 So we're very pleased to welcome Edward Keshishyan. Mr. Keshishyan is a long CV. I'm not going to talk about his past so much. I'll just say that he's a senior VP at Qualcomm, one of the largest IT companies of any kind in the world. And in particular in the semiconductor industry, maybe V, or one of the absolute leaders. And so a very fast-moving industry as we know. And there's semiconductors everywhere, literally and figuratively. And so today the topic we're going to hear about is the industry itself sitting at an inflection point, like never before. It includes an AI, machine learning, you know, the things, connectivity is driving new requirements. And so living through a paradigm shift is very exciting. It's very, well, fascinating to be in the middle of and to be on the other side of it as it happens. And so the automotive industry is what's propelling a lot of these changes. And so we're going to hear about sort of where is the world of automotive and sort of recent history and looking ahead how that sort of intertwines with semiconductors. So the requirements, these requirements are pushing the boundaries of what's possible in leading technology process nodes, product quality and cost and system deployments. In addition, of course, there are security privacy and policy issues, and all of these kind of come together to affect the semiconductor industry. So that's kind of my paraphrasing of the abstract we receive. I'll end now and pretty soon we'll fire up the presentation, but thank you. Thank you. Thank you for the introduction. While we're waiting, I'll do, how many of you have heard of Qualcomm? Okay, quite a few. Actually, the guys who work for synopsis don't count, so. Okay. And you know, I know, but some of the faces, I met them last year. I met them this morning. So what I would like to do is until they get the thing ready, maybe I'll talk quickly about Qualcomm, what we do, but also during the presentation, just to make it interactive. If you guys have questions, don't wait till the end. Just raise your hand, and I think it's better that way, because I don't have too many slides. I think it's about 16 or 14, but the PowerPoint is about, what, 35 megabytes, so it's very heavy content. So, and I tend to go, I tend to, usually I tend to go very fast with the material. That's why you have to stop me if you need me to slow down, explain something a bit more, so let's make it interactive. So Qualcomm has been around about 30 years. We're the one who started CDMA. It's a different way of encoding technology, encoding bits over the air, and basically that revolutionized the whole industry. So we have, we created the mobile industry pretty much, and now we're moving from that part of connecting people into connecting people to everything, and that's part of the talk I'm going to give about, give today. What's interesting about Qualcomm? I joined Qualcomm 10 years ago, 2007. I moved from Toronto. I used to be in Toronto doing graphics for ATI, and then I did a couple of startups. What was interesting for Qualcomm at first for me is the scope and the volume of things we do. So just to give you an idea, in 2016, we shipped close to 850 million chips across our product portfolio. Some of the products over their lifetime have attained close to 800 million units shipped. So you can imagine the extent of challenge we have, both from the development side, the design side, specifying the right thing, you have to have the right product specified so that people buy it in those quantities. But also everything that comes with it from the metrics we track, we have something we call the PPA, it's power performance and area metrics, and then any inefficiencies in the design, let's say if you're going to turn that into cost function, let's say you have a 10, 50, 80 cents or a dollar kind of inefficiency, 800 million units shipped, it's 800 million dollars that's down the drain pretty much. So the team is very much focused on, first of all, putting these chips out. We have capacity, we take out about 30 million, 30 chips, 30 new products every year. And by the way, if you have an issue with the nomenclature, I use lots of acronyms, just stop me and ask. My team does the large chips, so the last one we taped out was a week ago is close to more than 6 billion transistors. So you can imagine the amount of effort that goes into that. The team is very global, so I have locations in Canada, in the US, Middle East, Europe, Asia, and we deal with all kinds of aspects of the particular product. So we start from discussion with the foundries, my team develops all kinds of standard sales, memories, surveys, all the way to specification of the product, to architecture, to post-silicon validation, eventually commercialization. So it's a large operation. Qualcomm is about right now, before any acquisition or any merger, we're about 32,000 people, about 18,000 of them in San Diego and the US, and the rest is worldwide. So what else? Yerevan is one of our favorite suppliers, so that's why I can't say no to him, so he drags me everywhere. Two weeks ago we were in Dallas, but again, this has been a very nice surprise for me, being in Yerevan. This is my second time, I was here last year, and I realized the amount of IP, the amount of products, the amount of engineering that's coming out of this side, particularly from synopsis, and I was very pleasantly surprised. So you guys are doing a great work, and hopefully you continue this, and my first exposure to the university here, hopefully there will be many more. So let's wait for the real stuff now. Yeah, I can tap dance for so long. All right, okay. Do you guys have a clicker? Oh. Okay, all right, yeah. Thank you. All right, so like I said, Qualcomm basically came up, revolutionized the whole mobile industry, the cell phones, most of you are fairly young, but some of us still remember the large bricks and eventually the flip phones and so on and so forth eventually to these days where we have the smartphones, right, and the smartphone, the whole generation that came with it and everything that's associated with it. I mean, things are accelerating and the rate of change is accelerating, and that rate of change is a double exponential for you guys who are interested in this type of thing. And the last one, the last major driver, it's been the automobile industry. First of all, because they have lots of money, they can drive things, right, so they're spending lots of money and then they're offering something new for everyone, right, and the major, major impact is going to be how these experiences are going to change from being connected in isolation, you're with your cell phone, to being in environments where your whole ecosystem, your personal ecosystem moves with you, right. Let's go through this. Okay, so what I'm going to talk about today, again, if you have questions or you want clarification, please ask. What I'm going to talk about today, what the automotive industry wants from a company like Qualcomm, what type of devices they want, what type of systems they want, and then take that and kind of translate that and we'll put it in context of how it's shaping the way we look at things as design engineers, as semiconductor professionals. So if you look at, I mean, the most obvious thing at the top is really self-driving cars. So people are completely fascinated by that concept. And honestly, I think it's, once we're there, I mean, probably we need a decade or two, I mean, complete immersion, complete self-driving, it's going to be a major change. Think about it, how much time do you spend in your car and per day? First of all, from your time perspective, but also how much that particular appliance is costing you, right? You spend probably two or three hours in a car, depending on traffic, how far you drive, and that time in the car, usually, the best you can do is, well, some people text and do other things, but you shouldn't, but, yeah, the best you can do is maybe listen to some audio books or listen to music and things like that. So it's a completely wasted productivity as far as our personal time is concerned, but also from the cost of that appliance, again, if you look at the whole 24 hours, you'll probably, that thing is being used for four to six percent of the whole time, and it's a complete waste. So this is going to revolutionize, basically, the way we become more productive. Imagine you're in your living room and your living room is being transported to that appliance, that's moving and taking you places. So that's the one obvious thing, and it's driving, by the way, lots of technologies from all the machine learning, all the AI, all the obstacle recognition, diagnostics, self-repair, collision avoidance, and things like that. The next one is connectivity, and this is particularly very, very valuable for us because it's, at the core, Qualcomm is a telecom company, is a modem company, and this is where actually our core competencies is, and now we're seeing, with all these devices connected together, we're seeing all kinds of modems from all the way to the different Gs, whether two, three, four, or five, to things like Bluetooth, things like thin modems, things like wireless LAN and so on and so forth. The other one is, I already talked about it a bit, it's really, you're taking your whole space at home, the stuff you're used to, where you have your music, you have your reading material, you have your videos and everything else that comes with it, you're taking it in that space, which is, again, an appliance that's moving and taking you places. And the last one there is efficiency. It's very expensive to run that infrastructure, whether it's the infrastructure of the highways, whether it's infrastructure of the whole transportation system, it's very expensive, as I mentioned, to run the cars from maintenance to fuel efficiencies to just general cost of having that type of expensive equipment. And because of all the smarts that are being put in these cars, we started kind of exposing all the inefficiencies, whether it's your driving habits, how fast you accelerate, then you stop suddenly. We had that experience in the last few days. Our driver was very special. To everything from proper climate control within the car, matching the environment, there are electrical engineers in this audience, people always talk about impedance matching. So there's lots of areas where you want to match the car to the environment, to how traffic is flowing, how closely you're following the car in front of you, how closely the car behind you is following. Just to give you an idea, even in today's technology, there's about 300 microprocessor type chips or microcontrollers in a given car. And I'm not talking about the really expensive ones. I'm talking about just the car that average price car you buy. And there's a cost to it. And we're trying to bank on all of it. So I'm going to talk a bit more in detail about all of this. Okay, so, by the way, we tried to completely... This is not an Audi or a Volkswagen or it's just a generic car, right? No brand here. So what we're trying to do is basically the first paradigm is very, very important. I mean, today, as users of smartphones and mobile devices, we connect to people, right? You know, we connect either for voice calls or we connect to people through social media. Now, with this change in paradigm with people... I'm sure you guys heard of Internet of Everything or Internet of Things, the IoT stuff. You guys heard about the cloud. You guys heard about all the machine learning that's going on. So everything is becoming connected to everything else. So the cars, and I'm going to talk about that, the cars would become connected to multiple different things, either stationary objects, people, or other moving objects. So, again, this is creating a whole bunch of challenge for how to manage this very complex system that's trying to make decisions, you know, split, kind of split millisecond type of decisions. In car experience, transforming that, I mentioned this. Basically, you'll end up with what you're used to at your house or at your favorite workplace or if you're at a resort. So you'll have all of these... Everything that you're familiar with that you're used to is available with you, you know, in your car. And then eventually, all of it is going to lead to autonomous driving. There are countries where they're actually forcing their auto industry to remove all kinds of human interaction and vehicles driven by people within a decade. So Norway, by 2025, they want to eliminate all cars, all traditional cars. So everything would be autonomous, right? You won't have any control as a passenger in that particular device. So it's going to be controlled by AI, controlled by software, controlled by whatever's available in the cloud. And, you know, some of us, it's probably a bit scary. On this side, you know, you see this is the technology that it's going to make everything happen seamlessly. So Qualcomm has been developing all these... You know, G is a generation of a particular telecom spec. So we started from 2G. I mean, there was always the 1G, but, you know, no one really talks about that. Those were really the voice call days, the very early days. But we started with, you know, 2G, 3G, and the latest 4G, where you see you get... There's much more experience than what we traditionally used to for a wired connected device, right? I mean, if you had to connect your device to the Internet, basically to a wire, today we have the same experience on... You know, on our mobile devices. So it's very high-speed Internet. It's almost always available. And you can do lots of different things with that type of a paradigm. If you think from this whole compute area perspective and what's enabling this technology, if you follow what happened in the last maybe two and a half, three decades, it started with the microprocessor, right? Intel started it, then IBM with PowerPC, then eventually different companies. So the processing became very, very fast and very, very mature. Then eventually the memory itself, the memory speeds and technology caught up to it. So then we had fast memory and then very fast CPU. So you can do lots of processing, but it was very localized because there wasn't any connection or connectivity available. So then the telecom area and then all these high-speed links and high-speed interfaces came about. So then you were able to have very high performance, high-speed processing, and then you had multiple of these connected to each other that you can think about parts of a different brain and then connected and partnering to do some complex computations. So this has been really what's been driving all these different generations. And then for 5G, the interesting part about 5G is that there's no clear definition yet. Everyone is trying to have their own definition based on their product roadmaps and what they're trying to bring to the market. But one simple way of thinking about this is that... I don't know if some of you have experienced gigabit speed at the internet where your upload and down links are close to a gigabit per second, really fast. So 5G is basically... you will have that on your phone or on your mobile platform. So you have at least gigabits per second downloading and uploading speeds, which basically means that you can upload and download very large files. There's a hidden message that we got to speak faster. So basically it means that first of all, there's very high speed links. Secondly, the latency. Imagine this. If two cars are going very fast on a freeway and there's some decisions to be taken from an algorithm or from an AI type of entity in the cars, you can't wait for that decision to be delayed. The decision has to be very quickly and happened almost instantaneously. And some of the metrics that are feeding or data that are feeding into this decision are in the cloud. They're not in the car. They're not in the device. So there's always have to be this communication to the cloud where either the car is trying to fetch some data or trying to get some machine learning algorithm and then you need the response back very quickly so that it could be a decision to slow down. It could be a decision to move lanes. It could be a decision to avoid some obstacle. So think about 5G enabling all of that. Another view of 5G is imagine these large events, a stadium where there's a concert, where there's a sport event. You have 70, 80, 100,000 people there. And all they're trying to do is they're trying to capture the moment and share it with their friends and families. And the problem is with very large kind of crowds like that, there's problems with the signals. Just the network gets very congested. So 5G pretty much is going to enable that type of an user experience. Okay, so I talked about most of this. These are some numbers. This is very, very large numbers. And basically, within the next 20 or so years, you see how much money this industry is going to generate. So it could be anything from buying the appliances themselves to just developing the whole ecosystem. And I'm going to talk about the ecosystem and what it means. It goes from any connected devices to smart cities and so on and so forth. And the majority of this cost is going to go to really the infrastructure aspect of it. It's very, very expensive to build the cities that are smart, to build the highways that are smart, to build the power grid that is smart, to build everything that you see. And we take it for granted that that's very smart. And it's going to take lots of money. So this is basically going to drive the whole thing. Again, if you look at the top connecting with the high-performance compute available with high-speed connectivity available, it's opening all kinds of new possibilities for these connected devices. So things that people didn't think about before. And I'll talk about that. Suddenly, it's becoming a possibility. New services. I don't know some of it. I'm not going to talk about augmented reality or virtual reality, but think of it this way. You're on vacation somewhere. You take your phone. You hold it up. You have the GPS. Obviously, the phone knows the GPS location. And what you do is you scan wherever you are. And what happens is that your friends or from your social network would have left some clues for that particular geolocation. And then you can actually see it live. You can communicate with your friends. Basically, someone might say, oh, go eat at this particular restaurant or drink at this bar. Or there's a very interesting vista not far from here. So that's the augmented reality aspect. So that's a new service. Then you have the whole virtual reality thing, where you're immersed into completely something new. So all of these are enabled by this 5G high speed networks, low latency networks. And of course, new user experiences. There's a famous, well, not famous, but it's something that's been circulating a lot. You have someone doing a mountain biking fairly high speed. And then he has his helmet. He has his GoPro. And while he's going down the path, he's broadcasting his experience firsthand as a first person experience, because the camera is capturing what he's doing. And it's broadcasting it to his parents in whatever they are in their house, maybe far away, maybe across the globe. So these are the type of things that are becoming almost trivial. People are expecting that. This is an interesting, very interesting aspect of what's happening. Right? I mean, that 2.5 or 2.4 trillion, I explained. The one thing interesting, if you look at the V2X. V2X, V stands for vehicle. And two, I mean, it's just basically connectivity. It can be V2 another vehicle. So it could be V2V. It could be V2I, like IoT stuff, infrastructure. So let's look at the most basic one. Up on top is vehicle to network. The network, in this case, is really the cloud. So there's lots of very, very large petabytes of data sets that are going to stay in the cloud. And these things have all kinds of information about either location information, information about a particular car, information about particular type of interaction that might happen, because I'll go talk about what type of cameras and sensors are in the car. So that's very important. That your vehicle be able to connect to the network and be constantly on. Because think about this, for the newer systems and the newer cars, you will not have the possibility to adjust course, right? The other one is really vehicle to pedestrian, where you're trying to monitor what's happening around you, people walking. And you want to make sure that you don't go over someone, it's usually fatalities you want to avoid. Vehicle to infrastructure, whether it's traffic lights, power grid, different type of smart city type of thing where you're trying to find parking, let's say, and your vehicle immediately will figure out where to park or where parkings are available. And then all of these things are basically demanding high performance type of products. So that's why some of the products we're working on has eight CPUs, eight large CPUs with 3.5 gigahertz and above speeds. They have high speed links. They have graphics co-processors. They have network processing units just to process some of the artificial intelligence and machine learning aspects of it. So we call them NPUs for deep neural networks and all kinds of sensors and interfaces, basically. Okay. Two years ago, we were trying to find a specification for what our next generation device is going to look like. And someone said, you know, in the chip, we have something called a display controller. Basically what it does is it projects or it controls the display of what goes on, you know, what's displayed basically on a flat screen or whatever surface you're trying to show the data. And they said, we need the 11 types of display to be controlled in a car. And then we were wondering, because we're used to these type of things, right? You know, the display on the back seats where your kids or whoever's sitting in the back are watching a movie. So we started counting. So, first of all, this part, the whole instrumentation is going to be electronic. So, I mean, it's going to be digital. So you can change things, you can change the dial. So that's one display. Then you have the center console, that's two, where you have all your controls, whether it's climate control or anything else, or your map. Then you have a display by the driver, by the passenger side, so that's three. Then you have two of these. So that's five. And then you have to accommodate for the third row. So two more displays of this, so that's seven. So we're saying, okay, where's the other four coming from? Then you have the two side mirrors. That's nine, right? So basically the side mirrors are completely turning into camera-based displays, right? And then you have the review mirror. So basically, we're taking anything that's simple, that works, things like very basic, like mirror, 18th century technology, and turning it into something that completely blows your mind. It's everything digital. Not only it's digital, it's also, because it's digital, there's going to be all kinds of AI working on these, whether it's pattern recognition, or just processing of information, and so that the car can take decisions on them. So these are the things, by the way, if you have a chip that's driving 11 displays, so you need a larger CPU there, you need a larger pipe for the memory subsystem, you need a larger GPU there, and you need all kinds of different processing so that you can accommodate for that type of request. Okay, this is another interesting one. Again, you've got to think about, yes, what, okay. So the question is what we gain by replacing mirrors with cameras and with displays. Basically, what happens is that, think about it this way, eventually this is all driving or getting to a point where you will not have a driver in the car. So a review mirror won't meet, won't mean anything, right? Or side mirrors won't meet anything, right? So think about this, this is kind of an in-between phase where we're getting ready for a completely autonomous car. In an autonomous car, you don't need a mirror, right? Because the car is doing... You can have the display, but so it's kind of an intermediate product to go there. Like, if you look at some of the cars that are claiming they're autonomous, they still have drivers in there. So they still need all these cameras to capture. They have the back cameras, the side cameras, and then they're saying, because we're going to capture all of this information, we're going to process it, we're going to digitize it. For instance, think about it this way. You're driving at night, right? You have really difficulty in looking at the review mirror. I mean, they do kind of optical shading or grading so that you can still see something. But imagine this, your camera turns into an infrared camera. So now you have full visibility even if it's completely dark of what's going on around you, right? So things like that. So this is why some of the high-end cars already have that, right? So it's all for safety, collision detection, things like that. Over time, the need for mirrors is going to disappear, but the need for processing is still going to be there. It's just the fact that you're not going to display anything else on a mirror, but the processing is still there. Okay, another, we're in a kind of a cross-section within a city, so entering a lane. So this is some type of decision-making between two cars, right? What do you do if you're a proximity or a car? You've got to change lanes. So all of these communications have to happen. It's almost like the cars have to figure out who has the right of the way based on some algorithm. Finding a charging station, like finding a parking space, all these decisions have to be made without human interaction. Then you will have things like updating maps, where you have something might change or we have an accident. So things change on the fly that the algorithm would redirect traffic into different zones and different streets, if you would like, and pedestrians. I mean, that's the most critical one where you don't want your car suddenly to start going over people, right? It's very difficult. I mean, think about it. If you're driving today and you come to pedestrian crosswalk, what happens? There's almost like an unofficial handshake that happens where they look at you, they see that the driver acknowledged them, and then they have the confidence to cross. How do you do that where there's no driver? And then you still want these people to cross in a safe way. You have to have cameras looking at facial kind of patterns and then acknowledgement, which are cues that are not obvious. So how about other things where you have a shadow and then the car doesn't know if it's a shadow or an obstacle, whether it can go over it or have to turn around it? I mean, these are some basic problems that still people have solutions to that, but it's not perfect. So these are the type of things that are driving all these requirements from a processing perspective. Okay, this is something about where, you know, how this industry is driving our requirements. I think I talked about most of them. I'm going to talk about accuracy a bit. Today, we have GPSs that probably, I mean, military grade, you probably pinpoint within a few centimeters. The best ones, it depends whether you're on a GPS service or you're just triangulating between cell towers. So it can be accuracy, it can be anywhere from, you know, a few meters to, you know, tens of meters. Now, with the 5G and millimeter wave, you can actually pinpoint someone's location by, you know, granularity of a millimeter, right? So these are a type of things that you would need when you're navigating to some foreign terrain or, you know, a new city or you just want the car to be very accurate because you can't have your car driving and then the uncertainty being, you know, a meter here or there, right? You see that's pretty much a disaster. So that's one thing. It's also a technology that we're working on. I talked about, you know, everything else and I'll talk about security in a few slides. Okay, so those are the areas that you're going into policy issues and it's a serious problem. I mean, who do you sue? Do you sue the software driver? Do you sue the car manufacturer? Do you sue the software that's being uploaded? Actually, there's no, there was one interesting case in San Francisco. This was maybe six or nine months ago. A policeman stopped self-driving Google car and, you know, people were, so the guy was trying to write a ticket and then who do you write the, because there was no driver. There's no steering wheel. I mean, I don't know if you, so eventually the cop wisened up and he didn't want to set any precedent, so he let the guy go. But, I mean, these are at the heart of policy. Insurers are very worried. Now Google is actually creating an insurance business because they know that traditional insurers are not going to be able to deal with this. So they're creating... Right, right. So these are some of the open policy questions, right? So, okay, so this is a quick thing about Qualcomm. So we were engaged with all the major auto drivers. We have, you know, multiple design wins. What, I would draw your attention to these three words. Telematics is all the connectivity. Infotainment is basically the stuff I showed where you have a center console. You have the, your movies or your audio playing in the back. And then connectivity is all the other connectivity that's not the modern. So things like wireless LAN, things like Bluetooth. One study showed that a major cost of manufacturing a car is the wires in it. And not only they're expensive to manufacture, but also they're heavy. So that, in fact, hits your efficiency. So what these car makers are trying to do is everything that's connected by wires today changed them into wireless. So everything in your car would be connected by Bluetooth, let's say. So the car would become lighter. The connection has to be, you know, much more robust than always on. So these are the areas that, you know, people are working on. Okay, security. Quick thing on security. This is actually what goes in the chip itself, in the product itself. So if you look at the top right there, a key management and provisioning means that, you know, I talked about we shipped 800 million chips. For each single one, there's an identification that's unique to that chip. It's almost like a DNA. And that happens when we're trying to push these devices into productization. Storage security is basically when you have on-chip memory or system memory, how protected that data. So it could be your personal data. It could be information about where you're going. It could be information about just the general situation of the car or your location. So you want all of that secure. Debug security is usually when something fails and people are trying to figure out what fails. And that's where usually hackers get into back doors. And that's where we have most of our challenges. Hardware crypto is all these crypto engines that are usually published by government agencies. So basically, they make it very hard for people, almost impossible for people outside of the owners to look and see what's in there. Secure boot and trusted environment is when you turn your phone on or when you turn your chip on, you have to trust that that particular entity who's controlling your phone or controlling your car is actually a valid entity. It's not someone hacked in. So we spend lots of time in all of this. Actually, this goes back to your, it's part of policy. What if someone hacked the car and drove it into a wall? Now, who's responsible? It could be the chip manufacturer. It could be the car manufacturer. It could be the software. It could be the network, the operator who's allowing that thing to happen. So it's a very complex problem. Okay, some other enhanced driver assistance. Basically, we have some of the stuff today already. It's been around for a couple of decades. The ABS, the brakes, the lane changing, lately the one that the self parking and things like that. So that's already been there. And it's been very kind of ubiquitous where people don't even think about it. Then you have the vehicle diagnostics where you want to make sure that that car, because think about this, eventually it's going to be all controlled autonomously. So you don't have a human interacting. Well, if you have a human, it's probably like air traffic controller. They'll be sitting somewhere across town or across the continent and they're just looking at data, points on the radar screen. So you want to make sure that that particular car or particular vehicle is in good shape. Collision avoidance, I mean it's obvious, right? Where these things are moving at freeway speeds, you want to make sure that there's no accidents and everyone is behaving or every vehicle is behaving the way they should be. Hands-free communication, we have it most of the time today, but it's going to be more than that. I mean you will have your overhead, I forgot to mention the overhead display, right? I mean besides all the displays, now you can have information or map the turn-by-turn map projected on your windscreen, right? Stolen vehicle, by the way, this is a funny one because imagine this, you know, decades from now, people will not own cars. I mean, I don't know if you guys heard of Uber, right? I mean you call a car or you call a service and a car would show up. So there's no ownership anywhere, but today, I mean, still people want to steal those things. And then what else? And distracted driver alerts, where this camera, it's not that today you have this lane-changing warnings or if your car starts riffling from the lane that the car warns you. Now the high-end cars have cameras where they look at your facial expression. They look at if you're sleepy or you're tired or there's all kinds of enhancements like that. Pardon? Yeah. Well, yeah, I mean, that's another thing. Police officers might be some algorithms monitoring other algorithms, right? So I'm going to stop here. Now I know if you have some questions, this is the other stuff is about Qualcomm and then I talked about this, yes. No, okay, so I mean, the way these things are going, like I said, all cities or countries are basically, they don't want any cars driven by humans. So if you want to drive a car, you go to these special places where there's a circuit, well, Formula One circuit, and that's where you drive, right? That's the reality you're going to wake up to. I'm going to retire by then, so. I driven my cars for 50 years now. But I don't think about it this way. I mean, if you're driving every day, you have a commute that's two hours, let's say each way. In some cities, I mean, especially in North America or Asia, that's not far-fetched. So people drive an hour and a half, two hours in the morning. I have colleagues in India who each way is about an hour and a half. So you're sitting in that thing for three hours. And what are you doing? Nothing, you're doing nothing. So don't you rather take a car you own, because you still can own the cars, but there will be special editions, and go on a circuit and then go crazy on that, versus driving to work every day in and out. Okay, I'm not going to convince you one way. Or think about it this way. I mean, I like driving, but I don't like driving when I have to drive, when I have to go to work. I'd rather sit in something and read a newspaper or just sleep or anything, right? So these are the type of experiences that are going to change. Maybe when that time comes, can nobody drive to his work? Yeah, that too. That too, yeah, that too. That too, but yes. Thanks for the presentation. And I have just a question regarding the cost of the technology. So as we have seen, so a bunch of new high tech stuff could be put into a car. So they are staying in a, so a regular car, and now they say you have a bunch of new electronics in it. So how does it, how is it going to be reflected in the price of the vehicle itself? So, okay, good question. If you look at it today, obviously they can't make these cars too expensive, too prohibited to people too. I mean, you know, Tesla started this whole thing, right? And these were really high-end cars and people can afford them, bought them, and they were usually not their first car, they were second or third car, right? They had a traditional car. Right after that, there's the next way of cars that are coming in that are very affordable, right? So just because all this technology is there, they can't make it very expensive. And then the savings are going to come from other places. I talked about, you know, getting rid of wires. I talked about getting rid of all kinds of hydraulic systems. I talked about safety, because part of the stuff that car makers spend is on infrastructure. So, honestly, I think that the car prices, I look at the cell phone prices, look at the chip prices, I mean, they're just going to keep on going down, but yet you're going to have all kinds of great technology in there. I mean, you guys heard about Moore's Law, right? You know, double every 18 months, twice the performance, half the price. And that trend, it's almost like, you know, this is going to be an extension of that, where things are going to become much more powerful, much more sophisticated, yet the price keeps on going down. I mean, if companies sell 800 million to billion chips a year, the price is going to be there, the pressure is going to be there to lower the prices. And then the other thing is, I talked about, you know, 300 different components going into a car. Those are expensive. But imagine if you start aggregating and integrating more and more, right? You know, you guys are working on 7 nanometer, compared to what was, you know, even maybe 5 years ago at 65 nanometer or 90 nanometer. You know, I'm talking about 6 billion transistors plus, 10 billion transistors on a chip. So then instead of 300, maybe you'll end up with 100 chips in the car. So that's where, you know, the cost is going to come down. Much more sophistication, much more powerful cars, yet the prices, I mean, that's the beauty of, you know, economics and large scale, yes. We have it today, right? We have it today. I think honestly, I think that system is very dangerous. That system is going to fail, because, I mean, some of you everyone knows in San Francisco area when they came up with these cars and Google was testing them. So you see all kind of drivers on the freeway trying to kind of, you know, cut on the cars, see how they're going to react. And yeah, it's testing, yeah. It's part of testing. So honestly, for this type of complex system to work, you have to remove the one parameter that's the most unpredictable. And what's that? Humans. Yeah, you got to remove them, right? Not remove the human. Remove them from driving a car. You see, he doesn't like it, right? Yes. They're going to turn into programmers and algorithm. I don't know, honestly, I don't know. It's shifting. Actually... Okay, let me give you an example. I think, you know, you have a good point, but it's not just the drivers. I think there's, you guys have heard of Watson, right? And someone was mentioning about TA, you know, teaching assistant. The jobs, the Watson, the IBM big machine, the big blue that, you know, the predecessor kind of defeated Kasparov at chess. The people at risk today, it's very interesting. Are medical doctors? Yeah, because for diagnostics, you have these machines that can diagnose better. Because imagine there's a doctor with the whole Google information, plus you have, I mean, you guys watch Star Trek, right? I hope so, because you won't be in science. But, you know, these machines have all kinds of sensors that can detect micro changes in your face. They can detect in your skin color. They can see what's under. That the typical doctor cannot see. So those guys are at risk. The other guys are at risk from losing their jobs. It's really lawyers and paralegals who do all kinds of searches on cases. There's so many sophisticated algorithms that can recognize legal text and then set precedence and so on and so forth. And it goes on and on. So, yes, there are going to be lots of casualties because all of this. Yeah, well, even surgery, I mean, you've seen some of the movies, right? Which one was that? Prometheus? When they have this pause where they can do, you type the type of, you punch in the type of surgery and you just lie there and then it's going to perform. So before we get to worry about the billions and billions of drivers, honestly, this whole thing is shifting the paradigm into there's going to be lots of change about so-called traditional kind of careers that are not going to exist anymore, right? Body shops also disappear. Body shops? Yeah, probably. Less than less, right? Actually, these cars are going to be very affordable that if something happens, they get into an accident, there's damage, you just drop it and get a new one, right? Yes? Okay. Actually, if you look at it, it's a distributed decision-making and it's not going to be... Again, I mentioned, if a human is part of that decision, it's probably some new way of traffic controller that's sitting somewhere remote and looking at data and looking at, for instance, one type of decision could be there's a congestion somewhere and some of you have taken fluid dynamics, right? Remember how fluids flow into narrow conduits and things like that, right? Same thing for traffic control and those are algorithms or metals that were developed decades ago, so people are applying the same thing so someone can direct traffic into less congestion there. So that's one decision-making. Then the decision-making, as far as the V2X I mentioned, so it's going to be within the vehicle itself, you're going to have chips and algorithms that are going to take that type of decision. No, I don't need the slides. Thanks, everyone. So you have some chips in the car that are taking those decisions. I talked about these powerful devices. For instance, in our latest product, we have eight processors, each running at 3.5 gigahertz, eight of them, and then you have a graphics engine, okay? So then you have a network processor that's good at deep machine learning or neural nets. Then you have a DSP, a digital signal processor that runs multiple threads on it. So you have all these kind of devices or all these IP within a given product that are trying to take a decision and there's algorithms that run on a combination of these devices or a single one. So that's another decision center, right? Then you have things like where you have a smart city and it's directing traffic or parking to where to charge. So it's really going to be very, very distributed and removing the person, the one... I mean, people behind the wheel, because they're the single points of failure, which is the worst situation to have. Anything in a car fails because of that one thing that's behind the wheel. And then this paradigm is going to remove all of that. Now, whether you trust the software or not, that's a different thing. But these things, one Google study said that they had eight billion hours of testing that went on for all these cars, for all these algorithms. What, this way? Some of these machines or some of these algorithms, the simplest one, neural nets have been around for 30 years, nothing new. Neural nets, basically, you have a data set, which is a training set. The more you have, the better those nets become. So the more you train them, the more sophisticated, the more precise they would come. So when you buy a car, you know, the car won't say that this is how many horsepower anymore because we won't say this is the fuel efficiency. It's going to say this neural net in this car has been trained for 10 billion hours or 20 billion hours or 100 billion hours. So things are going to change, right? I mean, but yeah, there's... You're not going to drive a car in 20 years. Someone's going to drive you. And if you want to drive a car, really, you go to these special places where the race tracks... Why do you need a license? What do you need a license for? You just need an app. Honestly, I mean, today, like, lots of places I go, I don't even bother renting a car because first of all, it's too expensive, especially when I go back to Canada, wintertime. You rent a car, you leave the car outside overnight, in the morning, you have to scrape it, right? Because it's frozen. Where, you know, you have an app, Uber, where, you know, you download the app, you have your credit card in front of you. You guys have heard of Uber, right? So it's very convenient. Another situation, I was in... A few months ago, I was in Taipei, right? So I had an extra day. I said, okay, let me go to a museum. I was at the hotel, hopped into a cab, went to the museum, went to the wrong museum, but that's... It wasn't very interesting. So after like an hour, I want to go back to the hotel and then no one speaks English. And I don't speak Chinese, not yet. So then she's pointing me to the street. So I go up, it's a main, you know, large street, and there's cars zooming by. There's cabs, but no one's stopping. So what I do, I called Uber, a guy shows up, nice car, speaks English, took me where I want. So, you know, at some point, yeah, you don't need a driver's license. You don't need, whether it's for weather convenience, just financially, economically, it just, things are going to change. Yes. You are talking about the future where everything is being automated. I'm just wondering, you are investing in the industry and why you are sure that in the future they will be able to move from going out, coming through? Or they must become now... Because, okay, look at... Work, go home, and we can do it. Yeah. Look, I mean, look back. You can go back hundreds of years. Look at what's driving every human invention. What is it driving? It's cost. I mean, it's funny, because as engineers we say, oh, cost, you know, no one cares. That's something accountants do. It's really, everything we do is financed by saving somewhere. Someone's making money. So that's driving. That's one thing, right? Second thing is what's driving. And the order might be, you might choose one over the other. What's driving? Inefficiencies. Go back to the invention of the printer. That whole thing revolutionized, I mean, these are major, major paradigms. Why did people invent the printer? Because there was inefficiency. You can't write things fast enough. And then look at every invention, major industrial age. Automation. Ford, right? Ford created this whole industrialized way of producing something that was piece, then piece, then piece, back to back to back. So everything is driven by squeezing efficiencies out. This is one of the least efficient infrastructures we have. This is one of the least efficient appliances we have. Again, how much you drive, it depends. But on average, there's maybe four to six percent you drive during the day. And the rest of the time, that thing is parked, doing nothing. So that's one issue. That's from an individual perspective. The whole infrastructure is so expensive to build these freeways, then you have to maintain them. All kinds of things are happening. That's why traffic, congestions, fuel efficiencies, just pile it up. So those are the drivers that are going to make this happen. People are already investing heavily in this area. We can't push products fast enough to hit all of these areas. It's not just the infotainment. That's a simple thing. You want entertainment in your car. It's just the whole artificial intelligence aspect. It's the whole efficiencies. It's the whole having the algorithm drive your car. So that's why people argue against these things. Companies usually are smarter than people, but you can argue. But mainly, things are driven by the economy of scale, driven by squeezing inefficiencies out. Why do you think we keep on chasing smaller and smaller transistor sizes? Because we can put more transistors in there. It's more efficient. And it's cheaper, way cheaper. So it's coming. I mean, embrace it. It's a good thing, yeah. The cost of maintaining the roads, the high rates, etc. So is the industry going more to invest in air cars or hovering cars? That does not have to depend on the ground. Right. So we have some people. Actually, one of our previous CEO, he was the son of the founder. It's actually, when we were doing, we have this annual thing where we do these ventures. We're bringing these companies to think about the future five years, ten years. He basically said it's easier to solve the flying or hovering cars problem than solving cars driving around roads because of all the issues that we have. So people are actually thinking about that. Uber, out of any good, they have this drone. Which can carry about three or four hundred pounds. Basically, it depends one or two percent. It depends how much, maybe Yervan and I can fit in one. But they are thinking about this. Now, there's other issues, of course. I mean, this gentleman was asking about policy issues, privacy issues, but people are thinking about that. And honestly, in lots of these areas, the technology is there. I mean, if you look at what you're studying, what you're working on, the pieces are there. It's just a matter of someone finding a way to make the economy of it work. Someone finding that the policy of it's going to work, that they're not going to get sued. So all of these, one of the messages there was that it's opening new possibilities of different things happening, new user experiences. The only thing holding us, or these entrepreneurs now, is really if there's legal issues or policy issues, or maybe it's too expensive. But the technology is there today. And we have it here, our cell phones. Every technology, all these apps, all these recognitions, all these sensors. Yes? Okay, my first reaction is, first of all, I think it's going to be safer. And that's, I mean, one of the numbers there was people, people are worried when they're driving their cars. And if you look at some of the statistics, I mean, it's true. You get more accident, more fatalities just by people getting into accidents driving cars than plane crashes. It's one of the, I mean, think about it this way. There's multiple ways of looking what a car is. For us, because we sell chips, a car is an IoT device, where it's connecting to everything. In the one extreme, we've seen some of the classic movies, the car can be a killing machine. And unfortunately, what's going on lately in the world, you see, it's very dangerous. And forget about these extreme situations. But even going on a freeway, getting an accident, even if it's not completely fatal, you get injured, you get back problems, you get incapacitated. So for me, the most important thing is, this thing is going to make that whole thing very safe. Now, as far as the investment is concerned, again, some people, and this started probably 10, 15 years ago, people realized that there's money to be made and lots of money to be made. Because, I mean, 20 years ago, we started, we had this Nokia phones, cell phones. No one thought about how much money it's going to take to invest in devices where I haven't talked about the whole IoT thing and connectivity thing. And people didn't think that it's going to be very expensive to invest in infrastructure. That's going to carry 4G traffic or 3G traffic. That thing costs hundreds of billions, and yet people spend the money because money is coming back, right? Through subscription models and things like that. The same is going to be with the automotive side. People know they're going to take their money, and it's not huge money for the consumers, the sharing economy. Just imagine you get one car and then that car is being used 23 hours a day, and then the one hour is for maintenance, right? 23 hours a day. You went from 4% to 6% efficiency into 96% efficiency. So that's what's driving this whole investment, right? And again, it's sad. It's a reality. There's lots of people, and again, it's not just the drivers, but it's lots of people who have to be retrained into going to different careers. I mean, I just mentioned the medical doctors, lawyers, things like that, right? It's just a reality. Okay, now when you go home, when you're in your car, probably I ruined your driving experience now. Think about every small decision that your left side of the brain doesn't think about anymore, because when you learned to drive 5, 10, 20, 30 years ago, like some of us, that part became part of who you are, right? And then everything moved to the right side of the brain so you don't think about it anymore. So now everything you do, every decision you make to go from point A to B, think about all of this are going to be taken over by algorithms, by devices, which can take much more efficient decisions, much faster decisions, and you're going to benefit from it. And you're driving, you're sitting in a two-hour... First of all, traffic is going to disappear because of all these controls, right? All these fluid mechanics management type of thing. But think about it, you're sitting in your car for a given period of time, but you're enjoying doing nothing, reading, listening to music, sleeping, whatever you would like. And believe me, it's a good thing. And we're going to sell lots of chips, so...