Introducing Weebit. For those who don't know, we are in the semiconductor space developing a non-volatile memory technology called ReRAM. Non-volatile meaning that it doesn't lose its data even when you unplug it from power. And obviously that's a critical type of memory that practically any application needs, even just to store the program, the code, the basic data so that the first time you turn the machine on, it can start operating.
And you know, before I dive into Weebit and everything, I think it's important to understand also how this market evolved and where we are today. So, the existing technology that everyone is using for non-volatile memory is called flash. It's been around literally decades. I would say practically 50 years already. That's what you have in your USB stick and the SSD drive, etc., etc. But this technology, it's been already quite some time since 20 years ago that people realized it's going to be hitting the wall.
And there were many attempts to look for something that would replace flash. There was a whole list of different types of technologies. Basically one by one they fell. They were too expensive, too difficult to manufacture, really challenging. About 5 years ago, there were basically two technologies left. MRAM was already in mass production and running, and to a certain extent everyone was sure we already have a winner.
But you know, that's when devices with MRAM in them sometimes failed. People realized the M stands for magnetic and we are surrounded by magnetic fields. So, you know, it is a challenge. We ended up having the big consumer companies basically vetoing MRAM. And by then, ReRAM already got to mass production and TSMC started manufacturing it and selling products with ReRAM.
So, today, it is pretty much the industry understanding that ReRAM is really the best solution to replace flash. It is much cheaper to manufacture than MRAM, much easier. It just goes into a standard manufacturing line, doesn't need any special equipment. And the other thing that happened kind of over this time was that other companies that were trying to develop ReRAM, it's really not easy to develop ReRAM technology. And these other companies basically one by one started disappearing. So, today, Weebit is really the only independent supplier of qualified ReRAM.
And I think that's very important. Qualified meaning that it passed all the tests required to actually allow, enable it to be in mass production. So, you have TSMC in mass production. There are a couple of other foundries that are still trying to get ReRAM going, but ReRAM is really the only other company that can do that today. So, let's go to the next slide and basically look at where we are now.
So, now that ReRAM is in mass production and people know that it can actually work and it works well, the expectation is for very rapid growth of this market. So, this is from one of the leading analysts called Yole, and you can see basically the growth of ReRAM from 2024 to 2030. Both the market is growing rapidly and the ReRAM market share is growing. As I said, now it's basically understood that this is going to be the new standard. So, we're talking about a 45x growth in 6 years, which is really very, very significant. Weebit is feeling it right now, and we are sensing, you know, this huge demand, and it's really a very exciting time for us.
So, let's go to the next slide and basically look at the recent achievements at Weebit. And we've had really an unbelievable year in the past year. A year ago in March, we completed the AEC-Q100 qualification. This basically is a set of very, very rigorous tests defined by the automotive industry. And the key parts of it is showing that you can operate for 10 years at 150°C. We also demonstrated that we can achieve 100,000 endurance cycles, which is again a very important achievement.
So, the technology has been demonstrated to be very robust, very strong, and once you pass these tests, you're basically good for any application, not just automotive. The automotive is really probably the most extreme in its demands. After that, during the year, we had quite a few successes, and I think the end of the year was a spectacular one. We already have product customers that are designing products with our ReRAM in them. We achieved qualification at DB Hi-Tech. So, basically we showed that we're ready for mass production at DB Hi-Tech. Now, DB Hi-Tech needs to go through its internal procedure, etc., to officially make it available, but you know, that's going to be happening this year.
We made very good progress with onsemi. onsemi is one of the leading analog companies, a very big company in the US, in the automotive space, in the power management space. We signed up with them at the end of '24. By the end of '25, we already had the fully functional wafers that were manufactured in their facility with our ReRAM. This is really record timing doing it so fast. We're now going through the testing and we're going to be eventually getting to the qualification with them as well.
As Mark mentioned, the big highlight was of course managing to close an agreement also with Texas Instruments, the number one analog company in the world and also number one in so many other domains. This was a major agreement for us. It was three years in the making, and we're glad that we closed it, and we're now moving forward with them rapidly. Of course, all of this work with the customers, it ends up being reflected in the revenue and the cash receipts, and we're very proud of the growth on both of these sides. And I'll be talking also, of course, on the AI in a minute and what we're doing there. So, next slide, please.
And as I said, you can see what's happening now with Weebit both on the cash. The green bars are basically cash receipts. The blue line shows the revenue, and you can see the growth now. Of course, cash receipts are tied to reaching milestones, and some quarters have bigger milestones, some quarters have less milestones. So, you know, not every quarter will be record cash inflow or so on, but overall, we're very proud of the direction that these lines are going to.
So, let's go to the next part. And you know, as Mark was saying, AI today is obviously the hottest topic. We were at the Embedded World conference just a couple of weeks ago, and that was really the key topic of discussion. Weebit has been involved in this domain already for quite some time. You can see here back in 2018 we already made an announcement about collaborating with IIT Delhi. But you can see here there's been a lot of collaboration with a lot of research institutes around AI. ReRAM is really a great solution for AI, a great enabler for AI. And these research institutes recognized that a long time ago. So there's been already a lot of work going on on that front.
We actually had an AI demo that we demonstrated at Embedded World last year and also this year, showing how it works with ReRAM. And we've also been working a lot with national projects. Many countries in the world now are defining, you know, big national projects to create AI technology for use in those countries. Israel joined, I'm sorry, Weebit joined the Israeli consortium called Nemo in 2025, and that includes a lot of leading companies in Israel.
We joined the next AI consortium in Europe. This is a very big project of the European community to develop AI infrastructure for Europe, and we're very proud to be there and adding our ReRAM into this consortium. More recently we announced also the Korean national project. They have a big project to see how they can use AI and how they can use ReRAM in AI to lower power consumption, which is actually maybe the biggest challenge in AI.
If anyone doesn't know yet, this has become the biggest consumer of energy. 5% of the energy consumption on this planet is in data centers mostly running AI algorithms. So this is really a major challenge and we're very happy to be part of that Korean national project. So let's go to the next slide.
And you know, AI consists of quite a few domains, but I think that really the hottest topic today in AI and what you really, if you look in the literature and articles and everything, in-memory compute has become really the biggest focus today. The challenge that you have today is that the existing compute engines are all based on what's called the von Neumann architecture. So that's what you see on the left. You have a separate CPU and memory. The CPU, the processing unit, fetches data from the memory, does whatever it does, and then it writes back to the memory. It ends up that most of the time the execution time and most of the power consumed are actually just on these, you know, data movements from the memory to the processor and back. And this has become the major bottleneck.
Now when you're dealing with AI, AI is a very interesting application because almost everything that you do there is matrix multiplication. That's by far the biggest activity. And it turns out that you can actually take the memory and rig it in a different way so that the multiplication is actually done in the memory itself. Which is a huge advantage, obviously makes things much faster, lower power consumption. And what people are doing today is trying to make it work. Well, what they have is flash, and so there have been many attempts and people are using flash for this. But flash has a lot of limitations: the cell size is large, programming voltages are high, flash is just more expensive to manufacture. And so it has quite a few limitations.
That's where ReRAM comes in, and ReRAM is really, you know, very well suited for in-memory compute. You can implement very large crossbar arrays with very fast write speeds, low voltage. It is an analog memory that can enable multi-level values. One of the things about AI is, you know, you're trying to mimic the human brain, and the human brain is not digital, it's actually analog, and you want to have many different levels of data in each bit cell.
So ReRAM is really, you know, it's cheaper to manufacture, you can go down to smaller geometries, a long list of advantages and reasons why ReRAM is really good for in-memory compute. And Weebit today is really involved with many companies that are interested in this, and I think we can go to the next slide. We're really working on solutions for in-memory compute. I mentioned, you know, the most recent announcement which is the Korean in-memory compute program. It includes, beyond, except for Weebit, DB HiTek, which is the foundry where we'll be manufacturing those AI chips at. Analog AI is a commercial AI company that's going to eventually take this technology and commercialize it, and then there are several research institutes that are going to be doing the research on how you implement the AI and really lower power consumption with ReRAM. So the goal is really to show energy and performance improvement using ReRAM. Let's go to the next slide.
So at Weebit AI is now a major focus. So we have one big focus which is of course commercializing ReRAM as a non-volatile memory in general, selling, sorry, and of course growing the sales and customer base on that front. In parallel we set up a new team in Weebit to focus on AI. We appointed Diti Intrator as our VP systems and AI. He's a very, very experienced person who's done so many things in his prior life, being, you know, VP, an architect and CTO in so many companies. And his team is going to be defining our ReRAM offering for AI, working with the customers, and eventually also expanding our partnerships and bring out a very strong solution for AI. We are expecting already this year to have an initial offering and have initial customers, and in the future grow this dramatically. So I think this is the last slide if I'm not mistaken. And if we have a few minutes for questions, or... this, that's good.
