For Australia, so we will be in the conference in Sydney. Jumping right into it, with you guys, I can skip a lot of the intro slides. If this work. Oh, but did you do that or did I do that? Okay. Oh, no, here. I managed to do it. Okay. I am skipping the slides that you guys all know. Okay. Yeah, excuse me for getting the logistics working. I am skipping all of the intro since I think everyone. Now it is working. Now, oh, it went back. Okay. I will dive straight into the recent achievements. That way you guys will have more time to ask questions and have an open discussion. We just released our quarterly report, as you know, in Q1 of the fiscal year, basically release the cash and not the revenue was another AUD 6.1 million in cash receipts this quarter.
I always say, I will repeat many times, cash is nice, and it goes up and down. Cash is based on achieving milestones in the projects, et cetera. What really counts is the revenue that we recognize over time. The blue line there is the revenue. There were these rumors or whatever spread by ABC about us going bankrupt and so on and so forth. I do not know how, with AUD 164 million in the bank, and when the whole fiscal year of 2026, the net burn was less than AUD 20 million, how you can have those kinds of things going on. But that is how these things work. Anyway, Weebit is very strong financial position. We have the cash. We have the revenue. Things are moving forward, and as you know, we announced the use of proceeds when we did the raise.
A lot of money is going into continued intensive R&D, continuing to improve our ReRAM. We need to constantly. I sometimes say we are running a marathon at sprint speeds. We just need to keep running as fast as we can, and this goes on for years, but improving the technology and doing all of the work on AI. I will be talking about all of that as we move forward. The commercialization efforts, we have three main customers with. Is this a pointer?
Nothing.
Oh, yeah, it is. But it is very hard to control, so forget it. Did not mean to do that. Anyway, with DB HiTek, we know they have been working for quite some time. We are qualified. They are now promoting us to their customers. We have initial customers that are designing their products with our ReRAM in them that are planned or even being manufactured at DB HiTek, and we talked about the fact that some of these customers already have silicon in their hand. They are running software, constantly making progress towards that level of mass production. This is really with DB HiTek, we are in a very good position. Onsemi, we are making good progress. Whoever remembers the timeline that we presented in one of the previous AGMs, it is roughly order of magnitude of two years to transfer the technology and get it qualified.
Well, we're getting close to that two-year mark, and they are already working on the qualification. They have working silicon. Onsemi being an IDM and being a company that actually has its own product, a manufacturer that has its own products, and you can imagine they don't engage with a new technology if their products don't want it, right? They've been working in parallel with their products, and there's really good progress there. TI, which is more recent, we have obviously kicked off the technology transfer and making progress there. Each one of them is progressing according to the plan. We're very really expecting to see onsemi get qualified and TI. Now, these customers, and this is something that we already announced last quarter, but I think it is important to emphasize.
We had two of the big customers, we unfortunately can't always name who exactly we're talking about, but a lot of these things you can understand on your own. They expanded after they already worked with us. They learned the technology. They saw it from the inside. They worked very closely with our team, and they saw the quality of our team. I often say the milestone of having an existing customer expand agreements and add more business units and more products and more things, that's really such a strong endorsement. They saw us, they saw the technology, they've been working with it, working with the team. When they actually extend the agreement and expand it to additional business units, that's, for me, the best endorsement, and I think the industry has seen it as well. Really great.
I won't go into the details of the slide, but I think that's an important milestone for us, seeing this thing works. Yeah. When we talk about product customers, because at the end of the day, all of these manufacturers, this is really the infrastructure that we need to set up. We need to get many manufacturers on board. But at the end of the day, the big revenues will be coming eventually from royalties from the product customers, having many of those that can manufacture and move forward. One of them has already announced, and they revealed themselves, Overlord. Overlord is relatively a smaller startup, but the people who set it up, like many startups in the U.S., are industry veterans, well-known, with a good track record, and they have some really strong customers behind them.
They're working on a chip, and they jumped in relatively early, working with DB HiTek. They already have samples, actually, so not only do they have working silicon and they're managing to run their software on it, they even started giving initial samples to their customers. At the end of the day, you need to remember, this is a long supply chain. DB HiTek manufactures the Overlord chips. Overlord sells those chips to systems companies who actually integrate them into a system, and then they sell them to consumers or industry or whoever. There are several steps. We will eventually get royalties once product companies like Overlord sell their chips to their customers, and we'll get royalties on that. That's, in terms of the business model, how it works.
I really think, and I think I said this in the past, the number one milestone in the life of a semiconductor company is when the product goes into mass production. At that point, you cannot have any more excuses. We still sometimes have someone say, "Oh, yeah, but you qualified, you did all these great things, but it is still on a smaller number of chips," or blah, blah, or whatever. People like to make up all kinds of things. But when you are in mass production, I mean, that is it, right? You are there, and all excuses are gone. We are really looking forward to that. Obviously, we do not control, I think it is important for everyone to understand, we do not control when a customer goes to mass production. It is not our milestone.
It is the customer that decides when they want to go to mass production, how they want to do it. It is their product, and they are selling it. We get the royalties. We are hoping to see them go into mass production. We know their plans and whatever. But at the end of the day, it is not something that I can commit to, that Weebit technology will be in mass production at a certain date.
We can just work with the customers, support them, push them forward as much as we can, and we hope, we expect or believe that they will manage to do that, and they will do it-ish in 2027. Details, people always want me to commit to this more, which is really not something that I can commit. We have another customer that is a little bit behind Overlord, and they are already moving forward.
They have working silicon. They are running software already, et cetera. There are more and more products now ramping up. This list right now, we just have three customers here, but obviously, there are more products that are being designed and having our ReRAM embedded into them already. We do hope to start seeing a pipeline of these guys, and obviously getting them to production. So that it actually works. Okay, continue. Okay. Overlord, I already talked about them. Their application is a battery intelligence platform. It is a very smart, sophisticated power management application. It is going to go into all kinds of edge AI devices. It actually is a very interesting technology. Their target for production is 2027. As I said, we are really supporting them so that they can achieve that. Okay.
Another highlight, back in AGM last year, I announced three main targets for 2026. It was revenue, it was customers going I forgot what the second one was. The third one was to have a first AI customer. We are glad to say that we now have companies that want to use ReRAM for AI, and I will talk about the AI application in a minute. But in general, people are now coming and actually engaging already with Weebit on the AI technology. I will talk about it. This is still at the level of they are becoming customers, but they are mostly studying the potential of the technology. It is important to say there are customers, but it is not mass production yet, and it is always important for me to set expectations properly, and I will get into that in a minute.
Those are the recent news, the things that have been happening, and where are we headed. We talked about this. This is a slide that I showed in the past. We are struggling with the situation in semiconductors. We are not making excuses or anything, it is just that is how life is. Every several years, there is huge demand. This time it is really more than I have ever seen in the past. For semiconductors, most of the fabs that we talk to are telling us that they are between 1.5 and 2 times the capacity that they can actually manufacture.
Half their customers are dissatisfied that they are not getting their products, and when a company has so many mad customers all over it and then just yelling and screaming, "We want our chips, why are not we getting our chips?" The last thing that they can do now is start focusing on a new technology. We really are seeing that all the time. We are talking to these guys.
Good news is, and I said it many times, not a single one is telling us, "We do not want" Well, let us say the general phrase, "We do not want ReRAM." They are engaged. We are making very good progress with some of them. You know that these negotiations take years, and it is just moving a little bit slower than we expected, but we are definitely making progress with more manufacturers. The plan is to obviously have more of them on board.
Looking at the foundries and a little bit at where we are. When we deal, and this is in general because many people are asking me, what are the timelines, and how do we look at it? We did, I think it was AGM 2023 that we showed a detailed timeline of how these work. Just to remind everyone, when we are engaged with the foundry, it takes us time to close the agreement. When we sign the agreement, roughly a year to do the technology transfer to get the first chips taped out and start testing. We then have a long period of test and qualification and modifying the designs and things like that until we finally get to the qual. Even when we get to the qual, it never ends. You are constantly improving. You want to improve the yield.
You want to improve what is called bit error rate, or the both the—y ou will see it here in a minute. If you are talking about a customer of a foundry, they normally want to see the foundry is completely progressing and the technology is in a good position. Many of them will just wait until after qualification. A few which are more daring, like Overlord, are willing to take more chances and engage with us earlier. Then they need to go into the design. The design takes, depending on the complexity of the chip, anywhere between 6-12, even 18 months, really, the work on the design is quite intensive. Then you have the test. Apologize for all the technical issues. Okay. Are the people online are online? Yes. Okay, she is Okay. Let us move to the next slide, and focus.
I mentioned, and I think, some of you guys might have seen the slide already. When we did the raise, we said we are going to be focused on improving the technology, and this is really a race of constantly moving forward. You cannot ever stop improving the technology. Good. Okay. Yeah. Can you go to the next slide and put it in full-screen mode so we can see that? I gave a general guideline to the team. I said, "Hey, guys, whatever technical parameter you are thinking of, we need to be at least one order of magnitude better than any other ReRAM in the market." Very simple. Let us just focus on that. Obviously, it is easy said, not so easy to do, but the team is very focused on it. And we are working hard.
To give examples, if it is endurance that most of the industry is still at 10,000 going to trying to go to 100,000 cycles. We qualified at 100,000. We are now looking at doing 1 million cycles. Access time, having Yeah. Okay. This one just have it at full screen. Hopefully, we will manage to finish this presentation. I hope I am done. Yeah. So, okay, you can leave it even like this. So access time.
For sure. Most of the industry, the read time is greater than 20 nanosecond. We are already pushing down below 10 nanoseconds. And again, it does not really matter what these are. Just so you understand, Weebit is very focused on having the best ReRAM in the industry and on constantly being ahead of it. So, let me let you to I think if you just hit pitch down, it will be easier. This one? Okay. Let us go already to this one.
We are losing time, so let us just skip to this one. So of course, we always start by talking about embedded memory and where we are, and then I end up saying, "And by the way, there is also that little thing called AI," and Weebit is now very focused also on the AI side. I mentioned in previous presentations how there is this move from near memory compute to in-memory compute to neuromorphic. In-memory compute is something that is happening now. It is really getting a ReRAM embedded into the chip so that it sits next to the processor, very close to the processor, but still next to the processor. And having the processor read the coefficients from a ReRAM, that is something that today's technology, it is available, and people can use it, and people are looking at adding it and moving forward in that direction.
That is something that basically there is no special R&D to do, and it is available. The big thing is going to in-memory compute. Now, let us go to the next slide. Okay. And people always like to challenge me in presentations to see how I will manage to deal with all of these surprises. So nobody can beat Eran on these things. But anyway, let us talk about AI. Let us focus on it.
AI has two main activities. AI, there is the training, and then there is the inference. Training in AI is basically done once per system. You learn whatever you needed to learn, and you develop those coefficients. Inference is done all the time, again and again and again. You take those coefficients, and now you use them in real life. So as you can imagine, in the early phase of AI, most of the work was learning.
Now, most of the work is inference, and as we go forward, inference will become the much bigger activity in AI. You can see that this year, it is actually for the first time, inference is already bigger than training. By the way, that is in data centers and of course, in the Edge devices. It is really growing very fast, all of the inference activity. For Weebit, that is great news because inference, the thing that you are doing all the time is you are reading data. You are not updating it. That is the training, has to generate the data, and there is a lot of read and write and everything. When you are doing inference, you are basically taking the coefficients that were developed by the training, and you are reading them most of the time, and that is where ReRAM shines, in the reading part.
We can read, and we talked about access time now and how fast you could read, et cetera. That is really a very good application for us. We have been working on it for some time. You have heard it before, and we are making very good progress. Before I go into the details, last time, I showed you an initial version of a video. Let us put the video on because I think it really explains things, and now it has more information in it. It is the next slide in the current release.
On the video down here, but it is not working. We need to go to close scene.
Oh, you have it already posted? Wait a second. Wait, click here on the link. Oh, you did it. A very nice visual.
Effort.
Solutions.
The speaker should be from up there somewhere.
It was with him.
Artificial intelligence is rapidly becoming part of everyday life. Services to edge devices, vehicles, factories, and robots. Behind every AI application are enormous numbers of computations involving large amounts of data. In traditional computing, data must constantly travel between memory, where information is stored, and processors, where computations occur. As AI models become larger and more complex, moving data consumes more power and becomes a growing constraint on AI performance. As AI workloads continue to grow, so do the power, cooling, and infrastructure needed to support them.
What if instead of moving the data to the processor, we move the processing to where the data already lives? In-memory compute performs calculations directly within the memory itself, dramatically reducing the need to move data. Emerging memory technologies such as ReRAM enable memory arrays that can both store information and perform the computation AI inference relies heavily on a type of operation called vector matrix multiplication. At each stage of a neural network, incoming data is multiplied by a matrix of learned weights. The learned weights represent that. It is a. Here's the catch. Conventional memory can store the weight matrix, but it can't perform the calculation itself. In ReRAM, cell specific conductance level. The input values are applied as voltages to the rows of the memory array. Within each cell, the input is multiplied by weight, resulting in a proportional change in current, which senses conductance.
That's why ReRAMs are in place for the calculation at the fine grains. This is significantly energy efficient while enabling more capable AI systems. This enables more powerful AI in systems where power, performance, and real-time decision-making are critical, from vehicles, factories, to robots, aerospace, and defense. Intelligent edge devices. As AI expands from cloud infrastructure to intelligent edge devices, in-memory computing has the potential to become a foundational technology for a new era of efficient AI computing. Weebit Nano, bringing a new memory technology for the AI era.
Okay. Again, I really apologize for all the technical issues. First of all, this video is online and on YouTube, and you can watch it again, listen to it, but it really explains what happens. Actually, for the people who actually came here physically, you're going to enjoy seeing a real model of what it looks like. I'll pass this around, but basically you have here the memory, the vectors that go in, the voltages. It's really nice. You can actually see different levels of resistance in each cell, and then it comes out and it's summarized. You have basically the model of what happens in in-memory compute. You guys, I guess I can pass this around and if you guys want to take a look. Let's go to the next slide.
Really, what we see here is the fact that we can do this in-memory compute with ReRAM. We have now initial customers, as I said. They already are looking at this closely, analyzing, seeing how they can actually use this technology. We still have a lot of work to do. Again, I emphasize there still is quite a bit of R&D to make this into a real commercial product or technology. There still is work, but we're making good progress. We have good results, as you could see already before. This is a big focus for us. I think we've wasted already too much time on things. Really, maybe I'll stop at this slide and I'll open it for questions even. But we had three targets for 2026. The revenue, and as you know, that we achieved significantly over the initial expectation.
First product customers to tape out already not so long ago, but it just happened this year, and they're moving forward and already, as I said, the first one even having some samples for customers. Now first AI customer engagements and bringing them on board. With that, I think we'll stop with the technical issues and just open it to questions. We'll start with people here in the room, and then if you have people online also. From me. That is by far the number one priority of Weebit, and it needs to be very clear to everyone we're not taking our eye off the ball. We need to have that work.
We want to take over that market, become a major player, and generate a lot of revenues. AI still requires R&D work, and it's further down the road, and we can't afford to wait for that for everything, so we are focused on the embedded. Having said that, AI has a huge potential. It might even be much bigger than the embedded market for us. So there is a focus on it. We have separate dedicated teams that is focused on AI activity, but the number one priority by far continues to be generating revenue.
You talked about the bottleneck in the semiconductor industry. What's going to break that bottleneck to bring on that greater adoption over the next nine
I'll handle that one.
You think that's a question?
I'll hand it off to you.
Sort of bottleneck in the semiconductor industry, which is slowing down the adoption.
Right.
What do you think is going to break that's going to encourage the-
A few things. First of all, everybody understands there's this under capacity and the demand is much higher. For that reason, all of these foundries and fabs are now building more capacity, and this capacity is going to come in the next 12 months to 24 months. A lot of it in Taiwan, in China, in Korea. In the U.S., of course, there's a lot of new capacity being built. Once this capacity is available, they'll have to fill in all these fabs. The foundries that we are talking to and highly engaged with are those that are doing these expansions. They need to show their customers that their next-generation capacity has more technology included, so that's where our embedded NVM really shines. That's going to be the next thing.
By the way, I think it was two years ago, at the Semiconductor Australia Conference, I presented a slide, and I think I also showed it in this forum, that back then, there were $800 billion invested in building new fabs that year. You can understand, people already realized that this demand is coming, and they started building the fabs. There are many fabs in the process of being built, like Eran said. You can read in the news about all the activity in Arizona and in Texas now, and all of these big fabs that are being built. Eventually this capacity is coming. It is coming online, and as it comes online, that situation will go down. By the way, I will tell you, since I have been in this industry more than a couple of years, and these cycles happen all the time.
We see that it is not anything new. Every roughly four years or so, we get to the point where the demand is crazy, and then more fabs come on board. Now, if you can try to imagine, demand is going up like this. The capacity is stairs. You have a certain capacity. At a certain point, the demand is above that capacity, and then the capacity jumps because new fabs come on board.
You are now in a good position. You can supply the demand. But then, the capacity stays the same and the demand keeps going, and you end up being under capacity. This happens all the time, and that is how this industry has been working since I have been in this industry for 46 years, and it has always been this way. This time is much crazier. I have to say, I have never seen anything like it, but it is happening.
By the way, part of the reaction of customers, when you are sending in an order and you get only half of what you ordered, guess what you do next time? You just send double the order, right? To a certain extent, part of the over-demand is because people are taking this into account, and you have that feedback effect, and people are ordering more. Then when they get more, if they start getting more than they actually needed because they ordered more, they start stockpiling. Then at a certain point, they stop ordering because they already have some chips in their storage house, right? It kind of goes crazy a little bit, and then it starts stabilizing again. But all of those effects happen and we are seeing more capacity coming on.
Yeah. Okay. Yep.
This is a reputational question. The recent issue we've had with the media here, there's been very strong anti-Israel seen through a lot of the Australian media. Are you going to take some proactive steps with your public relations to counter that? In other words, to counter the attack before it comes? Because this won't be the last time, in my opinion.
I don't know if the ABC note that they made was because we're Israeli or not.
I know. I know. Believe.
As a public company, we need to be very cautious about any claims we make and what Some people said that it was connected to the rise in shorting, that a lot of people, shorting suddenly jumped, and then all of a sudden you have this. I don't know. We have no proof. I can also say, being Israeli, all of this crazy, fake nonsense that goes on in the news drives me crazy. I just can't believe intelligent people actually believe all of this stupidity that's just all over the place. As a public company, we need to focus on our work. We're not going to be the ones who start fighting political fights or things like that, and I really have to do that separation.
I remember even there was a point where I started making some comments on LinkedIn, I got all these requests from shareholders, "Please, your posts are on Weebit, we follow them, please don't add political posts." They were right, by the way. We have to make that separation. On the personal side, I can talk about it's just totally insane, but at the—
I suppose—
—company level, I can't really do much.
I suppose what I was suggesting is not necessarily political, but to be on the front foot with your relations, to put yourself out there, not to do with the politics, but to do with the positive aspects of the way the company is operating. That's what's important at the end of the day.
I agree, that's what we're trying to do. Actually, the fact that I'm coming out to Australia four times a year to do these rounds, to talk to people, present Weebit as it is, right? I think that's the best answer, you're absolutely right, we keep doing it. Eran is here, the Automic guys are here in U.S. as well. That's what we try to do.
We talk about shorting. I'm sure you guys looked at the register. Who's been shorting for the last one month?
Well, unfortunately, shorting is the type of thing, when people hold the shares, they need to reveal themselves. Shorters don't reveal themselves. I don't understand why it's that way. I can tell you, I have a lot of questions about all of the regulation around shorting. But the biggest issue that I have is people can short you never know who they are. They can hide, they don't have to reveal themselves, you just don't know who they are.
Does it show on the register?
No. That's the frustrating part. You know who hold the shares. You don't know who loan them to short.
Yeah.
Very frustrating. Yeah.
Just one point about the recent media speculation. I thought I saw a note, a correction note, a week or a few days after that actually said that although the analyst claimed that Weebit's risk of bankruptcy has gone up, it's actually from a low base. Also, the other point that they made was that basically still within the safety.
In general, I'll say, and the question was about the ABC, again, the ABC thing, and I don't want to take it out of proportion because it was just something that didn't make any sense. We actually, of course, took the action and asked them, "How did you reach the conclusion that Weebit is getting closer to bankruptcy?" They did post a clarification, and they did make some excuse. Let's say it this way, to me, nothing that they said makes sense. Again, the simple numbers that I showed, when you have AUD 164 million in the bank, and in the whole last fiscal year, the net spend was less than AUD 20 million, you're not near bankruptcy, okay? I don't care what anyone says, you're not near bankruptcy. And the revenues are going up, et cetera. And we just had AUD 6 million this quarter coming into the bank.
Let's put ABC aside maybe, and not give too much.
In question about the director selling a few weeks ago, or a couple of months ago. I think there was a lot of directors who just sold out.
Um.
$78.
Well, I think, first of all, we need to remember, directors, just like anyone in this room or on the video, they are people who have lives, who have all kinds of events happen, and sometimes they need money. In Australia especially, you don't have what's called 10b5-1, which is a very basic thing that I would expect boards to have, stock market to have. Directors, they can't actually sell in the future like you can on Nasdaq or any other market. When you sell, everyone kind of thinks, "Oh my God, what's happening?" But they have the tax events. They want to buy a house for their kids or whatever. They held, and for many years, no director sold. For many years, they held. The share price went up. At a certain point, it was a good opportunity. Some people sold.
I don't look at it as a major. By the way, people, I don't know how many people noticed, about a month ago, Dadi and I bought, right? The share price suddenly dropped. We looked at it. We looked at each other. We said, "Wow, this is a great opportunity for us." Dadi and by the way, for us to buy again, because we can't ever sell. The thing is, when we buy, selling becomes such a huge thing that it's such a huge challenge. But still, we looked at it and we said, "Wow, this is a great opportunity." People didn't notice that side of things. But again, I can't recommend anyone else what to do, but I took the decision that I thought was right.
Well, we've got some questions about, from now to end of March next year, are you expecting more product companies or any other major announcement that you guys like in the process of e nd of negotiation, anything from now to end of March?
I can't. Again, I try to be very cautious. Maybe some people say I might be too conservative. I don't like to issue forward-looking statements when I don't actually These deals go up and down and whatever. Do I want to sign agreements? Do we have some agreements in the pipe that can materialize theoretically? Yes. Will they materialize? I can't give a commitment that they will materialize. We are engaged now definitely with more than one, two, or whatever. We're engaged with quite a few manufacturers. We're engaged with quite a few product companies pushing forward. We announce it when we close the deals because these things, even the TI deal, the negotiation lasted for so many years. It had its ups and downs. I was sure that it was going to close two years before it actually closed.
I was confident I would've signed a paper on it. It was ready for closure, and then it took another two years to close. It's very hard. I know many people are asking me, "Can you give us some sort of pipeline? Can you give us some sort of prediction?" The number of customers, potential fab companies in the world is not that large, and each one of them, the negotiation is so difficult and goes up and down and whatever. Trying to give this type of guidance, it's very risky, and it's very difficult. I know it's frustrating. I can tell you right now, we are engaged in quite a few of these negotiations or evaluations, or whatever you want to call them. Do some of them have the potential more than one to close in whatever number of months? Yes. But will that potential materialize?
I have no idea.
Can you comment on the federal grant that was supposed to close in 2025 that just slipped into 2026? Is that still look positive, or is that kind of bottom falling?
Yeah. So that one actually, that's a great example of an agreement that I was sure we were going to close rather quickly, and then the whole situation in the market changed, and that's one of the companies that we're engaged with.
That's still in advanced negotiations as it previously was?
That one actually, like the TI one, when it suddenly I was sure I was signing it, and it went back, we have these ups and downs in all of the negotiations. It's not something special with them. Right now, with that specific company, it's progressing less fast than I expected. Others suddenly accelerated that I didn't expect to.
It's still progressing, nonetheless?
We're engaged with them, let's say it that way.
The two AI customers, is that purely speculative? Not speculative, sorry. Is it purely exploratory, or is there revenue behind it as well?
At this point, they came on board because they wanted to study the technology more and see the potential for the future. They haven't committed to actually do products based on this technology because we still have a lot of R&D. That's why it's important for. I think we need to maybe also give these guys a chance.
Sure. I've got a bunch.
Oh.
Step through them?
Yeah, go ahead.
Some of these are some of the more favorites too, Coby. Some of them. Do you anticipate tape out at TI this year in 2026?
Tape out? Part of my challenge, by the way, and I said it earlier, these are internal milestones of the customers that I cannot always commit with, especially when you are talking about the bigger ones. I will get in trouble if I will make public what their internal milestones are. I cannot commit. I can only refer again to that two-year timeline, and we are progressing basically well with TI. I cannot make a comment on the TI milestones.
Okay, next one. In regards to new fab partnerships signed, you have identified that current production facilities are running at capacity and therefore delaying progress. How do you model how far into the future these capacity constraints apply, and can you get in front of the situation to minimize further delays?
I think, without going into too much detail on these things, we know this market, and we know these customers. I can tell you with some of them, we are making good progress now. There was a period when everything seems to stall. Now we are again in a position where some of these discussions, negotiations are progressing nicely. I cannot commit to when they will close.
Next one. What is Weebit's relationship with Tower Semiconductor like? Do they have their own ReRAM? Their website makes reference to a ReRAM that is either available now or will be available in the future.
Well, first of all, Tower is an Israeli foundry for those who don't know. Actually, I think—w hat percent of our employees come from Tower?
25%.
I think, yeah, something like 25% of our employees probably, I'm guessing, come from Tower, including, you know that we have Yoav on our board. Let's see. Ilan, [Divas], Ishai—
Raz.
—and Raz. Four of our VPs came from Tower. So we have a very good relationship with Tower. At the same time, whoever follows the news will see Tower made a big shift in their focus to photoelectric type of—
Yes.
—products and manufacturing the photonics. I am sorry, photonics. Right now their focus is on photonics and ReRAM doesn't fall in that interest level. The comment about ReRAM, whoever knows the history of Tower knows that they have a very close relationship with Panasonic. If people noticed in the past, I did talk about the Panasonic ReRAM that UMC is basically trying to make work. Nuvoton acquired then from them. Tower, because of that Panasonic relationship, have whatever old ReRAM. I don't think they are offering it in reality to any customer. I definitely don't think they have any customer for it. Next one, the two new AI customers.
Yes.
Can you give any indication of the size of it? Is it small or sizable? They are small companies. I don't think we Do you want to comment on them or?
No. We cannot disclose any names, but these are smaller companies, looking at the in-memory compute. This goes back to the video that we are trying to show here. It's a new domain. It's a very promising domain. There are other technologies in other A lot of vendors look this in-memory compute market. These AI customers are a part of that ecosystem.
And—
It's very difficult for now to say how big that market is going to be, but the promise is just insane. It's 10x, 100x better in terms of the power and latency compared to what we see today. It's a very big promise.
I would clarify though, the deals right now, because they did them for They did pay us. That's why we call them customers, right? But at the same time, I don't want people to think that it was millions. They did pay us. They did take the technology to study and to see how to progress with it, et cetera. But like I said, we still have a lot of R&D work to actually make this into something that can become commercially available.
Next one, Coby. MemSilicon recently taped out at 1 GB , one 1T1R ReRAM chip at 28 nm CMOS. Would you be able to provide some commentary on that development?
MemS, that's the Chinese company that we saw. I have to say, I wasn't really aware of them before, and I am learning now more what they're doing. It looks more like academic. From what we saw, we saw things that we know the academia is coming to work with. We haven't seen any of this yet anywhere near the industry. So I don't know how much this is like what we had eight years ago when we were playing with Rice University. So yeah, we're still studying.
You've commented that foundries are at production capacity. Is it likely that the next foundry group signing calendar year 2026, if not, when?
As I said earlier, I really do not want to commit to a timeline. We are making good progress. Some of these can potentially close in months, but I really am not at the point where I feel.
Why were new fab ReRAM signings not included in this year's targets? It seems announcements just keep getting pushed out.
Well, it was not intentional in this sense, but actually, I guess I did not put that target because the markets changed on us and suddenly all this capacity issue came up. On the one hand, I think Weebit does try to give meaningful targets. The targets that we had, I think, were important ones. So yeah, every year before AGM and now before the AGM next month, we are starting to look at what kind of targets do we want to set for calendar year 2027. We try to pick two, three important ones that kind of signify the progress. But yeah, after a few years that we talked about number of deals, we decided to take another no special reason.
Given the bottlenecks, TI, onsemi, et cetera, also taking a slower approach to ReRAM adoption, or are they trying to create a gap over their competitors having already started the process?
I would say it's probably the latter. Once they engage and once they have. They paid us a lot of money already, and they set up these large teams that are working with us. They're already committed. They are moving forward, again, without specifying who and what, but I did mention that important customers have been expanding their deals. You can imagine. You can try to guess which ones and what and I think they're very committed and moving forward nicely, both onsemi and TI.
Very specific question here. Are you able to estimate the number of production wafers that could be shipped next year to some customers, and the approximate royalties per chip?
I wish. By the way, we do our internal estimates, but as I said, first of all, going to production is a customer decision, not ours. You can see how much control we have over some of these things. We do have our estimates. We know what our customers are planning and when they plan to go to mass production. We have our analysis, but I definitely can't share it.
Ben, I do want to go back to the previous question on existing customers, DB HiTek, onsemi, and TI. For these customers that already committed to our technology, for them, keep in mind they are running under capacity. They cannot fulfill their demand, and by adding a resistive RAM, they can actually add more value to their wafers and therefore charge more money. If you have a fixed amount of wafers you can ship every year or every month, how do you grow your business? You charge more. How do you charge more? You add more. That's what we are seeing with these existing customers. They actually want to add this technology onto their next wafers and add more revenues there. So that's—
That's a very good point. Yeah.
Thanks, Eran. On competition. Competition-wise, many things have changed since the early days—
Yeah.
—since you took a seat as CEO, and many competitors no longer exist. Who or what is the biggest challenger to Weebit these days?
Yeah, that's a very good comment. When I joined Weebit, to be very honest, I looked at Weebit and I said, "We're like number three." Number one for me at the time, when you looked at all of the press, and [Infravision] was Crossbar. They were all over the place, making so much noise, already going and talking to fabs to get the technology manufactured. They raised $200 million , et cetera. Number two was Adesto. Well, both of them are not there anymore. The Crossbar technology ended up in the hands of a Chinese company called InnoStar, and they're doing some side things. We're not really seeing it successful. Adesto is, I think, officially in the trash, the technology. Yes, the market really changed. Today, the number of companies engaged on a ReRAM is very small.
To the best of my knowledge, and I have to always add that qualifier, I am only aware of TSMC and Weebit that have qualified ReRAM. I do not know of anyone else. Even the Chinese, and there are three—
Two and a half maybe.
Three. This Chinese company that you just mentioned, we never heard of them until a couple of months ago. I do not know what they are. So we do not really know a lot about some of the stuff happening in China. I can tell you that outside China, there are, I have been public about it, we know GF or UMC are still struggling or trying to make their ReRAM work. Might be some other efforts that we are not aware of. There really are not too many.
Just got a couple more questions. I think we are running out of time here. So I think I know your answer to this one. So realistically, are SkyWater dead in the water to us?
I would say almost. I am an optimistic person. I like to think SkyWater changed their focus. Now they are focused on R&D services. Okay? As you know, Weebit is benefiting from that because we do a lot of R&D now at SkyWater as well. I do not think that there is a big potential of things changing, but they have been acquired now, and in the new environment, maybe something will change. It is not like we are not talking to them, and we are trying to see if there is any potential of getting them to offer our technology in whatever limited foundry services they can.
Going to the final question. What timeline do you expect Weebit's in-memory compute solution to be commercially available?
I think I mentioned today we can see at least a couple of years of R&D before it really becomes something that I will feel good about offering as a commercial technology. We are developing. We do expect to have more customers engaged with us, actually paying us so that they can actually have access to our technology and study it and move forward. But to have actual products that go to the mass market, there is still a lot of work. On that, we do have technical challenges we need to overcome. On the IMC side, in ReRAM, our ReRAM works, and we know it is qualified and the level of technical challenges, I think I can say it is negligible at this point. We constantly are pushing it forward to new frontiers, but basically it is working. IMC, we still see some technical challenges before a commercial solution.
Where is the other ones? I am just conscious of time, so just back to you.
You control my time.
I know.
Okay, any last question here from the audience? First of all, thank you. By the way, one request from you and from everyone that is online. Next month, I am going to be back here for the AGM. You might recall that in the last AGM, we received a strike. We have an issue with the proxy advisors not always understanding how the worldwide, the global technology market works and so on. We had discussions with them. We did change things. You will see it. The notice of meeting should go out, I think, the end of this week, if I am not mistaken. You will see that there are things that we changed also in the remuneration, et cetera. The bottom line is, I really need the support of everyone so that we avoid a second strike.
I do want to encourage all of the people here and the people that are online to vote for us in the notice of meeting. I know many people ignore it and do not really look at it. This year, it is important that people take a look, and it does not take a long time to vote and give us your support. Thank you.