 And here we have some huge ARM processors here at AD-Link. Hey. Hi. Hi, Nikola. Please introduce yourself. I'm Henry. I'm a product manager at AD-Link. And I'm responsible for most of the ARM designs. That's to say low-power ARM, the unpaired parts that that's from Hank. So maybe I explain a little bit. We started with the first release, which was based on a PX-30 about four years ago now. And the idea was to bring industrial parts to people that have experience with Raspberry Pi, but still want to move it to an industrial situation. Then we released another one, which is based upon the IMX-8M+. Similar idea. You have an industrial part, but still with all the commercial interfaces that Raspberry Pi brings you. And also we support the whole ecosystem. That means all the hats, all the cameras, you can run on them. Then in 2022, we did the iPiSmart RB5. This is based upon the Qualcomm RB5. That's the robotics chip. And some one of these? Yeah. This is the RB5. All right. So maybe let me explain you very fast what we have. So we do have HDMI. We have dual Ethernet. We have four USBs. We have 40-pin, 100% compatible Raspberry Pi header. We have another header where we have extra GPIOs, I2C and an SPI. We have dual LVDS or DSI interface. So on smart, this is a mixed interface. We have onboard four-lane and two-lane camera. That's the SMART standard. But in the case of the RB5, we have another four cameras on top of the module. Also, on the bottom, we have two M.2 slots. So it allows you to put storage or an extra 5G module. And we also have a SIM card reader and an SD card slot. So it's a very small but pretty complete board. And moreover, it's very affordable. It's really, let's say, entry-level. We try to target developers that do not have a very, very big wallet to stay compatible with all the designs because we also have an Intel-based development kit. No, it's not here, Elkhart Lake. We have a removable audio codec to stay compatible with all the socks we are working with because incorporating audio is a little bit troublesome. So we can remove this little board here, and that's the whole audio interface, iSqrS or HDA. So it's your carry board that goes through, and then you have these modules. Yeah. This is all based upon SMART modules. This is a defined interface, the SMART 2.1. That means I can move this MediaTek module from there and this Qualcomm module from here. It works. I can just switch to an XP module or even a Px30 module. It's a universal standard. Look at it as USB, right? You can take a USB drive from any brand. It works. It goes similar. And is the full bandwidth on the connector there? Everything can happen. The full performance goes out to the carry board? Yes, indeed. For example, on the Elkhart Lake, we have 2.5 gigabits Ethernet. While on the RB5 and the MediaTek and the NXP, we only have gigabit. But that's the limiting part is the CPU there. And the HDMI can be 2.1 with the 4K and everything, and even 8K maybe sometimes? It does 4K on the MediaTek on the RB5, but it only do 1K on the 8M+, and 1K on the Px30. But that's the limiting part is the SOC, not the design. All right. It's like 1080p then. Yeah. Yeah. So how about these modules? What do some of your customers do then after all kinds of projects are possible? And do you help them to make it done? Yeah. For example, we support the customer with carrier design files. So let me explain a little bit how it would go in theory. They buy a kit. Of course, not everything is there what they need. But Raspberry Pi has an enormous ecosystem of hats. So let's say you want to add RS485 and we don't have it. You buy an RS485 hat, you put it on top. You need some extra special cameras. We have the interfaces to do it. And then you evaluate. You make a proof of concept. And then you can say, OK, this proof of concept is passing in terms of speed, in terms of probably pricing. And then the customer starts to develop his own carrier board. Or we can do that for him. In case he develops his own carrier board, he gets design files from us. Once he finished his schematics and everything, he can send them over and we validate them. And then either way, he goes to an EMS to make it or we make it for them. What happens in a lot of cases? And he has a whole solution. Now you would say that's a bit troublesome. Why don't you go chip down with everything? Well, if you go chip down, it's not future-proofed. If you are going with a module later on, you can switch to a different part. You can go from, let's say, from an XP to MediaTek or from MediaTek to Qualcomm. And you don't need to redesign your carrier. That's quite important. And for example, medical solutions, where you want to keep the changes as small as possible. The module is one part. And that's easy to re-certify. And maybe you design something that should be good for 10, 15, 20 years. Yes. And then you just want to upgrade to the newer chips every five years or something. So the future version of the Qualcomm, for example. And just swap it in. Yeah. And it should just work. Qualcomm, for example, they go with a lifecycle of about 10 years. And XP, on the other hand, we have 15 years longevity here. It means even after 15 years, you will be able to find a module that fits your carrier and go ahead for maybe another 10 years. And that makes your ROI on your board development is very good. You keep on using it. Maybe it's like a big expensive machine. You put it in that gets future proof. Yeah. It's future proof. You could say like this. Most of those devices really are edge located. They're low power. We're limited to 15 watt on the standard. And the lifecycle there is not that long. You could say five, six years. But still we have customers that use it in some different applications. Let's say weighing, scales. These needs to be serviced for 20 years. And that's where they come into the module to make it future proof. What's the idea with iPie? Well, you have the pie. And we have the industrial pie, right? All right. And so how's the uptake? Are you very, very famous, very popular? How is a lot of people using working with you? In terms of modules, I would say, we have a wide scale of modules. We do Q7. We do SMARC. We do ComExpress. We do ComHPC. And in terms of volume, we're either number one or number two worldwide. No kidding, really. SMARC is really something that starts to ramp up. In three years, we tripled or quadripled the volumes that we ship. So it's really a standard that starts to catch on. And it eats away from three and a half inch boards that are fixed on a single CPU. People want scalability. They don't want to be locked anymore. And also people, I guess the last three years, really show that people want to have a choice. If the chip shortage is in one vendor, they want to switch to an available one. That's a good example. If we cannot deliver the SMARC module, you can go to any other vendor and buy the SMARC module for him. And we have quite some projects where with ComExpress we actually deliver 50% and another vendor deliver the other 50% just for supply chain safety. SMARC is the same. And you're happy to work with all your competitors out here, right? Yeah, we're one big family here, right? It's like one big ecosystem. And especially the ARM. It feels like ARM is an open competition and a good competition and a good innovation platform. There's so much happening pushing people forward. Yes, it's growing very fast. One of the things is, let's say, you have an edge device and you need to do some machine learning, right? And you take the Qualcomm parts here and it has 15 tops. If I want to emulate that on x86, I consume triple or maybe quadruple the power. Now, you calculate that over five years. That's the cost of the device, the difference. So ARM is free, basically. It's probably much more than the device, no? We made a calculation, so we compared the i3 with machine learning add-on. So I consumed quite a lot there, 90 watts. And with RB5, I consume about 10 watts, same situation. And over five years, with the cost of the electricity in Germany, I would have saved $1,560, something like this, US dollar, if I go with ARM. So that's a little bit more than the device. Of course, the electricity price is nowhere the same everywhere in the world. In the US, it came somewhere close to $400, still. It's quite important, right? Yeah, $400 is hard for a second-hand car for some people, I'm joking. So you say it's affordable, but it's still for the industrial market? This aims at the industrial market, yeah. So it's not $35, right? Like some of the Raspberry Pi was saying a long time ago. It's not $35, and the Raspberry Pi is also not $35 anymore, right? I think they sell them for $150 or something. Exactly, yeah. So if you go with a kit like this that has a lot more possibilities, I think we sell this one online for somewhere $250 or $260 US dollar. But it has machine learning, it has an NPU inside. You also can hook in the ecosystem of Raspberry Pi, and you have something that is future-proofed. It's not a single board. It's a modular system. So for some people, it makes a difference. And here we are at your office on a weekend. And for example, right here, there's the Ampere solution. Yes. And can you describe a little bit? Do you have the direct connection with your, and this is a real, right? Yeah, that's a real one. Yeah. Do you have direct connection? You know exactly where the engineers are capable of what they're able to do, and then you can implement new ideas, new projects very efficiently? Yes, we do. For example, the development team actually here are all the IDs born, and we brainwash them later the engineers to do it, right? But we are boxed. We are working in standards. So you need to work within the limits of what a standard allows you. You cannot go too crazy. If you go single board and out of band means you make your own stuff, it's different. Here we need to work within the rules that the standard has defined. So it brings some limitations, but it's also a safety for the customer later on that he can upgrade his system to something else. But there's a saying, less is more. So when you have constraints, you can be more creative to get more done inside those constraints sometimes. Yes. So what we are doing now, we are more innovative on the software side. We try to grow the software side that it's reusable over all our platforms. It means one software for everything, and we're working hard on this. Then also, of course, we go a little bit out of band with the standard for the RB5, for example. We can have six cameras, but we have two following this mark standard, and we have another four on the side of the module. But that's fine. If you can walk over there. So you have this website. People can just go with the credit card and buy all these different platforms here. Yeah. One click. There's some other you can. Yes. We have ComHPC. We have ComExpress. We have SMARC. Later on, I will also post the Q7 module there. So that people have an idea of what a standard can bring for them. That's the first one we made up, basically. Right. That's history. And there. And do you have the software guys somewhere? Do you have different floors? We have nine floors totally. So Idealink is not a small company. There's more than 1,000 people in this building. Here, 7th floor is actually ComModule and also EV. That is a little bit further there. And engineering is 5th and 6th floor. Where we have the layout teams. We have the double E's. We have the DQAC. So where we verify the design quality. We have the, let's say, halt testing. That's the first floor in the basement, B1, where we have big lab with machines to do accelerated quality testing. Cool. And the factory is the building next to here. There we also have, I think, four floors. With SMT and everything? SMT and assembly and everything. And high quality SMT solution? Yes, high quality. And also mirrored with our factory in Shanghai. We have a very big factory in Shanghai. So those two are basically a mirror from each other. Let's say a big earthquake happens in Taiwan. We can manufacture tomorrow in Shanghai. A big earthquake happens in Shanghai. We can manufacture here. So just to stay on the safe side. And you can push a button if there's a big demand and boom. There's just one more. You don't have out of stock? Supply chain, everybody has the same problems, but we really did our best. And during the two years that there were supply chain issues, we even grew. Because we have quite a good purchasing department. And we have a very big warehouse. Check out the Wyze. Wyze is such a smart system. I think I'm saving over a thousand euros every year. Because I use it everywhere when I pay for stuff. You know, you want to pay for stuff in the world. So you can check out my longer video for where I explain some more. Why I think this is amazing. And it's free. You can just sign up with my link down here. And you can use it on your Android Pay, your Apple Pay. It's like a prepaid Visa debit card, right? So you need to put a little bit of money on it. And you can just put money on it from your local bank account. And there's zero fees. The fees are so tiny. It's like five to ten times cheaper than your bank. Don't use PayPal anymore. Don't use Western Union when you send money to a whole bunch of countries around the world. Just use Wyze. It's way faster and it's five to ten times cheaper. So really check it out. It's cool. And I appreciate if you use my link below. Thanks for watching.