Pod Insights a16z Podcast Autonomy is no longer a moonshot—it’s a market About this episode Autonomy is no longer a moonshot—it’s a market. And increasingly, it’s a battlefield.
Recorded live at the 2025 American Dynamism Summit, this discussion features a16z General Partner Erin Price-Wright sitting down with Qasar Younis (cofounder & CEO, Applied Intuition) and Jason Brown (Head of Government Business, Applied Intuition) to explore how vehicle intelligence and autonomy are reshaping defense, industry, and the future of transportation. They cover:
- Why the next decade is about deploying autonomy, not debating it
- The convergence of commercial and military autonomy stacks
- How to retrofit 100-year-old defense platforms with cutting-edge AI
- Why dual-use companies can—and must—set the benchmark
- What it takes to win the software-defined arms race
From self-driving ride-hail in San Francisco to autonomous naval vessels, the future isn’t just autonomous—it’s already in production.
Follow everyone:
- Erin Price-Wright - https://x.com/espricewright
- Qasar Younis - https://x.com/qasar
- Jason Brown - https://www.linkedin.com/in/jasonmbro/
Timestamps:
00:00: Introduction
00:40: Founder Intros
01:18: The History of Applied Intuition
03:30: Productionization and Proliferation of Autonomy
07:55: Will The Future of Autonomous Vehicles Look Like Waymo or Tesla?
10:58: Autonomy In Defense
12:30: Supporting Retrofits and Legacy Systems
13:26: Driving Force Behind Applied Intuition
15:39: Beginnings of DOD Engagement and Commercial Benchmarks
20:49: Features That Benefit Commercial and Defense Customers
21:45: Navigating Technical and Product Requirements
24:08: Hiring Practices (Commercial vs Government.)
26:10: Improvements In How The DOD Procures Technology
30:34: Partnerships vs Direct Relationships
33:18: The ChatGPT Moment For Autonomy
36:06: How Close Are We To Autonomy Being Ubiquitous?
38:49: The Future of Autonomy, Mission-Driven Engineers
Transcript 0:00 Speaker 0 A mobile operating system versus an operating system that goes on a vehicle that's safety critical are different, let alone the autonomy application that's in the top of that operating system. And I think that's where a dual-use company like us can be very valuable because we are benchmarking on the commercial side. And I think if you're only serving sometimes a DOD, you sometimes don't get that commercial benchmark. We always get like baseline to the truth. If our products are not good, we're not going to survive as a company. Full stop. We got to win in that very competitive global commercial ecosystem where we get no favoritism being an American company or not.
0:39 Speaker 0 So my name is Casser Yunus. I'm the co-founder and CEO of Applied Intuition. Applied Intuition is a vehicle intelligence company. That means we build we make software and AI that ultimately makes it onto vehicles. So that's tools, that's an operating system, but it's autonomy. Yeah, we started it eight, nine years ago, and uh and we started our DC presence uh probably five, six years ago at this point. So it's been it's been a while as well. Based in Mountain View, California, but office offices all around the country and the globe, about a thousand people, mostly engineering. I'm a very engineering heavy company.
1:18 Speaker 2 Amazing. Yeah. And tell us about your role.
1:20 Speaker 1 Yeah, Jason Brown, I lead the government business for applied. So based here in DC in our DC office, we have about a hundred and ten folks purely dedicated to the government mission, but really the entire company works on government uh government work. Um and we are an all-domain autonomy company at this point. So we are working with every service in in the Department of Defense. We are actually expanding that government business internationally to uh U.S. allies and um have you know autonomy projects happening in all domains. So air maritime and land.
1:50 Speaker 2 Let's dig in a little bit more to your background. You emigrated from Pakistan, you sold a startup to Google, you know, you ran YC, all very impressive. What brought you to autonomy? Like how did how did you end up here?
2:02 Speaker 0 Uh well, so when I was at Google, that's when the self-driving project was starting. This is in the early teens, uh, 2011 is when Google acquired the company. And uh I was originally not a lot of engineers. The thing that you missed was I I went to the General Motors Institute undergrad and uh I worked at General Motors, worked in Bosch big big Tier One. And even when I was a Bosch in 2004, 2005, the early what they call you know, eight-ass uh driver assist systems were emerging. And it's really Google that really said, hey, we can do a truly autonomous system, and that kind of sparked the the you know interest in the space again. And then Peter Ludwig and I, my co-founder, uh, we connected again in 2015. He was one of the founding engineers on Android Automotive, which is taking Android from the phone
2:47 Speaker 2 uh
2:47 Speaker 0 to the car.
2:48 Speaker 2 The car operating system.
2:49 Speaker 0 Yeah, exactly. And uh so we were just talking about this intersection of software and vehicles and um and the one of the big disruptions within kind of that broader vehicle intelligence umbrella is autonomy and the car or whatever vehicle driving itself. And I think what we learned after we got into the business cruise started the company was actually the dynamics that exist in automotive are the same that exists in defense, they exist in construction mining, they exist in trucking, and so and then they not only exist on on in kind of on the land, they also exist in the air and in the sea, and so the company's just naturally expanded, uh expanded from there.
3:29 Speaker 2 So you've seen a really long arc of autonomy in vehicles from you know something that uh a generation where we almost probably wouldn't recognize it to today.
3:40 Speaker 1 Yeah.
3:41 Speaker 2 And I think we've been talking about driverless cars, et cetera, for you know, more than a decade. Yeah. Suddenly starting to feel real. Like maybe t tell us a little bit about that. Yes, it's a it's
3:52 Speaker 0 Waymo is like literally on the block, you can time it as a.
3:55 Speaker 2 I only take Waymo's now. It's incredible.
3:57 Speaker 0 Which is like a huge fact, which like, you know, so the past 10 years very much uh autonomy was like this almost like sci-fi, will it load it? It's speculative, almost like crypto. Is it
4:09 Speaker 2 people get excited and then they get let down and then they get incited and then let down.
4:13 Speaker 0 I think we're at that threshold where you know the next decade is very much the productionization and proliferation of autonomy and you can say broadly speaking, the intelligent vehicle. And uh it is also the monetization of this technology. And so we, you know, it's literally in the name. We're on the applied side of AI, and uh which is we take, you know, the stuff that's been kind of uh conclusively built and research and tested, and now we take it to the masses. And uh so the state of autonomy today is there's two gross simplification, but uh there's two prevailing approaches. You have the Waymo approach, which is the one that you which is like there's no driver in the seat. That's kind of the key hallmark. There's nobody sitting in the driver's seat. And then you have the Tesla approach, which is there is a driver in the seat.
4:59 Speaker 0 Um and they're just on a spectrum of the level of autonomy and the level of engagement that you need from the the driver. Obviously, there's no driver in the seat. The car's got to be able to do everything. So uh what Waymo has proven and awesome other companies around the globe have proven that hey, you can actually do that. And now it's just a matter of how do you scale that in a business model that makes sense. Yeah. On the Tesla approach and all the other car companies again using gross generalizations. They've proven a business model. People will pay for Tesla FSD. And I pay for it. Yeah. And they're and people pay for it and they're continuing to prove that product through uh uh iterative and process. But it can do everything at all times. It it still needs that driver of the loop. And so the next kind of 10 years is and I think what you'll ultimately see is like somewhat of a technical convergence because the Tesla system is a very sensor light system, it's basically just a camera system.
5:54 Speaker 0 Uh and the Waymo system is very sensor heavy, so it's very expensive, including the compute that sits inside of Waymo is very expensive. Um so it just makes it very difficult for a consumer to buy that vehicle. But as Waymo, you know, basically makes their technology more generally available and makes it cheaper. Uh I think you'll see it start look more like Tesla technology, and Tesla makes it more reliable and generally more available in you know different domains and without the driver's seat, it starts looking more like a Waymo. So there's a there's a conversion of autonomy. All of that is also, by the way, can happen in commercial trucking, in defense, and in construction and mining. All of these vehicles are just becoming intelligent. And I think the last 10 years is just a lot of promises. The next 10 years is
6:40 Speaker 1 the
6:40 Speaker 0 actually
6:41 Speaker 1 the productionization.
6:42 Speaker 0 What's happening in autonomy today, there, you know, this is a gross generalization, but there's two separate approaches. Uh uh, you know, one is the Waymo approach, which is you know you're seeing in San Francisco, uh, which can roughly be described as a heavy censored, expensive business model not figured out, no human in the passenger in the driver's seat. Key point, no human driver's seat. The Tesla approach is business model's been figured out, people pay for it, literally with dollars. Uh, part of the way part of part of owning a Tesla. And then uh but it doesn't go everywhere, and the driver needs to be in the seat because it does have disengagements. Uh whereas the Waymo doesn't, generally speaking, has tens of thousands of miles between disengagements. And so the next, you know, the last 10 years has been will self-driving happen. It's speculative, it's a research problem.
7:28 Speaker 0 The next 10 years is the monetization. We're all on this, you know, we as in the technology community who works in autonomy. Nobody's wondering, is this a research problem? Now it's all on the applied side. It's taking this into production, making it cheaper, cheaper, cheaper, generally available. And like literally is in the name of our company. We take these things and we, you know, on the applied AI side or we really proliferate them in through vehicle manufacturers. That's kind of our business. We want manufacturers.
7:58 Speaker 2 Like, how do are they gonna look more like Waymo or more like Tesla? Or what what is what like Tesla. Got it.
8:04 Speaker 0 And uh and that's not only US automakers, I would say all global automotive OEMs uh are are like that. And when we say more like Tesla, just to be very like and hit the nail on the head. It uh you know, we're a vehicle intelligence company. We provide the intelligence that goes into it. Um what Tesla Plus looks like, and you see some of this in China, frankly speaking, with a Fahway vehicles and a couple of other other kind of marquee brands there, is just a really intelligent in-cabin experience, a really intelligent driving experience and a really intelligent connected app ecosystem which connects to your home and the Xiaomi has these great kind of demos where you know when the car comes home, the blinds kind of open, close, the heating and all of that stuff, and and you can take media that's playing in your TV and then seamlessly played in your like that interaction is just very, very smooth.
8:53 Speaker 0 Yeah. So I think that's the well more of the future. The Waymo stuff is really a true, you know, it's like a ride-sharing product that ultimately is kind of a replacement for Uber for more than anything else.
9:05 Speaker 2 Are you starting to see like an a standard autonomy stack peak shape? Um with but I I guess both across your commercial and more your your your enterprise and maybe defense customers. Like is there starting to be a standard set of hardware? Does that make your job easier, harder? What what's the impact for a mind?
9:23 Speaker 0 Yeah, so so the any of these advanced technologies, it's false to say that there's like just a standard. But um I think probably a more accurate version is the abstraction away from hardware is becoming more and more common. And what I mean by that is um the autonomy system that we have running, let's say on a defense, you know, a vehicle, let's say it's uh that's a troop mover or a tank or something like that. It's like ninety percent similar to what we're doing in the cons you know on the commercial side and the on the on the passenger car side or on the on the uh trucking side. And that is not was not the case just You know, a few years ago. So what's happened kind of below the hood is there have been just changes in model architectures of improvements. And so I think everyone understands the proliferation of transformers as it relates to large language models and how kind of chat bots have gone from kind of really bad to really really great in in a short amount of time.
10:19 Speaker 0 There's a similar um let's say uh paradigm shift happening in autonomy and in vision systems as well uh with transformers. And so uh what that ultimately means in autonomy is the similar dynamics in large language models, you can put lots of data in and you get these nice scaling laws, the same thing is happening in autonomy. And people will roughly use this kind of end-to-end uh you know terminology. Yeah. It's both accurate and uh and and not accurate. We we don't need to get into that right now, but the point is the architectures are in are in creating capabilities uh that allow autonomy to kind of scale horizontally in ways that it just didn't, let's say even five years ago. And
10:58 Speaker 1 the approach is generally the same, right? For defense where you need to have the same sort of approach and simulation and the tooling that you develop autonomy for the commercial side is exactly what you need to do for defense. The challenge we have, of course, is condensing and and showing and demonstrating that this is critical for developing any of these autonomous solutions embedded within these vehicles in development. And the other key thing about uh like defense is that these vehicles stay in their inventory and out in the field for years and decades. All right. So many of these systems are what's the
11:31 Speaker 2 yeah, the lifespan of a tank is like you know, we're talking about. Yeah,
11:34 Speaker 1 I mean the B 52 will be a hundred years old. Uh, you know, here in, you know, a c and it'll it'll still be flying, right? And so what's interesting is a lot of these vehicles still require software to become more lethal, more mobile, more survivable. And so how do you then like take all of this technology that we develop for the commercial side for vehicles that are continually being developed and employed and then actually bring that and retrofit back into the vehicles that exist. It turns out it's it's similar approaches. You know, different stacks potentially, but similar approaches on how you enhance architectures and then you know, bring these capabilities on, even legacy systems. So this is one of the you know, one of the approaches we're also taking with defense customers, because we recognize that
12:18 Speaker 0 you know they're
12:18 Speaker 1 like new vehicles are are you know not happening every day, not happening every year. You have to figure out also how do you basically make all of these legacy systems you know more capable with software.
12:30 Speaker 2 So what's actually your balance between specifically for DOD between like supporting retrofits and legacy systems and helping make those extend the lifetime and make those modern versus you know some new systems that are coming out and have you know a different architecture that might be AI first out of the gate. Aaron Powell
12:46 Speaker 1 Yeah, generally it's about 50-50. So about 50 percent of our products are related to vehicles or or autonomy that's in development, and then the 50 percent we're focused on you know, essentially legacy vehicles and some of these systems need to be, you know, we we we use terms enhance that's that tends to be like less dramatic for a lot of folks, but in reality, we have to jailbreak a lot of these systems, like literally get the government to say, yeah, I know this was built by somebody else. I just need you to go in here and like jailbreak
13:14 Speaker 0 this thing
13:14 Speaker 1 and show us what you can do, show us how you can actually make this a better vehicle because it's it's really difficult to do it any other way, and then just you know, hand it to you and and let's see what you can do.
13:26 Speaker 2 So m maybe backing up, uh you know, you launched, you alluded to this. You launched your defense business in 2019, so six years ago or so now, before it was cool in Silicon Valley, squarely when it was not cool in Silicon Valley to be building for defense. Uh why did what you know, why did you do that? What drove you to build that side of the business.
13:46 Speaker 0 Uh so this is an A16Z, you know, podcast I'll I'll do some shilling care health for promotion. Thanks. Uh it was an EBC that we had um uh and it was General Dynamics who had just come through the valley and we're meeting companies. And uh we you know, we have three pillars of our business. It's it's tools, operating systems, and autonomy. And uh and we were showing showcasing our tools with what Jason is talking about is what autonomy really is fundamentally is the tools that you need to ingest this data, curate these data, make these models, and then test these models before you actually deploy them on a physical vehicle. A lot of that is just tooling tooling tool, right? And um it's a software factory that makes the makes this. So that's what we're showing to General Dynamics. That resonated, you know, and uh at a much much deeper level than we expected.
14:35 Speaker 0 And we just found out that, oh, actually the primes, broadly speaking, have the same problems that the commercial side has. And uh so you know, in the 2019s, I think when we actually set up the office and we really had it, but we were working with the you know in the ecosystem even even sooner. And then we recognize actually there's this whole ecosystem, right? Just like you know, it's if you're in the defense ecosystem, it's hard to imagine how diverse the automotive ecosystem is or commercial trucking or construction and mining. But each of these are their own economies with hundred billion, I mean, multiple hundred billion revenue companies. Um so then as we kind of got into the defense ecosystem, we realized, oh, there's like this directed DOD, there's the IU, there's this whole, you know, uh everything from you know a PAC and setting up a DC office.
15:24 Speaker 0 And I think we just over the last you know, six plus years, we've just built a real presence within DC and and almost like uh, you know, uh a division of the of the company. Um it's its own area of expertise. It's very different than the commercial side. Yeah,
15:39 Speaker 2 it's very different. It's super it's notoriously complex. It's uh you know, the procurement processes and cycles are really different. You alluded to this. So after that meeting with General Dynamics, like how did you get your first toehold into a direct DOD engagement? Trevor Burrus, Jr.
15:52 Speaker 0 I mean, DIU obviously plays a critical role there. I think I mean I my memory does really like we've got
15:57 Speaker 1 yeah, the robotic uh combat vehicle program
16:00 Speaker 0 with the with the Army was our first big one. Yeah, exactly. It was our first big one. There's also like sibers that are always around. So there's there's ways to engage with the DOD. I mean, we were a tiny little company with you know uh under a hundred people, and so you're talking about sharing which sources, and that's where you're kind of gear matching, you know, your rate your your gears are uh matchy. As you're a larger company, you're gonna actually have folks who are fully dedicated to this and and pursue larger programs of record. So I think my recommendation to founders who are out there who want to get into that kind of defense ecosystem or a dual-use company is you just have to start.
16:34 Speaker 2 So
16:34 Speaker 0 you start and there's different ways to start, and that's the way we start is not necessarily the way everyone has to start. But it was an effective way to just get the flywheel going. Trevor Burrus, Jr.
16:42 Speaker 2 And on sorry, go ahead. Trevor Burrus, Jr. Yeah,
16:44 Speaker 1 I was just gonna so Capsar, do you feel like at some point like you know in the history of the company on the commercial side, did you feel like you were doing much more market shaping and then it flipped at some point to market like there was actually a market that then you were serving. Was there was there some sort of like shaping aspect? Because this is definitely you know where we're at on the defense side. We are absolutely marking market shaping up until this point. We're starting to sense all of this starting to actually ship into yeah, absolutely. Trevor Burrus, Jr.:
17:13 Speaker 0 I mean, the the version of you know writing Waymo's in SF, if Waymo's had a huge presence here in in DC, it would change
17:22 Speaker 2 plug for Waymo to come to DC. Yeah, which is also
17:24 Speaker 0 uh uh uh portfolio company. Yeah, yeah. I mean, if they're here, it just changes you as a
17:30 Speaker 1 you could be working on the
17:31 Speaker 2 viscerally, yeah. You're like
17:33 Speaker 0 you don't have to be with autonomy and you're like you could be working at the Department of Education. You go in that and you're like, oh, this thing is real. Like this is not something that's fake. I just ordered it on my I didn't need any special specialization. And so in terms of I think that like us pushing to uh and pulling is happening because of the large industry. A good way to think about this this being autonomy is the way to think about almost like the mobile revolution, which was the last big kind of consumer revolution that happened. And imagine if the DOD was responsible for building a phone versus Apple or versus Android and you know, it's very hard to imagine they they can outpace the commercial side. Yeah,
18:11 Speaker 2 you can you can't really design a phone with a list of DOD requirements. And so at least a great phone. Trevor Burrus, so
18:16 Speaker 0 that there's a similar thing in autonomy where at some point people of the DOD recognize and the government represent broadly, we need to have a mobile strategy. We need to know how these devices are used, how how they're a fielded, both on just the regular bureaucratic day-to-day DC uh job, but also out, you know, for the for the warfight on the front lines. Think about autonomy the same way. How is autonomy going to affect the government broadly and Department of Transportation, et cetera, but then also the warfighter? And I think it's way more complex because the technology is way more deep, like uh a mobile operating system versus an operating system that goes in a vehicle that's safety critical are different, let alone the autonomy application that sit on top of that operating system. And I think that's where a dual-use company like us can be very valuable because we are benchmarking on the commercial side.
19:06 Speaker 0 Yeah. So yeah, and I think if you're only serving sometimes a DOD, you sometimes don't get that commercial benchmark. So imagine again, you're somehow a phone providing. Yeah. And uh we're always we have to we're always getting like baseline to the truth. If our products are not good, we're not gonna survive as a company. Full stop. We got to win in that very competitive commercial, global commercial ecosystem where we get no favoritism being American company or not. Yeah.
19:31 Speaker 1 Yeah, I just want to lean into the idea that the nature of autonomy is dual use. Right. So all those benchmarks are on the commercial side right now. So the DOD trying to actually develop and do a a number of things without referencing, you know, the commercial benchmarks, those who are familiar with what does it take from a ROI perspective from a progress and how many, you know, how many you know iterations of software do you need to actually develop before it's ready to ship. All of that currently is on the, you know. On the commercial side, all that said, you know, given that this is a dual use technology from its nature, from its core, what we've seen as a company is the kinds of things we develop for the Army, for example, off-road autonomy has fed back into our commercial business. So in the agricultural mining, you know, and we we just started to see this kind of flywheels start to affect it to take place.
20:23 Speaker 1 Now that we're in aviation, we're in maritime, we're in other domains. We fully expect that all this investment that we're making and the DOD is making and bringing us it on as a partner
20:33 Speaker 2 into
20:34 Speaker 1 autonomy in those other domains, it's going to feed back in the commercial just the same way. And so we'll start to see benchmarks while currently in the commercial space are going to start, you know, starting to essentially take that flywheel effect, you know, going back and forth.
20:49 Speaker 2 Yeah, I definitely spend my career before investing at Palantir. And uh, you know, a large reason why we were able to sell to customers like BP and others where you know people liked that Palantir was the software that they used to find Bin Laden. So you know, it made it easier to sell to some of these more traditional enterprises. So I think that flywheel effect
21:06 Speaker 0 Yeah, I think this is also just the reality of technology, which is whether it's you know Palantir or or applied or whoever um when you're working in a domain, you are compounding those res that that all that software that's being written, it it combs so on our businesses, you know, tools, operating systems, and autonomy, every year, we're adding way well, you know, we're we do something like a thousand features a quarter, which is kind of a mind-numbing number, you know, number that's even hard to understand. All of that benefits not one customer. It benefits all of our customers and all of our domains across both commercial and and the defense side. And so again, that's that's I think the power of dual use, you know. But
21:46 Speaker 2 for from like a, you know, one level down, I get like building blocks are the same. You can share features, you can share things across. But like, are there any big technical requirements or product, fundamental product differences, things you have to build differently or builds in one case and not the other?
22:01 Speaker 1 Yes. However, so I'll give you a great example of a couple different products that we just rolled out. And so one is called maritime SEM. So we do obviously a lot of simulations capability, all those tools for commercial automotive, run thousands of reps in order to find failure modes, et cetera. So we are working with the Navy on a couple different projects, and we wanted to basically build a maritime SIM. So literally uh a virtual world where you would where you have all the glint and all the C-spray and all of these things that you would experience if you were sent to the
22:31 Speaker 0 domain,
22:32 Speaker 1 yeah. Yeah, only in that domain. We went from zero, like the
22:35 Speaker 2 horizon, like just thinking about the fact that the horizon isn't straight, like,
22:38 Speaker 1 the earth is currently right, the Earth is covered. All of these unique and unique variables, but we went from like zero, no product to the first functioning use of a product in five weeks. It's all it took to be to essentially create that product. All because of the foundation that we have in our commercial space and our commercial tooling and the technology C approaches and the talent. And we have these very talented engineers who never did anything for the military before. And then all of a sudden, five weeks later, we have the first version of that, and then are actually using you know autonomy behaviors that we would actually put on a physical, you know, uh unmanned surface vessel, like running through that process right now. Yeah, I
23:18 Speaker 0 mean, just in like dollars and cents, our tooling that this is built on top of uh this maritime sim example is like we've we've uh spent like 750 million dollars of engineering costs to build it. These tools are extremely deep. So it the analogy really always ends up being is like you have this base set of technologies which are then optimized for different use cases. And those use cases are different. So I think if you're taking like a raw percentage of how much it is is interoperable, it's vast majority of it. But for each of these domains, there is uh augmentation, and then even within the defense ecosystem, especially when you get to classify projects, there is a you know, there is some optimization that happens. Yeah,
23:58 Speaker 1 and uh the punchline is the RD is paid for by the commercial sector, right? Not by the taxpayer. Yeah. And so that I mean that's really the key for dual use uh technology. Yeah.
24:07 Speaker 2 And how is that affected like just how you run your business in terms of who you hire, you know, do do do you segregate the company based on sort of I work on commercial or I work on government? How like how do you organize your not
24:20 Speaker 0 really? Um uh yes yes and no. Obviously, we have a DC office and and the entire, you know, working with uh DOD and the primes is just a different thing. That's a it's
24:29 Speaker 2 a whole
24:29 Speaker 0 skill set. You have that, but that's still a small minority of of the you know company. Um but where there is a difference is you work in classified work, you you know, we as a company are a company of immigrants. So we have you know Chinese nationals, we have Russian nationals, and so we are we have to be extremely uh organized, structured, and thoughtful about that. But it's absolutely possible. This is not in the realm of like, you know, we need to like cordon off these engineers into one area or one building. Sometimes you have to, but a lot of times you don't have to. And so I think what one of the I would say false uh kind of trade-offs people make is you know, all of this data or all of these tools need to be completely segreg or segregated and especially built for this.
25:18 Speaker 0 Actually, well, there's a lot more overlap and than we think. And so uh yeah, it's uh it's it's we do have processes, but it's smaller than you would think. Um,
25:29 Speaker 1 DIU were talking about DIU. I mean, the whole point of DIU originally was to build that bridge, rebuild the bridge between the Department of Defense and Silicon Valley. And you know, fundamentally you've got to recognize that you you need all of this work, you know, to be developed, you know, unclassified, most of it anyway, right? Most of the work can be developed unclassified, and then you can ship it into classified environments. That's exactly what we do. We obviously comply with all the you know the rules of the government. But you have engineers in in Mountain View who are working, you know, with for uh a product for a commercial OEM, you know, for first half of the day and the second half of the day, they're working on an Air Force project for us or a Navy project for us, for example.
26:09 Speaker 2 Yeah. Um I mean, you just touched on this. There is obviously a huge push right now to modernize and improve the way the DOD procures technology, especially software. It's been going on for several years with DAU, et cetera, but it's definitely accelerated in recent months. And uh my question is, you know, how is it going?
26:28 Speaker 0 Yeah.
26:28 Speaker 2 From your perspective as um a company selling into the government and where are the what where what do we still need to improve, whether it's on the process side, on the culture side, et cetera.
26:37 Speaker 0 A couple of things uh uh Jason is is is closer to the ground here, but I think when you it's kind of like the stock market. If you zoom in, there's a lot of fluctuations, you zoom out, SP 500 goes up into the right. And I think overall, there is improvements. Like we are not the same, you know, uh, we don't have the same challenges that even applied Cs from five years ago. So it's uh everything's gotten a little better. But I would say the the high-level point I would say uh is the more senior the folks who know software are in the Department of Defense, literally from you know, from the even all the way from the president of the United States. How many levels do you have to go down before somebody really knows software and they were a software developer? The higher we can make that in the Department of Defense or any of the services, I think procurement will get better.
27:23 Speaker 0 So people oh the knee-jerk reaction is to make procurement better, the process needs to improve. And that's like you know, the other way of saying is in order to make the bureaucracy better, we need more bureaucracy, right? And so really what we need is actually the the right people. When you have the right people at those senior levels who are you know, practic had practice software development in technical, commercial uh, you know, um kind of industry, they can cut through a lot of that and they can say that's actually a good product, this is not a good product, and they can streamline you know the work. We see the same thing on the commercial side. In automotive, in construction, and the companies that you can literally measure from the CEO, how many levels do you have to go to where they really know software of those companies are you take a Tesla's example, you lend those software.
28:09 Speaker 0 So you're right at the CEO level, therefore the product ends up being more software uh kind of uh you know drenched.
28:16 Speaker 1 Yeah. I guess I'll take a little bit more of a downer note, uh, but then I'll come then I'll come back to a much more positive note. I mean, if you if you look at the top 100 venture backed companies that are doing business with the Department of Defense, and you take SpaceX and Palantir put on the side, so now the top ninety-eight. Um that's $10 billion a year of business going to those 98. That's only one aircraft carrier, right? That's not a lot when you consider it's an eight hundred plus billion dollar, you know, annually is what the Department of Defense is spending. So, you know, roughly just over one percent. So that has to improve. And so a lot of the a lot of the challenges uh, of course, are
28:57 Speaker 0 the you know, and this is well known as well discussed is just
29:00 Speaker 1 fundamentally the DOD focuses on hardware and they vertically integrate all the software pieces to that. That's starting to change. There's different programs, there's software pathways, there's a number of different things. Some of the behaviors of the various folks kind of in the trenches doing the acquisition, they need to be comfortable just going direct to a software company, specifically a product company. And I, you know, the lot of a lot of people say, hey, we're a software company. They are a technology services company. They're not necessarily a software, a product company who knows how to scale from one to ten and a hundred. The nuance
29:36 Speaker 0 differences when it's a services company, they're just gonna charge you, the taxpayer, i.e., the government, um, you know, per head. Whereas in a product company, you're getting that subsidy. You're getting a nice elegance of, hey, we're at multiple customers are footing the cost of this, and then for each of the customers gets a cheaper product.
29:52 Speaker 1 Right. Uh so time and materials versus product companies want to focus on firm
29:56 Speaker 0 fixed prices is what we this is where we're gonna provide you. So
30:00 Speaker 1 Some of that, some of that absolutely is changing, but it's still like steering the Titanic trying to try to move in that direction. We just had the software pathways memo that Secretary Higgs just issued. That's that's uh like a beam of light. That's you know, like a Hallelujah moment. How it's enforced is going to be very interesting. DIU can't do it all, also, though like the capacity of DIU can't do all the software work. The services themselves have to learn how to do it. So there's there is absolute like positive, you know, uh progress on on all of these all of these areas, but uh long way to go for sure. Aaron Powell
30:34 Speaker 2 And maybe some advice from you both from your experience uh with applied for other startups who are selling into the DOD. How important are partnerships versus building direct relationships with DIU and the services? Aaron Powell
30:49 Speaker 0 Applied all of our values can be like uh, you know, uh distilled the two words, which is radical pragmatism. So like if you have people at the DIU and you come from DC, then go direct. If you don't, um and you need a commercial partner, like a prime to do that. Um so you know what it's whatever gets you in the game faster. Uh I think is important. I think though it's hard to have a real business um in you know, let's say not only the DOD, honestly, but also if you're selling to the government broadly without really having a presence in DC with you know uh folks you know who come from the ecosystem who literally worked in the ecosystem or wearing the services themselves because they really understand the business. It's like you know, me going to the General Motors Institute for undergrad growing up in Detroit, working at GM and Bosch is the you know biggest tier one supplier and automotive.
31:44 Speaker 0 That informs many decisions.
31:52 Speaker 0 Uh on that. So yeah.
31:53 Speaker 1 Yeah. The first advice I have is always as a company, no matter what stage you are, have a direct relationship with the government, with the actual government customer and understand what that government, you know, who the buyer is, who the users are, have a relationship with them. With a partner partnerships are a lot of work. I mean, they can be incredibly beneficial. We have a lot of different partnerships with different types of companies, uh at different stages, venture backed, some, you know, public companies, big primes, that sort of thing. They're a lot of work. They're absolutely worthwhile, but you've got to balance out you you really need to have that understanding of what does a government need. There's oftentimes an opportunity where you know, we have an opportunity to go direct to the government on behalf of the partnership, regardless if our partner is much more well established, much larger.
32:42 Speaker 1 There's there's just doors that we get into that they don't.
32:45 Speaker 0 And then, you know, that you know, recently
32:48 Speaker 1 has been even more common, right? We're we're able to have a conversation with folks because of where we sit, because we're dual use venture backed company. There's opportunities now that we have to talk through all these and benefit the partnership as a result. Yeah, and that
33:01 Speaker 0 answer is honestly different between uh dual use venture back company versus you know, uh research lab in Pittsburgh with eight employees. That's going to be a very different answer. So I think we're heavily optimized for this like venture back to a use, you know, uh kind of profile.
33:18 Speaker 2 Um it's been uh you know, a a wild 10 years or so in the world of autonomy, and we're starting to see we talked about way about the glimmers in the consumer space, it's coming, people can suddenly or people are feeling that it's coming, trucking and you know, more commercial industries, heavy industry, mining, construction, et cetera, the DOD, like how we had our Chat GPT moment for autonomy. What do you think that looks like? Is that even possible?
33:43 Speaker 0 I think it's a Chad GPT moment, just to make sure we're both talking about the same moment, uh, is the fact that there is a broad consumer appreciation for the power of the technology. In that sense, no. I think you go to a suburb of Indianapolis or Kansas City where people are just they're not living in the venture-backed or the DOD or the technology uh uh universe, and they're an accountant at a local, you know, at a local uh grocery store, they're not thinking about autonomy. They're thinking about Chat GPD. So that's in that way, no, the uh event hasn't happened. Yes, the way the event has happened is if you go to anybody in the you know in San Francisco and in the automotive industry, there's no debate about autonomy. So I think uh it's different because you know, the way a chat bot and search information retrieval works on the web is just very different.
34:36 Speaker 2 You can go to a web page a lot faster than you can buy a new car.
34:38 Speaker 0 Exactly. But it will happen next in you. I mean, that is that is every a 100% of automotive OEMs globally are working on autonomy solutions. There is not a single OEM who does not have a real effort in the space. So all that means and it's just you know, these are the complex technical problems that are safety critical. You know, the big difference between ChatGBT and a autonomy product is you know, one is safety critical and you're flying down the highway in a metal box going 80, 90 miles an hour. What one one of the ideally 70, you know, miles an hour. But uh the the one of the things that you don't realize, you know, I started my career in safety in airbags is uh when you're on the highway, you're also going 70 miles an hour. You get this false sense of security because you're belting in and there's like white paint on the ground, but you are just flying down the highway.
35:27 Speaker 0 And uh it's actually the most dangerous thing that you do by far every single day. There is nothing you're doing as dangerous as just getting in a car, right? I mean what can we you know, we can have this interview and then we could be in a really bad accident 15 minutes from now. Like that's a crazy reality. It's also one of the reasons I think autonomy is so important just at a national level in terms of saving lives and costs of of of the suffering that that that comes with accidents. So that's all gonna happen in the next five, seven years. You don't need to go on five autonomy rides to be kind of blown away. And so I think all of that all of that is coming. Uh but it's gonna it's not gonna be like this one click. Aaron Powell
36:07 Speaker 2 And what about outside of outside of you know me buying my personal self-driving vehicle in in in industry whether it's an industry or we're in the government, how close are we to autonomy being ubiquitous? Aaron Powell
36:19 Speaker 0 I think I'm in government, I'll let I'll let Jason uh talk to about it within within those contexts within construction mining, ag these other other verticals similar to PASCAR, passenger cars, where every major OEM globally is thinking about uh infusing this technology and then broadly speaking, this kind of oper new modern operating systems in their vehicles to unlock a lot of these features. So uh yeah, there's no longer there's no debate. There's not a single person in the vehicle manufacturing universe who works on technology that thinks this is not something we should be putting uh resources towards.
36:53 Speaker 1 Yeah, you can look at Ukraine, for example, but even beyond Ukraine, Ukraine's been used as an example a lot. Um look at the Middle East, you look at Yemen. All of this technology is fairly democratized, right? So it's kind of founded on a lot of open source capabilities and and you know there's been an interesting, and if you really start to pay attention to what's happening like over Yemen, for example, like the Iranians have provided a like a loitering surface air missile system. So I used to be the director of intelligence for Air Force's central command. So this used to be like this is a problem that I nerd out on, right? And so this is all like an AI-based, you know, very cheaply made sort of uh weapon system that keeps shooting down our drones. And then the Houthis are launching these very cheap drones at destroyers who are shooting them down with these missiles that are a
37:42 Speaker 2 million bucks apiece. Yeah.
37:43 Speaker 1 And that and
37:44 Speaker 0 I think you we can all say that Houthis are not like the most technically forward in a fighting force. Absolutely not. So you have folks who are not very technically advanced who are using fairly advanced technology. Trevor Burrus, Jr.
37:55 Speaker 1 Right. And so currently, most of these things that we've seen like attacks on Israel and everything else, these drones come across, they're not networks, they're not really effectively networked, but the reality is is that what is being exposed or new problems that all kinds of people throughout the world are trying to solve on this on this autonomy space, right? Collaborative autonomy, which is really kind of where all of the defense work is gonna like lean toward, yes, for the US, but everyone else. It's kind of whoever figures out how to effectively network these swarms to where they can actually behave collectively, that is going to be what's the same thing. Yeah, and that is inevitable. That is an inevitable like reality of where this technology is going. And so the first time that that collaborative swarm is used anywhere, it's just a matter of time.
38:42 Speaker 1 That is just going to completely upend if it hasn't been already the entire defense industry globally. Trevor Burrus, Jr.
38:48 Speaker 2 It's an extremely exciting moment. What's next, what's on your minds? What do you what keeps you up at night? Aaron Powell
38:53 Speaker 0 We're uh you know, deeply technical company, vast majority of our employees are our software engineers. Um so what's on my mind is you know the culture uh within the company and making sure they're productive and they're pointing out the right problems and we're making progress on those problems faster than our competitors, and then uh, you know, recruiting people. So like uh if you're listening to sounds interesting, uh whether you're on the technical side or the commercial side, you know, just uh, you know, on the on the business development side, we're hiring, so you can always uh always reach out. Yeah.
39:25 Speaker 1 Yeah. I mean, one of the really cool aspects of our company is if you're curiosity, we have all these folks who are very curiosity driven engineers who really love hard problems. We've got loads of them and a number of
39:38 Speaker 0 different,
39:39 Speaker 1 you know, aspects of uh developing software and you know, doing various types of engineering uh to solve it. And then we also have just a lot of mission driven folks who really focus on how do we get this capability in the hands of war fighters and in the hands of you know, governments, US government and and uh and allies and uh see get this really Really cool kind of dynamic of all these people coming together. Um, it's really neat to watch, you know, all of them become like basically both curiosity and mission driven at the same time. I've watched these, you know, Silicon Valley base engineers who had worked on, you know, a lot of commercial products, like fly out to our site in Florida where we do a lot of flight testing for the Air Force, meet with a lot of our mission-driven engineers from our Epic team, and then actually build something together that they hadn't built before, you know, around collaborative autonomy.
40:29 Speaker 1 That's been really cool to watch. And that's, you know, fundamentally uh, you know, the direction we're going to keep going.
40:36 Speaker 2 Amazing. Okay. Well, send engineers your way.
40:39 Speaker 1 Yeah, yeah.
40:40 Speaker 0 Yeah. And thanks for having us.
Previous Drone Warfare: The New Rules of Combat Are Here Next The Dual-Use Founder: Vets Now Building For America Page updated August 25, 2026 · transcript and summary are machine-generated
