Pentagon Ramps Up Tech Development for Replicator Program

Replicator, the Defense Department’s all-domain attributable autonomous system program, has pushed the scale and speed of software development to new levels ahead of its Aug. 2025 deadline to fully deploy the program, said Rear Adm. Christopher Sweeney, director of the Integrated Warfare (N9I) Office of the Chief of Naval Operations, at the 2025 AFCEA West Conference in San Diego, California.

“A really important thing to consider in the context of Replicator is that it’s not another sat project. It is meant to get to production, meant to field systems,” Sweeney said. “It’s a lot of technological development. It’s a lot of sprinting. It’s a lot of taking … a pretty wide and diverse set of systems and a wide and diverse set of software and smashing them all together at a pace that is more akin to commercial software tempos.”

Rear Adm. Seiko Okano, commander of the Naval Information Warfare Systems Command, told the audience the development of Replicator drones has pushed the envelope dramatically in terms of speed, scale and feedback.

The Replicator program has created an accelerated feedback loop, changed risk postures to be more iterative and shifted decision making to the fleets to determine which capabilities are scaled.

The Defense Department announced its first tranche May 2024. The pentagon unveiled its second tranche of all-domain attributable autonomous systems in Dec. 2024.

The second tranche, dubbed Replicator 1.2, “includes systems in the air and maritime domains, as well as integrated software enablers that will enhance the autonomy and resilience of other Replicator systems.”

Replicator is designed to connect systems in different environments, while lowering the manpower needed to monitor them, said Aditi Kumar, deputy director for Strategy, Policy and National Security Partnerships at the Defense Innovation Unit (DIU), at Defense One’s State of Defense Business event last year.

“The end goal here is that we want to be able to field this map of systems, be able to communicate with them in all sorts of different environments, be able to reduce the number of operators that are required to field them, which inherently means increase the amount that they can do autonomously and increase the level of collaboration between these systems, so that, for example, you could have the aerial system identify a target and communicate that to a maritime system that is better placed to prosecute that target,” Kumar added.

Okano spoke of a “collaborative, teaming arrangement” between industry and government over the course of two-week sprints. Okano described the process as “DevSecOps in real life,” breaking the traditional mold into a new mindset centered around learning.

“You go out and do two-week sprints, and you find things don’t work, and the operator gives the developer feedback right then and there. We do another sprint, we iterate, and we’re doing that at pace, working on fast programs,” Okano said. “This is how the future needs to be. It really is a fundamental shift in how we’re developing and delivering capability.”

Sweeney emphasized the DOD’s need to sync up across the enterprise to achieve software capable of managing autonomous or command and control systems. As long as the conditions are flexible, Sweeney said that sailors would be able to work within them.

“I think our sailors, if you give them the conditions and the tools and the tech, they’re going to figure it out, because they started plugging all kinds of stuff together, and we got a lot farther than we originally thought,” Sweeney said.

Considering the presidential transition, Sweeney has his office to needs be in “lock step with” DIU in order to drive change across Naval fleets.

“We’re going to get that moved around a little bit with the new administration,” he said. “I think we need to be lock step with DIU, so we can put our money, and double down, [where] it makes sense.”

Okano, along with Capt. Alex Campbell, DIU’s Maritime Portfolio director, said cultures and ecosystems as the biggest challenges to tackle while scaling Replicator.

“It’s important to create not just the technical ecosystem that integrates the capability, but to also create the ecosystem where we’re able to readily communicate at the appropriate levels with our industry partners, so that we can affect this end-to-end chain where speed of development, the speed of integration, and ultimately the speed of fielding, [is] really end-to-end,” Campbell said.

Okano said that despite Replicator’s speed, old processes for acquisition, development and testing might still be useful for select programs.

“There has to be a realization that the government has to be ambidextrous in this. If you’re building a nuclear submarine or something, absolutely go through the classically trained PPBE process that makes sense for that model. But I think we had ‘a-one-size-fits-all’ for everything. I think now we’re learning that there are systems that we can do in a more rapid piece, it’s really differentiating the work to our organizations,” Okano said.