In this episode of the Judgment Call Podcast Amit Mehta and I talk about:

• How the automotive industry is structured (i.e. tiered) and how innovation works in that industry
• Why Electric Vehicles (EVs) are so expensive in the US?
• Will Apple ‘corner’ the market for Electric Vehicles soon?
• Are autonomous vehicle ‘just around the corner’?
• What is the discussion about LIDAR sensors and radar all about?
• What role will AI play in the automotive industry?
• And much more!

You may watch this episode in 4K resolution on Youtube – The Judgment Call Podcast Episode #37 – Amit Mehta (The future of the automotive industry).

Amid Mehta is the Head of Innovation at North American Lighting and host of the Mobility Now Podcast.

You may reach Amit via LinkedIn.

 

 

Torsten Jacobi: Okay, I’m here today with Amit Mehta, and Amit is the head of innovation at North American Lighting and the host of the Mobility Now podcast. I was really curious. We had a chance to talk a couple of weeks ago, and I’m really curious about the whole field of self driving cars, and from what I learned, and that’s what you told me last time, the company North American Lighting isn’t actually about lighting anymore, right? So you guys have started to do something completely different. Tell us a little bit about that.

Amit Mehta: Sure. So let me give you a little background about North American Lighting itself. So North American Lighting is a fully owned subsidiary of the Coito Group. The Coito Group is the world’s largest tier one supplier of exterior automotive lighting. One out of every five vehicles you see on the road, that’s us. So we have a massive market share. I think you know that ever since Tesla started to kind of populate the roads, their intention wasn’t to provide a vehicle. It’s to provide a solution, a technology, a product that the customer can engage with, a product the customer is happy to get into. It’s not just a finite transportation platform. So what we saw after Tesla created this evolutionary choke point, you could say, is all the OEMs rush into this field. And the real tipping point happened in 2016 when GM made a huge decision to invest or buy cruise automation for $1 billion. And that’s when everybody woke up. That’s when Silicon Valley woke up. That’s when OEMs are all around the world woke up and said, wow, there’s other companies out there that are buying these automation companies. And that’s when everybody started coming to the Bay because the talent, you get automotive manufacturing talent in Detroit, but you don’t get the user experience, the HMI, the talent that’s required to make a vehicle into a solution rather than a product anywhere except for Silicon Valley. Of course, that’s changing now with COVID and the pandemic. People are moving different places. But the point is that’s when Coito, who’s 20% owned by Toyota, made the decision that we need to be in Silicon Valley to support our customers, to help them learn how lighting will change in the future as well. 

Torsten Jacobi: Yeah. Well, one thing, maybe we have to help our listeners a little bit with the terms, right? OEM, I think a few people might understand. And then also the tier one, the whole way that the automotive supply chain works, I think most people don’t know much about that. I didn’t know until I started working in that industry a long time ago.

Amit Mehta: Good point. Yes. So let me explain a little about that. It’s basically a pyramid. So we have tier three, tier two, tier one, and an OEM. OEMs are at the top, tier three is at the bottom. Tier one is an integrator. So we provide a full component to the OEM and they assemble this vehicle. So for example, we provide lighting solutions. Those lighting solutions are comprised of plastic injection molded polycarbonate pieces. They’re comprised of aluminum die cast heat sinks. They’re comprised of wire harnesses. Wire harnesses, LEDs, and these small injection molded pieces and heat sinks, they come from tier twos. So they provide those solutions to us. We are integrators, put everything together, make sure that we meet regulations, and then provide that to the OEMs. And OEM is an original equipment manufacturer, which is a Ford, a Toyota, a GM, a Honda, basically the big automakers, including Tesla. Finally, a tier three, a tier three is a materials provider. So they’re providing materials to the tier two, and the tier two is making these components and then providing these small solutions or specific solutions in particular areas to the tier ones.

Torsten Jacobi: The one thing I think we have to explain a little before we go into the sensors, which I think is really exciting in the whole way they can pave the way for self driving cars. And I only learned this when I had a startup, when we sold products and software to the automotive industry. What actually happens is that a lot of innovation and development for the next car is not actually done by the specifications, say, of BMW. BMW just goes to Bosch and says, here’s what I want. I want 20% lower consumption of energy. I want this thing to be 50% smaller. Good luck. The deadline is two years from now. So the discussion is more and more involved. But basically, there is a space that’s only being held for Bosch. And then Bosch, I think that’s what every tier one supplier now does. It literally, BMW has nothing to do with this component. So the lights, how they work out, they obviously got to check it, and they are involved, and they were doing their process, they have some kind of quality control. But the whole process of innovation is completely out of their hands. And from their point of view, it’s something less they have to worry about. They can just outsource innovation, so to speak. And hopefully, it works. So you have to have a lot of trust in your suppliers, because they can mess it up. And then, well, your car works, but it has crappy lamps. And so I can’t drive. I can’t see anything in the dark. So it is a huge risk they take on one side, but on the other hand, I think it spawned a lot of possible innovations because the complexity has gone down a little because the car is almost like an airplane now. It’s really complex, a lot of computer engineering. But if you define these spaces and the APIs, it becomes much easier to manage. And that, I think, gave the opportunity for guys like you to really say, OK, we can do much more. Are you able to pay a little more? The first mandate can come from, do you know that GM approach and Toyota say that, OK, we want more? Or you said, no, no, no. We can put sensors into lamps, which sounds a little counter intuitive.

Amit Mehta: Yeah. So that’s a good question, and that circles back into what you said earlier. What is Coito really doing? So since we do make lighting solutions for vehicles, our goal is to adapt and put the sensors into the headlamps and rear lamps for future AV vehicles. And I think the question you’re asking is it a push strategy or pull strategy from a market standpoint? And the answer is really three years ago, it was a push strategy from tier ones, which I explained earlier, provides these components and solutions to automotive companies. But now, as we are getting more involved in putting these sensors into our locations that really feature our asset, now it’s becoming a pull strategy. We are seeing OEMs more open to the idea of this innovation, of putting these sensors into lamps and providing this solution to our customers.

Torsten Jacobi: How did you get into this? I know you’re the head of innovation now, and you mentioned to me before you went to Silicon Valley for this position, right? What’s your love affair with all things automotive? Or how did you get into this specific trajectory here right now?

Amit Mehta: Yeah, that’s a good question. So I started my career at Bosch, and I was originally part of the automotive area in Farmington Hills working on throttle bodies. From there, I went to the galvanization steel mills in Pittsburgh designing bearings for galvrigs. Then I came back to Bosch Power Tools for six years in Chicago, and I was working on different solutions for table saws and miter saws. Then finally, I made a decision to go back to Detroit, and that’s where I got involved with NAL. When I got involved with NAL, it was kind of a tipping point for me, and I remember this exact tipping point, and I think I’m going to detail this to you. When I was at Bosch, we were in the engineering department, and we were working on a new type of saw where if you touch the blade, the blade will drop, so it doesn’t cut your finger. Basically, the blade is rotating around 3,400 RPM, and as you get closer, there’s capacitive coupling between the blade and your finger, and it can detect it before you even touch it, depending how fast you’re going, and drop this blade. So you save a finger. We worked on this as engineers for six months creating patents, and all of a sudden we had marketing come in and tell us, you need to change your entire direction. At that point, I got a little upset, and I said, okay, well, we’ve been working on this for six months, and all of a sudden, you guys changed your mind, and I didn’t understand how the market reacts, and I was getting mad at marketing, but market reacts, markets change, everything’s dynamic. At that point, I made a decision I need to understand more of what marketing’s doing, and I did my MBA, and from there, I went to North American Lighting after a couple years, and they hired me as a technical market or senior technical marketing engineer, and I was in charge of looking at how lamps were constructed and explaining this to different OEMs and letting them know exactly how lamps are made. Then in 2017, I got the request, would you like to go to Silicon Valley to kind of help start that division and see where it takes us? And I said, sure, that sounds like a very exciting opportunity. At this point, I wasn’t too fascinated by cars, I was very fascinated by lighting, and the reason I was fascinated by lighting is, I think you and I and everybody on this call, the only way we can recognize a car at night is by the lighting, and it was fascinating to me to see the different faces and different styles that lighting really created for these images, for the consumers that are purchasing these vehicles. But as I got to Silicon Valley, I saw revolution, and the revolution is so evident from here, maybe not in Michigan or anywhere else in the world right now, but everything is moving towards electric vehicles. And we are at this point, this frontier, this revolution, we’re at this tipping point where today only 2% of the entire United States population drives electric vehicles. But that’s going to change, everyone’s invested a lot of money into electric vehicles, and now my fascination is growing, especially as you mentioned Bosch, and there’s other big ones too, like Continental, there’s ZF, there’s TRW, the guys who supply the internal combustion engine components. Where are they going next? That’s a big market for them, Bosch is a $77 billion company, and a lot of that comes from automotive components, especially in the engine. What is their next move? And I think they’re struggling to find that next move, whereas I see lighting, we have a future, it’s a required future, and it’s going to be there for a long, long time because we will have to drive at night. And it makes sense to me more and more, and as I study it more, I’m more passionate about lighting because that is going to become a very strong solution for OEMs. We’re going to deliver something as a white glove delivery to our customers. So that’s where I’m getting more and more, I guess, fascinated and thralled by lighting itself.

Torsten Jacobi: Yeah, you strike me as a very serious engineer, and I think that’s good, this is often required, right? We hear a lot of from marketing folks, and they get excited, but literally they don’t know the technology, they don’t really know where to get excited, they get excited because that’s their job, so to speak. So I think you’re definitely onto something there, on a personal level, and I find this interesting the way EVs work. I find it interesting how we see them introduced in the US and how we see them introduced in China and in Europe. The pattern seems quite different. So we had this story from Tesla in the US, which seems to be extremely up marketed. They basically did a supercar, so that’s kind of the first Lotus that they did was $150,000 and it was made with laptop batteries, right? Since then the prices come down, it’s still around $40,000, $50,000 on average, right? There’s a cheaper model, but a lot of people go for the nicer model, and I think they make wonderful cars, and I love that we had the idea to have, like every single car, it doesn’t have to be, but could be in EV. Our cities would be quieter, but the progress has been slow and it’s extremely expensive still for the average consumer, because most cars in the US are really cheap, right? You can get decent cars for $20,000, and they work for 10 years, or they make excellent cars, right? You can always upgrade, but I’m saying for most of the US it seems out of control, expensive and unnecessary, so to speak. A Ford F150 is the same price, but you get way more car, not more efficient, but way more car. And then there is the Chinese strategy, and I was really blown away when I went to China, I went to Wuhan actually just two months before the big outbreak in 2019. And China doesn’t have a lot of cloud in a worldwide brand for, as an OEM brand, of making cars, so they might be involved in lots of components, do you know better than me? Probably way better. But I went there and didn’t expect much, and then I went to many cities in southern China, and it was so quiet, it was eerily quiet, and there was traffic right in front of me. Every single two wheeler was completely electrified, all the buses were electrified, basically all the trucks, there aren’t that many trucks in the cities, and it was like, like traffic in Singapore just quieter, there was a lot of traffic, but I’d say half of the cars were also electrified, the cars were the least electrified, but everything else was 100% electrified. And I’m like, this is amazing, you literally, I didn’t have to put my AirPods in because it was so quiet, and I had like the little scooters coming up on the sidewalk, I’m like, this is insane. Like I couldn’t hear them because it was so quiet, I could hear the birds in the middle of the traffic, could hear the tires, but it was slow moving, most of the roads are pretty traffic, they were so moving. So I thought China hasn’t really worked out in a mass adoption cycle because what they did, they made it really cheap, most of the electric vehicles, obviously there is subsidies, most of them are really, really, really cheap, even the cars are just released, the car that’s like under $10,000 fully electric, and they probably had that before, it’s just a new version, a better version of that. And then we have Europe, which I think is quite different from the third part of the developed world, where everyone was, most of the economies in Europe rely on the combustion engine, right? The Germany is all about combustion engine and big V6 and V8 on the Autobahn, and really tuned for that usage of high speed chasing, so to speak. And they didn’t want any electric cars at all, and they said it’s never going to happen and the Tesla is going to fail. And then suddenly a couple of years ago, like three, four years ago, they suddenly switched from, oh, by the way, now we changed our mind that maybe the Volkswagen scandal was part of that, we changed our mind and now just 10 more years and we’re going to have no more combustion cars, right? We go from like 0 to 100, so to speak, so we overtake everyone else and we do this by government policy, like Sweden is ahead of this in all way, but I think all the other countries will follow. What do you think is the best approach? And I think we all agree that we want the electric cars, maybe we don’t want mandates for it, we don’t want subsidies, some do, some don’t. I think we shouldn’t distort the market, but I do love to see electric vehicles everywhere because I think it’s wonderful for our cities and the places where we live in. What do you think is the best approach? Who do you think is eventually going to win out with their approach? Is it that China is going to take over the world or do you think the Tesla is going to take over the world or it’s going to be the next BMW?

Amit Mehta: That’s a good, very good question. So China has a couple of great OEMs, which mean vehicle manufacturers that are involved in electric vehicles. The big one is NIO and we’ve seen NIO’s market cap explode. I think at the bottom it was $1 share and now it’s up to around $65 a share since last year. NIO has a very interesting business model and I think I believe China will definitely be the leader in EV and AV and the world will follow the trend and there’s a few reasons on that. Number one is NIO’s strategy is not only are they offering the vehicle, but they’re offering battery replacement centers. So you get about 600 miles of range on your vehicle. Then if I drive from Wuhan to let’s say Nanjing, for example, I can just replace my battery based on a subscription based model. Instead of waiting there for 45 minutes or two hours to charge my car, I can just replace it and leave. Another thing what China really has going for it is it’s got the regulations to push all of this AV and EV through and the government’s really working hard on that. Then finally, the biggest thing that China really has is the infrastructure. They have the ability to make the correct infrastructure for AVs to actually work. With this, many other car companies such as Tesla and even the new ones coming out, Mercedes says they’re going all electric on research and development. GM is investing around $27 billion into AVs and EVs over the next five years. Ford is investing $29 billion into AVs and EVs over the next five years. If we look at those numbers, they’re huge, huge numbers. Those will be introduced into China as well. I think China has the ability to actually make that adoption happen because 10, 15, 20 years ago, a lot of people weren’t driving cars in China. To maneuver them from even a marketing standpoint, if you try to sell an EV F150 to a consumer in the Midwest, they’re going to laugh at you because they are used to their gas guzzling, gas powered V8 F150s and they believe that’s what works. It’s like when Andrew Carnegie built the first bridge, no one would take the bridge until he brought an elephant and he showed, okay, this is how an elephant will go over this bridge so you’re safe going over it. That’s what we need. We need that kind of adoption. I’m going to show everybody that range anxiety is not an issue. You can accelerate. Your battery won’t deplete that much. Degradation is not a huge issue. Those have already been introduced to China. The generational thought process behind EVs in China is already there where we’re lacking here is in the United States because we still don’t believe EVs are the future, especially in the Midwest area or even the coastals in New York and all the way through to Miami and different areas. Of course, in California, everybody has that thought process. We already know that EVs are the future. We saw the hybrids coming with Prius. They did a great job when Toyota sold the Prius and now we’re seeing the next revolution which is EVs. There will be a revolution after this. There’ll be fuel cells. Fuel cells may be the end all. Right now, they’re mostly beneficial for very high load equipment such as trucking, but they will have a benefit to come down in cost and price at some point. The beautiful thing about fuel cells is the energy transfer rate. So when we go fill up gas, our energy transfer to go put gasoline into a vehicle is about one minute. If you go charge your electric vehicle today, it takes, I would say, depending on if you’re using a supercharger, but let’s just say you’re using it at home, a regular 240 volt, you’re probably going to take at least six hours to get a full charge, seven hours or so. But the fuel cell is about four minutes. So it’s a beautiful process. People are still afraid to have a big hydrogen tank in the back of their car, but that makes sense. But I think that’s where the world is headed. So we’re seeing these kind of ICE dropping down, fuel cells taking a little bit of higher approach eventually, and electric, of course, being the big winner in the next 10 to 15 years.

Torsten Jacobi: Well, I mean, this week we have the Cold Wave in Texas, and I think everyone who has an electric vehicle that he can’t charge, but having an F150 that runs and you go to the gas station, it shows you the ruggedness, right? It’s a technology we know how it works, and we can control it almost 100%, and we have to deliver a framework. So I think this is going to stay that way. EVs will always be a little less reliable for time coming, and then they will come like digital cameras and analog cameras that took 20 years, so that I think is something we have to watch out for. And I think one thing we have to just, and I think I’m probably also not knowledgeable enough there. So when we talk about EVs, we talk about electric vehicles, right? When we talk about AVs, we talk about autonomous vehicles, correct? And does an autonomous vehicle, does it have to be electric? No, right? These things are not necessarily related. It’s just the computerization that’s often required for an electric vehicle also helps making it more autonomous, correct? That’s why Tesla is ahead of this kind of by definition.

Amit Mehta: That’s right. So yeah, you’re entirely right about that. One thing I want to point out about Tesla, everybody’s arguing about the market cap of Tesla. It’s the biggest discussion I’m hearing everywhere after GameStop, of course, but you’re hearing everything about Tesla. And some people can’t comprehend why the market cap is that much, and some people can. But if you look at it from Tesla’s standpoint, if Tesla is and was the Google of vehicles, there is no Google of vehicles right now, and everybody’s trying to get in that area because the profit pools of vehicles is expected to grow to $326 billion in 2035. And vehicles today for just a market are over $2 trillion, just the entire market itself. Tesla, their biggest growth or value is not their batteries or the car themselves, but it’s the ability to capture data. They’ve been capturing real life data for the last 10 years, and they continue to make their cars better and better over time. And if you think about the markets that can really come out of that vehicle, it’s amazing because you have the ability to go from point A to point B, capture all this data, sell that data to other startups, provide that data to businesses. If someone’s hungry, just like Google recommends Berger King because they pay some extra money, Tesla can recommend the same thing as you drive. They can recommend different restaurants, they can recommend different hotels, and this is actually a value chain for Tesla to make more and monetize that data. So when you look at EV and AV and ICE, you don’t necessarily have to have an EV vehicle with AV equipment. You can definitely have an ICE vehicle with AV equipment, but if you’re going to buy a car that’s AV today, or even in the next three to four years, and you think about it, why would you not buy a Tesla versus other OEMs? It doesn’t make sense because you’re getting this self driving package that other OEMs just don’t have available yet. So that’s kind of my opinion. They really need to catch up. OEMs have a particular brand value, brand loyalty today, but Tesla is kind of eroding that. And as they introduce their cheaper and cheaper cars, such as the 25,000 model that’s supposed to be coming up, it’s going to even take away market share from Camry or Accords in the future.

Torsten Jacobi: Yeah, Tesla, it’s an interesting case. I can’t make up my mind about Tesla, to be honest. For a while, I was really on the side of the short sellers, and I always feel, and I still is, Tesla feels like an accounting fraud. I’ve seen through all of them in the 2000s, and they are like the textbook example of this. But on the other hand, and I still think they are, on the other hand, I think they are driven by an engineering spirit that doesn’t just talk to talk, they also walk to walk. They have a bunch of misses. They always announce things a couple of years too early, but that happens to a lot of us. And I think if we blindly give money to an accounting fraud, it should be Tesla. If there is someone who has big ideas and really changes the world a little more than just doing another app and another app and another social network, which I don’t think help us that much. They help us a little bit. I’m not saying they use less, but I’d rather give a trillion dollars to Mr. Musk than give a trillion dollars to Mr. Zuckerberg if I get the choice. So I’d rather donate, so to speak, my money in Tesla stock, and instead of having companies that I don’t even know why they are so big, and nobody really knows. Nobody can answer this question sometimes about Facebook. So that’s kind of my weird way to think about Tesla, and they do a beautiful, they do create beautiful products. So it’s sometimes hard. There’s a lot of good things and bad things baked into one thing, and you don’t really know beforehand if what is actually, is it a driving story that’s just literally a driving story? Is it coming out? Is it a driver for this company? Is it a good thing or the bad thing that comes out? And I was so wrong in 2000, but there’s something strange that’s going on in the market these days. But I want to ask you about the Apple Car. So we hear a lot, and there was an article I saw today, Hyundai is apparently readying production for it. So it seems to be on the horizon. And what they, nobody has a real specs, there’s a lot of speculation. Do you, do you think this would be a full driving vehicle, a full self driving vehicle? What would that entail if it is? So what does, do they actually mean by this? And do you think we’re going to see this in the next two years or is it just speculation?

Amit Mehta: So Hyundai, Kia and Nissan have both, all three have denied reports that in the recent, maybe yesterday, since yesterday that they’re going to actually build the Apple Car. Based on that, Apple does have a lot of key indicators that indicate they are working very aggressively on developing this autonomous Apple vehicle. And I think we’re not going to see it in a couple of years. If we do see it, it won’t be before 2025. And the reason is they’ve taken a different approach in my opinion. A lot of companies take the approach of let’s build a car, put it on the road, and then try to get data and take that data and then sell it to the public after we start selling the car. So they’re monetizing on their CapEx. But Apple has the money to first build up the data, build up a car in private, and then sell an entire solution or package, making sure that it’s fully connected with this ecosystem. Because their goal is to monetize on that ecosystem they already have, which is the iTunes store and the Apple store, et cetera. Seeing that, when I read an article recently, it basically said that one third of the market of the United States is a luxury market. And Apple is going to go after that market very aggressively. And even if they get just 2% of that market, that’s $26 billion. And that’s just on vehicle sales. If they start adding their subscriptions and their other iTunes, Apple TV to that, can you imagine how big that market can actually be? So if you were Apple, if you were Tim Cook, would it make sense to you to sell a car? Probably not. It makes sense to you to sell the ecosystem around the car. Because you’re heavily investing the CapEx there, but how do you make sure you can get the subscription to fulfill whatever that cost of the car is? So I think the car will be subsidized in some way by Apple. And they’ll really make a lot of money around the ecosystem that drives that vehicle and makes, I guess, Apple fanboys, as you call it, really geared towards buying Apple products.

Torsten Jacobi: Yeah, I’m surprised they seem to be so far ahead with this. And because Apple is typically a bit of a late adopter, right? They see what’s out there, they take someone else, take the blame, and then they put the engineers on. It’s a very German approach, right? German engineers like that, too. They wait until something bubbles up somewhere, usually in the US, then they make it much better and then they sell it as a luxury product. And for that, you need the initial wave to go through and then you basically attach yourself to it. And the initial wave, strangely enough, we see very little adoption, maybe it’s changing now, so I’m curious on your perspective, but the adoption of electric vehicles seems to be, in the mind of most people, to be Tesla, which is very expensive. And I think this is really reducing the upside because they don’t want to cover, and nobody else seems to want to cover that market on the low end, which China does so well. Where do you think is the US market? And I know you obviously want it to grow, but realistically, where do you think is our electrification that has some autonomous features like the self driving cars that’s fully self driving, but highway self driving that the Tesla already features? Where do you think this looks like in five to 10 years from now? Will we have a bunch of different brands? Where would they be? You mentioned NIO earlier, I guess that’s one of the manufacturers you would bet on.

Amit Mehta: So that’s a good question. I want to start with a little thing from McKinsey. McKinsey says, and it’s proven today, the average person pays around $543 per month for their vehicle. As an electric vehicle costs somewhere around $900 a month. So we have a $400 gap to get to where the average person wants to spend their money, to make sure that they can afford an electric vehicle. And no vehicle is actually there today. But electric vehicles are a lot easier to make than internal combustion engine vehicles. They require about 25% less components or labor, I should say. So the industrial line, especially for, let’s say, Volkswagen or Toyota are 25% smaller. That being said, we had a big hiccup this year where a vehicle sales went down from last year, I think, I believe it was around $17.5 million, or the year before, sorry. To last year, it was somewhere in the range of about $14.8 million in the United States. So huge drop. I mean, we were shut down for two months, even our factories in North American lighting. If you look at that and you say, okay, when will we recover? How will we get to the point where we need to be? The biggest indicator of car growth or people buying cars is people getting licenses. There’s a 0.86 regression correlation for our value for people who get licenses versus how the growth in vehicle sales. The second biggest is unemployment, which is 0.79, and the third biggest is population, which I believe is 0.78. So if you look at those three values, the population is expected to grow from 320 million or 30 million to 360 million by 2030. The unemployment rate, I’m not sure it’s very hard to predict, but right now we are in a tricky time, but I’m sure it’ll start to go down as we go over the next 10 years. And finally, the first one I said is licenses. As you go up in population, you’re going to see people that actually increase the number of licenses they have. The problem is, when you look at car vehicle sales, the prediction is to be around 17 and a half million in 2030, whereas we just sold 17 and a half million in 2019. So we’re going to take 11 years to recover. That’s the biggest question. What is happening? We’re seeing a population growth. We are seeing license drivers increase as a forecast. The only thing that could really affect that is unemployment. So are we predicting that employment should go down over the next 10 years? Potentially, because we’re adding all these new functions such as AI, we’re adding a new industrial optimization, we’re adding electric vehicles, which I just said need 25% capacity. When you start adding that, these kind of, I would say, good value jobs where you were making 70, 80,000 in the Midwest, they’re not needed anymore, and they start pushing them to very low value jobs. And that might be an indicator that we’re up for a tricky road ahead in the United States. How do we get past this 10 years? So we can only hope that the only way that we can actually push electric vehicles through over the next 10 years, or even an internal combustion engine, is by stimulus or support from the government. So tax credits and these benefits that now the government’s offering Tesla and GM, they’ve reignited these credits that they’ve expired a couple of years ago. I think it was around $7,000. It’s coming back. So now we have to artificially boost these electric vehicle sales. And I think that’s going to be the big question. If you take out the market manipulation or anything like that, yes, electric vehicles are going to come. We’re going to add more infrastructure, more charging stations. And if we look at the regression analysis of the population growth, it makes total sense. But there is a case of unemployment and AI will eat some of that up. And as we get more efficient and move the electric vehicles, the automotive industry will actually decrease in the number of labor required. So high end jobs will increase, lower end jobs will increase, but the middle jobs will kind of be pushed out. And I think that’s what we need to focus on. So it’s a very hard question for you to answer when we’ll see this revolution take place. And it could be pushed back or forward, depending on how the government plays out on this.

I like how you just explained the world by looking at the number of car sales per year. That’s fascinating, that’s a fascinating way to go from where car sales are, where there will be and then like retrospectively explaining the world that’s fascinating. So I think it’s a spot on analysis, by the way. And I think we’ve had a bunch of different guests and we talked about AI in particular. AI will be getting better and better really quickly. And what it will do, it will give us in the first round, I think, of AI that’s already happening. It gives us better decisions basically for free, for very low costs, because once you make a decision generating algorithm, it’s relatively difficult like a chip, like an Intel plant, the factory. It’s hard to build and relatively expensive, but then you just need sand and labor and energy and it churns out enough chips as you want and then you have to discard a couple years later because it’s just not state of the art anymore. But in the second part, and I think this is still quite science fiction, but it might happen in just 10 years is that we’ll see is kind of something that kind of resembles the decision power of a human. I might not have the physical stature of it, so it won’t be a robot that walks around, but it is something that as an employer, you wouldn’t know if this is a freelance on the other side of the world who does a good job for you, or does something really difficult that you couldn’t do yourself, or if this is a complete AI. And I think what we will see is that we will go from 10 billion people, 9 billion people will suddenly have the labor, but labor in the sense of it’s digital creation and digital change of the world, we will go to 100 billion or 200 billion. So the job prospects for everyone who is still human, but has probably very different costs associated with that will be a real problem. And that will happen in the next 10 to 15 years, but we are not that far away from it that for a lot of AIs, it’s very difficult for us to see, is this a human or not, from afar, right? As closer we get, as more likely we can see that, but for a lot of tasks it’s not required. It’s just a decision, the correctness of the decision is what’s required, not if that’s a human or not. So I think you’re a spot on there, and I think this is something we have to go through, and it will be unpleasant. And I don’t know if car sales is the only way we see this, but we’re going to see this in unemployment, we’re going to see this in the way this society is going through this upheaval. And we do it in the U.S. first, but I think it will capture everyone in the world. A lot of people are laughing at the U.S. and say, well, you guys are a banana republic and they’re right, right? We have like these crazy things going on and people are all depressed. But you know, we call it the Great Depression 90 years ago in the industrial revolution, because that’s what happens when you have some massive change that puts you first out of business, and in your lifetime it might come back, but you might be 20 years older by the time the golden times return. So I think that’s the big danger people have to grapple with. And I think having a car, well, it’s great. There’s so many alternatives now. Like I didn’t have a car in about a decade, and I used to have like all these great cars. I didn’t own a car anymore because wherever I am, there’s Uber. If there’s no Uber, then you know, there’s generally an alternative app. And I think there’s a fascination with cars that I still share. But like living in San Francisco, it’s extremely expensive. I think $500 is hard to find a car on this because parking is $400 and then you just leasing is whatever, and now the $400, and then you can’t park it anywhere because literally during the day, we have now roaming bands of thieves just going through the city and breaking the windows and trying to steal what’s inside or just trusting your radio. So you can’t leave it unattended in the city. So the car ownership on one side at least in coastal cities, it’s definitely in Texas, but car ownership is a nightmare. I don’t know why anyone would go through this trouble anymore because there’s such cool alternatives. Right. And you’re totally right. I think you’re totally right, but think about this. Your utilization of a vehicle today is 4%. If you were to drive normally, it’s approximately 4%. So 96% of the time, your vehicle is sitting idle. But as Musk just introduced through his tweets or his texts, he said that by the end of 2021, he will have level five available and ready on his vehicles. Now if you comprehend that you don’t own a car right now, but what if your car, like your house turned into an Airbnb when you’re not there, what if your car could be something that produces money for you when you’re not using it? Would that make you want to buy a vehicle? I think so. I think that adds a lot of value into you buying that vehicle, even if it is a couple more hundred dollars more expensive because you think you can make that money back. But we already have that, right? We have two row. We have get around and they kind of do that to give you, you rent out vehicles to strangers and they do the insurance and they deal with all of this. And I felt it’s so much hassle. I mean, maybe you can make $500 a month, it’s possible, but it’s on the high side. Say depending on your car, you may be able to make 200 to 300, but they have strangers renting it for an hour, smoking weed inside and you can throw away the car, right? It’s so easy to ruin a car, but it’s very difficult to maintain its value. So it’s doable. You can make money with this, but I don’t think it changes the equation much to bring up the utilization because there’s a reason utilization is so low because the highest effectiveness of a car is if you drive with yourself and you park wherever you can and there’s nobody else you have to coordinate with, right? That seems to be in our minds and of course, the coordination of Uber has changed already. But in our minds, that seems to be the easiest because I go, it’s waiting for me downstairs. Maybe it’s heated like that, the Tesla and I don’t have to worry about that. But then I drive just myself somewhere else and park it right in front of where I want to be and I can go just anywhere I want. For me, that’s the highest utilization of a car in the best utilization, not high utilization in a mathematical sense, but that’s how I want driving to be. I agree with you on a certain set and it makes total sense as we move into the pandemic. I think car sales are going to grow because if you have a child or you have an older parent, you no longer want them taking an Uber or you no longer want them taking public transit because you don’t want any contamination risk. So you purchase these, I would say, not low cost, but $10,000, $15,000 vehicles for your parents or for your child. So I think car sales will grow in the future, but you’re right. Something to add a little bit more context, which I think will change everything. What is Uber? Uber is a middleman. What is Turro? Turro is a middleman. Basically, they’re taking 20 to 25% of what you’re paying to the driver as you travel from point A to point B. There was big news a couple of days ago, Tesla bought $1.5 billion of Bitcoin and eventually they’re going to start accepting Bitcoin as a token for purchase of a vehicle. And if you think about that, what is a Bitcoin? It’s basically a ledger. It’s a debit and a credit. You take money from one side, you give it to the other person. And if you think about the value Tesla can bring if they add Bitcoin as a payment system on their vehicle, they’ve automatically removed the middleman, which is Uber. Now it’s a trustless based decentralized financial currency system that can be used with impeccable trust because now you can trace the entire blockchain, whoever has ever used or purchased anything in the past, you can see the entire transaction history. And I think that changes your mindset to buy a vehicle even more and maybe rent it out as a secondary income. So you own your own private vehicle. Now you own a second vehicle that you use as an income generator. And that could be used to make 20 to 25% more than you would with Toro or Uber. Yeah, I don’t know about Bitcoin. This is a whole another can of worms. So maybe we can open it in a second. They have lots of thoughts of this, but let’s hang on for a second to, I have two more things or like a couple more things and maybe we can go into Bitcoin in just a moment. One thing is, there is this long standing debate that I have really no clue about how this actually, how this started. And what I know is that, that you and your company and many others are big fans of LiDAR, which is basically an adaptation of radar from what I understand. And Tesla is a big fan of cameras. And what the other way around, I actually don’t know, no, it’s the other way around. There’s these two orthodoxies in the market and they seem to be developing independently. And I wanted to find out where you stand on this, maybe have to first figure out how they actually work. I have to understand that better. Yeah, let me give a high level overview of what each of these systems are. So camera and LiDAR are optical systems. That means they need a field of view to see what’s in front of them, send out a light, especially with LiDAR, sends out this laser signal, creates this point cloud based on reflectivity. You’re wearing a black shirt today. You’re going to be less reflective than your white earpods because it’s just a more reflective material. So it creates this point cloud and provides this information back to the vehicle. So imagine a laser going, hitting all these spots and then coming back and refreshing around maybe 15 to 30 frames per second. Then you have camera. Camera is also an optical sensor. I think we know how camera works. It creates an image of the every 60, let’s say 60 frames per second, 30 to 60 frames per second and sends that back to the vehicle. Now, LiDAR is a very interesting component. It sends out a frequency, a chirp, I would say, of anywhere between, today it’s 24 gigahertz, in the future it’s going to be 77 gigahertz. So it sends out this radio frequency. It waits for something to bounce off of and it receives that signal again. But this is not optical. This is a non optical sensor that can be hidden behind something such as a bumper or even a headlight. In the back of a headlight. When you think about Tesla’s view, what are they saying when they say we only need camera and radar? What he’s really saying is it could be very correct for him because he’s been building a neural network using his camera and radar for the last 10 years. So every time you drive a Tesla, you gather data for him, you go to your home, you have this over the air updates and that sends him back the information. Now he’s creating a better and better neural network crowd sourcing all the Tesla owners around the world. So his network is much better. When you look at OEMs getting started today, how do we catch up to Tesla? Because they have a much better neural network than us. They can recognize objects and make decisions better at this point. Well we need a tertiary sensor and that’s LiDAR. So when you look in the day, you have camera, which uses, it needs light to see unless you’re using a thermal camera, but if you’re using a regular camera, you need light to see because you’re in the visible spectrum. Then if you move to LiDAR, you’re now in the 850 to 1550 nanometer spectrum. That’s not visible to us. Our spectrum is around 450 to 720 nanometers. That’s why we see the colors of the rainbow. But LiDAR is in a spectrum that we can’t see. So it’s a solution that is needed to make an image or to sensor redundancy for the future of autonomous vehicles. Is it the end of solution? Maybe not. And the reason is, I’m personally tracking about 82 LiDAR startups today. And a LiDAR is basically made up of a laser followed by a beam steering mechanism. So you shoot a laser and there’s something that steers the beam because you want to cover a certain field of view, whether it’s 60 degrees or 120 degrees. Then it takes a reflective image off of, let’s say, a car and there’s a photo detector. That photo detector detects either a single photon, such as you have in your new iPhones. It has a spat array, a single photon avalanche diode. So it can detect just one photon and create an image. Before you have something else, there’s SIPMs. There’s all kinds of different ways to detect light. And then it creates that image based using some algorithms to create a field of view for the vehicle. That’s basically something that’s a very high cost component today. And even in volume, I know a lot of OEMs, a lot of startups are saying, oh, that we have an opportunity to make it $100. Maybe. Maybe in volume, it would somewhere still be around $800 to $1,000 when I say $100K of LiDAR. It’s a very expensive solution. And I think as you as a car owner, you’re not interested in $1,000 solution to make your vehicle autonomous. Radar, there’s about seven or eight radar companies that are working on high definition radar. The key difference between radar and LiDAR is radar doesn’t need light. It can work at night. It still sees the same way. But the angular resolution is very bad. So today, it’s about one degree of angular resolution. What that means if Torsen was 100 or 200 meters away from me, I can’t tell if Torsen is a cat or a human or a bag. I just know something is there. Whereas LiDAR can tell me that is a human, do not hit him, especially at night. But there are companies, seven or eight startups that are working on high definition radar. And they’ve gotten the angular resolution down to 0.5 degrees. Today LiDAR’s angular resolution is 0.1 degrees and can drop a little bit further. But OEMs want at least 0.1 degrees of angular resolution. So if you can imagine, if you just look at the relativity, angular resolution of radar, high definition radar is about five times higher at its best than LiDAR today. If they come up with a solution to make angular resolution of radar match LiDAR, you’re going to start cannibalizing the LiDAR market. Because LiDAR is a effective proven solution that’s already on vehicles today, and it’s used by many OEMs, and it’s a lower cost solution to provide than LiDAR because of the laser diode and the photo detector that you don’t have to make and the optics. So I think, I don’t want to downplay LiDAR, I think it has a lot of potential. There’s companies out there such as Ouster who are saying that they’re going to have around a 10 billion TAM, total addressable market, in the next five or 10 years. And there’s many other companies out there working on infrastructure LiDAR, railway road LiDAR, detection LiDAR. At Coito, we also have 30% market share of traffic lights and street lights in Japan. There is privacy issues of using camera because you’re detecting somebody’s face and you’re just not allowed to do that. LiDAR doesn’t detect your face, it only detects that you’re a human and moving in a certain way if you’ve ever seen a point cloud. So that’s another benefit of LiDAR. But is it the end all solution for vehicles? We’re not going to know yet. It’s just something we can’t answer. Musk hasn’t been wrong before, really. He’s done something that no one else has really done ever. If I was a betting man, I wouldn’t bet against him. So we’ll see what happens in the future. But there is a market for LiDAR. How big that market is, I can’t really say today. That’s really interesting. I mean, finally, I understand the difference between the Windows technologies and the philosophy behind it. And I think it’s stunning. If I put my entrepreneurial head on, one thing that immediately seemed obvious to me, I guess you can always, in order to get a better resolution, you can have a better sensor, but you can always just stack on a couple of those radar beams. Instead of sending just one, you send 10 or 20, and then you just compile the best image out of those slightly shape shifted signals. You can do this in 3D, this is how 3D radar works in a bigger resolution. Is that a question because is that not working yet because it’s a chip issue, because the chips are expensive or because the radar wave is so big that actually that makes it harder to see, even if you go through it 10 times, you still don’t know what a small item actually looks like. Is that the problem with radar? The problem is, is there something called a local oscillator? I’m not an expert on radar, but there’s a local oscillator, meaning that there’s something called FMCW technology. What that is is frequency modulated continuous wave. So one of your questions should be, if I send out a signal with my radar, how do I know that I’m getting my signal back? What if everyone in the world has radar? How do I know I’m not getting your car signal, and that could make me crash? I say, oh, that’s Torsten’s car, I could be crashing, that’s not the right signal. The way they do it is they use FMCW technology, which is available only on radar and 1550 nanometer LiDAR, and I’ll explain 1550 nanometer LiDAR between, it’s a difference. But the radar solution sends out a signal. When it sends out that signal, it keeps a little bit of its signal to itself. So maybe it only sends out 99% of the signal and keeps 1% for itself because it’s waiting for that signal to come back to make sure it matches. And that’s on a local oscillator. But as you send out more and more and more signals and you want to get higher angular resolution, this local oscillator has to get bigger and bigger and bigger. And as you get bigger and bigger and bigger, it doesn’t fit on a vehicle correctly. So that’s the tradeoff in physics. As you go down in angular resolution, your radar system starts to become more complex and bigger. So there’s a big tradeoff in cost and in benefit, especially for automotive. If you’re asking military applications, yeah, no problem because there’s no budget there. But we start looking at automotive, there’s a big budget, so there’s a lot of cost restraints. But there are solutions working on it that we’re thinking of radar purely as a hardware component. What if you start thinking about it as a software based solution, where you take what you have as hardware and multiply the effects of that using software, like we do in everything. We scale everything using software. There’s a lot of companies specifically based in the Bay Area working on not only hardware solutions but software solutions to kind of proliferate this signal to make sure we can keep the physics smaller while going into a more smarter solution for software. But you’re still adding cost because now you’re going to add processing power. So the tradeoffs are, there’s so many tradeoffs that we’re not going to start seeing this type of technology on vehicles until 2025 or 2026. So that’s one area that I think is going to grow. And I just want to explain the discipline 1550 LiDAR and 905 LiDAR. 905 LiDAR is, as I said, the wavelength is very close to what we can see our visible spectrum. You can’t pulse that laser with too much power because it’s not eye safe because it’s too close to the visible spectrum. So you have very limited power budget to work with. When you move to 1550 LiDAR, you have a lot of power that you can pulse this with because it doesn’t affect the eye. And what that means is I can get 200 to 300 meters of distance versus 100 to 120 meters of distance because I’m using 1550 Nm LiDAR. But as I said before, the physics behind it make 1550 Nm LiDAR four or five times more expensive than 905 LiDAR. Now if you’re thinking 905 LiDAR is $800 a unit, if you look at 1550, you’re looking at about $3,200 to $4,000 a unit. That’s in scale. So there’s a lot of tradeoff still happening. I know there’s a lot of LiDAR SPACs and a lot of interest around it in the last year and there’s more coming. And a lot of those LiDARs are working on 905 and 1550 solutions. So if anybody’s ever interested, just research those two solutions and you’ll see that the market is kind of growing for those startups today. I find it super fascinating. I remember the times when we had modems, and I don’t know if you were around, and we had modems and then we came to a physical limit. So there were physicists that were going on TV and said, okay, this is it, and I don’t know, it was maybe half an ampit. I forgot what it was and said, there isn’t much you can do beyond that. And it was based on audio waves, right? Because if we go lower, then we can’t distinguish the signal from the noise anymore. And it was like two years later, someone came around and gave us ADSL. And then Fiber came around and basically at the same time, it’s just it’s more expensive and you have to restart from scratch. You can’t use the same alliance and then we realized, oh, we can use just the old cable. And we can pump a couple of gigabits over cable now. So I find it so fascinating and it’s the same race with the iPhone that we had these very expensive sensors that the iPhone really wanted to have in the iPhone, but it was like a few hundred dollars for gyroscope. And then two years later, it suddenly a 50 cents thing. And this I think is an underappreciated way how we made the world a better place really quickly because we think of Moore’s law and we think about big desktop computers or maybe now we think about the iPhone and the new A14 chip, but actually these sensors and the making something relatively complicated, that’s maybe few components of hardware and software and then it gets integrated and the crazy part about that is how cheap it gets. Like it’s for Moore’s law, we think, okay, every 18 months, it doubles for the same price. But these sensors, I feel like it’s like it goes down a thousand times in like two years. Like the rate of progress seems to be so much higher because it’s a concentrated problem and you don’t have to worry about, you know, the old, I think with the 486 architecture was a big deal for Intel to upgrade to a better architecture because I had to be backward compatible. You can literally just redefine the API and really outscale Moore’s law because we all think in Moore’s law terms, right? We know what the iPhone is going to look like in 2030 and it’s going to be crazy. But its sensor can be, you know, we can have something that’s, I think that’s what LiDAR, everyone is a bullish about LiDAR right now. We thought it’s something that you have to put on your car. That’s the old first cars that I saw in San Francisco and Google had a couple of those. They had this massive $100,000 set up on their roof and now you say it’s $1,000, that’s pretty stark. I mean, three years and, you know, do another three years of those and it’s like a $1.00 equipment, right? That’s the hope, I guess, for the one survivor in that race who manufactures sensors for pretty much any car in the world. Do you guys have any, as a developer, do you look at these startups and you’re just, you’re trying to predict who’s going to win this race or you just, you kind of play it safe and you buy a couple of those depending on the price point. What’s your role? Can you fund startups that have these sensors that you want to sell to OEMs? Yeah, we put $50 million into a Silicon Valley based startup in February last year called Septon. It’s a LiDAR based startup in their Series C and we see a lot of progress on how they can produce these LiDAR systems and provide them to OEMs and they’re a very smart team. Years of knowledge in LiDAR based solutions and you have multiple startup exits. So the team was great. We also put $24 million into a company called Brightway Vision in Israel. They’re working on gated camera technology, very edge case based. Gated camera technology is basically, if you’re driving in San Francisco over the Golden Gate Bridge, there’s a lot of fog. Do you continue driving? Well, you can’t because you can’t see, but gated camera can see because they’re sending these snapshots of pulses of light and taking images at each pulse. So one millisecond, 10 milliseconds, five milliseconds and sending this information back to the vehicle so they can clearly see what’s ahead of them. But if we don’t know who’s going to win, I’ll be very honest because it’s impossible to predict right now. Just like when you had VHS and beta, for example, or BluRay and what was the other one, HD? I forget the competitor BluRay, but that one died out. So it’s not going to be solely based on technology. It never is. It’s always based on technology, the business model, and the relationships of that company with different tier ones. So the fact that we invested in the SEPTON is SEPTON the best solution out there. So they send a laser to something called a micromotion technology for their optics. So they’re scanning their laser and they’re getting a solution back. Is it the best out there? Maybe not. But the fact that a strong tier one has agreed to take this startup and coach it on the two most difficult and costly things in automotive life cycle, which is validation and qualification. Those are the most expensive two words you’ll ever hear in automotive to qualify and validate the technology and bring it to market is a big key point. So if you’re going to look at a solution from a standpoint as an investor or just trying to understand the market, don’t clearly look at the solution because it’s not the way to look. It’s to see who’s backing that specific solution because that’s the one that will grow and proliferate and bring the cost down because they have the ability to work with the startup and make sure we can validate and qualify those technologies. It’s the old boys network at it again. I mean, I’m on two sides of this. There’s one side of this podcast where I go into the value system of the old boys network, so to speak. I get a lot of criticism for this. And then on the other side, I feel like the old boys network is really reducing meritocracy and it’s doing us all at this service because we’re not looking at merits. We’re looking at measurements that we infer could have this value, but we don’t understand them enough. So we look at what someone else who is the old boy in that sense, it’s an OEM or it’s a T1, has a validated already. And then if you buy IBM, you will never be fired. So we go back to the safe situation, although we are in a race for something really innovative. And I think this is really strange sometimes how this happens and I learned that when we even we had our software product and we sold mostly to the automotive industry and procurement solution, it took us I think five years to sign up Daimler Chrysler at the time now Daimler Vance again. And it eventually happened, but the amount of effort you have to put into this that has nothing to do with your software, right? It has nothing to do with does it work because we knew it is on day one, maybe not day one, but like 12 months later, it was like software that worked perfectly. But going through all these steps, building this trust, creating also an entry level is something completely different that I learned being an entrepreneur and I don’t want to put this bad label on it, but it’s entrepreneurship is not and I was debating this with Daniel Ross who runs the Pioneer App Accelerator, I felt entrepreneurship in many cases is as much psychology as it is technology and meaning the technology of your customers, your investors of the people who are stakeholders to putting them all in the same trajectory, obviously you have to hope for the market to actually go the way you bet on it. But the technology is an important ingredient, but it’s not, as you just said, it’s not necessarily the decisive factor. And I think a lot of entrepreneurs are underestimating this and for one thing, because technology we have control over a little bit at least, but we don’t have any control or we feel we don’t have any control about our environments is this really a wave of technology that’s useful or is it just a mania? We don’t know that before we start, they all look like they all look great, but some of them be five years later say, oh, this was boring, why do we even look into this? And now there’s we’re so excited about. And then you have very little control about, will it be adopted in a way that you think it is by other stakeholders in that game, right tier one supplier, or that’s like, which social network actually goes viral? I’m grappling with this finding a real predictor for this. Like I don’t know if there is, but I interview a bunch of VCs here on this show and everyone is like, they have their own rules that they’re publicly saying, right, they’re very open about that. We don’t know the future. But they have to make that call, right? They have to make that judgment call and say, okay, this is the company I want to invest in. I only have 100 million, so I put 10 million into this company. And then sometimes I’m not sure, is it a gut feeling or what I’m trying to, you know, get to that point where they feel confident with their decision. And I’m honestly struggling a little that maybe people don’t even know. That’s what I sometimes feel. They don’t really know why they made that decision. And then in retrospect, it’s becomes clear. But in the moment and for the first couple of months for these investments, it’s completely it’s completely unclear to themselves why they made that decision. Right. Yeah, I totally agree with you, I was listening to Elon Musk’s new podcast with Joe Rogan and he said, you know, a lot of people are investing in SpaceX. But if you look at it, there’s no market in space. I go to space. What’s the market? There’s zero market today in space. Yes, there’s a potential for a market, but there’s absolutely nothing that’s going to prove that there’s a market in the future. So what are you investing in? The first person is a team, you’re investing in Musk. He has a proven record of what he’s done in the past, right? The next thing you’re investing in is his ability to bring business models to the market. He’s disrupted the entire chain of the automotive industry. Specifically, my favorite part is dealerships. His ability to eliminate dealerships, the real middleman. Of course, he can’t do it in Michigan. They’ve lobbied against it. But dealerships are not anywhere else. You can’t go to a dealership and buy a Tesla, but everywhere else you have to still go. And where do dealerships really make money? They make money on new car sales. They make money on financing. They make money on service and they make money on aftermarket. Their most money is on financing and service. Well, he’s taken that out of the equation. So if you’re going to invest in somebody and you look at a startup to invest, you’re looking at the team, you’re looking to me, you’re looking at the business model, and you’re really looking at the market. Then finally, after those three things, if you’ve unbundled them and kind of unpacked them and said, okay, this guy has created many startups in the past, he’s created a business model that actually works and he has a market that needs to be, there is a market for that. Finally, is there a market disruption? Can he disrupt a market and create a new market out of that? So that’s how we approach our systems when we’ve been investing in the past. Yeah, you’re a real disciple of Elon Musk. I can see that. I think he’s revolutionizing things that have never been done before. And I think it was time to do that. When you see what’s happening entirely around the world and how it’s being affected by all these political influences and what we’ve seen in the past, I think there is a need for change. And this is actually happening. We’re moving forward. Yeah. Yeah, it started from very humble beginnings. One thing that I wanted to pick your brain is, and I read this fascinating book a while ago, and hopefully we can edit it in later. I forgot the name. The title, what it does, it illuminates a really interesting case. Actually it’s about two things. One is, there is an AI that becomes to an extent self conscious, but that’s later in the book. And what it actually starts out with is there’s a collision between two cars, both of them are self driving and on an autonomous mode. And it is for the investigators, it’s completely fictional. This is science fiction. For the investigators, it seems not clear what could have caused that accident because it seems like from the outset, this accident cannot happen because this technology is error. It’s free of errors. This error has never happened. It cannot happen because the algorithm cannot predict it. And then even if it would have happened, the people who were in the car with more passengers should have survived and not the single person on a motorcycle. So the ethical question that the author raised, and I think this is something a lot of people ponder about, and there’s definitely no easy solution. What do we do if no humans are involved in this ethical decision about life and death? So when you were sitting in a car and you see the computer predicts an accident, you will have to figure out, are we saving the people inside the car or the people outside? Are we running over the pedestrians? Are we killing the people inside the car? And the difficulty, that’s another one of those judgment columns. Now we do this in a car. We don’t want to make these decisions, but I think all of us have made those decisions subconsciously at some point. And it seems to be not a big deal to us. We are kind of ready. We’re not ready, ready that we want this to happen, but we know we would make that decision in a heartbeat, and it’s correct for us. And most people around us, I wonder what you think once we have cars that basically run like all the algorithms, how do we give them that knowledge of what we would actually do in that situation that’s cultural appropriate, right? I think there is no perfect solution, but what’s culturally appropriate is the self driving car industry. Is it thinking about that? Yeah, it is. So there’s about 40,000 deaths per year in the United States with traffic, roadway fatalities. That includes cyclists and pedestrians and automotive fatalities as well. And what I heard, and this is from Jim Adler, he’s the head of Toyota iVentures. One of the things he said was, we empathize with a person who’s been in an accident. So if I was to get in an accident and run over somebody’s dog, I empathize with that person because I know that could have been me, I could have ran over that dog as well. But when you give it to a machine, and a machine runs over the dog, you have no empathy for that machine because you can’t feel that a machine should be able to do that. So the algorithms that we really need to build into the AI to make sure that it works in the future to avoid these kind of solutions is going to be very difficult for us to make those decisions for the machine because there will be an issue that occurs. For example, if you remember in 2018, there was an Uber driver that was testing a vehicle in Phoenix and this person ran over somebody with a bicycle and killed them. It wasn’t an Uber driver. There was an Uber autonomous vehicle that was testing with a person inside who was supposed to be monitoring the activity on the road but was instead watching Hulu on their phone. And she didn’t pay attention and somebody was crossing the road and it hit this person. And she died. Immediately within three months, Uber shut down all activities because there was such public uproar about how could Uber allow this to happen. So Uber had to shut down all its self driving activities and then slowly restart them. Finally it wasn’t profitable so they sold it to Aurora recently. They sold their ATG advanced technology group to Aurora. But the point is no one is really ready to release that and make that decision on how that AI should be trained. And that’s going to be something that no one can actually make that call today. We’re just going to have to see what happens and make a decision on do we continue on with this and continue to push self driving cars or do we take a step back? And so far what we’ve seen our only test point or test case is what happened in Phoenix with Uber. Well, I think the general issue is that we don’t have an algorithm for this but crazily enough it’s because it’s so inbuilt into culture. We kind of know what’s right immediately, right? So cops arrive at the scene, talk to a couple of people and they know okay this was an accident or this was like someone who was negligent, right? They literally, in that moment they make that decision to take a suspect home or not. And it’s not hard for them. I mean, you know, there’s errors but in general people look at the scene and 99% will agree this was one of the options they, maybe it was the right call, maybe it was a bad call, maybe this was an negligent homicide, maybe this was just an accident. And on the other hand we have AIs and that when you work, you probably have done this too. I’ve worked a lot with AIs recently on algorithms. There’s a couple of problems involved, one is they are really, it’s really hard for them to get to 100%. They’re really good at getting to 90% but the last 10% is real hard work very often they never get there. And I think this is why we haven’t seen full autonomous driving because as more parameters you add as further away you move on 100% and that’s why city driving with the Tesla will be hard for remaining autonomous driving with the Tesla will be hard for the time being. And then the other problem is AI by design doesn’t know why it’s doing something. It literally doesn’t know it can’t explain anything. It doesn’t have a concept. It’s a pure statistical model. We think actually we don’t really know. A lot of AI’s appear, they feel like magic. A lot of people say even the people, the researchers were like we don’t know why it works but it works and I’ve got to publish this white paper and that’s great. So but the rules when we just talk about it, they read like a law. If there’s like a history, like the computer code we’re used to, if then. And there’s a couple of exceptions and there’s else if and else if. So it’s something that we can code in a law but what I’m trying to say is how do we make it apply in that moment because technology is written in a completely different way. It’s written in a Behesian way, it’s pure statistics and we kind of see the dots and that’s what we believe, which is probably fine for most things. But in that scene where every human knows, okay, distinct counts and it’s really important we get it right. AI’s are a huge disadvantage because they never get it 100% right. They never know it’s important, right? No AI knows today it’s important. That’s why they make these silly mistakes, like they run into standing trucks because they didn’t even know it’s important. Everyone else would see the situation coming up, say oh it’s important, but a focus put my eye on it and you’re good. So we don’t have a focus solution. We don’t have a priority solution for AI and I don’t even know how the framework is for making that decision even start to the automotive industry just using AI technology on behalf because it goes counterintuitive to the exact problem that we are looking at. Yeah, I mean AI is such a broad term, right? We’re just saying AI and then you have deeper levels of AI where you look at deep learning and then you get to unsupervised learning or supervised learning and you get to unsupervised learning. So the key is going to make sure that we get to unsupervised learning, that this machine can learn itself and say okay, I understand this is how humans should react and this is the decision I should make because as we keep feeding for supervised learning, we got to annotate all this data, feed labels, keep going, going, going, but how do we make that clear distinction between supervised learning and unsupervised learning? And once we get to unsupervised learning, I think that’s when we’re going to start to see a real change and as you said, it’s a black box. We don’t really know what’s happening inside, but that AI will be able to make that distinction change. If this is how a human mind works, this is what they’re thinking and this is how they’re going to make the decision. Unsupervised learning in that sense could be having a lot of accidents. It will figure it out, but maybe it needs to run over a few thousand people, a few hundred thousand people. I hope not. That’s what I would think. I hope it’s simulation only. But in the end, we only pay attention to a real accident. We play GTA and it doesn’t bother us at all, but then we see someone being killed in the streets and we’re like, oh my gosh, this is a real crime. And it’s literally the same thing where our eyes look the same, especially given the technology we have today. So we’re able to do this focus, but it’s a massive issue and I hope there’s good solutions and maybe we find out that the solutions that we have embedded in our culture, they’re actually not very good. I think this is what a lot’s happening with AI and in our culture currently is that we are redoing our mental model and sometimes we come up with something and say, oh my gosh, this wasn’t actually a good idea. Right. That’s very true. I totally agree with you. And I think that’s why over the next five or 10 years, we’re going to see a lot of autonomous driving on highways because it’s a very safe use case for this application before we start seeing it in cities. One thing that I have to ask is you’re a very talented engineer and you know so much. I’ve been talking so much about the singularity and it really relates to driving as well and the way technology develops. What do you think happens if you say go to Ray Kurzweil’s, now it’s 2038, it used to be 2045, things have changed. What do you think happens with driving and the way we interact with say transport technology? Do you think it’s going to change completely or are we still going to have a vehicle, it’s just going to be autonomous? So, my personal opinion is I don’t think we’re going to see a high market share of autonomous vehicles until about 2060, 2070. And it doesn’t just come from a shot in the dark. When I say that, I mean on all the streets in the United States. It comes from if there’s I think a very talented individual that works for a company called Main Mobility, Main Mobility is a kind of autonomous shuttle working on geofenced areas. And he wrote an article about basically describing Moore’s law but looking at it from a standpoint of autonomous vehicles. And if the capability of the AI doubles every year, he’s saying we’re getting to autonomous vehicles about 2035, 2040 somewhere there. But that I think is still in a geofenced area. So if we look at that and we say that’s a geofenced area, we still need a lot of political and infrastructure influences. And that’s a generation to get to. So that’s why I think 2060 and 2070 is the appropriate standpoint. But to answer your question, your question basically is how do I see transportation changing? I think I see it changing from a more finite or discrete to an infinite type of solution. So the moment I wake up to the moment I sleep, I’m continuously transporting myself from point A to point B. But how I do it is going to be a multimodal solution. So I might get up in the city of San Francisco, I might go downstairs, take my scooter to my parking garage, which is a little further away, or I might take a, my autonomous vehicle might come pick me up at my house, or I might go on a robo bus, a robo taxi. There’s going to be multiple solutions for us to take and interact with everything around us. And infrastructure is going to play a big part in that, because as if I’m driving and 20 miles away, I know today we know there is accidents using Google Maps or Waze. But how do I know there’s potholes or there could be a car that’s just stalled in the middle of the street somewhere? How do I know what the exact issue is? That’s where infrastructure is going to be used. So I think we’re going to change from what we see today to a very seamless transition into how we approach our day, and everything will feel basically natural like it does when we get up and say, hey, Alexa, what time it is? What time is it? And that’s how simple the solution is going to be in the future. That’s a beautiful vision. It kind of reminds me of the 1970 science fiction books with the flying cars. I mean, we know how that ended, right? But no, I mean, there’s a lot to it. I think I’m just making fun of it. I think there’s a lot to that. And it’s just a diversity of transportation options will make life much, much better. The one thing I wanted to add is there is something exciting happening in the field of AI, and the doubling rate is not 18 months anymore. It used to be four months, and now it’s going down to two months. And obviously, it takes advantage, of course, of bigger clusters of better and easier cloud computing and access to it. But David Orbanon, and he had a term for this, I think he called it jolting, what he means by this is not just going up at a certain logarithmic rate. It also speeds up the logarithmic factor over time. And if that holds true, it might be over tomorrow, right? So I’m actually worried, and I’m surprised you’re so conservative that it’s 2060, 2070. I felt like we are in a good trajectory. It’s now finally that it has taken off a little. We will see this in the 40s or maybe earlier than this. But that’s really interesting. I mean, because in the 70s, that would be way past the so called singularity. And then there isn’t a lot of predictions. I’m really curious. I started asking a couple of VCs, what would they say, past the singularity, how your business is going to work? Because the definition is you can’t predict it, right? So what are you investing in 2035? Yeah, so I think you invest in anything that predicts the future. Yeah, that’s the key. You want to invest? 20i, right? Yeah, exactly. So anything that predicts the future, you invest in, whether it’s Industry 4.0, whether it’s something for healthcare that can predict cancer, that can predict any of the diseases that we are facing today, or even vehicle sales or vehicle solutions, anything that could predict the future, that’s what you invest in. And that’s been our solution for how we approach investing as well. You just wrapped it up for me. I mean, thanks for doing this. Thanks for taking the time. Thank you very much. It was very enjoyable. That was awesome. Thank you, Thurston. I hope we get to do this again. Yeah, I really hope so. I think I’d like to have you on our podcast one day if you’re available as well. Yeah, I might not have a lot of knowledge about sensors, but I’m very interested in the world of the world. I know it’s a big topic for you guys, but I can bring in some broader ideas maybe. Yeah, just about mobility. So I think that’s a good idea. I’m going to schedule it with you. Absolutely. I’m looking forward to it. Okay, thanks a lot. I mean, all right. Thank you. Talk soon. See you.