WeBit Nano CEO Coby Hanoch discusses ReRAM technology, AI applications, and 2027 mass production milestones
Summary
Coby Hanoch, CEO of WeBit Nano, explains that ReRAM is a non-volatile memory technology designed to replace Flash, which has reached physical limitations at 28nm. He claims ReRAM offers superior speed, endurance, and lower power consumption, with analysts projecting a 70% CAGR for the technology. Hanoch outlines partnerships with Skywater, DB HighTech, Onsemi, and Texas Instruments, noting that mass production is targeted for 2027. He highlights WeBit's in-memory compute solutions for AI, which reduce data movement and power consumption. Hanoch dismisses recent bankruptcy rumors as absurd, attributing market volatility to broader semiconductor corrections and short-term trading dynamics.
Key takeaways
- 1 Analysts are expecting the usage of ReRAM to grow CAGR of almost. 70% in the next several years.
Unidentified speaker (speaker_1) · 3:32
- 2 Uh so re-ram has a long list of advantages over flash. Uh beyond the fact that it's uh about a hundred times or orders of magnitude faster than flash, uh significantly better endurance. It is again orders of magnitude, lower power consumption than flash.
Unidentified speaker (speaker_1) · 4:11
Episode sections
- 0:02 – 2:14
Introduction and Company Overview
Henry Jennings introduces Coby Hanoch, CEO of WeBit Nano (WBT), highlighting his 46 years of experience. Hanoch explains that WeBit develops ReRAM, a non-volatile memory technology intended to replace Flash, which has hit physical limitations at 28nm.
- 2:14 – 5:27
ReRAM Advantages and Market Growth
Hanoch contrasts ReRAM with MRAM, citing MRAM's high cost and magnetic susceptibility. He claims ReRAM is faster, has better endurance, lower power consumption, and is cheaper to manufacture than Flash, with analysts projecting a 70% CAGR growth.
- 5:27 – 11:26
Commercial Partnerships and Foundry Relationships
Hanoch details partnerships with Skywater (R&D), DB HighTech (qualified, mass production next year), Onsemi (qualification phase), and Texas Instruments (technology transfer). He explains the distinction between foundries and IDMs.
- 11:26 – 14:44
Business Model and Revenue Streams
WeBit generates revenue through manufacturing license fees and NRE from foundries, and design license fees plus royalties from product companies. Hanoch notes that royalties offer nearly 100% marginal margins, justifying high IP multiples.
- 14:44 – 18:37
2027 Milestones and Mass Production
Hanoch identifies 2027 as a critical year for mass production milestones. He states that while TSMC has ReRAM, it is not an independent provider, leaving WeBit as the sole independent option for other manufacturers to license the technology.
- 18:37 – 24:34
In-Memory Compute and AI Applications
Hanoch explains that data movement consumes 60-80% of AI power. WeBit is developing in-memory compute solutions where matrix multiplication occurs within the ReRAM, offering orders of magnitude faster speeds and lower power for AI inference.
- 24:34 – 30:01
Financial Position and Market Dynamics
Hanoch addresses a recent $164M raise and dismisses reports of bankruptcy as absurd. He attributes a lack of new agreements in 2026 to fabs being overwhelmed by demand, prioritizing existing customers over new R&D integration.
- 30:01 – 37:10
Share Price Volatility and Long-Term Outlook
Hanoch attributes share price drops to broader semiconductor market corrections and short-term trading dynamics. He highlights insider buying by himself and the chairman as a vote of confidence, emphasizing a long-term focus on technology development.
- 37:10 – 38:18
Closing Remarks and Personal Anecdotes
The host and guest discuss the guest's brother, a yoga teacher and photographer. The host congratulates the guest and wishes him success for 2027, noting the speed at which the year has arrived.
- 38:18 – 39:18
Discussion on Market Volatility and Media Accuracy
The guest criticizes recent negative news reports about his company, citing absurd mathematical errors in the reporting. The host agrees, suggesting the reports may have been generated by AI without proper fact-checking.
- 39:18 – 40:16
Final Thanks and Future Contact
The host thanks the guest for appearing on the show and mentions that insider buying by directors provides a vote of confidence. They agree to speak again in approximately six months.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| NVIDIANVDA | equity | neutral | Uh Nvidia Intel are talking about uh three, two nanometer even going now to 1.8 and 1.4. |
| AppleAAPL | equity | neutral | That's why even the biggest players like Apple, Google, Facebook, Nvidia, et etc., cetera, don't have their own. |
| AlphabetGOOGL | equity | neutral | That's why even the biggest players like Apple, Google, Facebook, Nvidia, et etc., cetera, don't have their own. |
| Meta PlatformsMETA | equity | neutral | That's why even the biggest players like Apple, Google, Facebook, Nvidia, et etc., cetera, don't have their own. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
Coby Hanoch has nearly 46 years of experience in the semiconductor industry.
Unidentified speaker (speaker_0) ·0:03
How to verify: Check Coby Hanoch's LinkedIn profile or WeBit Nano corporate biography for career start date. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
Flash memory hit a physical limitation at 28 nanometer geometry.
Unidentified speaker (speaker_1) ·4:47
How to verify: Review semiconductor industry technical roadmaps (e.g., ITRS) or technical papers on Flash scaling limits. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
Analysts expect ReRAM usage to grow at a CAGR of almost 70% in the next several years.
Unidentified speaker (speaker_1) ·3:29
How to verify: Search for market research reports from firms like Gartner, IDC, or Yole Group on ReRAM market forecasts. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
Primary sourceDB HighTech is one of the top 10 foundries in the world.
Unidentified speaker (speaker_1) ·6:32
How to verify: Check industry rankings from SEMI or TrendForce for global foundry revenue market share. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
WeBit Nano raised $164 million in May.
Unidentified speaker (speaker_0) ·24:54
How to verify: Check WeBit Nano's ASX announcements or SEC filings (if applicable) for equity raise details in May 2026. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
Primary sourceWeBit Nano had $164 million in the bank and spent less than $20 million net in the last fiscal year.
Unidentified speaker (speaker_1) ·25:12
How to verify: Review WeBit Nano's latest Annual Report or Quarterly Financial Statements for cash reserves and net expenditure. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
TSMC is the only other company officially qualified for mass production with ReRAM besides WeBit.
Unidentified speaker (speaker_1) ·17:18
How to verify: Check TSMC's product portfolio announcements or independent semiconductor industry news for ReRAM qualification status. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
Data movement in AI consumes 60 to 80 percent of power consumption.
Unidentified speaker (speaker_1) ·19:48
How to verify: Review academic papers or industry white papers on AI hardware efficiency and memory wall challenges. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
Data centers consume 5 percent of the planet's power.
Unidentified speaker (speaker_1) ·19:48
How to verify: Check data from the International Energy Agency (IEA) or similar bodies on global data center energy consumption. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
Intel's share price came down 45% in the last couple of weeks.
Unidentified speaker (speaker_1) ·31:41
How to verify: Check historical stock price data for Intel Corporation (INTC) on financial platforms like Yahoo Finance or Bloomberg. For recent events, verify 2026-08-09 through 2026-11-09. Use another period only when the speaker explicitly names it.
Do your own research
- Verify the 70% CAGR projection by checking the specific analyst reports cited by Hanoch for their methodology and source.
- Confirm the status of DB HighTech's qualification and mass production timeline through their official press releases or SEC filings.
- Review Skywater, Onsemi, and Texas Instruments' public disclosures regarding their R&D or technology transfer agreements with WeBit Nano.
- Check WeBit Nano's (WBT) latest 10-K or 10-Q filings for details on revenue streams, NRE fees, and royalty structures.
- Investigate the specific 'absurd mathematical errors' in the negative news reports Hanoch referenced to assess their validity.
- Analyze insider trading reports (Form 4) for recent purchases by Hanoch and the chairman to verify the 'vote of confidence' claim.
- Compare ReRAM performance metrics (speed, endurance, power) against MRAM and Flash using independent technical whitepapers or industry benchmarks.
Full transcriptexpand
- 0:02
- Unidentified speaker (UNKNOWN): Well,
- 0:03
- Unidentified speaker (speaker_0): hello and welcome to the latest episode of On the Couch with myself Henry Jennings from Marcus Today. And today I'm delighted to be joined by, I guess, an old friend of this podcast, Kobe Hannock from uh WeBit Nano. He's the CEO with nearly 46 years of experience in the semiconductor and related
- 0:22
- Unidentified speaker (speaker_0): industries. So I I've gotta got to say, Kobe, looking at you, uh, I wouldn't believe you've been in this business 46 years because you you look very young and youthful still. So clearly uh clearly you're not stressed enough
- 0:36
- Unidentified speaker (speaker_0): in terms of the business. But uh it's uh it's delight to have you back on the show. You're previously a CEO at Packet Light Networks and has held VP Worldwide Sales roles at both uh Veracity and Jasper Dire Design Automation. So welcome, Co Kobe. Thanks very much for coming on the show today. And I'm looking forward to our chat.
- 0:56
- Unidentified speaker (speaker_1): Yeah, thanks for having me here.
- 0:58
- Unidentified speaker (speaker_0): My pleasure. Now, just before we kick off, I just must mention, of course, that uh this is general advice only. So please do your own research, contact your own financial advisor regarding any of the thoughts, ideas, or insights in this podcast. Kobe, we bit nano. Uh um WBT for those people that uh
- 1:18
- Unidentified speaker (speaker_0): are listening at home or wherever they're listening to, that's the stock code. Um run me through for those that don't know the story uh run me through the WeBitNano kind of elevator pitch uh non-volatile memory uh re am what what's we WeBit bit nano all about
- 1:36
- Unidentified speaker (speaker_1): so uh keeping a it short, uh in semiconductors, in any electronic component uh or device, you need a memory. You need something to hold the code, you need something to hold the data, the coefficients of AI or whatever. Um, and you want it to be in a memory that
- 1:56
- Unidentified speaker (speaker_1): even when it loses the power, it the data doesn't uh go away, right? Uh I often say every electronic device has the magical moment that you turn it on for the first time and you expect there to be some software and data there. And that's non-volatile memory.
- 2:14
- Unidentified speaker (speaker_1): In our industry, for decades, literally decades, the dominant technology has been Flash.
- 2:22
- Unidentified speaker (speaker_1): However, Flash already 20 plus years ago, people realized.
- 2:27
- Unidentified speaker (speaker_1): Flash is going to be hitting the wall, starting to have more and more limitations. And for the last 20 years, people have been looking for uh what's going to come after Flash. Uh, cutting the story short, there were many different technologies, many attempts. Uh, bottom line is there are two
- 2:47
- Unidentified speaker (speaker_1): technologies really left today. Um One is MRAM that has been already in mass production for well, I would say maybe seven, eight years almost.
- 2:59
- Unidentified speaker (speaker_1): But it is very expensive. You need a special manufacturing line for it. The key issue is it's also the M stands for magnetic, and uh there have been cases we're surrounded by magnetic fields, and there have been cases where. devices with MRAM started malfunctioning. So,
- 3:20
- Unidentified speaker (speaker_1): ReRAM, what WeBit is developing, is now I would say recognized in the industry as
- 3:29
- Unidentified speaker (speaker_1): the next standard technology.
- 3:32
- Unidentified speaker (speaker_1): Analysts are expecting the usage of ReRAM to grow CAGR of almost. 70% in the next several years. So really very rapidly to grow. And we're seeing it now happening in the market. So it's very exciting times for us. Our technology is ready now for
- 3:52
- Unidentified speaker (speaker_1): mass production and getting very close to it.
- 3:56
- Unidentified speaker (speaker_1): And uh yeah, so it's it's really exciting times now with uh with re-RAM. rem.
- 4:01
- Unidentified speaker (speaker_0): So in in terms of re-RAM, well what what are the what are the big differences? What what's your big advantage, I guess, over the existing technology?
- 4:11
- Unidentified speaker (speaker_1): Uh so re-ram has a long list of advantages over flash. Uh beyond the fact that it's uh about a hundred times or orders of magnitude faster than flash, uh significantly better endurance.
- 4:27
- Unidentified speaker (speaker_1): It is again orders of magnitude, lower power consumption than flash. But I would say the two most important advantages are first, you might know in our industry there's what we call Moore's Law. Everything is shrinking all the time. And uh we're going down to
- 4:47
- Unidentified speaker (speaker_1): smaller and smaller geometries. Uh Nvidia Intel are talking about uh three, two nanometer even going now to 1.8 and 1.4. Flash basically hit the wall at 28 nanometer. It can't really shrink below that. So the advanced devices today don't really
- 5:07
- Unidentified speaker (speaker_1): have even the option of had of using flash. Uh re RAM am can go down to the small geometries, but at the end of the day, the number one criteria for everyone is always cost. And re-RAM is simpler and cheaper to manufacture than Flash, which I think that's at the end of the day, that always ends up being the number one
- 5:26
- Unidentified speaker (speaker_1): criteria.
- 5:27
- Unidentified speaker (UNKNOWN): It
- 5:28
- Unidentified speaker (speaker_0): does, doesn't it? Money always talks. And you've got some impressive names that have come on board and are they doing some uh some testing. Uh, The likes of Skywater, DB, Hightech, High Tech, uh, technically. and Texas Instruments as well. What's the next stage with the relationship with these uh with these big players in addressing this? Well, I would imagine is a pretty huge market,
- 5:48
- Unidentified speaker (speaker_0): Kobe.
- 5:50
- Unidentified speaker (speaker_1): Yeah, it is a very big market. Uh, and we are very proud of the progress that we've been making commercially. We started It off uh with a relatively small um foundry in the US. Skywater.
- 6:05
- Unidentified speaker (speaker_1): When you're in this industry and you're dealing with things that are so complex, you don't want to start off with the big guys.
- 6:12
- Unidentified speaker (speaker_1): If you end up burning yourself, you want it to happen on a small guy. So we we went with Skywater. I'm glad to say it was a big success. Uh The challenge was Skywater did a redirection and they became more of an R. RD facility than a foundry. And um we're using them for RD.
- 6:32
- Unidentified speaker (speaker_1): But then we moved on. Uh we engaged with DB HighTech. DB HighTech is already one of the top 10 foundries in the world. And just to clarify, uh maybe what a foundry is, uh, because I'll be using that term a lot. Uh in in semiconductors, people don't realize just how. expensive it is
- 6:52
- Unidentified speaker (speaker_1): to manufacture semiconductors. A fab, which is where you manufacture semiconductors, an advanced fab, is tens of billions of dollars, not millions, billions.
- 7:07
- Unidentified speaker (speaker_1): That's why even the biggest players like Apple, Google, Facebook, Nvidia, et etc., cetera, don't have their own. manufacturing facilities, and they end up using companies that are called foundries that offer manufacturing as a service. So Skywater and DB Hightech are
- 7:27
- Unidentified speaker (speaker_1): both foundries.
- 7:33
- Unidentified speaker (speaker_1): Korean company. We engaged with them. We transferred the technology to them. It's a long process. We got to the point where the technology passed all the tests and eventually got the stamp of approval that's called qualification, which is basically passing a
- 7:53
- Unidentified speaker (speaker_1): series of very tough tests that were defined by standards bodies. So with DB High Tech, we're qualified, we have already product companies.
- 8:04
- Unidentified speaker (speaker_1): that are developing products and embedding our technology into their products and sending them to db high tech to manufacture.
- 8:15
- Unidentified speaker (speaker_1): And one of them is already talking about publicly talking about going to mass production next year. Others are following. So things are moving very nicely forward with DB High Tech.
- 8:31
- Unidentified speaker (speaker_1): And then we continued
- 8:34
- Unidentified speaker (speaker_1): to go to larger and larger companies. The next one in line was Onsem. On SEMI, many people know it better by its old name, Motorola Communications. It's a big company, very big. It's an SMP. SP 500, very big on NASDAQ,
- 8:53
- Unidentified speaker (speaker_1): etc.
- 8:56
- Unidentified speaker (speaker_1): They are what's called an IDM. They not only have a manufacturing facility, they actually design their own products and manufacture them in their manufacturing facilities. So, to a certain extent for us, it's a big advantage because we have the product company. and the fab all
- 9:16
- Unidentified speaker (speaker_1): in one and you can move faster. Uh I would even say an IDM would not engage with a company like Weavid if it weren't for their product teams wanting to use this technology, because otherwise they don't really have a reason to engage with us. So we engaged with them, we
- 9:36
- Unidentified speaker (speaker_1): worked through the technology transfer again. that's a long process requiring big teams from both sides cooperating, working together for roughly two years. We're now in the qualification phase. So going through all of these tests that are
- 9:56
- Unidentified speaker (speaker_1): required to get the stamp of approval that we're ready for mass production. You know, obviously. they have some products already also that are working with us. And
- 10:09
- Unidentified speaker (speaker_1): we hope to get that qualification and go into uh mass production after that with them. So that's moving forward very nicely. The last one that we signed with is already one of the top five uh companies, uh, Texas Instruments. Uh that's a huge company. Again, an IDM, they have their
- 10:29
- Unidentified speaker (speaker_1): own products and they manufacture them. Um, and with them, we're moving forward uh through this technology transfer. Uh so overall, with with all four of them, there's good progress. Uh, we earlier this year announced that two of the big customers, I didn't we
- 10:49
- Unidentified speaker (speaker_1): didn't say two, but we said customers,
- 10:53
- Unidentified speaker (speaker_1): expanded um their agreements with us, bringing in more um more business units and more activities. So uh you know there's really once these guys have have gone in, learned the technology, uh deeply lear got to know our team, the
- 11:12
- Unidentified speaker (speaker_1): whole, you know, the R RD and D team, not just the management or the salespeople, they saw that this is what they need and they expanded their agreement. So for me, that was a big uh vote of confidence in Weebin.
- 11:26
- Unidentified speaker (speaker_0): It it clearly takes a long time to go through the qualification stage and and move into mass production. In terms of the of the business model, um what what's the business model? How do you make money? Is it a royalty stream or is it a licensed deal and a royalty stream? How how does WeBit uh leverage that mass production?
- 11:48
- Unidentified speaker (speaker_1): So I I mentioned actually two types or even three types of customers that we have. Uh on the one hand, there's the foundry or the fab uh that manufactures, uh, and we uh uh sign uh a manufacturing license agreement with them. So they pay us uh a license
- 12:08
- Unidentified speaker (speaker_1): fee and what's called an NRE, non-recurrent engineering,
- 12:12
- Unidentified speaker (UNKNOWN): uh
- 12:13
- Unidentified speaker (speaker_1): for all of the work that we do in technology transfer. I I mentioned it's you know it's roughly a two-year project with many people involved. So up to now, all our revenue was obviously from uh or practically all the revenue came from manufacturing license fees and
- 12:33
- Unidentified speaker (speaker_1): NRE.
- 12:35
- Unidentified speaker (speaker_1): In parallel, we work with the product companies. They, you know, because everything has been shrinking so much, they can put now a full system on a single chip. So the way that these guys work is you know they design the core of that chip, but then all of the
- 12:55
- Unidentified speaker (speaker_1): periphery, you know, if they need a processor, there are several processors out there. You take an ARM or something, they need a Wi Fi, they need an HDMI, they need, you know, all of these components have already been designed by other people, and they just go and take them and plug them in. It's like a big puzzle. They need a non non-volatile volatile memory, they need someone to give them. a
- 13:15
- Unidentified speaker (speaker_1): module of a non-volatile memory. So we actually give them that module, they plug it in. And that's why we call it the embedded market, because they embed our technology into their system on a chip. And then they take that chip and they go to a foundry to manufacture. So with the product guys, we sign
- 13:35
- Unidentified speaker (speaker_1): a design license agreement. They pay us. a license agreement. And then when they go into mass production, that's when the real fund starts. They start paying us royalties from the revenues from those products. The third type of customer I mentioned
- 13:55
- Unidentified speaker (speaker_1): earlier are IDMs like On Semi and TI, where the product company and the manufacturing is all. in one. So we have a special type of agreement with them that covers both. So that's the business model. The real ramp in revenue starts when you get to mass production
- 14:15
- Unidentified speaker (speaker_1): and then when you start getting royalties. I would actually clarify royalties by nature are practically 100% a hundred percent marginal. uh margin. Uh so that's why IP companies like us have very high multiples um
- 14:35
- Unidentified speaker (speaker_1): in the market, because uh we actually work in in much higher uh margins than many other companies.
- 14:44
- Unidentified speaker (speaker_0): Well that that that that's kind of good to know. Uh as far as um you know twenty twenty twenty-seven seven was what's nearly upon us isn't it we're we're we're very close to that to that now new year um what what should we be looking for in terms of milestones heading into that production in 2027 and I guess how confident are you with um with the next 12
- 15:04
- Unidentified speaker (speaker_0): months in terms of that that uh production capability
- 15:08
- Unidentified speaker (speaker_1): uh actually 2027 is going to be a very very exciting year for us uh at least what I'm seeing right now. Uh, we can already see products getting to mass production uh next year, uh, which mass production is really the number one milestone
- 15:28
- Unidentified speaker (speaker_1): in the life of a semiconductor company. Um you know, knowing that uh your technology is now going into real products and going into mass production. and and
- 15:40
- Unidentified speaker (UNKNOWN): you
- 15:40
- Unidentified speaker (speaker_1): know starting to be sold uh in in large volumes, that really shows that the technology is is robust. Uh today uh some people are saying, Yeah, you already qualified, you qualified at Skywater, at DB Hitech, at Leti, at this, um, but it's still small
- 16:00
- Unidentified speaker (speaker_1): volumes. Yeah, we we want to see large volumes. So mass production is really the big milestone that we're expecting to achieve
- 16:10
- Unidentified speaker (speaker_1): in 2027. Obviously, this is a milestone that is in the hands of our product customers because they need to decide when they're ready for that. But they are pushing forward, and at least one of them already announced they're expecting to get to that uh next year. Um that's the big thing, of course. In
- 16:29
- Unidentified speaker (speaker_1): addition, we expect to sign additional um uh license agreements both with uh with fabs uh and with product companies. In db high tech, we want to ramp up more activity with product companies, we want to sign up more fabs. Uh
- 16:49
- Unidentified speaker (speaker_1): today I mentioned the industry knows that re RAM is really the new standard.
- 16:56
- Unidentified speaker (speaker_1): Everyone's looking for re RAM right now. I'm I'm not aware of any manufacturing company, a fab or an IDM that is saying we don't want re RAM. We don't think we'll need re RAM in the future. And WiiBit WeBit is the only independent provider of ReRAM re RAM today. I think that's also an important thing to
- 17:16
- Unidentified speaker (speaker_1): point out.
- 17:18
- Unidentified speaker (speaker_1): Today there are only two companies that are officially qualified to go to mass production with ReRAM. There's TSMC, which is the largest foundry in the world. And they're offering it to their customers. They're in mass production already. And But they obviously will not give their technology to other
- 17:38
- Unidentified speaker (speaker_1): manufacturers.
- 17:39
- Unidentified speaker (speaker_1): And then there's WeBit,
- 17:42
- Unidentified speaker (speaker_1): which is,
- 17:43
- Unidentified speaker (UNKNOWN): you
- 17:43
- Unidentified speaker (speaker_1): know, we're independent, we can operate with everyone, and we're engaged with many, many, many other uh big uh manufacturers. Um the negotiations obviously take a long time on these kinds of things. Uh, when you sign an agreement with uh such a manufacturer, it
- 18:02
- Unidentified speaker (speaker_1): it's literally for decades. It's uh, I mean, you know that you'll be there even in 20 years and beyond uh once you actually are in there. So these decisions are very tough decisions for them. Uh, but we're making good progress, and I believe we'll be signing more agreements in 2027. Um,
- 18:22
- Unidentified speaker (speaker_1): so those are the main things that we're looking at. The other thing. which is now ramping up nicely, and and I want to talk about that is uh in-memory compute and the whole AI side of things. So uh
- 18:37
- Unidentified speaker (UNKNOWN): if
- 18:37
- Unidentified speaker (speaker_1): it's okay with you, maybe I uh I
- 18:40
- Unidentified speaker (UNKNOWN): should
- 18:40
- Unidentified speaker (speaker_1): explain what we're doing there.
- 18:43
- Unidentified speaker (speaker_0): Kyby,
- 18:43
- Unidentified speaker (UNKNOWN): the platform is
- 18:44
- Unidentified speaker (speaker_0): yours, my friend.
- 18:46
- Unidentified speaker (speaker_1): Yeah, so um
- 18:48
- Unidentified speaker (UNKNOWN): you
- 18:48
- Unidentified speaker (speaker_1): know when a a little bit of background so that people understand what I'm talking about. Um in today's compute in in basically in every electronic device almost, they all use a similar architecture. There's a an a a compute engine and or an
- 19:08
- Unidentified speaker (speaker_1): execution engine and there's the memory. And what you're doing all the time is you're taking the memory, bringing it to the compute engine, and then that engine does whatever it does. Now, uh, the thing is, when you start looking at it, transferring the data from memory to the compute engine
- 19:28
- Unidentified speaker (speaker_1): is what takes a lot of time and a lot of power. And today power has become a major thing, especially in AI, which by nature, you know, the data. is kind of spread all over the place. Um, it takes uh in in in the AI world, 60, even 80%
- 19:48
- Unidentified speaker (speaker_1): of the time, and more importantly, 60 to 80 percent of the power consumption is taken just by moving the data around. Now, uh, I don't know how many people realize it, but today data centers are already consuming five percent. 5% of the power of this planet. So it's it's
- 20:06
- Unidentified speaker (UNKNOWN): really
- 20:07
- Unidentified speaker (speaker_1): a major issue. When you look at at um at AI, there are two main activities that you do. You have training and you have inference. Uh first of all, you train the system to know how to do something or recognize something, and then you enjoy
- 20:26
- Unidentified speaker (speaker_1): infer and you use that. So training is done once, you teach it once, and then you infer all the time. In the early days, the only activity that was done was training, but now there's more and more inference. As time goes by, inference is becoming the dominant activity in AI. The nice thing about inference is you're
- 20:46
- Unidentified speaker (speaker_1): mostly reading from the memory. You're not writing to the memory a lot. You're made in the mainly reading, which is where REM shines. That's where RAM is really very strong. So
- 20:59
- Unidentified speaker (speaker_1): now, in inference, when you analyze a little more what actually is going on, you realize that
- 21:07
- Unidentified speaker (speaker_1): most of the time you're doing what's called vector matrix multiplication. And
- 21:13
- Unidentified speaker (UNKNOWN): if
- 21:14
- Unidentified speaker (speaker_1): you look at it, you can actually rig the re RAM memory so that the matrix is in the memory. And when you write the vector into the memory, the multiplication is done instantly on the spot without moving data anywhere, and you immediately get the result of that multiplication. So now
- 21:34
- Unidentified speaker (speaker_1): no data movement and initial results, and we actually we had uh an American customer take a look at this, uh, use our re RAM uh in evaluation mode, and you know they did very specific tests, but you know, the results are that as expected, orders of
- 21:54
- Unidentified speaker (speaker_1): magnitude faster and orders of magnitude lower power consumption with our re rema. RAM. Um so we we have now really recently announced uh that we're providing uh we call it the matrix. It's uh it's a memory matrix to hold the uh uh the
- 22:13
- Unidentified speaker (speaker_1): matrices and and play with. So this is for researchers who are researching um AI and specifically doing inference. And uh they can take our re-RAM remain and and uh and use it in their research. They're committed to give us uh the results. We have
- 22:33
- Unidentified speaker (speaker_1): a lot of interest in this. Actually, before we ever announced it, we already had a long list of companies call us up and say, hey, we we know you're doing re-RAM. Can we have uh uh your memory to to try it for uh inference? So we already had a lot of requests before. Now we officially announced it and and we're
- 22:53
- Unidentified speaker (speaker_1): expecting uh to really have many people ask for this uh matrix. We already have a few that have engaged. Um and and the results of their uh
- 23:05
- Unidentified speaker (speaker_1): uh of their research uh they'll be sharing them with us so we can use them when we're looking at how to do uh inference we have a team internal in WeBit that is now working on different options of how you can apply uh in-memory compute to
- 23:25
- Unidentified speaker (speaker_1): re RAM. Uh there's there are still a lot of uh obstacles. I have to be clear. This is something that there are a lot of technical challenges to do this. Uh the initial good results are great, but we still have a lot of. work to do. And
- 23:43
- Unidentified speaker (speaker_1): we expect it to take time until we manage to find ways to overcome these challenges and actually have a solution ready for the market. But we are very focused on developing a commercial solution for inference.
- 23:59
- Unidentified speaker (speaker_1): We actually, when we did the raise earlier this year, we we said there's going to be a chunk of money to improving our remote, and maybe I should talk about that as well. And then another large amount that goes into developing the in-memory compute.
- 24:16
- Unidentified speaker (speaker_0): So so this in-memory compute, this is this is the research program that you announced uh what, three or four days ago. Um and uh I'm just reading the announcement. now. So uh that's the IMC research programme which does look quite exciting. Uh looking at um
- 24:34
- Unidentified speaker (UNKNOWN): other
- 24:34
- Unidentified speaker (speaker_0): the other side of things and you talk about the money side of things, uh I noticed y you have done you did a raise um not that long ago and you do have uh I guess quite a lot of money uh at the moment and yet you seem to be quite frugal, Kobe, I have to say, in terms of the the money spent. So the the the the new money that you have
- 24:54
- Unidentified speaker (speaker_0): raised uh you've got 164 million bucks in in in the bank uh having raised in May uh can you kind of outline where that money is going or what it's earmarked for
- 25:07
- Unidentified speaker (speaker_1): Australian we're just about bankrupt now so uh you know
- 25:12
- Unidentified speaker (speaker_1): yeah
- 25:12
- Unidentified speaker (speaker_0): that that was odd wasn't it
- 25:15
- Unidentified speaker (speaker_1): That was like one of the most uh absurd things I've ever seen, you know. They they they announcing that we're almost bankrupt when we have a hundred and sixty four million in the bank and you know, the whole last fiscal year for the whole year the net spent was twenty million less than twenty million. So it was uh it was amazing,
- 25:35
- Unidentified speaker (speaker_1): you know, that comment that they made and how much noise that uh did. But yeah, we uh we do have money in the bank. Um we are using it and uh mentioned in memory compute. We're also using it uh to improve our embedded technology. And uh I mean to
- 25:55
- Unidentified speaker (speaker_1): to state it in a simple way. I told the team, listen, every technical parameter you can think of, we need to be at least one order of magnitude better than TSMC and others. So we're working very hard to really be, without any questions, the best re-RAM in the market. So that, I mean,
- 26:15
- Unidentified speaker (speaker_1): there's this big vacuum that's opening up now. And uh people want re-RAM. Re RAM. I just want everyone to, whenever they want re re-RAM, RAM, not even think about it. Oh, re-RAM, re RAM, that's weird. WeBit, of course. We're calling WeBit. So we're working very hard now to improve the technology. Um
- 26:35
- Unidentified speaker (speaker_1): and and we're making very good progress uh with that. We we really are on on different parameters already getting to the point where we're an order of magnitude better than others. So we're using the money, but this is shareholders' money. And I'm very conscious of the fact that the shareholders gave it to
- 26:55
- Unidentified speaker (speaker_1): us for a specific reason. So we're spending it, but I'm also, you know, I mean, let's face it, we haven't signed an agreement since the beginning of the year. And, you know, that's kind of the elephant in the room, which I'd like to address because I know many shareholders are asking, what the hell is going on? Why aren't you signing agreements?
- 27:16
- Unidentified speaker (speaker_1): Um, and uh I think people need to understand what's happening this year in the semiconductor market. Demand is huge. I think you hear about it everywhere. Uh, I mean, even in non uh industry news, uh, it's the talk. Um, and
- 27:35
- Unidentified speaker (speaker_1): the fabs are now really, really
- 27:39
- Unidentified speaker (speaker_1): not managing to deal with demand. Most of them are telling us demand is 50 or even 100% higher than the capacity that they have. So they're just trying to deal with all of this customer demand. They're trying to keep their customers happy. Um, everything's going crazy, and it's just very
- 27:59
- Unidentified speaker (speaker_1): difficult for them to uh to commit now to. I mean, I mentioned it's a big team, a big project to bring rem on board. And they're basically saying, hey, just let us let us focus now on addressing the customer demand. We want re-RAM. We'll be talking to you. Just please not now. Let us just deal
- 28:19
- Unidentified speaker (speaker_1): with this craziness. And by the way, I mean, now they're selling their silicon wafers with huge margins because of the demand. They're cap, they want to capitalize every wafer that goes through the Fab. They want it to be for customers. I mean, starting to run now RD wafers with re-RAM doesn't make any sense when
- 28:38
- Unidentified speaker (speaker_1): they're not managing to address the customer demand. So I understand them very well. I would, I think I and you would make the same decision and delay getting the re-RAM re RAM on board. So we were fortunate to have the money in the bank to go through this period. I mean, obviously this period won't last forever. There are many
- 28:58
- Unidentified speaker (speaker_1): fabs that are being built right now. There's more capacity coming up.
- 29:05
- Unidentified speaker (speaker_1): And we, you know, once this happens, we expect to actually have you know many of these guys come back and engage. That's why I said next year I'm expecting. already to have more agreements. So uh I mean uh while maybe you think I don't look that old, but I have been around uh for uh for more
- 29:25
- Unidentified speaker (speaker_1): than 40 years in this industry, and I know how easy it is to just, you know, you heard other people of uh who won the lottery and three years later were on a bench in the public park, right? I didn't
- 29:38
- Unidentified speaker (UNKNOWN): want to
- 29:39
- Unidentified speaker (speaker_1): be that person. We're we're managing the company in a conservative way. Um I think we're doing the right things. We're improving the technology. We're ramping up in memory compute and um you know we're moving forward. Uh we try to deliver on all of the targets that we announce and
- 29:59
- Unidentified speaker (speaker_1): uh and that's what we're doing now.
- 30:01
- Unidentified speaker (UNKNOWN): Um
- 30:01
- Unidentified speaker (speaker_0): can Kobe, be I mean I guess I I guess there was um you know that that um report um I think it was the ABC wasn't it um suggesting that you guys are bankrupt, which was complete and utter rubbish, as you say, and and just just crazy. But the but the share price-wise, it it's it's never really kind of recovered. Uh uh
- 30:22
- Unidentified speaker (speaker_0): which is it is is weird. You you think uh there would be some sort of covering. Is what's going on at the moment with this AI kind of revaluation across the board and some of the volatility we've seen in the US, is that um do you think that's partly responsible for um for for the um the sort
- 30:42
- Unidentified speaker (speaker_0): of the uh the drift in your share price to lower levels?
- 30:46
- Unidentified speaker (speaker_1): Uh at the end of the day, we're part of the semiconductor industry. Yeah. The semiconductor industry earlier this year, uh, when when the demand was going up, it went crazy and share prices were going up hundreds of percent. Uh, and we went up with it. So uh you look uh I
- 31:06
- Unidentified speaker (speaker_1): don't know, a year, a year and a half ago, we were at one and a half dollars or or less than two dollars. We went up with everyone as uh as a pre-profitable company, you always go more extreme than the rest of the market. So we went up very high. Then the whole market, the whole
- 31:25
- Unidentified speaker (speaker_1): semiconductor market, uh came down very strongly. I mean, the semiconductor index itself of the whole industry would came down more than 50%, I believe. Uh, Intel just in the last uh couple of weeks. came down 45%.
- 31:41
- Unidentified speaker (speaker_1): Uh, you know, Intel, not a small company, and and many other you look at all of the big semiconductor companies, they've lost uh, you know, 40, 50, and even more percent. So again, we are pre-profitable and we take the hit stronger, and we came down. Um, I think that's one
- 32:01
- Unidentified speaker (speaker_1): thing that happened. Uh you know, another thing that many people are telling me is is common in Australia when you enter ASX 200 and uh the index funds come in, you know, the share price goes up when they come in and then uh they lend it to shorters and you go down. So I always tell people I don't
- 32:21
- Unidentified speaker (speaker_1): manage the share price and I just focus on managing the company, on pushing it forward. I believe that eventually the share price reflects the state of the company. The only thing that I can say is I think it was a month or a month and a half ago, both uh Daddy, our chairman, and I, and that was when when it was at three and a
- 32:41
- Unidentified speaker (speaker_1): half dollars, we looked at each other and we said, Wow, this is a great opportunity for us, and both of us bought shares. So that's the only thing I can say. I'm not allowed to recommend. anything to anyone, but that's what I recommended to myself.
- 32:55
- Unidentified speaker (speaker_0): Well, Kobe, money talks as always, and and and if um you know investors see directors buying shares in in the company, it's always uh a pretty good sign. I I actually I looked at the short position and I was quite surprised in some respects because um there are many companies at the moment that have massive short positions in them and we WeBit was
- 33:15
- Unidentified speaker (speaker_0): only around 10 million shares or 4.1% of the company. I mean, it's still significant, but it's certainly not up there with some of the uranium companies that we've currently got, which are 20% shorted. So
- 33:28
- Unidentified speaker (speaker_0): I guess that's something in some respects. And we we've seen some of the volatility that comes with these short positions in terms of uh when they decide to cover uh things can get quite volatile. quite quickly. So um hopefully uh things will improve. Uh as far as um
- 33:45
- Unidentified speaker (UNKNOWN): I'm
- 33:45
- Unidentified speaker (speaker_0): I'm gonna ask you a weird question here because you you you're on to some extent you're on the outside looking in. You know, you're you're in Israel, and and there's an obviously a lot happening in the US in terms of what's going on there, and the US market is absolutely well, it it's become the AI has been. become not the trade, it's become the market.
- 34:05
- Unidentified speaker (speaker_0): Is there things happening in the US you look at and you think oh you know, that that that's not a good sign? And what we saw that massive volatility uh uh just after the um the blue wave at the elections. Is there is there things that are happening in terms of valuations with anthropic maybe that are keeping you up at night thinking that you know this this could this
- 34:25
- Unidentified speaker (speaker_0): could end badly or or is the market kind of just going through this this this period of um revaluations and getting getting some of the hot air out of the thing
- 34:36
- Unidentified speaker (speaker_1): um I'll I'll start with the basic I believe the world needs semiconductors and will need even more semiconductors and no matter you know you you can have these uh specific events that happen, the elections in the US or the riots in France or
- 34:56
- Unidentified speaker (speaker_1): the war in Iran or whatever. At the end of the day, no matter how we look at it and whether we like it or not, the world is moving more and more to AI. The world needs more and more semiconductors. Conductors. And you know, long term, the demand will just keep growing. You can have
- 35:16
- Unidentified speaker (speaker_1): these things that happen in the middle that drive the market crazy. And I know you know a lot of the traders react so strongly. And today, with all of the uh social media and with all of the the fact that you can trade so quickly, people react like instantly. They get the news, and before they even realize what
- 35:36
- Unidentified speaker (speaker_1): the news is, and before they really digest it. they're already reacting and the markets jump up, jump down, you know, go crazy. I'm in this for the long run. WeBid is developing a technology that the market needs, that the market will need. Again, as I said, and I'm serious about it,
- 35:56
- Unidentified speaker (speaker_1): it's decades. Um, so we um uh we need to really look at the long term. And yes, I mean as a person, I look at all these things that are happening now. Um, and uh we uh we really um I'm you know it's disturbing or whatever, but as a
- 36:16
- Unidentified speaker (speaker_1): company, I'm just focused on uh moving forward and uh and and eventually as I said the markets will reflect the the value of we did. be.
- 36:27
- Unidentified speaker (UNKNOWN): Uh
- 36:28
- Unidentified speaker (speaker_0): Kobe, I I know this is um audio and not video, but I'm look I'm looking at the the picture behind you and I'm kind of fascinated as to what it is. It's it's a it's it's kind of an an orange haze on a on a very black background. I'm I'm sure it's something particularly ethereal or very
- 36:48
- Unidentified speaker (speaker_0): you know, tied up with AI or re re-ramp. ramp. Is is that the case or is it just uh a lovely
- 36:53
- Unidentified speaker (UNKNOWN): piece
- 36:53
- Unidentified speaker (speaker_0): of art?
- 36:54
- Unidentified speaker (speaker_1): My my brother is uh is also a photographer.
- 36:58
- Unidentified speaker (UNKNOWN): What you're seeing
- 36:59
- Unidentified speaker (speaker_1): behind is actually a couple a couple dancing in long exposure. So
- 37:04
- Unidentified speaker (speaker_0): okay. It's a terrific pitch. When when you say also, does that mean you're a photographer too?
- 37:10
- Unidentified speaker (speaker_1): No, I mean that he has a lot of things, he's actually mainly uh a a yoga. teacher. He he's a yoga he teaches yoga teacher trainings and stuff. So yeah, that's his main job. But photography is also a passion of his.
- 37:25
- Unidentified speaker (speaker_0): It's well it's certainly a cool picture. I I was fooled there. I I did think it was some sort of um AI or technology uh thing but uh there we go it it's it's purely analog analogue and just beautifully created so um
- 37:37
- Unidentified speaker (UNKNOWN): so I can
- 37:37
- Unidentified speaker (speaker_0): crack congrat natural congratulations to your brother. Um Kobe, it's been a delight having you on the show again and um I wish you every success in 2027, which has seemed to have come round very quickly. We only spoke to you in I think it was March this year. Um and uh, you know, so 2027 is looking at uh That's being
- 37:58
- Unidentified speaker (speaker_0): um hopefully a very exciting year for you guys. I mean it it's always hard, I guess. to uh to pinpoint things, especially I guess when certain things are out of your control in terms of how quickly um the the fabs and the um and the customers move. So um good luck for 2027. We'll certainly be watching uh with bated breath and and hopefully uh
- 38:18
- Unidentified speaker (speaker_0): you won't get too many other stupid reports um in uh in the next few months, which I know think just
- 38:25
- Unidentified speaker (speaker_0): you you kind of look at these guys. and they really I mean come on, you know,
- 38:30
- Unidentified speaker (UNKNOWN): they
- 38:30
- Unidentified speaker (speaker_1): don't know how to do simple math. I mean, 164 divided by 18, something, or 20, you know, how can you be going bankrupt? But it was like a so so absurd. I can't believe you know, uh such an important uh news channel that the reports report something like that without even thinking or checking basic math. Well,
- 38:50
- Unidentified speaker (UNKNOWN): you think they put it? through you think
- 38:52
- Unidentified speaker (speaker_0): they put it through Chat GPT just to check.
- 38:55
- Unidentified speaker (speaker_0): Even if it's
- 38:56
- Unidentified speaker (UNKNOWN): just a spelling and
- 38:57
- Unidentified speaker (speaker_0): some of the some of the facts
- 38:58
- Unidentified speaker (UNKNOWN): it
- 38:59
- Unidentified speaker (speaker_0): just seems mad. Anyway, uh I guess it is what it is, but uh the vote of confidence with you and other directors buying the shares certainly uh does help. And as I say, I wish you uh every success in twenty twenty seven and we'll uh we'll circle back, as they say, in terrible corporate speak and and we will uh we will chat again no doubt in twenty twenty seven so good luck with uh with
- 39:18
- Unidentified speaker (speaker_0): with the moves so thanks again for coming on the show
- 39:21
- Unidentified speaker (speaker_1): uh thank you for having me and looking forward to talking again in whatever six months or so
- 39:27
- Unidentified speaker (speaker_0): cool we'll we'll we'll it's a date all right thanks Kobe
- 39:31
- Unidentified speaker (speaker_1): thanks