# Anduril built one software stack for many autonomous systems

Markdown copy of https://realalpha.app/s/anduril-built-one-software-stack-for-many-autonomous-systems/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Trae Stephens](https://realalpha.app/people/trae-stephens/) · Trae Stephens is cofounder and executive chairman of Anduril Industries and general partner at Founders Fund.
- Host: [Shawn Ryan](https://realalpha.app/people/shawn-ryan/) · Host, The Shawn Ryan Show
- Show: The Shawn Ryan Show
- Episode: [Trae Stephens – Anduril’s AI Weapons, Palantir Myths, Antichrist and War with China | SRS #245](https://realalpha.app/v/qyT2OZpFDog/), Oct 16, 2025
- Moment: 58:10 to 1:04:45 of the episode (7 min), playing on https://realalpha.app/s/anduril-built-one-software-stack-for-many-autonomous-systems/; the episode from there: https://realalpha.app/v/qyT2OZpFDog/?t=3490

## Quote

> “the **exact same backend** that we used for every subsequent product.”
> — [Trae Stephens](https://realalpha.app/people/trae-stephens/) at [1:00:03](https://realalpha.app/v/qyT2OZpFDog/?t=3603) · The Shawn Ryan Show

## Takeaways

1. The first product needed to be fast to field and politically urgent.
2. Lattice grew from the tower’s AI and operating system.
3. One software backend supports systems across multiple domains.

## About this segment

Trae Stephens explains how Anduril’s first border-surveillance tower led to a shared AI platform for drones, interceptors and other systems. Reusing the software lets new products build on earlier R&D.

From “Trae Stephens – Anduril’s AI Weapons, Palantir Myths, Antichrist and War with China | SRS #245”, Oct 16, 2025. [Watch the full episode](https://realalpha.app/v/qyT2OZpFDog/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[58:23](https://realalpha.app/v/qyT2OZpFDog/?t=3503) we made is, like, you have to build something that is going to be pretty fast. Like, you have to be able to get it out into the field quickly. You don't want it to be a science project. And you want there to be a significant amount of political urgency around the thing that you're going to build so that you can drive a different kind of decision making rather than business as usual procurement policy.

[58:46](https://realalpha.app/v/qyT2OZpFDog/?t=3526) And so for us, that meant that we landed on Century Tower, which is now deployed not only at military bases, but also on the southwest border in the United States with a border with Mexico. Obviously, in the first Trump administration, this was a big concern. whether or not you have a border wall you need to know what's going on and so we built this autonomous century tower that was you know completely off-grid mobile so you could pick it up and move it around had, you know, a solid-state metamaterial radar, an infrared camera, optical camera, communications network, and it would just persistently look around itself and flag for the Customs and Border Patrol whenever something was crossing the border that they needed to take a look at, and then it would enable them to make a decision.

[59:40](https://realalpha.app/v/qyT2OZpFDog/?t=3580) Like, yes, that's something that we care about that we're going to go and take a look at. No, that's something that we expect to see. And over time, you train that model, and that has now grown into a massive program. We cover an enormous swath of the southwest border, and we're expanding constantly, delivering more towers to get more and more coverage on the border. But the important thing about that initial work is that all the lessons that we learned from that, the AI computer vision system, the operating system for surfacing tracks to the users of the platform ended up being the exact same backend that we used for every subsequent product. And so it was kind of the, you know, the AI brain was the real product.

[1:00:24](https://realalpha.app/v/qyT2OZpFDog/?t=3624) And then whether you made a century tower that flies, which was our next product, which was called Ghost, to, you know, interceptor missiles, to submarines, to autonomous fighter planes, They all have that same operating system on the back end that we call Lattice. And so we're still getting amortized research and development out of the very first thing that we started building eight years ago. Damn.

[1:00:50](https://realalpha.app/v/qyT2OZpFDog/?t=3650) Where did you guys come up with the list? It was just, you know, sitting around with Palmer, me, Palmer, Matt, Brian, and Joe. Just, you know, riffing on what we think would make sense for us to build, leveraging that combination of software and hardware. And, you know, I'm honestly kind of surprised that we ended up doing almost exactly what we said that we were going to do. That's crazy. Do you have that slide with you by chance?

[1:01:16](https://realalpha.app/v/qyT2OZpFDog/?t=3676) I'd love to put that up. Yeah, yeah. I can show you the slide. Perfect. So what did come next? So we went, the next product after the Sentry Tower was Ghost, which is just a flying version of the tower. It's a modular payload drone that could do autonomous identification, detection, and tracking of targets. And so you can kind of think about it as like a helicopter with a pilot where you just tell it, hey, I want you to go and do this mission.

[1:01:45](https://realalpha.app/v/qyT2OZpFDog/?t=3705) Like, go and see if you can find things. that look like this and then it would just go and conduct a mission and then report back to you when it was seeing things that were interesting and these are now deployed all over the place that are deployed with us forces abroad they're also deployed in ukraine so uh it's become it's become kind of a core product line for us um and then we went from there into counter drone we started building uh interceptors for you know small drones we have a project called anvil that is the first version was not explosive it would just fly like a bowling ball into a drone and then we've continued to level that up into Roadrunner which is kind of like a reusable interceptor, Barracuda which is you know like a missile more like an interceptor cruise missile and kind of expanding that out every time but the goal is really like all domain like you want to

[1:02:37](https://realalpha.app/v/qyT2OZpFDog/?t=3757) do you know work in space work in air work on ground work on the surface of the sea subsea And Palmer's new obsession is subterranean, which, you know, not a lot of details that are shareable on that front yet. Damn. What are you guys doing in space? Actually, all of our space stuff is classified. So that's like the one domain that I can't really say much about. All right. How about under the sea?

[1:03:02](https://realalpha.app/v/qyT2OZpFDog/?t=3782) Yeah. So we have a whole platform called Dive. So different sizes. Dive LD, which is the large diameter dive, is like the size of a pickup truck, basically. Again, same software, you know, you send it out on missions and it can do everything from pipeline inspection to, you know, in theory, you could do kinetics. You could, you know, drive it into the side of an enemy warship. We started a joint venture with the Australian Navy to build an extra large version of Dive that's called Dive XL, a really clever name.

[1:03:36](https://realalpha.app/v/qyT2OZpFDog/?t=3816) And we just completed the delivery on a program of record for Australia where we'll build a bunch of these for them as they're kind of transitioning over to the AUKUS treaty where they're getting the delivery of nuclear submarines from the United States. so they obviously have real concerns about their the the sea space around the continent um and uh and we're spinning up a bunch of work there with them on that as well we also have a product called seabed century that is kind of a platform that you can drop into onto the floor of the ocean floor and it does the same thing as a century tower would except uh And so, we also have Copperhead, which is kind of like a modular payload delivery vehicle undersea, which is kind of a euphemism for a torpedo without saying torpedo. But yeah, we're, you know, we're always building new stuff in all these domains.