In this installment of the AEye Insights series, AEye Founder and VP of Strategy & Partnerships, Jordan Greene, interviews Reilly Brennan, General Partner at Trucks Venture Capital and author of the popular newsletter, Future of Transportation (FoT). In this discussion, Reilly and Jordan pull out their crystal ball to talk about COVID’s lasting effects on logistics and delivery, the challenges in store for “structured” autonomy, what’s behind the flurry of AV and sensor SPACs, and why AV engineers are like chefs.
JG: Welcome to AEye Insights, where we talk industry trends with proven business leaders. Our guest today is Reilly Brennan, General Partner at Trucks Venture Capital and author of the popular newsletter, Future of Transportation. Reilly, welcome and thanks for joining us.
RB: Thanks, Jordan, it’s great to talk to you. Usually our conversations are texts in the middle of the night about something big that’s going on, so to put it on Zoom here I feel is a little funny, but I’m happy to be here with you.
JG: This is true. It’s usually very candid off-the-cuff discussions, so I’m excited to riff about where the transportation market’s headed and I think that you’ll have some really interesting insights given the very holistic approach to the market that you take in the newsletter that you put out. I think the obvious way to start, and one of the most common subjects that’s discussed, is what is the impact of COVID on all of these different areas of the market or segments of the market? How do you think it’s going to play out with autonomous, ADAS and all the supporting functions in the automotive value chain?
RB: Sure, well, first of all, thank you to the team for having me on. Before I get into your question, AEye was one of the first investments we made at Trucks Venture Capital back in 2016, right when our fund was getting off the ground, so it’s been great to be a partner of AEye for now over four years. My background coming into this discussion is: we run a small venture capital fund focused on transportation that’s very early stage. So, usually we’re giving the first check to a company as they’re getting off the ground. And it’s interesting because COVID has accelerated a lot of things that people think about, like software and things related to delivery, while also probably tamping down on some of the expectations for things that shared an asset like shared scooters and things like that, which had, in the beginning, a more difficult time. I think the big change with COVID, though, is all the behavioral changes that are probably going to be closer to permanent. There’s that old saying that it takes 28 days to form a habit. We’re well past the 28-day point with COVID, so all those things that people have been doing around delivery, particularly, I’m really fascinated about how those stick after a vaccine. I think if you look at the way that most of those services work — for example, we have a Shipt subscription, which you buy for 12 months, or Instacart would be the same way — so even if the vaccine gets delivered and everybody is “cured tomorrow”, there’s going to be a lot of people who have already paid in advance for a lot of the stuff. And I think a lot of those behavioral changes will remain, and that’s going to be interesting to watch over the next few months.
JG: It’s on the consumer buying side of things that you see those behavioral changes will have impacted people, and I’m curious about what that means from the technology side when you start looking up the value chain from there: what are the implications that has on what people start to invest in to make those more streamlined logistics and everything else?
RB: Well, I guess one of the maybe big opportunities is, are we going to continue to operate in this universe where Amazon becomes the primary mechanism in North America to get you the goods that you need? A lot of people now think Amazon first and Amazon, of course, is building out a really interesting logistics network, potentially, with their Zoox acquisition, and potentially more with commercial vehicles down the road. Maybe the bigger question is, are there other ecosystems out there that are going to drive as much enthusiasm, and where you think of them first? And in that regard, I think about Shopify and other platforms that are out there in the ether that are knitting together a lot of parts of e-commerce. And ultimately, although it sounds kind of insane right now, would Shopify eventually have a logistics network? Would Shopify have a need for vehicles for trucking, for commercial delivery? I think you could probably imagine that’s not too far away for a Walmart in the United States or potentially a Loblaw in Canada or Costco or any of these other big retail entities to support systems like that. And so, I think about first the software that knits together the commerce and ultimately getting that to your door. Those are some of the things I’m thinking about to your question.
JG: And there is, I would assume, multiple levels of vertical integration within that logistics model, because I’ve seen everything from the companies that do just the logistics side of things and are trying to set up the ways to track items from the ports through to the consumer to the actual first-mile last-mile delivery vehicles and everything in between. I’m very curious what you see as the greatest opportunity there, because there are people like Amazon, who arguably are going to want to control everything end-to-end, and they started from the touch point of the consumer. I’m curious how that will impact companies and how they want to fit into the chain. What are your thoughts on that? And have you seen anything really cool?
RB: Yeah, well, there’s a couple of macro trends to think about. One is, so much of commercial logistics is moving to short haul: things under 500 miles. So, whereas historically, you might move a bunch of goods from a big assembly facility or distribution center thousands of miles to get it to a local distribution center and ultimately put it in a retail store, now there are so many more moves around a retail location. A company might have multiple distribution centers, and that just as the background of this discussion is really fascinating from a real estate perspective. But it also means all the really different points along the journey that different entities or different vehicles might move a good. Instead of one long, sort of like linehaul trip to a distribution center (DC), you break that up into many other different trips, so the value of real estate, and the people who are holding these goods is really fascinating. There’s a company I’m sure you’ve heard of called Prologis. I’m always kind of keen on following what Prologis is doing because many times they’re Amazon’s landlord. They’re the people who will actually go and break ground, put in a warehouse, and then invite customers like Amazon, etc., in to use their warehouse, and they’ve been putting those facilities in some really interesting places. Then think about all the systems and need to service the moves between the distribution centers. We have a portfolio company called Gatik, which does exactly that. They move middle mile trips between DCs. Then we did an investment in a Canadian company called Swift. They basically will allow a retailer to do same day delivery by having a courier go and pick up things directly from a warehouse. There’s a ton of opportunity here, because if you look at the movement of how commerce is changing, coupled with consumer preferences and all these new trends around ecommerce and delivery, there’s so much movement under the hood of getting a package to you that’s actually way more interesting than Robotaxi. I know that when people got into autonomous vehicles four or five years ago, the dream was always around people. But I think for many years, many people considered the commercial delivery part of AVS, I always considered it like people thought of it as the Sancho Panza of autonomous vehicle investment. Then in 2020, everyone was like, oh, wait, that’s actually going to be more important. So, you’ve seen a rush of things into delivery, commercial goods, and that’s not going to change.
JG: I’m curious to hear your thoughts. First, I’ll start with Gatik, which has a big partnership that you can talk to that I believe is with Walmart, and they are doing some of the logistics side of that, but I know that a big part of their effort also starts to dovetail into the automation of the vehicles themselves as well, and how to create these logistics systems that are automated. Our prediction at AEye was there was going to be an emergence of all of these distilled down or more constrained versions of autonomy that are more limited in function, rather than automated vehicles that we were promised and people are still somewhat pursuing that go everywhere all the time. But there’s derivatives of it. There’s long-haul trucking, there’s middle-mile, there’s first and last mile. There’re all these different ways. And usually it’s about a technology reduction in scope. It’s like if you’re automating a truck from the hub, it’s a lot easier of a problem than trying to automate something that goes anywhere all the time. And similarly, these applications start to work themselves out because there’s business opportunities that exist with considerable demand and with a much more down scope technology problem. What are some of the applications that you’re seeing in that regard that have piqued your interest? What are some of the trends you’ve seen in those areas? Gatik and Swift are obviously examples of that, but I’m sure you have similar predictions in the space.
RB: We came up with this term, I think it was in 2016, of “structured autonomy” and the thought was, look at all of these huge vertical markets. Everybody’s focused on Robotaxi, but what about ag and mining and construction and trucking — and actually trucking is like a Taco Bell seven-layer burrito. There’re all these different parts of it you could pull apart and you could just have one company with amazing leadership and incredible value in only one part of that. So trucking is almost too broad a term. But we would like to have the number one best company in all those different structured autonomy environments. We have made a handful of investments across these. I don’t actually think that we have ever publicly disclosed this, but we invested in a company called Teleo earlier this year, which does automation in construction, and they use a lot of teleoperation. That’s actually where a lot of the core IP is, but it’s focused on the structured environment of construction. We’ve done in ag, Bear Flag Robotics, which is on farms, and they’re doing some specific tasks there. I think one of the fascinating parts about structured autonomy is thinking about, is the vehicle movement the hardest part of this, or is it a bunch of other tasks that are integrated into the application? What I mean by that is Bear Flag, which automates farm stuff: moving a vehicle on a farm is actually really simple because there’s no pedestrians, there’s no cross traffic, there’s no traffic lights, there’s no rules of the road per say because you’re on private land. Getting the tractor to run autonomously actually isn’t that difficult. But what is difficult is knowing where and how to apply that vehicle within an environment. Do you want to move something behind the tractor? Do you want to pull a tool or some sort of application? Do you want to spray? Do you want to mow? Do you want to do something else? That’s actually where the value is. So, when you look at these applications of structural autonomy, it’s always like the autonomy part, plus this other thing, and together that actually creates the value. So, in trucking or in delivery, for example, it’s not just good enough to move from point A to B, it’s also integrating with the package itself or letting somebody with the right authorized control open up the back of the vehicle. It’s all those things, and I think that’s actually the harder part. It’s not just about the vehicle movement piece, though that’s also quite difficult.
JG: Even, not just at the opening of the truck, but also the actual unloading and offloading of the containers, all the different components that go into that, I could see that being a big part of that. But obviously, and I have a biased interest here in moving the discussion to this, but there’s unique sensors that you’re going to need for every different one of those applications, because for automation, there’s different types of sensors and there’s different types of sensors regarding not just the automation of the vehicle, but also all those other very key discrete components of how to access the containers or the equipment or whatever and actually create actionable information and to then carry out that action. I’m curious to hear your thoughts on how that is evolving and what you see in the space, because you clearly follow it very closely, and I’m sure you’ve seen a lot of the recent announcements in the market.
RB: Absolutely. Engineers are putting together vehicles for these tasks, I think of them like chefs because they’re looking at a particular task and they’re trying to figure out the right ingredients to achieve the goals of the system. So, we’re making cookies here and you could use different recipes, you can approach a cookie from a different way. You could have a little bit more flour here, a little more baking powder there, a little bit more sugar, and depending on what you really are trying to do, the mix of that is quite different. And the same is really true of sensors. There’s no dogma that says you must deploy this type of sensor if you’re going to have autonomous vehicles or even ADAS vehicles. The better approach to that is to have a better matched sensor to an application. You might not need the most powerful AEye sensor capable if you’re only doing ag applications, but on the other hand, if you’re doing long haul trucking, you might want the full bore, the best that Luis and team can give us, AEye sensor, because that’s what we’re doing. And that’s actually the really interesting part, I believe, of this next era of automation is you will have customers — and ultimately, in time, the stock market value some of these sensors and some of these things, depending on what the most valuable applications are. On-road applications at high speed, you clearly want to have incredible range at low light reflectivity for which AEye builds a great sensor. For these other things, you’ll have this other set. I’ve always been fascinated and I think I’ve learned a lot of this from you guys, the incredible dogma around sensor types. And it’s actually kind of a folly to argue over what the right sensor is. It’s better to know what the application is first and then work backward to the sensor.
JG: I remember very clearly, when I was in my old venture capital role, I was looking at the value chain and at a number of different applications for UAVs, back when there was a whole craze about UAVs. And the interesting thing was when you looked at the value chain, there was a similar concept. It was trying to create actionable data for these applications, whether it was a topographic map that you were trying to create for agriculture or whether it was a pipeline or wire or some type of power line monitoring so that you’re making sure there’s no degradation of those wires or pipes. Accordingly, you’re trying to create actual information. It became ironic because there was never an evangelist argument about which sensor was the best sensor it was, which was the right sensor. But the right sensor comes with a number of different associated variables that have to do with how it can be configured and the ways in which you use it. And so, there’s the implications of what you’re going to use a camera for, multispectral camera or thermal camera, or what you want to use a radar for. Ultimately, I like to say that at AEye at least, we’re not evangelists about the sensor type. If you say you wanted to remove radar ultrasonics or camera from the car, I’d say you’re removing one of your senses as a human. Why would you take away your sound, you’re hearing, or your eyesight when you’re trying to drive? It’s a silly concept to try and promote one or the other. But then there’s another aspect of it, which is that sometimes people choose the wrong sensor and try to out-hardware with software. It’s finding the right fit, you’re 100 percent right.
What is interesting to you about the market emergence, and I’m sure you’ve seen all of the SPAC announcements, especially in the sensor space? What do you think about that market and what is there to expect? Because there’s a lot of people that are preaching their sensor or their discrete widget is the right one that’ll solve all problems. What do you think there?
RB: Well, it’s absolutely springtime for AV and sensor public companies. Obviously, a lot of this has been pulled in from the EV side because of Tesla. And I was actually thinking earlier in the summer that the Tesla impact on the public markets would have a negative impact on anybody doing something in LiDAR who wanted to go public because all along we’ve heard that Tesla’s fearless leader doesn’t like LiDAR. I thought, if all these other stocks are really impacted by what Tesla does, what will happen when the LiDAR companies come out? In fact, the opposite has happened. The market is making a couple of different big storyline assumptions about where the vehicle business is headed, and in the macro, they’re true. One is that electric vehicles are a dominant force, both because of consumer acceptance — we’ve seen a quarter more people want to shop for an EV now as opposed to 24 months ago — and also policy wise. You have governments of significance who now say by 2030 or 2035, the only cars that we want to see sold are EVs or EV hybrids. So, there’s this huge thing with EVs — that’s a macro story that’s absolutely true. The individual stocks and the individual companies are going to bring us there. I don’t trade public equities as my day job, so thankfully, I don’t have to have an answer to that.
On the sensor side. I also think the markets are accurate in saying so much of the value of these vehicles in the future is going to be on their intelligence. And so, whether that’s on the supercomputer side, like in NVIDIA or the perception side, those are accurate claims. These are the same arguments that you would have found in venture capitalist offices making early stage investments in 2015 or 2016 that now, six to seven years later, the public markets are starting to trade on, and so I think those are right on. Individual companies in this space, it’s going to be really interesting how their contracts play out, like how will we see the contract value of these companies as new vehicle production platforms get announced, and that’s ultimately how they will be judged. You don’t walk in and buy Innoviz at the auto parts store today. The way to judge these public companies, the ones that are coming out, is on the content value in the new platforms, and so I’m really excited about that.
JG: Absolutely, and I will make a bold prediction that it’s going to be the value of your ability to fit into the value chain and leverage that, because to your point, you’re not going to buy an Innoviz or an AEye. You’re going to buy a Continental product within an overall larger OEM bundle. I agree with you and think that there will be a need for a lot of different sensors and we will see how they play out. But the flexibility and the ability to fit into the value chain are going to be key in growth and expansion to the broadest possible market, similar to what Mobileye demonstrated.
On that note, unfortunately, we have to end our conversation since we are out of time, but it is always great to talk with you. As you said earlier, you were an early investor, and we’ve always been very proud of our partnership with Trucks. We’re very excited and thankful to have you joining us. And we hopefully can do this again next year to see how our predictions and our discussions stack up.
RB: Thanks. Jordan.