 We're here at the Lenara Connect here in San Francisco, and who are you? Hi, I'm Peng Cheng from Thundersoft. I'm the CEO of Thundersoft. So Thundersoft, right here, you are a solutions company, software company, what is Thundersoft? Okay, so we founded about nine years ago by a bunch of companies, Hikers. So our business started from day one. We're focusing in this so-called smart device, printing system, and also platform, like Kuginix or Android-based or RTOS-based. So you work on all these Linux platforms? Yes. So founded by a bunch of Linux hackers in China? Yes, we also have presence in Korea, Japan, US here, and also Europe as well. And you're one of the founders? Yeah, I'm in charge of the technical division of Thundersoft. So also strategy, what to do in the future? Yeah, especially what are we going to do in the future? And so there's all these development boards right here you talk about, for example. And so how many people in Thundersoft? Currently we have three thousand people, mostly engineers, like I myself, and we went public two years ago in Shenzhen store market. So we're a public company right now. So mostly engineers, that's got to be a very interesting office, right? Yeah, yeah. And how do you organize, how do you figure out who does what? There's so many projects, right? Oh yeah, yeah, yeah. So we have different approach. It's kind of a mixed the procedure model and also all these bizarre model. And internally we have four business unit, we have this mobile business unit, we have the IoT business unit, we have automotive, BU, and as well as the AI-BU. So four groups working on different project, but they have something in common, is what we call this a smart system, smart device. They share the similar operating system, the similar media, multimedia code, multimedia code or media wire. And right here are some of the platforms you are on. You are on CoolCom, Texas Instruments, a lot of ARM solutions. Oh yeah, yeah, yeah. I try, yeah. Mostly is ARM, right? And also Intel, I have to say. But because we started with mobile, so that's why we're doing a lot of ARM based devices. In terms of quantity of platforms is ARM because there's so many. It is. And so, what do you think about the ARM ecosystem in Linaro? How long have you been part of the Linaro? Yeah, we have been in Linaro long ago, actually from the very beginning. Actually when Linaro founded, we have signed some engineers, we are one of the early participants from Linaro. And now we are part of the student committee of 96 boards. And mostly IOT based activities in Linaro right now. So there's some, right here you're showing some development boards. What are these, for example? What is this one? Okay, so this is the board we did for drone. So if you go to the Apple store, you can buy some products. For example, a product called zero, zero drone, you can buy this in Apple store. And this is what Qualcomm is on here? This is a Qualcomm Snapdragon 800 board. This board only weight about 16 grams, very lightweight. You can see it's very small, just the size of my finger. And so with everything included, including the camera, wifi, everything. So you can basically put some of a projector and you are ready to fly. So very lightweight, powerful. Does your company do PCB designs too? Or you work with companies to do that? Yeah, although we call us Thundersoft, so software is our major business. But we do provide this kind of module or system module for our customers, ODM, ODM brand owner, so that they can build the devices, mostly IoT devices, much easier and quicker into market. Like drone, this one is for drone. And this one is for VR, AR, type of application. That's A20, A20 based. And... So that goes in VR? That's VR and AR. So we're... And hand mounted this solution? Yes, yes, yes. I try my D, I try my D type of device. So that's another example. So customer can hook up their lens camera and very easily to put into market their VR devices. This one, another example is the camera is a Qualcomm A053 based IP camera or smart camera. You can also run a lot of fun stuff, like deep learning stuff on top of this little board. And that's the... Which chipset is that? This A053 chipset of Qualcomm. All right. Yeah. And so you have, for example, a session about automotive solution. What do you do with automotive? Yeah, so we started this automotive business three years ago and that's why we have this automotive BU. And I will have an overview. Okay, so that's a good slide. And this is what we do. So generally speaking, we provide this... The code is a smart cockpit, total solution to either tier one or car makers. So that smart cockpit will contain several things. Mic, some infrared us. So we did that for Baidu. Later, I can introduce more about the works we did for Baidu self-driving car. That also includes the interface 3D engine. We provide the 3D engine and the 3D tools so that you can have a very fancy look. Interface in your car for both cluster and the infotainment. So we provide infotainment as well. And together with the digital cluster, you are more consistent and professional look for the... Isn't it related with the 3D UIN? Yeah, yeah, yeah. You can see this one is a digital cluster, digital cluster meter. And traditionally, you only use a mechanical, it's more like a feature form or feature car. Now, if you go to the smart car, you can do a lot of cool things and interesting and informative things, smart things on your cluster. So this is an example, we have the 3D meter and also I can project some multimedia or human maps on top of your cluster display. You work with the self-driving cars in Baidu solution. What is this self-driving car solution? I will show some video, maybe that's a good example. So this is, Baidu announced a self-driving car in their Baidu World Conference. So this is ARHUD or ARHMI project we did together with Baidu and Bosch. So it's in Chinese, but you can see what it's doing. Is it just cameras or is it using special sensors? Yeah, you can see here, we have this AR projector and information like this line and all this signal for the traffic. And that's what we did. We provide this solution like this tablet, still later we will have an HOD and project all this visual assistance to the driver. It's just based on the camera? Based on the camera only. No need extra sensors, no need at all? No need at all, you can see this is just a tablet. And it's driving by itself? Yeah, with one camera. With one camera, with the off-shelf camera. But need a good camera? Of course, yeah. Good camera and what's the chip set or is it a secret? How do you... The chip set is up to the Baidu to start. Can you run it on the call come like this or you need more? Yeah, certainly. Currently actually it's running on some immediate platform and depending on the customer choice, you can run this of course on call come or other chip sets. So how soon is for example China gonna regulate that every car need to be self-driving car? You know, maybe it can go very quickly, right? I'm from the technical side, so... It's not part of the technical side. I can see China is moving much faster, I put this way. So, you know, people have some concern about the regulation, about safety, but the people also want to get all these benefits and also those innovations. So China, in some sense, a little bit more aggressive. Not from the business sector like Baidu. You can see what Baidu are doing. Actually they're driving this in Beijing, right there. And you can see... And it's fine. Not totally fine, but technically it's fine. Our government also behind that, I have to say. So they're putting more resource, capital, and of course they build all this driving environment. You have to type this, right? So the government also invest a lot so that the business sector can test their self-driving car. I think, of course, it would be great to have zero accidents, but I think the most important is to have less accidents than people driving. And it feels like here in the USA, for example, it's like, no, we don't agree with less. We just want zero. So that means it will never happen. It will never happen. So it just needs to be better, and that's the common sense. Your solution is already better than a human, maybe. Yeah, yeah. The government has to play a very crucial role here. Government, actually, our government is fully behind this, although I'm not from the government sector, but you can see the result. And also, China, we also want to solve the problem of the environment. We have some pollution there, so that would be a key to resolve that. So our government is fully behind this. So there's a lot of future in the automotive. What other sectors are you talking about? If we can go back, can you slide a little bit, maybe? And so this is from technical side. Actually, in order to reach this smart car, this is the architecture we build. So you need to have a good hardware for the smart car. Review smart car is more like a smartphone. It's just a smartphone with four wheels, with actual safety and actual power. And you need hardware. You can see most of those hardware come from the mobile business, mobile sector, like a Qualcomm, like Renaissance, like TI. And then you need a good hypervisor to provide all this security and regulation certification. And also reliability? Of course, very important. That's enabled with hypervisor? Yes. The hypervisor, we provide that as well, together with our partner. And like, for example, the 2X or Green Hill. And on top of that, we run Linux, our Android, or any Autos on top of that. So that, and also, of course, also middleware, especially like the multimedia, like the Kanzi is the one we, actually that's a company we're acquired early this year, it's called RedWire in Finland. So they provide a pretty cool UI engine and the tools for those smart cars. And also the AI engine. So built on top of those middleware, you can have all this infotainment, or cluster, or actually these things. So do you think you can have all the NANARO talking about maybe helping in this, working in this automotive space, right? Is it important to have an open source ecosystem doing a lot of things in here, right? Yeah, it's very important. Actually, yesterday we have the talk and also the presentation and also the discussion. Especially in areas like the hypervisor or Autos or even the Linux, like NANARO is also organizing this special inches group for automotive. We're part of that. And we hope to get more deeply involved as well. And for open source to jump into this new sector, especially self-driving car or this smart car, it's very crucial to have all this innovations and all this problem to be solved in a community type of fashion. What do you think about Android car? Android auto? Android auto. Is this something you can work with? Or is this something Google want to do everything for everybody? How do you optimize Android and customize? Okay, actually, for example, like Android auto car and we are fully involved for some customers, tier one customers. And I think it's similar to the smartphone. So for smartphone, of course, it's dominated by iOS or Android but people are also doing, actually, we're also doing a Linux-based smartphone. Even for that particular domain requires a much stronger ecosystem. For smart car, actually, we do both Android and Linux. For Android, we enhance this Android auto car as what we did for the smartphone years ago. We think that the history has been repeated. Actually, like six, seven years ago, you have this Android for oil or this early. What did you do with this? For smartphone or? For the smartphone. Oh, for smartphone a lot. Okay, if you remember, actually, not long ago, like five, four, maybe six years ago, and even you do not have a full Chinese support there. Arabic is not support there. And the characteristic certification is not there. And the multi-window is not there. And the fast-boot is not there. A lot of things not there yet, but the market cannot wait. And so we did a lot of enhancement works and for this kernel-side, mid-wire for Android. Now, Android is getting more and more mature than before. But for car, it's the same story. It's just the early stage. You need a lot of security patches. You need the UI like what it did. You need the certification. And also, you need this much more mature multimedia, quite different from what you got for your smartphone. So a lot of things need to be done. And different customization requirements cannot be only solely provided by Google. So we did a similar thing years ago for smartphone. We're doing this right now for automotive. And you mentioned multi-window. One thing I'm very excited, I mean, I want to see more of is Android for productivity. I'm hoping someday that it will be useful in a laptop. You know, like, have you done stuff with that? Yeah, actually, this is what I can show you, something like connected. So this is something we did multi-window for smartphone. And for car, it's even more important because in your car, you have many display. You have cluster, you have IVI, you have backseat display. Do you want to do things like running different operating system, different UI, and cannot talk to each other? That's not the way you want to go, not the way our customer want to go. So that's why we provide this so-called contact. Which means as a developer, you only develop once, for example, a media player. But you can easily configure this to control either through your backseat or through your cluster, control something that you are running on your IVI. And without having to program differently for those different platforms. So that's what we provide this so-called Kanzi Connect. So that for all these display, you'll have one media wire, one education, and one consistent look and feel to the end user to the customer. And actually, Audi did one demo with Google running this Kanzi Connect technology, which we provide. They think that as a future, you're going to see more and more of this kind of a consistent, connected display in your smart car. So over the last nine years, how many products were shipped with ThunderSoft technology inside? Is it like millions and hundreds of millions? Or what is it? In terms of volume, we provide those solutions and also services. So the total volume is approximately more than millions because we did a lot of smartphone reference design support. I will use an example here to show some products we built. It's a good example, actually. For example, this QRD is the first Qualcomm reference design we did in 2010. You can see that's how Google Android looks at that stage. What is QRD? It's Qualcomm reference design. So you worked on that? Yeah, we provide the software. You were one of the first to provide the Android? Yes. Android for smartphone kind of solutions? Yeah, yeah, yeah. Actually, before this one, actually, we started from Intel, I have to say. We started from Intel MID. And that's when we built on the Linux. It's called Midinix. Actually, that's the root of Amigo or now called Tizen. So that's how they started from the Nokia MIMO project. There was also based a little bit on the TI OMAP that kind of started Amigo. Yeah, yeah, yeah. Before it started with the TI for Nokia. And before that, we ported that on Intel iPhone. But then when the smartphone kickoff like the 2007 iPhone comes to life, and the Qualcomm switched here from Linux to Android, we do the same thing. So we do more and more Android. Now IoT and automotive. What is this camera turnkey? Camera turnkey is like this. So the camera, similar to the smartphone, is getting much, much more smarter than before. Before that, it's only dumb camera. You put there and you're streaming to the cloud. Hopefully, you do not watch this. You only watch this when you see some instance. If something got stolen, you go with that. But that's not the thing people are looking for. At this stage, people are looking more and more intelligence, like the AI things. So you already do smart IP camera in 2012? Yeah, yeah, yeah, yeah, yeah. So the smart decisions of what to send to the cloud? Yeah, like this one. So this one, actually, you can order from our website. It's based on this module. Actually, that one running exactly on this module is 053, this module. You can run the AI applications. You can stream out all these 4G, 4K content. And it's very powerful. And that's designed for the smart camera. But this is recent, no? This is a recent. Recent version. Yeah, yeah, yeah. Recent device. Yes. And you're also in the robots and smartwatch and? Yeah. And VR, AR, and the drone, like this one, yes. So you're all over the place? Yeah, because operating system is everywhere. So we're going with them. And if we go back here, there's something about the IoT turnkey with the narrow. Yes, yes. So that's one of the big fields right now with the narrow is the light group. Yes. Doing something in there. Are you a part of that? We're part of this 96 boards. We're a steering committee member of 96 boards. So we're heavily involved with their activities. Yeah, yeah. All right. And then in the future, there's so many more opportunities in this world for using technology in a very smart way to do more. Then suddenly maybe what is the next iPhone? What is going to be the next killer app? You might be part of this. You're probably already working on it. So next one will be something like this. Of course, everyone is talking about this, not that guy. This guy. AI. So you can see everyone, I will talk about AI. But we're the one that actually really doing it. This is a demo we did earlier this year for Qualcomm. It's running on their 835 platform with our algorithm and using their Snipey neural processing engine to show the good performance. So does the 835 have an SOC, an AI silicon? Or is it just using? It's running on their high silicon. Hexgon, DSP, and also their GPU and CPU. All right. But now the trend is that more and more SOCs will maybe come out with AI on the silicon. That will bring much more work for you. Yeah, that's what we like. For example, for Hexagon, if you watched the news, actually last week, they have an announcement. And also previously, Huawei Mobile also has an announcement in German for IFA meeting. So if you watch the news, you can see we also have the joint announcement and the joint demo with Hexagon to show this full detection and also the face detection using their Hexagon AI core. So I would like to see a detection of everything inside indoors, detecting everything. That's going to be very exciting. That's going to change the world, right? Yeah, it's getting there. And especially, it's very important to detect all these things on the edge, on this module, not to send everything to the cloud. It takes too much time, sacrifice too much privacy or security issue. So that's getting there. And we provide those AI algorithms. And most importantly, this AI enabling layer. So we bury the gap between the silicon people and the algorithm people. Those people sometimes hard to talk to each other in this big different language. So we are sitting in the middle as an operating system guy. We know both. And we bury the gap between those two type of one. Are you kind of making the operating system for AI? Yeah, it's kind of AI OS that I put this way, more like a marketing term. But also technically, we think it's right because you need this new AI OS to schedule the resource, like a DSP, GPU, CPU, or even AI core more intact in a way, more efficient way. So you need a scheduler. And you also need a monitor to monitor the performance and the thermal. And you also need even the IP address store, right? There's no way to provide you, to provide the algorithm store like what Apple did for their education. So that's actually operating system. That's what we're doing right now. And everything you've been doing for the last nine years, everything is open source? No. No? Well, you know, we started from the open source community. We're a bunch of hikers. We have this passion for open source, for freedom. But frankly speaking, after this nine years of this startup, we're still a startup. And most of people go to their forties. We reached this stage of combine this idealism with all this, I don't know how to call this. Capitalism. Ferguson. I think that there's some space spot we can do the job. If you go to further left or further right, not good. We have to find something in the middle. In Chinese, we have something called the Zhongyong. I talk about philosophy, but... Zhongyong. Zhongyong means not open source. Zhongyong confuses? You know, you know, confuses? Or... Confuses? Yeah, yeah, the philosophy. Most of the Chinese people agree on, including I myself. So that means you have to find the internal peace. You cannot go extreme to either side. I know people are enthusiastic about open source, so open everything. It's not good to open everything, I have to say. Although I'm in the open source community. Some, but it's also not good to close everything. Like, you know, Apple way or whatever. So there's something in the middle. So something they open, like for example, for automotive, we need open source hypervisor and autos for those low layer and so that people do not duplicate the effort. There's absolutely no reason to spend so much money, so much effort to some similar, requires so much security. We just need one or two very good operating systems like Enix in automotive. But we also need some algorithm and applications in the middleware and so that we grow the ecosystem. Those things does not have to be open source. It's perfectly fine for people to close their algorithm, fire the IP, the patent to get money. That's how American got bigger and how we think the product like this can get more attention and more market as well. Okay, so that's.