their own products, but those are the relatively rare. And if you look at the ReRAM landscape or at the nonvolatile memory landscape, I started talking about how everything evolved. I mentioned already the other technologies were too expensive and difficult and disappeared and MRAM. MRAM has its niche and people already started manufacturing with it. They're not going to stop very soon, and it's not easy to switch, as I said. But except for that, the market is really ramping up now on ReRAM. At the same time, the other ReRAM companies and you go look at who the market considered the large ReRAM companies 5 years ago, they're gone. They -- each one of them for different reasons, made different mistakes. They're just CrossBar, Adesto, et cetera, they're not there anymore. Even Intel and Samsung had big ReRAM projects and they stopped them. By the way, why? Not because they didn't have money, obviously. They have the money, they have the resources. They didn't have the management attention. This is such a complex -- such a challenging technology to develop. As I mentioned, all these disciplines of very, very strong people. You need to have very focused management that makes sure that they all work together, that they all cooperate, that everything flows. You hit so many roadblocks on the way, and you need to get over them. And for the big companies, they just can't have that senior management attention to it, and that's where they normally fail on that. So today, there's TSMC, I mentioned they're in mass production. Weebit is ready for mass production and hopefully get there very soon. On our side, we're ready. Now we're waiting for DB HiTek. But except for that, there are 2 other foundries that are trying to get ReRAM really out there. There's UMC from Taiwan and GlobalFoundries. They have yet to -- at least I'm not aware of any customer using their technology. I don't know everything, but it seems like they are still struggling with getting the technology to work. And that's it. That's really -- there are 2 players in China that are trying -- that are doing some things, but what we hear from Chinese customers is that it's not really there. Bottom line is Weebit is the only independent supplier today of ReRAM. I think that's the key point. TSMC, UMC, GlobalFoundries, let's be optimistic and say that they'll actually manage to get the ReRAM to work. Those are foundries. They will not give their technology to other manufacturers. So whenever you have another manufacturer who wants to manufacture ReRAM and now more and more, almost all of them already do, they really have one address. Now Weebit not only is the one -- I don't want them to come to me just because I'm the only one. I want them to come because they really think that we have the best ReRAM, and we do. We already, in many parameters, our ReRAM is the best. We are continuing to push very strongly the limits and the customers are forcing us. A company like TI obviously comes with its demands, et cetera. So we're constantly improving the technology. We have a very, very strong technology. Last year, by the way, one of the big achievements you mentioned, we did what's called AEC-Q100 qualification. AEC is the Automotive Electronic Consortium or committee, something like that. And they defined very, very extreme criteria in order to be eligible to go into the automotive market. We passed that test. We showed we can operate for 10 years at 150 degrees Celsius. That's not an easy task. We showed we have much higher endurance than Flash. And so today, I think that opened the eyes of a lot of people. And once you meet those criteria, you can go into any application. Now you look at -- by the way, people think, "Oh my God, you're going to be competing with TSMC, this huge giant and what's the future of a company that tries to compete with TSMC. So the very simple answer is TSMC is definitely not my competitor. TSMC, if anything, is a potential customer. I actually clearly think of them as a potential customer. And why? I mean, let's go back to the basics. A company like TSMC, what do they make their money off of? These guys manufacture semiconductors. They make money off of selling silicon wafers. That's what their revenue is. Why do they need something like ReRAM? Well, remember, their customers do systems on a chip. A company that does system on a chip needs all of those modules around their core before they can go to mass production. TSMC wants to remove all obstacles so that they get to mass production as fast as they can. So TSMC and the same goes for the other foundries. They have what we call an enablement team. They go and collect all of these IP modules and put them in a big library, I call it the big supermarket. So that when a customer comes and says, "Yes, I want to manufacture a TSMC, but I need a processor. I need a WiFi, I need an HDMI, I need this. They just go,"Oh here, this is for you, this is for you. Let's get to mass production quickly. And they realize people will want ReRAM. They put ReRAM in there. However, ReRAM is not their core technology or something that -- that's not their core competence. They have it there as an enabler. We already have customers who came to us and said, "Hey, we went to TSMC, we wanted to use their ReRAM. But they told us, "Hey, here are a few options, pick one. And the customer said, "No, but I needed something else. I needed something more specific for my system on a chip. And TSMC said, "Hey, guys, this isn't -- we're not a ReRAM company, remember? And the day will come when a big player will go to TSMC and say, "I want to manufacture here, but I want Weebit's ReRAM because it is better in these parameters that I need. TSMC, they won't mind. They'll only ask one question. How many wafers are you going to manufacture? Okay? That's all they care about, right?

Even if they pass all of the cost -- all of the fee through to you guys in licensing, I know it doesn't work that way. But even if they did, it would still be worthwhile them doing.

Yes. They have in their library their own SRAM. And then there are, I think, 3 or 4 other SRAM options in their library today. So I mean, there are already examples of this there. So yes, so Weebit is feeling very bullish. We are now basically the leaders in this market. I think on the technical side, we will very soon be recognized as the technical leaders because we really are pushing the limits in so many ways, getting better technical parameters than TSMC. Now the market can already see the big players are adopting us huge DB HiTek and Onsemi and Texas Instruments, people can see that TI, let's talk about TI and the deal that we did with TI. We were engaged with TI in 2023. I remember the last time we talked, I was already saying we're engaged with the vast majority or most of the big guys are already engaged with us, and it was absolutely true. And people said, wait a minute, you're saying that statement for so long. And when are you going to close a deal, but these deals take years to close. And TI, they got to the point where they said, wait a minute, Weebit is insisting on terms that we don't want to pay. And they went and they looked around and we just had to wait patiently while they went and I know some of the places that they went to look for ReRAM. Eventually, they came back to us and they needed ReRAM, they needed to move forward. We have a very nice deal. I didn't go into the commercials, but it is a very nice deal for Weebit, and we are engaged. They're a great partner to work with. They're a great company. They're a great partner to work with. They're planning to use us. I believe this type of company doesn't engage with you if it doesn't have a clear product interest, right? Yes. So we started already working with them. We're moving forward very nicely with Onsemi. With Onsemi, we signed the agreement in the end of '24. We already have wafers in our hands of working silicon. So we made really good progress in just 1 year. We taught them how to manufacture. We already manufactured the first batch. We already have some good indications. So things are looking really good. Yes.

So on that, Coby, I mean, 1 million questions. The -- it's a theme that comes up so often is -- and particularly with early-stage tech, you can say, look, look, it works. Here's all the lab results. But what you're saying to me reinforces what -- something I look for, which is the social proof. There is a customers want something that's derisked, not from what a lab report or a study will show, although that goes a long way. But when you say we have an agreement with Texas Instruments, our chips are being used in X, Y and Z product. there is something about that, which just feels as though it's a safer bet even if there's -- I know this isn't the kind of technology that someone makes in their garage. But even if there's someone over here in their garage go, look, what we've got is like they've got to go through all of the testing, all of the negotiation, then they've got to actually like have it out in the field that they can point to and sort of say, look, it works. We know it works, but now you know it works, too. So to me, that feels almost the most significant thing is -- I mean it's, of course, enabled by everything that got you to that point, but it's that, that sort of says this is legitimate.

Yes, definitely. In semiconductors, we call it silicon proven. There's a term for it. We say people want to see that something is silicon proven. They don't care until they actually see it in silicon, in mass production. And that's where we really struggled for so long. We had so many deals that we could have closed back in '23 or '24. We were ready. But the customers -- that situation where everyone is looking at and saying, we know this is the future. We know we're going to need it. We know everything. But someone else should be the first one.

Yes. No one wants to go first.

But then eventually, you have that someone else and things start moving because people realize that, hey, this is really a good technology. It is orders of magnitude faster than Flash, orders of magnitude lower power consumption than Flash. It has so many other advantages. It's cheaper than Flash. Flash is limited, cannot scale down below 28-nanometer. Talking about Moore's Law. We're down now -- NVIDIA is working at 3 and 2-nanometer, right? So people need a nonvolatile memory below 28-nanometer. That's where a lot of the advanced things are. That's where AI is, and we didn't even start talking about AI. But they need nonvolatile memory, and you just don't have Flash there. So that becomes a whole other issue. So we have so many advantages. And now that they see some of the big players starting to use it, I don't want to say everyone is there like, okay, no risk, let's just run forward. Every company has their own issues, has their own considerations about timing, about all kinds of other things. But I said many times, every deal makes the next one easier to close.

Not no risk, just much less risk.

Yes. The perceived risk level has gone down. It still is there, and then you have the issues where some fabs are saying, well, we're now at full capacity. We just can't look at this. We'll come back to you later or all kinds of other issues. But it's moving forward. We're feeling very bullish in terms of -- we have a great technology. We really do have a great technology. The situation in the market, as I said, we're feeling very strong now. We can insist on getting royalties. One of the things is that the big guys don't like to pay royalties and they start pushing back on you and come on, we're big this, whatever. You need to really go and insist on it, and we're strong enough. Our financials are finally showing that things are moving. As you mentioned, in fiscal year '24, the revenue was $1 million. In '25, it was $4.4 million. We're projecting more than $10 million for fiscal '26. You could look at the 2 last quarters and the cash coming in, we had -- in the first half of fiscal '26, we had roughly $15 million coming in.

Almost $16 million. I'm looking at it now, yes.

It's -- now I have to say not every quarter is going to be that amount of cash. The cash coming in is based on milestones. In some quarters, you have more milestones and some quarters, you have less.

It's a good point.

You can see the direction. I think what's important is to look at the bigger picture in our -- looking at the bigger picture and what the direction is and where it's going. And again, this is a strong company. We've spent the last year. The thing that people don't really look at are sometimes the things that are most important. We've spent the last year in building the infrastructure to growth. We realized that, hey, we're going to have a lot of customers in parallel, and we really need to set up that infrastructure. We cannot manage with Excel spreadsheets and whatever you need to put in. The real project management tools, coordinated, be able to say to the auditors, hey, this person put these many hours into this project and that many hours into another project and all of these things. So there's been a lot of work at all levels of building that infrastructure, including, by the way, in the Board level, and you can see that our Australian directors with Naomi and Templeman-Jones that joined. You can see the whole -- the whole way that the Board is functioning now is much stricter, much better governance, and we've been working on a lot of things that sometimes we don't really have to. I just told everyone, hey, let's just think ASX 100. We're going to be there. We might as well start planning for it and build the infrastructure so that we're ready for it. And there's really a lot going on. By the way, the biggest thing that we didn't mention, 2 things that we should mention. One is there are already product companies that are designing products with our ReRAM in them. The biggest revenue source in our industry is royalties. At the end of the day, the products hit mass production and we get royalties. That's when people always say that's when you hit the hockey stick. The clock is ticking. The clock is ticking because there are companies now that are already designing products with our ReRAM. Yes, it will take them time until they get to mass production, but it will happen. At least some of them will be getting to mass production and royalties will start coming in. So you can -- again, you can see actuals. It's not just like you said, it's not just presentations, it's not lab work, it's real life now. It's happening, and it's very exciting. And that's even before I start talking about AI.

Before we get to AI, I just -- I was going to -- you mentioned the financials. And while I've got them here as well, the -- so the balance sheet seems like it's in pretty good shape, too. There's $82-odd million that's there, which gives you a pretty good runway. One of the things that's limiting when you look backwards with companies and you're looking at a company that is in that transition from development to commercialization, what you see looking backwards is very different to what you see going forward in the cost mix -- the cost base and the mix? And usually, there's sort of like, as you said, there's a ramp-up in sort of sales and marketing and middle management and all the infrastructure you need to take the orders and manage customers. And there's a bit of a reduction or at least a less of a focus on the R&D and that kind of stuff. But it's something that never ends at the same time. So I guess my question is, how do you try and manage all of that? Do you feel as though the tech is sort of at a point now where you can really sort of prosecute the commercialization without going too hard on the R&D? Or is this a game of where the stakes are so high that if you take your foot off the accelerator on the R&D front, there is a risk that in 5 or 10 years' time, someone overtakes you. It's a hard question to answer because you don't know what's around the corner, but how do you think about all of that?

For me, it's actually an easy question because in our industry, the moment that you just start even slightly lifting your foot off the gas, you're already falling behind. So you have to constantly push forward. You have to constantly improve your technology. You have to constantly -- and there is so much more to do. One of the things that I sometimes see is -- and I won't talk about specific companies and whatever, but people don't realize even when the ReRAM is qualified and qualified means it passed all the tests, it's ready for mass production. The work doesn't stop there because customers want you to have faster, lower power consumption, smaller memory bits, lower voltages. It's like -- it's a never-ending story, and you need to constantly improve. And some of them -- they actually demand it that, hey, we're going to sign this agreement. You need to commit that you will have a ReRAM that does A, B, C and stuff. So we we are constantly moving forward. I mean, you can always throw more and more money at R&D and try to get even more. But -- and that's where really the balancing act comes in. We have a very experienced management team, and we look at the balance there. We try to invest as much as we can in efficient R&D and in ways to really improve the technology in parallel to pushing on the sales and marketing. Yes, it's not an easy task and sometimes we are getting closer and closer to the point where today, people refer to it as FOMO. When the customers will start realizing, hey, it's not that I don't want to be the first. I actually don't want to be the last. And at that point, everyone jumps on you and everyone is all over you because they all want to get the ReRAM before their competitors. They realize it's already working, it's proven, and now they just want to get their hands on it as fast as they can. So that moment is going to happen. Again...

And hence, bulking up the resourcing. Yes. I mean you want to be ready for when the phone starts ringing, I suppose rings more. Okay. We have to talk about AI before it goes too fast. So it's actually -- look, as I'm sure you can appreciate, it's been a big area of focus for the world, particularly in investing. We've seen some brutal share price reactions with concerns over what that does. So I think our audience really gets the technology or the significance of the technology. What they perhaps don't understand, Coby, is how does ReRAM fit into that? And why is ReRAM significant when it comes to AI?

Yes. So it's funny. I was talking to some institutions, and I realized they were thinking of Weebit somehow as some sort of a software company or someone who's going to take a big hit from AI and they were concerned in everything. And then I said, wait a minute, there's the companies that take the hit from AI. And then there's the companies who actually enable AI. Weebit is an AI enabler. Weebit is a key player enabling AI. And it suddenly dawned on me. People didn't really understand even that basic thing. So let me try to explain a few things that are happening in the AI side. And there's so much happening and Weebit has so much to do in so many domains, but I'll try to focus on maybe one, which is the key and the easiest to understand. So in AI, first of all, you have 2 key parts. You have the learning or training and then you have the inference. The inference today, people have pretty much reached the conclusion. It needs to be done in the edge devices. I mean the edge devices have all these sensors that are collecting data. If you try to send all that data to the data center, it takes a lot of time, a lot of power. It just -- it isn't practical. A car driving down the street and seeing something in front of it, you don't want to start sending everything to the data center and wait until it reacts and oh, the communication isn't that long, isn't that good, et cetera. So I mean, you just need a quick reaction. Is that a human or not? Do I actually stop or not? Now inference, AI as many advanced technologies, it's done already below 28-nanometer. It's done at 22 nanometer and below. And we mentioned flash just can't go there, period. So they need to keep the AI coefficient somewhere so that when they turn the system on, they have them, right? They need a nonvolatile memory. So what they do today is they have an AI chip and then they have a separate chip with flash where they keep the coefficients. And when they turn the system on, they have to copy all these coefficients into the AI chip and then they can only start running, okay? Now that means, a, you have 2 chips. So obviously, that's more expensive than one; B, the boot time until the system is actually ready to run is long because you need to copy everything from the Flash into the chip. And by the way, while you're copying, you're susceptible to real security risks because it's so easy to eavesdrop on that communication. So it's a big issue today, and there's a really simple solution, and that's what people are already working on today is just in that AI chip, you put a ReRAM and you can have all of the coefficients in it. So you have a one-chip solution, it's instant boot, you remove that security risk. You have a lot of advantages there. By the way, ReRAM is nonvolatile. Now that chip becomes nonvolatile. And if the sensors aren't sensing anything, you can just turn the system off and save battery time. A lot of advantages. I won't even go there are more advantages, but you can see where it's heading. So that's something that's immediate. That's something that is easy to do now. The big thing in AI, and you might have read about it, is called in-memory compute. So in-memory compute now, it's really -- you have so many headlines about it and everything. What does it mean? Basically, in a standard computer today, you have a strong processing unit and you have a memory and the processor keeps reading and writing data from the memory. Most of the time and most of the power are spent on that data movement from memory to the processor. Now when you are doing inference, you are basically -- and it's not -- by the way, not just in inference also in learning, but I'm focusing now on the inference chip. The inference is basically matrix multiplication. It's a very simple function that you do over and over and over again. Now for that specific function, you can actually rig the memory in such a way that the multiplication is done inside the memory and the data never needs to leave the memory. That gives you a huge advantage because now power consumption and speed are just proof. They're orders of magnitude better. That is something -- if you go back and you read Weebit's announcements from 2018, 2019, we've already been analyzing and learning it and working with research institutes about it on it and so on. So this is something we expected to happen. And now that it's happening, we're there. And we are working with companies on this. And this is not so far away. This is something that I believe will be ready for the market. You'll see it in the next few years. This is going to be a big boom on AI. And Weebit definitely wants to be a key player there, a leader in this field. And we just appointed one of the industry veterans as our VP, Systems and AI. He's now leading that activity. So this is something that's becoming a big focus for us moving forward. And I can't stop here because the holy grail really of AI is the day that the computer memory will just be able to mimic a human brain, what's called neuromorphic computing. And again, Weebit has been working on this already for 8 years with research institutes. We've been giving them our ReRAM so that they can develop their neuromorphic algorithms. ReRAM is an ideal solution for this. It's cheap memory. to manufacture. You can have -- you really have a big flexibility there. And a bit of ReRAM operates in a very similar way to a synapse in your brain. So it has that basic potential, and that's what people are working on. Again, there's a lot of more R&D and a lot more development, but these things are really moving forward now at an unbelievable pace. And it's really so fast now. And we definitely want to be leaders in this domain and key enablers to AI. So that's the big activity. And people ask me, so wait, what happened to the discrete chips? Well, there's just so much I can focus on right now. So we do look at still at technologies for the discrete chips, et cetera. But the big focus of Weebit is, first of all, there's this huge vacuum, huge vacuum in the embedded space. As I mentioned, we're basically the only independent provider of ReRAM for systems-on-a-chip, and this is a huge market, and I just want to capitalize on it. I mean the nature doesn't support vacuum. If you're not there, someone else will eventually fill it. So I want to be the first one to fill as much as I can. And the second thing is all of this AI stuff is just mind-boggling and there's so much to do there. So we have a lot to do, and it's going to be a very exciting year for us.

Yes. I mean I definitely get that sense. So I don't want to end on a downer at all, but I mean, what are the challenges and the risks that you see? Because there's just so much work behind you to get here, big opportunity in front. But the world of business in general, technology, in particular, is a lot of companies that get right to the edge and then just fall at the last minute. And sometimes it's sort of own goals. Sometimes it's just things out of left field. What are the kind of things that you kind of -- that you all really alert to, to kind of think, we can't muck this up. We're right here now. What are -- you know what I'm getting at? Like what are the working?

Definitely. I told my team that so many companies have failed in the moment where -- I call it the tornado when everyone jumps on you at the same time and wants that technology. And that's the peak of the success, but you're not able to answer that need and everything fails and people talk about the website crashes, right, and all that. So we have been preparing for that, and that's why I was talking about last year building the infrastructure because that's where a lot of companies actually at the peak of your success, you go under. And I definitely don't want to be in that place. I think we've done amazing work already to set up that infrastructure. We hired some key people with a lot of experience in this domain. By the way, I'm really proud of our VP Customer Success. We hired someone who was the manager of a fab. So I mean, nobody knows what the issues of a fab manager are better than her. And she is doing an amazing job at working with our fab customers and understanding their needs, addressing them and et cetera, again, resolving these issues because many times, you just don't understand what the customer is asking for. You're trying to give something that is not what the customer really needs, not what they say they need, but what they really need. There's a big difference sometimes. And so we've been really working on that. I think that's the biggest issue right now. We are already at the point where the technology works. So the biggest concern is now behind us. We can show the ReRAM works. Our ReRAM works. It works well. It does AEC-Q100. Big companies that did significant evaluation ended up buying it or paying for it already. But hey, my Chairman is Dadi Perlmutter from Intel and the motto is only the paranoid survives.

That's great. I love that. It's such a great saying. I love that saying.

And Daddy keeps reminding me of that. So yes.

I mean it's a great mantra. So do you feel as though that the team -- I mean, we will always grow. And as the business continues to scale and have more success, you would just grow naturally. But -- in terms of sort of like the real key gaps that you had sort of seen maybe previously, do you feel as though that infrastructure layer that you've talked about is really starting to bulk up to the point. I guess what I'm trying to get at, there'll be a lot of people watching this, and we all love to model, right? It's what we do as investors. We try and extrapolate trends and the rest of it. What does the cost base look like? Or maybe more generally, it was sort of like what's the nature of growth for the cost base as best as you can estimate it, not in a percentage way, but like what's the gut feel as to how that sort of evolves over the next 3 to 5 years, say?

The company is definitely going to have to grow. Obviously, with more customers like Onsemi, TI, et cetera, you have these big technology transfer projects. Now obviously, we finished DB HiTek, so that team can work on TI or on other stuff. And we move people around, but still we'll have more and more companies in parallel, and we will need to grow. We are working very hard on automating, on improving our internal procedures. Like I mentioned the VP Customer Success. She came in and she set up this whole template that now every time we have a new customer, we just duplicate everything. Everything is there. Everyone knows where everything is. Everyone knows the naming conventions. Those are simple things, but sometimes those are the critical things, right? So we did that. We're automating a lot of things. We're developing memory compilers and other stuff that we will do things in a more automatic way. So yes, it's -- you asked me, I'm working very hard now. And even when I'm on vacation, I'm on a semi vacation because there's just too much happening and you just need to really keep things going. I have the most amazing team. I mean, literally, that is the secret of success, a team that when we do hit the obstacles because they're always there and many of them, the team knows how to bypass them, knows how to work together in order to resolve the issues and move on. The risks are always there, but I just believe that when we'll hit them, we'll know how to bypass them. And the potential is huge. And yes, it's just really exciting.

It is definitely exciting. I'm definitely picking that up and understandably so. I'll let you go, Coby. But I guess what's -- a good final question is, because you obviously speak to a lot of investors, a lot of stakeholders, and I'm sure you get a lot of the same kind of questions. But what do you think the market tends to miss most? Like if you were to sort of leave investors with one insight that's sort of like what they need to focus on, what's the North Star for someone who's outside of the company to keep a track of to really sort of get a gauge of momentum and progress? But no one looks at, that is underrated in terms of something to watch?

Well, I mean, I'm sure you use the term, do your own research a million times. And if someone just starts doing some research here, they can see -- I gave you the story, but it's a story that you can go and verify all of it, right? I mean you can see who are the players in ReRAM right now. What is the competitive landscape? Where is every company and so on. And go and check it out and see what's going on out there. I think that's a key element. Look at what analysts are saying and look at the analyst reports that have been coming out. For so long, analyst reports were very, very cautious about ReRAM. And all of a sudden, now they're showing this big growth in the expected market share of ReRAM. And yes, look at what Weebit has already announced, and you can start seeing where this is heading. So I think all -- being the -- really literally the only independent provider today in this market is something that I -- I mean, you need some luck in life, right? It's not -- but I look at the other companies that just -- they made mistakes, they were trying to use the wrong materials. They were doing things in the wrong way. They invested a lot of money and ended up not getting very far. We've been extremely cautious in our spending and in the way we control things. And yes, I mean, overall, the data is -- we are in a really fast-growing market. Analysts are expecting the ReRAM market to grow 45x in the next 6 years. 45x, people get excited about...

Not 45%, 45x.

Yes. So I'm saying people get excited about 4x, and we're talking about 45x, right? It's mind-boggling. And we are the single player. We are the enablers now for AI. Yes, I think you need to look at all the data and make your decisions, right?

I think that is -- it is excellent advice, and I will double down on that. I think if -- it's true in life in general, but particularly in investing, if you're going to follow the crowd, you're going to get the average results. You literally have to think independently and for yourself. Otherwise, what are we doing here? So I think it's a great advice, and it's a great place to end it.