 Hi, and welcome to theCUBE, the leading source for insights into the world of technology and innovation. I'm your host, Donald Klein, and today's topic is the market for autonomous vehicles and the ecosystem of suppliers looking to tap into this brave new world of autonomous capabilities in our daily commute. To have this conversation, I'm joined by Rudy Berger, managing partner at Woodside Capital. Rudy, welcome to the show. Thanks, Don, it's great to be here. Great, so look, why don't we start off, Rudy, why don't you tell us a little bit about Woodside Capital and your role there? Great, so I founded Woodside Capital about 20 years ago, having started five different companies of my own, one of which I took public. We are a specialist M&A advisor. We work with so-called growth stage, often venture-backed companies, and help them find buyers that are usually much larger public companies. Our clients are usually US or European companies, and we find buyers in the US, Europe, or Asia. Excellent, excellent. Okay, and why don't you talk a little bit about your kind of specialty areas? So I focused my career, and certainly the work at Woodside Capital, on imaging technologies and as an enabling technology and the products and markets that are enabled by imaging and increasingly computer vision. So nowadays that is autonomous vehicles, consumer technology, security, surveillance, and digital health. Excellent. So enabling technologies, the computer vision is the theme that binds those together. Okay, well, the thing that's on everybody's mind these days is autonomous vehicles, when we're going to get them, very high-profile for sure. Before the show, we were talking about the kind of two key ingredients to making this happen. The AI software, which is kind of the brains of the operation, and then also the sensors which enable all of the AI. So why don't we talk about the sensor world first, okay? A lot of discussion about there, sort of does the brave new world of vehicles need LiDAR, does it not need LiDAR? Are there other types of sensors coming along? What's your sense of that market and how it's looking for all the different players in it? So, Don, I look at it from a fairly basic standpoint. Humans have two very capable image sensors and a very powerful processor. And the degree to which the automotive manufacturers and so-called Robotaxi developers have decided it's necessary to sprinkle every sensor known to man, and I'm talking LiDAR, LiDAR, ultrasound, thermal, and of course, cameras, is to some extent a degree to which image sensors are not as good as our eyes today. Now, there are some areas in which we will probably always have technology as a help. For example, humans are not very good at seeing in the dark, whereas a thermal technology can do that very well. But my overall belief is that it's never a good idea to bet against an incumbent technology, and in this case, I'm talking about so-called CMOS image sensors, which are the sensor that goes into pretty much every camera in the world now. It's never a good idea to bet against the incumbent technology being able to scale into a new market. Every time people have done that, they've been wrong. Back in the early days, the debate was whether CMOS image sensors would ever be good enough to replace CCDs as the sensor technology. And of course, now everything uses CMOS image sensors. In other markets, there was a long period of time in which people were thinking that LCD panels would never be large enough to replace for television, for example, 50 inch and so forth. It's never going to happen. So we needed plasma TVs, we needed rear projection TVs, but slowly but surely the incumbent technology LCDs expanded to that market. So my belief is that CMOS image sensors will evolve to a point at which they will replace the need for LiDAR in most applications. Interesting, so that's a very controversial statement, right? Because you've certainly seen a lot of emphasis on the development of new generation LiDAR companies. Over 100 LiDAR companies started over the last three, four years. And of course, many of them will not be happy to hear me say that. There are two distinct markets. And one is the so-called robo-taxi market. And the other is more of the consumer vehicle ADAS market. And I think we need to think about those separately because the economics behind both are very different. Interesting. If you look at the robo-taxi market, those vehicles tend to be much more expensive and are relatively price insensitive. So if they can improve safety a little bit by putting a LiDAR on there, great, let's do it, multiple LiDARs because these vehicles will be in operation 24 by seven, right? And if each vehicle costs $200,000, $250,000, fine. When we talk about the mass market for automobiles, type of car that you and I might go down and buy, very different thing. And automakers sweat the pennies. And so putting a $1 or $200 LiDAR in a vehicle, big decision. Interesting. And to the extent that they can replace the need for that LiDAR with a much less expensive camera system, that's what they'll do. Bear in mind that Mobileye, which has been the biggest success story, acquired by Intel for $13.5 billion, second largest acquisition Intel ever made, they for the most part still run on one camera, forward looking camera, that's it. No LiDAR, no LiDAR, no thermal, one camera. So the clever use of image processing, computer vision and one image sensor can do a great deal. Interesting, okay. Well, so I want to talk about the software in just a second, but just to kind of finish this point. So if you were advising a sensor company that's developing some next gen capabilities, whether LiDAR or other related technologies, is the point you're making here that there are certain segments of this industry which are going to be more attractive to your technology than others? Absolutely, yes. I mean, the first thing to recognize is that the automotive industry has never really been a particularly comfortable fit with the economics and timeline of venture capital. VCs need to invest and recoup and redeploy back to their LPs on an eight year cycle. But the automotive industry moves quite slowly, perhaps Tesla are accepted. And what the first piece of advice I would give these companies is, it's probably going to be three, four, five years before even if you have the right technology before that technology really starts generating any significant volume and revenue. So for many venture-backed companies, that's too long. And what they need, so the first piece of advice is find pockets of revenue, beach heads if you will, where you can land your technology and start generating revenue before you get to the automotive market. And many of these LiDAR companies we just talked about are not going to last long enough to get to the automotive market. It's not only does the automotive market move slowly, but the autonomous vehicle market keeps on getting pushed out to the right as the industry realizes that this is a big hairy problem. Interesting. And so I would say, what is it that your technology can do? An order of magnitude better than any other technology. Focus on that and find some opportunities for revenue outside the automotive industry that will sustain the company on its way to the Holy Grail. Interesting, yeah. So find that alternative revenue source to get you to base camp, then when the market's ready, climb that Everest too. I've seen so many companies basically go out of business because they've set their sights on either the automotive market and it's go for broke. We're not interested in it. All these other things are distractions. Entrepreneurs don't have a plan B or this. We're going to get our technology into a smartphone. That's it. And there are possibly some other opportunities, but it takes so long and it's so difficult to get your technology into a smartphone that they go out of business before they ever get to that point. Interesting, okay. So good advice for people looking to kind of apply their technology in this kind of a very difficult market, right? Very complicated market. All right, well then let's switch to the other side of it. So we were kind of talking about the key ingredients, right? Sensors, but also AI and the software around that, okay? And there's some very big players developing the software. Tesla's had their autonomy day where they've showcased their technology. You've obviously got Google with their capabilities developing software. How do you make sense of this overall landscape? Because we do see a lot of smaller providers also trying to develop software here. So the first thing that I find fascinating about the automotive industry is that for the most part, there is no software market. There's perhaps one exception of any scale that's BlackBerry that sells the QNX software. They found a point within the entertainment console where they can license their software. But for all of the development and capital invested into automotive software, nobody is actually generating revenue, making a living by licensing software. And one of the main reasons for that is that, you know, the automotive market really since inception has been a hardware business. You know, this is a business of bending sheet metal, internal combustion engines, and software has really not played that bigger role up until relatively recently. So, you know, even those companies that do have software technology have ended up selling it into the automotive supply chain as a piece of silicon, embedded on a piece of silicon, not as, you know, here's my software on a USB stick, right? I think that the whole software licensing model hasn't so far fit well, fit comfortably with the automotive industry. And the other reason is that there's no standard platform. If I were to develop a piece of software, you know, I can, in the PC industry, I can develop for Windows, I can develop for Mac, I can develop for an iPhone. There's no such thing in the automotive industry. And particularly in this new world of autonomous vehicles, there is no standard platform. There are many different processes and video, you know, has staked an early claim there. And the reason that most of the companies developing autonomous vehicle technology had developed the so-called full stack solution, everything from, you know, code running on the processor to the integrated to the sensors and so forth is for that reason, there is no standard platform. So each company has developed the whole solution for themselves. And, you know, there are many of them around here that have raised hundreds of millions of dollars, some cases billions of dollars for that purpose. So, you know, there is today no software market for automotive in the same way that we think about it in, you know, other industries. I understand, I understand. But in terms of the companies that are actually pushing the envelope on these kind of capabilities, right? So we're taking the best of AI, we're applying it to, you know, big data sets, and then hopefully being able to extract that to create kind of capabilities for these vehicles, right? What's your sense of how far that's come along? Well, it's come a long way, but here I'm going to push the boat out a little bit. I don't believe that the so-called deep learning technology, which is, you know, the current state of the art for AI, it's the technology that has allowed computers to beat humans at chess and go. I don't think that that flavor of AI, that approach to AI is ever going to get us to safe enough autonomous vehicles. And that's because it works extremely well in fairly well-bounded rules, rules-bounded, you know, games or any scenario like that. But can you imagine trying to teach your 16-year-old how to drive by showing them images of every situation that they might encounter, right? Impossible, you know, it's an infinite, you know, it's not a well-bounded set. And so that, and that's so difficult because we really haven't developed the technology to allow computers to learn, to have things like common sense, to infer, you know, well, this happens, so this is likely to happen. So I think we're going to need a whole breakthrough in AI before we get to what is generally considered safe enough vehicles. Interesting, well then maybe if we kind of apply your kind of previous thought about sort of, you know, robo-taxies as maybe being the segment where you're going to see the most use of kind of these newer sensor technologies, exactly. What about maybe, is that sort of the same rules apply there for maybe the AI providers that they're? I think so, and that's why they're all focused on that. I mean, from Uber to, you know, Waymo, they've all made the same calculation which is if you're running a fleet of vehicles, you know, and so for example, in Uber's case, the driver takes 80% of the fare and only 20% goes back to Uber. But if you can replace the driver with a computer, you can keep that vehicle on the road 24 by seven and you can keep 100% of the revenue. You don't need to pay the computer. So that's the calculus that they're all going through. But I think that there are many of them are making a fundamental mistake. And I predicted recently that I think Uber, my prediction for 2020 is that Uber is going to divest its autonomous vehicle business and get back to, you know, the business that it should be focused on. Uber generates about $14 billion a year in gross revenue. So 20% of that, which is the piece that Uber keeps after, you know, the drivers take their 80 is what, 2.8 billion. Uber should be able to be an extremely profitable business on, you know, on 2.8 billion of net revenue. But they're spending a huge chunk of money every year on R&D. Now, I would argue that, you know, Hertz and Avis have successful businesses. They're in the service, they're in the transportation business, but they didn't decide that they had to build their own cars in order to be in that business. My view, personal view is that what Uber should be doing is saying, you know, that's not our business, right? We are the world's best at managing this sort of peer-to-peer, you know, network, crowdsource transportation, if you will. And when some company, some Silicon Valley startup comes up with, you know, safe enough technology, great, we'll use it. But we don't have to develop that ourselves. Well, then maybe just to play the devil's advocate here for a second. What about, let's say, RoboTaxi-type technologies being applied in bounded areas within metropolitan areas, where the rules could be more... I think that's where it will start. But, you know, I think part of the problem is that we have, perhaps in part due to all of the media hype around autonomous vehicles, we've been misdirected to thinking about autonomous vehicles as a replacement for the car we drive to work every day. And I think that's the wrong way to think about it. I think that autonomous vehicles are going to show up in the market as an extension of public transportation. Interesting. Right? Okay. You know, I get off the train and there's an autonomous vehicle waiting to take me for, you know, the last couple of miles to my office. And those last couple of miles, you know, would be sort of a regulated space where the AI is more than capable of functioning. Right. And that, you know, yes. And so it's better to think about autonomous vehicles as not being a revolutionary technology, but much more of an evolutionary technology. And in fact, most of these technologies are showing up in so-called ADAS technologies, which are designed to make driving your regular car, you know, safer, lane assist, you know, keeping you a safe distance. Maybe just explain that word ADAS and what that means. So ADAS stands for Automated Driver Assistance Systems. Okay. So, you know, one of the first was cruise control. Sure. Right? Everybody's familiar with cruise control. And so to some extent ADAS are just, is just building on cruise control. You know, in addition to maintaining a constant speed, you can now stay in the lane. In addition to maintaining a constant speed, it will now automatically slow down if you get too close to the car in front. And so you can see ADAS, you know, as, you know, collision avoidance and so forth. Not full autonomy. Still have to have a driver in the driver's seat, but evolving year by year until, you know, one year we wake up and yep, my car will actually drive me all the way from home to work without me intervening. All right. It's going to happen in that way. So incremental improvements. Incremental improvements. ADAS, as opposed to kind of revolution of autonomy coming from nowhere. Exactly right. Understood. Well, then let's pivot from that then. Okay. So let's talk about the automotive industry as a whole and your thoughts on how this is all going to play out. Yeah. So there's some very interesting dynamics playing out in the automotive industry. Firstly, as good news, you know, as a result of all of this money and innovation in the automotive industry, Detroit's actually coming back. You know, I go there once or twice a year and you know, you can feel the economy coming back in Detroit, but it's not going to come back around, you know, bending sheet metal. And the challenge that the automotive companies have is so much of their infrastructure and expertise has been built on, you know, construction, building a car, production lines to, you know, bend the metal, install the engine and the internal combustion engine itself. And by complete coincidence, to some extent, we've got this confluence of, you know, all of these autonomous technologies and electric vehicles happening at the same time. Electric vehicles are much easier to make than internal combustion engines, far fewer parts. It's one of the reasons that, you know, China has spun up about 20 different electric vehicle companies recently. So, you know, I think that long-term, my prediction is that the automobile industry will go the same way that the computer, personal computer industry went. When the PC first, you know, was born by IBM or, you know, Apple in some sense before that, there were dozens of companies producing different PCs. And it was, you know, very much, they were expensive products and, you know, relatively unusual. As the industry matured, the supply chains matured and it became apparent there were really only two companies that were making a lot of money out of the PC industry. The companies that developed the software, operating system, and the companies that developed the process. And all of the manufacturing went over to, in this PC's case, in Taiwan, right? And I think that exactly the same thing is going to happen with the automotive industry. You know, Tesla today still actually makes cars, but I don't see them long-term being in the, you know, car business because they're really a technology company. It's the reason I don't think Apple is ever going to get into the car industry. You know, they make fantastic margins selling computer products. The gross margin selling a car is miserable. You know, it can be single digits or teens that would completely tank Apple's gross, you know, blended gross margin. Understood. So my prediction for the industry is there will be a few small pockets of very profitable businesses, particularly around the operating system by which I mean the intelligence or the AI intelligence. And then the processor, whether it's a Qualcomm procedure or a Nvidia processor or an Intel processor. And as with the PC industry, they will, most of the profit will go there and most of the manufacturing will end up getting outsourced because that's not the value add, you know, bending metal and so forth. Interesting. So in the kind of compute market today, right? We have this notion of sort of, you know, cloud native. Yes. Right. Okay. And that many of the companies that are developing apps as we're relying on cloud native infrastructure have a kind of technology lead that's going to be hard for some of the legacy providers to actually catch up on. Now other people say that that's not necessarily the case and et cetera, right? Can you make the same argument for the electric car market that some of the electric natives might have a kind of sustainable advantage here? I should have added, you know, today the cloud infrastructure companies, cloud services, SaaS companies in the PC world, you know, very profitable. And I can see a similar cloud services model developing for the automotive industry. However, other than Tesla, the interest, it's very difficult to change the automotive channel to support that. I give you one example. Everyone that owns a Tesla is very used to the idea that sometimes on a daily basis a new bunch of software operating system software is downloaded overnight to your vehicle. You wake up in the morning and some new features being turned on, right? Tesla can do that because they bypass the entire dealership channel that has a complete lock on the rest of the industry. So for example, if GM wants to do the same thing as Tesla and do sort of what's called over-the-air OTA updates, software updates, they can't do that because their contract with the dealership network states that if there is service to be done on the vehicle, the vehicle has to be bought back to the dealership, right? And the dealerships consider updating the software on the vehicle as service. So their contract with the dealers actually prevent them from doing something that basic, right? So it's not just a technology issue. The whole, you know, channel and way vehicles get sold is going to have to change. Interesting. So that's the advantage of some of the new generation of vehicle. I would say that Tesla has a five-year lead, technology lead because they, like Apple, are vertically integrated. They're doing everything from user interface, fit and function, all the way down to the semiconductor. They're developing their own semiconductors now. So they have become a fearsome competitor in the electronic vehicle space because they've been doing it for longer than the other major auto companies. They've figured out a lot of the tricks and techniques of how to extend mileage and so forth. And so they have a substantial lead in the industry at this point, despite the fact that over the next 12, 18 months, every automotive company is going to be coming out with their own flavor of electronic vehicle. So it's more than just about having electric drive trains, et cetera, right? It's about the whole suite of capabilities. It's a systems engineering challenge. Interesting. Okay, all right, well, Rudy, we're going to have to leave it there, okay? But I think everything you've told us is, it sounds like some good news for some of the Tesla stock holders at the moment. I think so. Okay, well, we'll pass on making an opinion about that. But great conversation. Thank you for your insights. Okay, this is Donald Klein, host of theCUBE here with Rudy Berger, managing partner of Woodside Capital. Great, thank you, Don.