Weebit Nano Limited ($WBT)

Welcome to another episode on the couch with myself Henry Jennings from Marcus Today. And in this episode, I'm really happy to be joined by Coby Hanoch from Weebit Nano. Happy and more than once because obviously times are tough in Israel. And Coby at the moment is as he said to me a refugee in South Africa with his mom. So I'm really delighted that he spare the time today. And talk to us about what's going on in storage chips and RAM storage specially DRAM storage. Which is going to be big performance in the last few months. So welcome Coby thanks every much for coming on the end.

Thank you.

Now many of you will know Coby from previous podcast, but Coby, I have to say, you don't look like you've got 45 years' worth of experience, Coby. You still look very young. You're obviously doing something right with your life. But Coby is a very experienced professional in semiconductor and related industries, previously a CEO of PacketLight Networks and held VP worldwide sales roles, a variety of companies. So very well qualified and holds a Bachelor of Science from -- in system design from the Israel Institute of Technology. So really looking forward to this chat. And as always, this is general advice only. So please do your own research, contact your own financial adviser regarding any of the thoughts, ideas or insights. So firstly, Coby, it's good that you're somewhere safe and your family, I hope, is safe at the moment with the current situation. Can't be easy running a company in Israel, I guess, at any time, but now especially difficult, I would imagine.

Yes. It's definitely a challenge. Unfortunately, we're used to these challenges, I guess, since October 7 and since the day I was born, we've been in these challenges all the time. So we did -- like the rest of the industry is set up for this. The employees are working now mostly from home. Most of them have shelter rooms in their homes. And so they can just focus on work and go to the shelter whenever the siren goes off. Our lab is in a shelter room itself. So the employees that go to the lab are protected there. It's definitely not a normal situation, but we've set up everything, and we've done all the business continuity plans and all the other procedures for how we operate in these situations. So I think I can clearly say it's not affecting the work. And if anything, the Israeli industry is always super focused on delivering and meeting the expectations even more during these kinds of times.

Sure. Well, I hope everyone stays safe there at the moment. Now Coby, it's fair to say that you guys, Weebit Nano has had a pretty good run of recent times since we last spoke, which I think was probably about a year ago. And a lot's been happening. There's been -- what's it, RAMageddon, is that what we're calling it with the shortage of ReRAM and storage. What's happening in the chip and the memory market at the moment?

Well, I think you heard the term AI, right? So.

Yes. I remember that.

There's that little thing there.

There is.

So I mean, AI is the type of thing that no matter how much memory you throw at it, it just needs more and more and more and the demand is crazy. But I do want to be exact. I don't want to be misinterpreted. There are 2 main markets. When you talk about memory, there are really 2 main markets or application areas, and we need to be very clear here. There's what's called discrete memory, which are memory chips that are 100% memory. And that's right now when people talk about this whole crazy situation, most of them are focused on the discrete chips on what Micron and SK Hynix, et cetera, are doing. This is a domain that Weebit does plan to get to, but we're not there yet. The other domain is embedded memory, which also has huge demand right now because AI, many of the AI chips are actually looking for embedded memories, and they too need more and more memories. And so the demand is on both sides. But most of what you hear in the papers and everything what you read in the papers and hear in the news is about the discrete. But we are very, very happy to be in this domain, which has huge demand right now on all fronts.

And your share price has been going pretty well in the last year or so, what was it $2 or something, and now we're up to close to $5. So things have been going pretty well. What's been happening in the Weebit Nano world?

So we've really met a lot of very important and critical milestones along the way, technically, commercially and financially. So I think we can say that on the technical side, the biggest milestone was about a year ago when we announced that we achieved AEC-Q100 Qualification. What that means is that we have shown to the market that we are meeting one of the toughest tests in the industry. We showed that our memory can operate 10 years at 150 degrees celsius. So really harsh demands, and we showed very strong endurance results with that. And that really demonstrated to the market that this technology is ready for production that we're not just talking, but we actually have real results. If you can meet those demands, you can meet basically any application demands already. So on the technical side, I think that was the most important achievement. On the commercial side, at the end of '24, we signed an agreement with onsemi, which is a major player in the power management field and the automotive field. And then at the end of '25, we signed an agreement with Texas Instruments, which is already the #1 analog company and a very, very strong player in automotive and in many other domains. So those were 2 very, very important achievements, not only because of their financial or cash impact, but because of the impact on the market. When people see these kinds of names that are using our technology, obviously, they -- it makes an impact and it makes it easier for other people to decide to work with us. We also, on the commercial side, started signing up the first product companies. So Weebit actually has 2 kinds of customers. And I'll remind quickly for those who don't remember how it works and how the business model works. We work on the one hand, we license our technology to companies that manufacture semiconductors. We give them a manufacturing license. We bring them up to speed on how to manufacture it in their facility. And we then also license it to the product companies, to the companies that embed our memory into their system on a chip. And then those product companies go to the manufacturers and manufacture there. So we -- TI and onsemi are actually a very rare case of companies that are called IDMs. They manufacture their own products. But we have been working with companies that are just product companies and they're designing our technology into their chips. And that's a very important milestone because it means that the clock is now ticking towards getting royalties and royalties are really what you want to get to when you're in a IT company. And then, of course, there were the financial results, which you could have seen in terms of cash coming in, you can see the cash reports. And in the last half year, it was about $15 million coming in. You can see the revenue numbers that are ramping up. So overall, we're feeling very bullish. Weebit is in a very, very strong position in the market. We're the only independent supplier of ReRAM. And the demand, as you can understand already from the opening of this talk, the demand is really high, and it's a very exciting time.

So Coby, in terms of Texas Instruments, what's the next step there? Is it just more of the same in terms of commercialization there and royalties? Or what's the next step there in terms of the long-term revenue potential?

So when we engage with a company like TI, we basically go into an order of magnitude of 2-year project of ramping them up on how to manufacture our technology, manufacturing the first demo or test chips, then we go into testing them and verifying that everything is okay. And eventually, we go into qualification, which basically puts the stamp of approval that we're ready for mass production. So right now, we're in that phase with TI. TI being an IDM and obviously, a company that has its own products that is going to manufacture, their product teams are obviously involved, and we work with the product teams also so they understand how to embed our technology into their products. So there's a very intensive project going on with TI. Eventually, we will reach qualification, and we will reach mass production and, of course, the royalty side and everything. So that's an important project. But as I said, in parallel, the impact of this agreement is not just the direct cash implications or revenue implications, it is the fact that TI is a very well-known name in the industry, a highly respected name. And so other companies are taking notice, and that's also an important part of it.

Sure. Now you've guided to FY '26 revenue of around $10 million. Is that right? What do we need to see happen in the second half to make that a reality?

Well, we need to keep executing. Weebit is very conservative in the way that it recognizes revenue. Some companies, as soon as a payment -- a license payment comes in, they recognize it on the spot and they have these big spikes of, oh, wow, we've got so much revenue, I believe in being much more conservative and methodic. So Weebit recognizes the revenue that will come from a project like TI over the span of that project. So the license fee, the NRE, we actually recognize it over the span of that project. I mentioned 2 years, order of magnitude or whatever. So with TI, with onsemi, with DB HiTek, it's -- with all of them, that's how we recognize revenue. And that's why I believe our revenue will be much more consistent and with less bumps along the way. So we need to continue to execute with all of these guys. Finish the technology transfers, get to mass production eventually. And the revenue will be recognized as we progress with these. Obviously, we need to sign more. And the whole idea here is to bring on more customers so that we can see the revenue continuing to grow.

And you're highlighting multiple agreements with product companies. How close are these to meaningful commercial revenue, do you think going forward?

So the product companies, it's actually in a different way, it's kind of similar to what we do with the manufacturers. A product company, when it decides to use our technology, it then goes into the design phase and it needs to embed our technology into their design. Now it really depends on what the product is. Some products are relatively simple and you can finish the design maybe in half a year. Some products are much more advanced and complex and it can take 1.5 years or 2 years to design. But I would say, on average, a year to design and then they need to go and manufacture the prototypes and they need to test the prototypes and they need to get to the point where they feel that they're ready for mass production. And no matter how you look at it, you end up again with that magical number of about 2 years from the moment that they engaged with you until they reach mass production. So we're following up with the first ones. We're signing up more along the way. And we're really working on getting them to the point where they can go into mass production. In parallel, of course, we're pushing forward, especially now with DB HiTek that we already achieved qualification. So on the technical side, Weebit has achieved the qualification with DB HiTek. Now DB HiTek needs to go through their internal process to actually announce mass production, but we're supporting them there as well. And hopefully, we'll start seeing actual products going -- taping out and then going into mass production later as fast as we can.

It's funny, isn't it? We think of technology is happening so quickly. It seems that every other week that NVIDIA is releasing a new chip or there's some new advance in the AI chatbot world or the large language models. But it does seem if things move slowly elsewhere. And I guess when you're gearing up for full commercial production, that's inevitable that there is not quite as huge a rush, I guess, to bring it through to production. Is that the case?

Yes. I know it's an anomaly. People think of HiTek as this thing that, oh my God, every year, we have a new version of the iPhone or we have a new version of this and everything is moving forward so fast. The underlying infrastructure is so advanced. And so complex that it actually is very, very difficult to move forward and the manufacturing process. Again, I gave these -- in the past, I used to give these examples so that people would understand the complexities of semiconductors. The places where you manufacture semiconductors, the fabs, they need to be something like 100,000x cleaner than where you do open heart surgery, okay? Just think about it, 100,000x cleaner. They cost many, many billions or tens of billions of dollars to set up. The manufacturing process is so delicate and so difficult and the machines there are just -- they do magic. So the whole process and everything is very, very, very difficult, sensitive and you want to bring in new technology into it, you need to be very careful about how you do it. The manufacturing process itself, by the way, is normally -- I mean, when I want to manufacture a new chip and I send it to the fab, it will take between 4 and 6 months to get it back. I mean if I really pay a lot of money and rush it and whatever, I might get it after 3 months, but that's already very extreme. So it's not easy to manufacture semiconductors. It's not software that I click a button, I compile it and after whatever minutes or an hour, I get a new result. And it is challenging and people need to understand. This is this is something that takes time. On the other hand, once you are in there, and that's something that people should understand. When someone signs an agreement with Weebit, when a manufacturing facility signs with us an agreement, this is for decades, okay? This is literally an agreement for decades because just like it's so hard to bring us in, it's almost impossible to take us out once companies start manufacturing products with our technology. In the automotive industry, many countries have a requirement that you keep spare parts for 17 years. So people will need to continue to manufacture these chips for a long, long time. Many of these chips do stick around for a long time. Once companies learn how to use ReRAM and embed it into their design, it's hard to move to something else. So it is hard for us to get into the market and you can see how long it's taking us to close these deals and to move forward. On the other hand, it's very important to understand once we're in, we're in, and we're in for a very long time. And as I said, the agreements with these fabs are agreements that last decades literally.

And in terms of your memory, the DRAM and the AI uses that it can be put into, what's the focal point? You talk about car parts being kept for 17 years. What's the focal point, I guess, of the commercialization of your storage chips?

So I guess 2 points just to clarify. First of all, our technology is called ReRAM. DRAM is another one. So just to make sure that we don't confuse people.

I'm getting the RAMs confused.

There are many different types of RAMs and DRAM is actually a volatile memory. So it's a completely different type of thing. But ReRAM has many advantages. So in general, before I go into the AI side, in general, practically every application needs a nonvolatile memory. At the end of the day, any application has what I call the magical moment that people turn it on for the first time. And that device needs to have some software on it, needs to have some data on it, and that has to be in a nonvolatile memory. Up to now, people have been using Flash, but Flash has a lot of limitations. So our ReRAM is orders of magnitude faster, orders of magnitude lower power. It is cheaper to manufacture than Flash. It's very important to remember, we have that advantage as well. We can go down to the small geometries, et cetera, et cetera. We have a lot of advantages over flash, and that's why no matter which application you'll talk about, we have a list of must-have reasons why ReRAM is better than Flash and why we should use ReRAM. So in the automotive space, I already talked about the fact that we can operate at very high temperatures. If you look at IoT, we're very low power consumption. We're cheap and cheap to manufacture. So obviously, for battery life, also, we consume much less power. Again, medical, and I had a medical company that said, we do this medical device pacemaker. And we don't want to operate on people just to replace the battery. So it's kind of -- they wanted that lower power consumption and making sure that the people will live longer than the battery. So different applications have different demands, but we really address them well. Obviously, now the hot topic is AI. And ReRAM has a very big advantage. On the one hand, it is an analog type memory, which means that it can -- it actually can store multiple values and not just a 0 and 1 if you work on it properly. And it is cheap to manufacture, and it can go down to the small geometries. So in AI, AI is the more advanced designs normally. And those designs go -- most of them are done below 22-nanometer, 22 and below. Flash cannot operate there. So you have that basic issue that if you want to have a nonvolatile memory on your chip in an AI chip, you normally just don't have an option of flash. And you need ReRAM. Now there are -- I normally look at it as a 4-stage solution. The initial stage is happening already right now, and it's very straightforward. People realize in AI, there are 2 main activities. You first teach the system something and then there's the inference where you actually use that knowledge. The inference is -- I think today, there is a consensus already. It needs to be done in the edge devices. It needs to be done in all of those devices that are out there. You want fast reaction. You don't want them to send the data to a data center and wait for a response, et cetera. The edge devices are chips that are done at less than 22-nanometer. And they need to have those -- the coefficient of -- that are used in AI in order to do all the computations in a nonvolatile memory. Today, what people do is they end up having a 2-chip solution. They have a chip for the AI stuff. And then they have a separate Flash chip where they keep the coefficient. When you turn the system on, you need to copy everything from the Flash chip into the AI chip. So you start off with the fact that it's a 2-chip solution, which is obviously more expensive than 1 chip. You have a situation where the boot time takes long because now you need to copy everything into the AI chip. And while you're copying, by the way, you have a security issue because people can ease drop on that communication very easily. There are a lot of limitations there. If you would put a ReRAM into that AI chip, now you don't need the Flash. You have a one-chip solution. It is instant boot without that security issue. You make that chip now nonvolatile. If the sensors aren't sensing anything, you can turn the system off and save power and most of these things work on batteries. There are a lot of advantages. And this is something that is pretty straightforward and things that are -- that people are looking at today. But the really -- the hot topic today in the industry is what's called in-memory compute. The thing about computers is everyone's been using what's called the von Neumann architecture for decades, for many, many decades now. You have a processor or some compute engine and you have a memory and the processor needs to keep taking data out of the memory and operating on it, et cetera. Now all of that data movement between the memory and the processor, that's where most of the power consumption goes. That's where most of the time is spent. And that's a huge waste. In the case of AI, most of that -- most of the computation that is done is actually matrix multiplication. So it's a very specific type of activity. And for that specific activity, you can actually rig the memory in a way where this matrix multiplication is done in the memory itself and the data doesn't need to move anywhere. So you end up having a solution which is orders of magnitude faster and lower power consumption. And if you read anything about AI, the biggest issue with AI right now is power consumption. So in-memory compute is going to make a huge impact, both on the learning side and on the inference side. And ReRAM is a great solution for in-memory compute, Weebit recently announced we set up an AI division in the company. We've been researching this. By the way, go back to our announcements from 2018, 2019, and you'll see how many announcements we already had then about working on in-memory compute and learning the topic. So this isn't new for us. We've been waiting for this to happen. And now we're defining the offering that we'll have in this space to the market. We're working with customers on it. And we expect -- I think we announced that we expect to have a first customer in this space this year already.

Excellent. Now you got nearly $83 million in cash. How -- the question, I guess, everyone always ask is, how long is the runway in terms of the cash burn for that $83 million? Is that going to last until you get meaningful revenue through the door?

Well, if you look at our reports from 2025, you will see that our total expenses were roughly $35 million. However, we did get cash coming in from customers at about $20 million. So we're in a good position right now. The expense rate is going up. We do need to invest more in supporting the customers and improving the technology, but also we expect the cash coming in to go up. So overall, I think we're feeling very confident with our cash position. And we definitely want to grow the side of the revenue and the cash coming in.

That's fair enough. Now Coby, I'm going to ask you some opinions without notice really because I'm an AI expert, and I read so much in the papers about what's going on, how it's going to eat so many jobs, et cetera. And the amount of money that's being invested into this, can you see a good ending to this? I know you're in the business, but when you take it your Weebit, Nano hats off, can you say, hey, you know what, this is going to be absolutely brilliant and everyone is going to make huge strides and profitability and productivity, et cetera. Is it going to be great for human kind?

I think it's really hard right now to answer that question, to be very honest. Like most of the big inventions in time, it can do wonders to help humanity. And at the same time, it is very dangerous, and it all depends on us how we use it and how we control it. And obviously, there will be the "bad people who will take it and use it in a bad way". When they invented Dynamite, they were thinking about all the good things that they could do with it. And then people also use it for some bad things. And AI is no different. It can really help us. And I think today, the average person already uses ChatGPT to check things or to consult with on stuff or to generate all kinds of things. So it's going to be a great companion that enables us to do so many things that we really couldn't do before. At the same time, if misused, it can really cause also a lot of damage, and we'll need governments to control it, and we'll have to see how it evolves. It's a very good question, and I really hope that we managed to control it properly.

Now Coby, just one final question because this came up, I was giving a presentation to the Australian Shareholders' Association last week. And one member said to me, if you were advising your kids on a career path, what, would you -- they ask me what I would advise, and I'm no expert. But if you got the same question, what would you advise the 18, 19-year-olds of now to focus on career-wise? Would it be your kind of business? Or is it something more, I guess, hands-on. I talked about carpentry and plumbing and being an electrician, jobs that seem to be relatively immune, I guess, to AI, although I suspect it won't be long before a robot can start I'm plugging your. Any advice for the young people listening out there?

I think that, yes, the tasks which are relatively easily replaced by AI. And today, I guess, we're seeing the impact on the software market where we used to always say, oh, go into the software market, that's a very safe bit, et cetera. Now AI can generate the code for you. And in many tasks, I think we're going to see AI going into more and more. I would say, well, I'm in the semiconductor space. And again, the whole manufacturing of semiconductors and this is a domain which you know that Weebit has 19 PhDs in chemistry and physics who are working on this technology and experimenting and there's so much experimentation that needs to be done and so much I -- it's harder for me to see the AI getting into this, the whole semiconductor manufacturing and everything that fast, but who knows. I would say yes, it's a challenge right now, but many of these traditional jobs, maybe they -- the jobs that are not so easy to learn fast, I think those are the ones that I would look at.

Right. Okay. Well, hopefully, there's a few young people listening to this one. Now Coby, we caught up a year or so ago, and we're now catching up now. If we fast forward a year, where would you like to see Weebit? What were the milestones you'd like to say, I'll kick those off this year. We're well advanced. We've got more money coming in, in royalties and commercial revenue. Is that what you'd like to see? Or is it something else that really pops up on your radar as the major achievement you could knock off the list this year?

Well, just like I analyzed last year, and I said there are 3 domains that are important for me. On the one hand, we obviously, on the financial side, we want to sign or the commercial side, let's say, we want to sign more customers. We really want to ramp up the activity in the market and have it reflected on the financial side and see more cash coming in from customers and the revenues and everything. But I also want to see us continue to invest in improving our technology. At the end of the day, HiTek is such a domain where the moment that you stop and you stand still, you're actually falling behind. So you need to constantly improve I want to see Weebit recognized as hands down, no questions asked, the best ReRAM technology in the market. And that's what we're focused on constantly improving the technology and getting to the point where it will be a no-brainer for people to just go inside, "Oh, we need ReRAM, we go to Weebit".

Well, Coby, good luck with your next year. I'm sure given the success you've had in the past year, I'm sure you'll be able to sit down next year with us and tickle all those boxes. So thank you very much for chatting to us today. Please stay safe. I know it's a difficult time and certainly a difficult time for your family. So I hope everyone stays safe wherever they are and best of luck with the project this year. Thanks very much again.