Oh yeah, just I guess we can probably kick it off. So thanks everyone for coming and everyone that's online. I understand we have quite a few people online. So welcome and I apologize that this presentation was actually released just a few minutes ago, a technicality, but I'll kick it off and of course open to people to ask questions, etc. The introduction slides I'm going to skip. Everyone here I think knows the theme and knows the key things and the history of non-volatile memories. We have an updated analysis from Yole that came out more recently. No big surprises. You can see the size of the non-volatile memory constantly growing every year. They add to the prediction. You can see here RAM definitely making the large market share, 50% in their expectation. Now by the way, yeah, RAM is just cheaper to manufacture and everything. When you look at the number of wafers, and this is what you have on the right hand side, you can see the number of wafers is actually much, much larger. It's just that ReRAM is cheaper and that's why it's only 50% on the left. But I think this is really showing what we're seeing, what we're feeling in the market. It's very much in line with what we're seeing at Weebit. Advantages of ReRAM. Again, I think all of these things you guys know and I don't really have anything to add on this slide. So, you know, we'll talk about what's been happening more recently. And again, going mainly to the right hand side of the screen. So, you know, obviously we did the raise back in March. We announced that we have three product companies tape out and I have a separate slide on that. So I'll focus on that progress and we just recently in the quarterly announced expanding the agreements with key customers. We don't have that many key customers so everyone can do their guessing. I'm not allowed to mention names but this has been a very important milestone for us.
And I guess the result is, you know, looking at where we are and the 4C that just came out and the announcements, we are in a very strong position with 167 million in the bank. So definitely enough for several years moving forward. We upgraded our guidance. So with all of these expansions that we've done with the key customers, we upgraded the guidance obviously in the yearly report in the end of August. That's when you're going to see the final number, but it's been very bullish. So you can see the line over there. Of course the cash and the revenue are, you know, we're showing you the current status. Anyone who does any extrapolation to the future, it's, you know, we're not doing that here. So you can see the numbers. I always repeat the revenue numbers are much more important than the cash. Cash depends on achieving milestones and some quarters you achieve a big milestone and there's a lot of cash, some quarters there are less milestones so cash will go up and down and change, but I think the revenue is what's really important because that's something that we recognize over the life of the project in a much more organized way.
So I talked about the three product customers taping out and I think our shareholders already know what tape out means, but I want to expand on it a little bit. So first of all, just kind of, you know, trivia question, why is it called tape out? When I started in this industry many years ago, we actually would prepare a tape, in those days those were the storage devices, and when the design was ready for manufacturing, we actually put everything on a magnetic tape and we shipped the tape physically to the manufacturing facility. So that's why it's called the tape out. I want you to understand why tape is so important in our industry. You know, many of you might know more, you know, the software industry where, you know, you write your code and then you say, okay, you know, I'll compile it, run a few tests, oh, I have this bug, I have this issue, no problem, I go fix something, recompile, run it again, recompile, run it again, and you go through that, you know, you can go through that tens of times in a week because it's so simple and easy. In semiconductors, it's not easy and it's not simple. When you send to manufacturing, first of all, manufacturing, you know, if you pay a lot of money to have it really go through the fab fast, you get it in three months. If you have a more normal cycle, it's four, five, six months. So it takes a lot of time. It's not a few minutes that you compile and you have the results. And it costs a lot of money because you have to pay for silicon wafers, you have to pay for the masks, you have to pay for everything. So unlike software, in the semiconductor world, when you get to the point of tape out, you actually have gone through very, very intensive tests of your design. You actually are testing for power, for speed, for voltage, for, you know, a million different parameters and making sure that everything is working properly. So when you get to tape out, the chances that you have bugs are pretty slim. You know, I won't say zero because obviously there are always some bugs, there are always issues, but you're in good shape. So what this slide basically means is a product customers believe ReRAM is going to give them value and they actually went through the whole process of signing up with us and paying license fee, going through the whole design cycle and going through all the testing and everything and getting to the point where they actually shipped the design. Today it's not tapes anymore. It's through the internet to the manufacturing facility to manufacture.
We actually already said that one of those chips came back from the fab and is already running software. So it's in a much better shape. Now this is a very important milestone, but it's a very important milestone on the way to what is the number one milestone in the history of a semiconductor company. The most important milestone that you can have in a silicon company is getting to the point of mass production of your first product. And that's our goal right now, to get to that mass production, what in the industry we refer to as silicon proven. Because until you get to that point, people can still claim, yeah, you have silicon and it's working, but it's probably under very specific conditions and maybe you have all kinds of issues and blah, blah, blah, whatever. When it's in mass production, it's in mass production. You know, all of these kind of background noises disappear. So we're very focused on getting to mass production. You know, we're not giving indications. It's really confidential information of our customers when they're expecting to get to mass production. Excuse me. But we are on the way. You know, no matter how you do the math, you can see that in 2027, we should be getting to mass production. And I think that's kind of the important thing. So this is an important milestone for us. Three products, it's already not just one, you know, is it or whatever, it's already you can start seeing that it's becoming more systematic and we're very proud of this milestone. And actually one of these customers is already willing to be named. Like, you know, this is a classic Silicon Valley type story. It's a younger startup, you know, a relatively small company, but the people who started this company are very experienced from what you know from the industry, known, and they're doing a battery intelligence platform. So that's what their chip is. They're using our ReRAM in it and, you know, they're targeting first silicon, actually they already made it public so I can talk about it, they're targeting to have mass production in a year. So we're working with them and making this happen and, you know, this is very exciting for us.
Now the big thing is that we have the existing customers expanding usage. So for us this is a great indication, you know, when we're talking about the existing customers and we're talking about key customers, again, I'm not allowed to mention names, but I'm sure you guys are realizing who they are. These guys have started working with us. They saw the initial results. They're happy with the results and now they've been expanding to additional product lines, additional business units and this is what really pushed forward also the raise in our guidance once we started seeing the results of this. So, you know, they sign up, you know, they sign the initial license. We work for a lot of months on technology transfer. They build confidence and as they build that confidence, they get to the point where they expand the agreement. Remember, you know, our key customers have many product lines, many, you know, their product list is pretty significant, you know, and now it goes really into all of these, a very strong partnership, strengthening the partnership with new applications, new products in additional business units. And eventually what this really shows and it shows that to the market is there's a very strong customer endorsement because they wouldn't be expanding their agreements with us if they wouldn't be happy, if they wouldn't be seeing the value of our ReRAM. And obviously this leads to larger opportunities and increasing the revenue, etc., and strengthening the long-term relationships with them.
Now, where are we heading and what's happening? So, I think you guys know, you know, the slide. I've already been showing it a few times. The, from the Crossing the Chasm book. We've signed up with the early adopters and now we're kind of in what's called the bowling alley where we're still working on one customer at a time. We're still focused on, you know, knocking one pin at a time, but, you know, and I'll talk in the next slide about exactly what's happening with all of these deals. You know, we are expecting to get to the tornado. You know, my basic feeling is we're getting much closer. Mass production obviously is a key trigger to that. But I, you know, before I go into the detail of where we are on these deals, let's first of all understand something that's happening in semiconductors and the semiconductor market is undergoing huge transformations right now. Oh, I think many of you are following the Weebit share price, but if you guys are following what's happening in the semiconductor market, you can see that Weebit is just following the trend of the rest of the market. You know, the AI demand is huge. Okay, it's huge. I've never seen something like this before. Now in our industry, again, to get you to feel this industry, demand for semiconductor is always growing. You know, it's always, you know, this line that goes right and up. And that's been true, you know, since I joined this industry 46 years ago already. Now at the beginning, you know, there's enough capacity in the fabs and everything's fine. But as the demand grows at a certain point, you cross the available capacity and now suddenly there's over-demand, there's not enough. Everyone goes crazy. The companies start building new fabs and that's a step function. You know, you suddenly have a lot more capacity. So you're addressing that demand again. Everyone calms down. Everything becomes normal again until that line crosses the new capacity level. And again, you have the same thing and that normally it's about a four-year cycle, three, four-year cycle that, you know, every three, four years we get to the point where there's over-demand and everyone, you know, scrambles and builds new fabs and so on. What's happening now is the demand because of AI is way more than what we've ever seen before and the scrambling and everything is much crazier. Now you see here the revenues of the semiconductor industry. Just at the end of last year, if you would have asked anyone in the industry, the prediction was that the industry will reach a trillion-dollar level in 2030. That's what everyone believed. Guess what? We crossed it already. We crossed it. The expectation now is that this year we will already get to 1.5 trillion. So doubling the industry this year. And that's a huge thing. I mean, it's not the normal growth that people were saying, okay, 800, 900, whatever. It's jumping all of a sudden. The demand is crazy.
And what happened to Weebit is we signed an agreement with TI. We were sure that, wow, you know, now everyone's going to start moving faster and everything, but now all of these fabs are so swamped with customer demand. They're going crazy. They're saying, leave us alone. We know we need ReRAM. We know we're going to be using ReRAM, but right now we just have so much to work on. We can't pay any attention to a new technology. We can't work on it. Every second in the fab needs to run wafers that are generating cash. We can ask for these huge margins now. The prices that they're asking for are just unbelievable. And they're getting these great margins. So we're not going to start running wafers with test runs of ReRAM when every wafer that runs through can generate so much cash. So it's kind of a crazy situation. The market is going crazy. By the way, I can tell you unbelievable stories that I heard from people in the fab, from managers in the fab. They tell us a customer came and gave us a check for a billion US dollars. He put it on the table. He said, "Cash it now today. The money is in your bank. Just commit to give me all the wafers that I need." They had to throw the check to the trash. I mean, it's crazy. It's crazy when you think about what's going on right now. So for Weebit, what this means is that suddenly, you know, nobody has attention span for us. We're kind of trying to move forward. We're, you know, everything's ready, but the market is just not a fast place. So, but this is like the slide says here on the bottom right, it's a temporary situation. I mean, there are so many fabs being built and we will get back to the norm and the market will stabilize. So right now we are really focused on building our infrastructure and preparing. So, you know, if we're not closing the deals now, what can we do so that when the market comes back, we'll be ready and we'll be able to run as fast as we can. And that's what we're basically focused on now because, you know, that this technology transfer takes two years and it's a lot of work. We are now looking at how do we improve, build our environment so that we have a lot of, and you can see it on the bottom right for the people on the video, on the bottom right, reuse. We're building a lot of prototypes and a lot of things that we can use later so that when we sign up with customers, we can ramp up faster and not start from zero. So there's a lot of work on building a library of templates in order to enable us to deal with that. We are working, you know, looking at the top left, at strengthening our leadership. We talked about that when we did the raise. We are today, I believe, already the best ReRAM leader so that when people think about ReRAM, they won't even think, they'll just say Weebit. And even when they go to TSMC or whoever, they'll just say, hey guys, Weebit is so much better, we want to use Weebit. So we're working a lot on this. That was one of the three main things in the use of proceeds that we talked about. Working on this, looking at future, you know, big steps, not just improving the technology, but really doing things, you know, you know that there was work on the selector and everything and looking at advanced ReRAM technologies here. So there's a lot of things that we're looking at for, you know, for the future. Obviously automating and building, you know, streamlining our infrastructure and everything. So we are now in the mode, we raised money, we have the money, we can really strengthen our infrastructure, we can really build that gap and of course focus also on AI, etc.
Now when I talk about building the gap, you know, I decided, you know, I was hesitating, but I said I will put one more technical slide. I won't go into the technicalities and it doesn't matter what these things really mean, but I want to give you a feeling for what we're doing. So, you know, the market today, for example, the access time, the read speed, the common read speed is 22, 25 nanoseconds. By the way, this is what Weebit has right now. However, we've been doing R&D work and we found ways, really creative ways, to bring the access time to below 10 nanoseconds. And that's a huge thing when you think about, you know, all these systems that need to run as fast as possible, AI and other systems and whatever, you know, access time is critical for them. And this will really enable us to expand, especially in the microcontroller market, which is a huge market, but also in other markets. So again, what access time exactly means and what it means 10 nanoseconds, you can just see the gap that we want to build. The same thing goes with endurance. The market today, the standard is 10,000 cycles. We already qualified at 100,000 cycles and we are now on our way to a million cycles. We know how to get to a million cycles. It still requires a lot of work. That's what R&D is focused on. But again, there are applications where having stronger endurance is critical. You know, when you talk about temperatures, the market today is, you know, in the 105, 125 degree level. We qualified AEC-Q100. We already showed that we know how to work at 150 degrees and we're even demonstrating already 175 degrees. So we're, you know, this industry always moves forward. If you stand in the same place, you're basically falling behind. And we want to do exactly the opposite. We just want to keep running as fast as we can. We want to improve the industry so that anyone who's trying to develop ReRAM will just realize that no matter how much they work, they'll never catch up to Weebit. And that's the key message. Nobody will, no matter how much they invest, they won't be able to catch up to us. The same thing here, you know, automotive architecture. Right now, kind of the common knowledge or understanding is that if you want to go into automotive, you need to use an architecture which is 2T2R, basically two ReRAMs per bit. I won't go into why and how. We know how to do it with a simple 1T1R and we've been demonstrating it. That's how we qualified AEC-Q100 on 1T1R. People sometimes have difficulty to believe that, but we actually did it. So, and I won't go into all of the other parameters, but I just want you to understand Weebit is constantly paranoid, constantly running forward, constantly improving. Not only will we not let anyone catch up with us, we're going to broaden the gap so that even the customers of these foundries that are trying to develop ReRAM or have ReRAM, they will just go and say, "Hey guys, you know, we know you're working on your own ReRAM, but we want to use that." And that's the big goal here.
You guys know the potential markets. There's a lot of potential in different markets. I've shown this slide already many times. But let's talk about, of course, AI because you have to talk about AI and Weebit is focused, as you know, on AI. Yeah, one of the very nice things is that the same ReRAM that we're working on right now is actually applicable for both embedded standard NVM and also what's called memory-centric computing, both, you know, near-memory compute, in-memory compute, neuromorphic and all of that. So the technology that we developed is applicable. We might end up, and this is definitely a potential at a certain point, trying to develop a version of this which is even more tailored to AI, to in-memory compute, etc. But the base technology is already good for it. We can already move forward. We can already proceed and ReRAM really is a great technology here. Flash can't do the scalability. It can't go to the smaller geometries. So people are doing these things, trying to do it with flash, but, you know, the scalability is really hitting. You know, MRAM can't do in-memory compute because of the way that technology works. So ReRAM is really the, you know, the kind of immediate response that can address all of these domains and we're very excited about all the work that's going on on AI and, you know, looking forward to being able to announce some of these things. Now many people have been asking me, you talk about inference at the edge, but what about data centers? What's going on with data centers? So ReRAM, first of all, it's a technology that is really tailored for inference. So whoever knows in AI, you have the learning and then you have the inference. Now learning requires a lot of reads and writes and whatever the endurance levels and stuff are way too much, you know, we're not looking at that. But the inference side, that is actually what's growing the fastest, even when you look not only in the edge devices where basically AI is inference, but even in the data centers, this light blue is the AI inference. You can see that even in the data centers, the fastest growing domain is AI inference and it is rapidly becoming the biggest share of usage in the data centers. So we are very focused on inference. We're very focused on in-memory compute and making it more efficient and this potential market in the edge devices and also in the data centers is just, the potential here is huge and we definitely plan to capitalize on it. Now I want to help you understand a little bit. We talk about in-memory compute a lot and so the team prepared a small video kind of to explain a little bit what exactly is...
