Autonomy at the Edge, with Scott Sanders (CGO of Forterra)

Thanks for having me.

Oh, that's awesome. Where did you get a retro Sun Valley poster?

Um, yeah. I think I had mixed feelings on Sun Valley as a kid. It's kind of a unique place to grow up. Um, you know, things like the Allen and Company conference happen there every year and like it's just like a part of life that like most of the world's wealth shows up and hangs out near town and people there don't react to it. Um, but it's a pretty small place. Um, my grandparents have been going to Sun Valley since right after it opened. Um, and so my parents made the choice to move there in the 90s to, you know, take kind of a sabbatical and, you know, they were going to move back to Seattle and, uh, never moved back. Uh, still in the same place.

But Sun Valley was ultimately the reason that I joined the Marine Corps. Uh, was from watching a ski movie in Sun Valley that Warren Miller did in what would have been 2001 or 2002, because the Marine Corps decided that the type of, you know, person they wanted to join the Marine Corps was the same type of person that would be into like backcountry skiing and freestyle skiing. So they sponsored these Warren Miller ski movies and as a part of that they did a bit where they sent professional skiers to Mountain Warfare Training Center in Bridgeport, California. And I was like, 'Oh, I can like ski around with a gun. Like that sounds pretty cool. I want to do that.' Turns out that's not at all what the Marine Corps is like. That was a total lie. Um, yeah, that's growing up in Sun Valley. You get a weird mix of stuff that comes out of there.

Yeah. Philippines, let's call it 2014 to 2017. Um, the US basically decided that we'd won, right? Just a P had been stood up about the same time as the rest of the war on terror. Um, and had kind of ran its course and like, yep, it's done. Mission accomplished. You got George Bush on the aircraft carrier. It's over. Let's wind it down. Um, and so over a period of time between 2014 leading into 2017, the US presence, especially in the southern Philippines, was kind of down to what you could categorize as individual augmentee teams of like, yep, we're still going to have some people down here, but you guys aren't going to really do anything. And I was coming off the back of several, you know, prior to going to MARSOC, several Afghan rotations and was like, you know what, I'm good. Um, I want to go to First Raider Battalion because they're going to go to the Pacific. There's no one fighting each other in the Pacific. Sounds nice. I hear they're surfing in the Philippines. Uh, let's do that. And so our team gets fragged with the mission about a year before we go. And like the reason people join the Marine Corps and the reason people join SOF, especially in the GWOT, is like they kind of want to go fighting. Um, if you're joining for college benefits in the middle of the war on terror, like, got news for you, man. Like it's going to be a different experience than you signed up for. And, you know, the way we thought about it was like, hey, this is going to be a good reset. This will be a different, interesting deployment. It's not going to be a combat deployment. And, but let's treat this as a way to like kind of upskill the team in terms of how to do intelligence collection, how to do human intelligence, how to do imagery, and we'll make our deployment basically a class in intel operations and like everyone's going to, you know, get upskilled from this and we'll make this, what will be, this is going to be a very boring deployment. This will make it interesting.

What we didn't plan on was discovering that there was a fairly deep-rooted ISIS network that was emerging in the southern Philippines. Um, and it was going to be a really big problem. So, we kind of like stumbled upon this because we just used a real-world example to kind of teach intel collection and because there was no one focused on it, like really focused on it, uh, it kind of came out of nowhere. Um, and so we deployed in January of 2017. We wrote one assessment of the current environment. We said, 'Hey, you've got a SOCPAC. You have a problem. Like there is a radicalization that is happening now that just P has wound down to the Western Mindanao element. Just a bunch of guys in a shed. Um, ISIS is going to take advantage of this. They're going to roll up Jama'atul Islamiyah, the BIFF group, the MILF, the MNLF. Um, and they're going to form with effectively a crime syndicate family, the Maute family. They're going to form ISIS Philippines and it's going to happen in this very narrow time period. It's going to happen in this very specific place.' Um, because intel officers love writing assessments. You can't be wrong. It's like there may or may not be terrorists in this location and then again, maybe not. Right? This was like a very, like, this is gonna happen and we got like laughed out of the room at SOCPAC. They're like, never gonna happen. That doesn't happen here. Um, we said, 'Okay, we're actually pretty confident in this and you're not going to let us go fight with them. So, how do we put together the technology to go win this?' And we tried to do it the kind of normal way. Here's my requirement. Here's my problem. Who can resource this? Again, everyone laughed at us. I'm like, 'Okay, this isn't that hard. I need like find my friend for bombs.' So, like we kind of have a version of this. It's called ATAK. Um, and I need some communications architecture, some communication systems. Um, and I'm going to go beg, borrow and steal and duct tape this thing together. Uh, in partnership with some guys from the SF community, notably Eric Rolls, who had this kind of brainchild for doing this in Iraq. And we built this system which became a program of record down the road called RAVAK, the Remote Assistant Virtual and Complement Kit. It's like how do I take unclassified comms, secure them enough, and have awareness of where my partner forces are so that when I'm back behind the forward line of troops I can see where they are and communicate with them. I can support them with, you know, SIGINT awareness, fires, all these things, all this JADC2 stuff people are trying to do in the US DoD. It's a lot easier when there's basically no rules. Um, and you're a foreign military force that wants to get something done. Um, so we ended up building that, deploying it, um, and then fighting in about a 165-day war in the southern Philippines that most people, you know, didn't know about. Um, which is kind of the point of special operations. Like you probably shouldn't know about them. Um, yeah. But that was the foundational, like if you really want to get something in that acquisition time and space, you kind of had to go do it yourself.

Uh, yeah. So to be very clear, when I quit in the middle of that deployment, uh, after getting web orders that we're going to have, for those of you who don't know what web orders are, it's like how the Marine Corps specifically like tells you where you're going next. Um, it's kind of a, I would categorize it as a relatively inhumane process sometimes or like, here you go. Um, you want to do a bunch of other stuff. Got news for you, buddy. You're going to camp in North Carolina. Um, and you just show up in 30 days. Uh, I had gotten web orders on that deployment and was like, 'Hey, I like, I don't want to go to North Carolina. Like, I'll go to North Carolina for a couple reasons. None of them are the ones that you've given me or I'm happy to stay on the West Coast and do one of, you know, five different jobs.' And basically got a response from the Marine Corps that was like, 'We don't negotiate with terrorism. Um, you can take these or leave it.' And I was like, 'Sick. I'm going to leave it. I'm going to get out.' Yeah. Uh, and actually never wanted to work in defense again. Like that was my hard and fast rule. Uh, like I had friends who were going to Palantir. I was one of the original, one of the first Palantir users, like power users back in like 2009 when it first rolled out with the Marine Corps. But I was like, no, I don't, I don't want to work in defense. Like this system is completely backwards. It's never going to get fixed. I really like working in tech and I can't code well enough or I'm going to get paid for that. So, I can go be someone who can do operations or the sales thing sounds really gross. I don't want to do that. But like helping optimize some system for delivery here sounds cool. Uh, so every job that I was looking at was actually in the Bay Area in operations mostly on tech. And then just happened to run into a friend of a friend who was friends with Trey and he was like, 'Hey, based on like the story you told me, you might want to check out this company. It's like seven guys in a, you know, warehouse next to John Wayne Airport. They're doing something with defense. It's like super unclear what that is, but they're 20 minutes from where you live. Like you should just go meet them. I know you don't want to do defense, but you should go check it out.' Um, and went and checked it out and was like, all right, this is pretty cool. This is like a pretty good idea, and also I don't have to move. Like, it's down the street. Um, and it was legitimately a, I mean, it was a really cool vibe back then. It was a warehouse next to a baggage handling facility. It was a bunch of people just trying to build something that would, you know, you could deploy on a reasonable timeline, solve an actual end-user problem, not go through the same sort of requirements build that you would on a typical program. Um, but I wish I had the, you know, the alternative thesis here is like, oh, I learned how to like build a program from the ground up from the inside. Like, I'm going to go take that in the defense tech world. It's like those are actually not correlated events like whatsoever. Um, it became very obvious.

Yeah, it became like a very obvious thing like six months down the road. I'm like, wait, we're trying to build programs of record. Like I started as like an ops person. Like I log into Gusto on the first day of Anduril and the options on the dropdown for employment are like founder, co-founder, engineer. And I'm like raised my hand in the corner like I'm not any of these things. Like what do you, is there like another category? Like I don't know. Um, uh, and so it took me kind of a minute like, oh wait, this playbook we ran on RAVAK to get the thing that we wanted as operators, it's actually not wildly different than what Anduril needs to do. But at the time defense tech was not a thing. There was no one doing this. Palantir was like in the maybe we're a commercial company, maybe we're a defense company, like kind of unclear time.

Kind of wild. Yeah. This is back when there's like no defense, like defense tech wasn't a thing. There was no conferences like... four years at Anduril and never went to a defense tech conference. It's great.

It might be different today, but it was definitely true when I said that.

There's a lot of things that like have changed for the better and there's things that have I think actually become harder because there's so much more. I mean, I remember as we were like signing our first defense contracts at Anduril, uh, it was really exciting because like we were the underdog. There's no one doing this at least in this kind of playbook. There's always been small innovative tech companies in the defense space, um, but not kind of running this playbook. And, you know, America loves an underdog story, right? We just got to relive one, uh, with an Olympic hockey team, right? Like what if we don't pick the best players and see if we win, um, just to get a little edge. Um, but what I quickly learned is like there is no such thing as like you're selling to the Department of War, the Department of Defense. Like the department isn't a single monolithic entity. It is a myriad of fiefdoms that have their own procurement strategies and those strategies change as people move around and as the world changes. And so I think there's always this vision quest of like what is the playbook to go work in the department. Uh, the reality is there's not a really clear set of rules or playbook. It's like a loose set of guidelines of like just generally how you should try to interact with your customer. Uh, I think the biggest thing that I've learned is like it's just really hard to change an org. Like not that we have any influence over changing the department, but like obviously the Congress has tried to change it. Secretaries of defense have come in and tried to change it. We've been talking about defense acquisition reform since Rumsfeld was the SECDEF, right? Like this isn't new, but it takes a long time to move a very big organization. Some parts will do it faster than others, but at the end of the day it's about the people. It's not shocking information, but like each human that runs an acquisition strategy is like a very different person, so the way they want to do it is very different. Um, and trying to build a team and a culture and a company that can react to that is hard because it's not like you're just doing enterprise SaaS, like this is the playbook we run to capture.

No. One, I didn't know what venture capital was, which should be embarrassing as someone who majored in that field of study and came from a family that worked in finance. I was like, I don't know what venture capital is. Um, so no, never really crossed my mind. Um, I also don't think I'd be very good at it because I'd probably say no to everything. Um, yeah.

Yeah, I think there are two parts of this. One is venture is a hammer, like it is a specific tool that does a specific thing. I think far too often we try to, we the society tries to use venture as the right capital tool in a lot of situations where it's not the right capital tool. I, this might be slightly controversial, but I don't think that if you can't make a multi-product and a platform out of your sort of initial offering, like if it doesn't kind of neatly roll into that, the ability for it to have a venture outcome, I think, is basically zero. And there's a lot of companies that raise capital on like a really good product idea that's like, you know, they got the timing perfect. I would say a lot of things in the counter-unmanned space portfolio fit into this. Maybe some less on the unmanned space, more on that specific, you know, I need to shoot a thing out of the sky. I need to make a missile to shoot something out of the sky. And it's like, I've got the demand right now. I have the backlog right now. I built this great product. I'm going to raise venture capital to scale this thing. Well, if you start raising at venture multiples, your ability to get acquired in the defense space for something that looks like a return to your investors and to your employees isn't great. Um, and it's just probably not going to work is my fear. Um, like the department needs like two or three Andurils. You know, an argument that says like they can't afford that. And you'll hear an argument that like, well, we actually need a bunch of these like small innovative companies that like feed into an ecosystem. It's like, well, the capital profile for that doesn't really close. And so, when you're looking from an investment standpoint of like, is this an investable company? It's not just like how big is the TAM for loitering munitions. It's like, well, what's product two? What's product three? Can they address a different market within a department? And how fast do you think you can do that? Because you need to, in order to survive in this space, you need to have diversity of product very, very quickly, um, because there'll be ups and downs.

Yeah. A little bit more of the backstory that will help frame this is that going from Anduril to Vannevar, like I walked out of Anduril on like a Friday and walked into Vannevar on a Monday. Um, for those who are looking for career advice, I highly suggest you take time off between companies, especially if the early-stage startups. But I was like, I'm good. Like we're going to do this. And, uh, when I left Vannevar I was like, okay, this time I'm going to take some time off. I'll consult for some companies and like, you know, help out along the way and I'm going to do some more skiing and like enjoy life a little bit. Uh, and a couple of friends of mine and I, uh, were like, 'Hey, we should go start a company.' But this is still kind of before defense tech was defense tech. Like there weren't a lot of companies doing this, but it was like starting to build. But peak pessimist in me was like, we've peaked. There's already too many defense tech companies. It's not going to work. It's like 2021. Um, and we had thought about, hey, you know, got some really talented engineers. I think we can figure out a go-to-market strategy. Let's go build something we think everyone needs that can be a big scalable platform. Uh, ultimately decided not to do it. And when I was consulting, one of my first clients was, uh, what is now Forterra. And most of the companies I consulted for, with the exception of maybe one or two, Forterra being one of them, weren't very what I would call, uh, engaged, right? They're like, 'Hey, I want advice, but I'm not actually going to go do any of the things that you're suggesting.' And so, it was like this Groundhog Day of like, 'What are we, why am I doing this? This is like a waste of time.' But what stuck out to me about Forterra was like I would talk to Josh Rajo, who was the CFO at the time, and like, 'Hey, we're going to fix your pricing strategy. Give me all these docs so I can go figure out what does it cost to build this thing? How do we go to market with it?' And I would ask him that at like six o'clock at night, East Coast time, and I'd have an answer at like 10:00, like every day. I was like, 'Oh, you guys get it.' Like, 'You need, you know, you need to move fast to do this. It's really exciting.' And then I started spending more time with the company and I realized if I wanted to go start something, like this company is what I would start. Like at the time, maritime was just getting going. Air was already beginning to get saturated. The reality is we love land army so much in America. We effectively have three of them and no one is focusing on this edge problem of like, yeah, counter-UAS is super cool, but like everything that was built prior to 2021 and even today is like mostly around fixed sites. How do I get it to combat? How do I get data at the edge? How do I do the comms infrastructure to support these network and distributed systems? Like it's going to be on vehicles. And if we're using most of our manpower in the military to drive a truck, that's probably the wrong answer. And if you think about what a self-driving car is, it's a bunch of algorithms and a bunch of data and a very hard data problem. It's safety-certified edge compute. It's incredibly complex networking that has to work across a variety of different modalities and enclaves. And it's all the payload capabilities that vehicle is actually doing. And so individually, each one of those things are things that can be standalone venture-backable businesses, at least by today's standards. So if you can nail the self-driving vehicle piece, which is very hard to do from a can-I-get-a-scalable-product-into-a-user's-hands perspective, you create this very hard mode from a hardware customer acquisition perspective and a technology perspective that you just can't do with a SaaS product, especially now. And it immediately becomes not just your first product, it becomes your first platform and it becomes products two, three, and four because you already had to do them. Now you just take your networking system off and you go do it everywhere else. I can take my edge compute, I can sell it individually, I'm going to make AWS at the edge with basically very minimal, um, NRE to get there. And so to me that was the exciting piece of like this company had been around for 20 years. They'd already learned all the lessons of how to not do some of these things and could rapidly get mature hardware and software into the field and had the culture that felt more like the first six months at Anduril than a lot of the other companies that had come across. And so I was like, all right, this is the thing to go do. Let's go scale this.

So most people know us for AutoDrive and the self-driving side of the house. Um, I like to remind people that my first two years at Anduril I'd walk around the Pentagon and people would be like, 'Oh, you're working for that border security company.' You're like, 'Yeah, and Amazon sells books.' Um, like, got to start somewhere, guys. Um, so that's a majority of what people know us for, but we also are bringing out new capabilities based on what's already on that platform. So, we're bringing our next-generation communication system out in the market. We're working on being able to bring our edge compute products out into the space. We also have an EW business that we stood up in a November-December timeframe that's coming to market now. Um, we also acquired GoTenna, uh, to bring, uh, burst mesh radios to the edge because we think the future of unmanned systems are all autonomous. They all work at the edge and you don't need, you know, there's a time and a place to have a very capable Silvus Technologies radio. They make incredible products. You don't always need that. I need something that's dramatically lower cost and doing a different mission. So, that's kind of the family of mission systems that we build today and we'll probably bring another two or three out by the end of the year. See how it goes.

Yeah. I mean, it's one of those mildly polarizing things where there are some places that are just like, 'Yeah, we don't like the...' We've had a customer straight up tell us, 'We want all the things in the subscription model and we're happy to pay for it. You have to call it something besides subscription.' Like, cool, you know, rename this. Um, and we actually, we sell the product a couple different ways. Um, some customers love that model. They're like, 'I don't want to deal with a, this hardware component breaks and I gotta, you know, buy one and like I have to wait for it to come through inventory. Like I want you to feel the pain of that system being down.' Um, and there's...

A time and a place for as-a-service models in the department. We've tried to make a pricing model that is not one-size-fits-all. Like there are many ways you can buy capabilities from us. At the end of the day, it's really hard to sell software that has to continuously change and continuously be updated without having some sort of out-year price for it. But there's ways to make those trades with your customer, like, hey, you're going to get this many years of updates for this firm fixed price and after that it's still going to work for you. You're just not going to get any updates. Like, are you okay with that? Like, yeah, I want that. Okay.

Yeah. But it's weird because like you'll talk to a contracting officer who's like, "We're never buying something as a service." And like the next minute there will be a war notification for someone doing something that like I never conceived you could do as a service, like a physical radio. Like the Army just bought like $20 billion of radios as a service. You're like, what? [laughter]

Yeah. You really have to tailor how you are selling it. The reason I, as a taxpayer — I've worked at defense tech companies that primarily sell things as a service — the reason I, as a taxpayer and a former Marine, like an as-a-service concept is that I want the vendor to have as much skin in the game as a thing working as I do. And if it costs them money every time it breaks and they have to fly out somewhere and fix it or send me a new thing, I know that like businesses need to be efficient and they're going to make it so they don't have to do that, right? And it changes the cost incentive from a cost-plus fixing model to say for a certain defined firm fixed price, I'm going to deliver a capability to you. And if I don't, then I'm getting penalized. And if I do really well and it never breaks and I never have to have an FSR on site, then I should be allowed to have more profit. Seems like a pretty straightforward economic solution. Not everyone's fully going to get that. So you have to have a couple different ways to meet the demand from your customer.

Yeah. And we're happy just to sell, you know, the autonomy hardware and you can buy a single license for a fixed number of sort of years of updates. No different than you buy enterprise operating systems these days. Or you can buy it all as one, or you can buy the software as like an enterprise and say, "Hey, I've got like, you know, 400 platforms. I need this much capability. What does that look like from an enterprise configuration and management standpoint?" It's really flexible. We based all this on how we sell this product commercially, which I think is a pretty big difference for most defense companies. Like most defense tech companies don't actually sell anything commercially. They sell it as a commercial product to the DoD, which is a different definition of what commercial means. But we base it on like this is what we think closes from a one, cost efficacy calculation of what does this buy you; two, what does the commercial market bear and how do we sell it there. But then reserving enough flexibility to say, all right, well, if you want to acquire it this way, we can do it that way. So, happy to sell hardware and you have hardware kind of margins and license software in innovative ways so that the customer is getting their, you know, the thing that they need, and package it in ways that are easy to be broken up. But the issue you run into is like now your pricing model is a lot more complex than, hey, I need an instance of Salesforce and like I just get the price that I charge commercial people when you do it by a seat.

So you run the risk of making it too complicated, too fragmented, and then you spend all your time explaining pricing.

Mostly what we would refer to as inside-the-fence outdoor autonomy. It's probably not the best description of it, but it's things like ports, yards, distribution centers where you're not on road, you're in a confined space, a controlled environment, and you're operating in very similar problem sets where there isn't defined rules of the road of how you're going to navigate getting things from point A to point B. And you're also going to run into kind of the same constraints and restraints you see in the department where it's dusty, it's rainy, it's got to work in all weather. You have to deal with a dynamically changing environment. And so we look at that as like here in the yard space it's very similar to defense and that makes it a relatively straightforward commercial outlet of technology versus being, you know, the dream dual-use company of like half your revenues come from commercial and defense. It's more of a we're defense-first, you're happy to sell into commercial where that makes sense, and look at that as a way to, you know, bring down the cost of the system to the department so they're not burying all the risk on their own.