Until recently, Gen. Frank Donovan ran the Defense Autonomous Warfare Group, the white-hot center of the Pentagon’s drive for affordable mass and battlefield robots. Now he’s in charge of U.S. Southern Command, which is working hard to put the DAWG’s products to use. Defense One sat down with Donovan during SOF Week in Tampa. This interview has been edited for length and clarity.
Q: You’re an expert in autonomous warfare, as a former leader of the DAWG—for which a nearly unimaginable $50 billion has been requested in the fiscal year. How do you want to develop and use it at SOUTHCOM?
A: It's embarrassing to think that I'm an expert on autonomous warfare, because there are folks here that know so much more about the tech and the science and how it all works. I don't know all those things. I've learned a lot about it, but I've really focused on how you actually synchronize those things and bring it to bear, because I think my concern is right now, what I'm sensing—and you know, three years as vice commander of SOCOM, I got to be in the building watching three [program objective memorandum] cycles build. I come up here as a Marine infantry officer, reconnaissance, special operations, but I'm going to talk about what matters. It's budget and resource, and applying those resources to what we actually really need.
And so, what I started seeing is that even though Ukraine is going on, we're learning some lessons—and that's a whole side topic, which lessons we're learning from Ukraine—but we're seeing things in the South Red Sea, we're seeing things in the operational SOF environment, things I've faced, and I'm like, there's something different here, but how does it compete in the [Pentagon] with the services that hold most of the strengths? They hold the relationship with the defense industrial base, they hold a relationship with Congress. That's just how our government works, and it's healthy, and it's good, but are we going to be able to embrace autonomy? And they then embrace autonomy, not autonomy platforms, because I think we get caught in this a little bit, you know. I don't really care about platforms, I care about autonomous warfare, and are we really willing to take a step forward and embrace autonomous warfare. I think there's definitions, and so three years as vice commander at SOCOM, I saw this tension between what the joint force needs out front— and I'm going to say the joint force, not what our Army, Navy, Air Force components need out front, it's what the joint force needs to fight—and how those autonomous needs actually enter back into the Pentagon, and then get built into a service to actually come out and end up back with the war fighters. That's a misconnect.
I call it the two Olympic rings. Those two Olympic rings don't touch. When we had it as a very short window, nine months with the Defense Autonomous Warfare Group, I worked for Gen. [Bryan] Fenton [and] Adm. [Frank] Bradley was my boss for SOCOM, but I was working for [Deputy Defense Secretary Stephen] Feinberg. He held the resources. And that's what gets everyone's attention in the Pentagon: who holds the resources? So we could take the needs that came out of Replicator tranche one and two, and then quickly turn and say, ‘What can we bring to bear quickly with what's out there?’
And so we started to see if you match the actual true joint autonomous requirements, your actual needs, with service acquisition, there's something there, there's another ring in the Olympic rings that could be added there, and so what we saw in the DAWG formulating, we then said, well, if we come into SOUTHCOM, how do we actually create the SOUTHCOM Autonomous Warfare Command to address that gap, to address that need, and drive those requirements back up into the DAWG? So that's where we're planning on, and that's the journey we're on with SOCOM.
Q: You’ve talked about how battlefield networks will enable autonomous-warfare concepts like distributed swarming. And when I talk to Ukrainians, they wish they had such networks. But, of course, Russian electronic-warfare forces work hard to prevent that. How are you approaching this problem?
A: I think the operational data enterprise—operational data environment, whatever term we want to use—that we have to kind of encapsulate, and that's—the Marine colonel we're bringing in for the SOUTHCOM Autonomous Warfare Command, we talk about this. We don't talk about robots, we talk about the data environment with the different data layers that we need at the very forward edge, so our SOF and our conventional force teammates with an ATAK or a cell phone, that they can actually plug into that data network, and whatever robot shows up with the capability, they can leverage it instantaneously. It doesn't come with a priority stack or a company that we vendor-locked on. It is truly a fully capable system that we can use in selecting the needs, whether it's kinetic, when it's non-kinetic.
I think for us in SOUTHCOM, most of the systems we're looking at primarily are domain awareness systems. And for us it just magnifies—because if our partners, who have the access and placement where they live, where they operate, the environments that they have to work in, in the rough terrain, the jungle, over the horizon, thousands of miles at sea— we're working with our partners, going after these [Designated Terrorist Organizations], we have to enhance their domain awareness, but they have to also be able to plug into this environment in a cheap, easy, and very fluid way. And I think if we think about the data layers, the data environment, that's the first thing that we are focused on right now, is setting the environment. Because we can match the robots to the environment, I mean, whether it swims, it flies, it has feet, whatever it does, we have to make it do what we want it to do when we want it to do it without someone telling us, ‘Yeah, it's great only if you use it this way, only if you use my service stack, and only if you connect it to that.’ Unacceptable across the board.
Q: Are vendors still bringing proprietary systems, or has the open-architecture push actually taken hold?
A: I think we're starting to see improvements in that. And I would say two years ago, not at all. Everything was solely focused. And the concern is that you get a vendor with well-meaning folks, and a lot of them are retired folks, they got out, they moved on, they want to pitch a piece of kit to a commander, and they get all excited about it. And the problem is: it's great for a specific event or an exercise, but it doesn't have a path forward.
The more we, as military leaders, demand open architecture, we have to make sure our demand is clean: “Hey, this is what I need this thing to do for me.” And that's not always clear either, because I think part of it is: folks my age, we're not sure how to embrace autonomy and what this means, and to really give freedom down to the lower edge, that tactical edge, all the way up to lethal effects, without, you know, always a human in, on, to the loop, but we'll always have that, because there's nothing truly autonomous, there's always someone involved. But we have to think about delegation and empowering ways that autonomy makes people nervous. I mean, if you have a one-way attack, lethal one-way attack system, it's not that we're going to—that's why I'm little concerned that we get over infatuated with FPV.
Actually, I'd like to move away from FPV entirely, but every time we do, we have someone saying, “Well, what about collateral damage, what about the final go, no-go?” We’ve got to start thinking very differently. The approval to launch the system, or even put it in place, is lethal.
What happens often is that we don't come with a clear signal to our tech industry and our vendor partners with what we really want. We just compound it and ask for different things, and all of a sudden, “explosive boat” turns into an ISR platform turns into something el