More often, co-investment means the Navy waits for companies to mature a product on their own before buying it, rather than funding early research itself.

The stage of company the Navy buys from has shifted, too. “We mostly buy Series D through F type companies,” Fanelli said, adding that the Navy used to fund its own early-stage research to cover the seed-through-Series-B gap.

It’s trying to hand that job to commercial investors instead, which is part of why the priorities document exists at all. “The cost of that, or the responsibility, is for us, if we’re not going to do it ourselves, to cast a cleaner signal,” he said.

As for recent purchases, Fanelli — a former Air Force cadet whose career has spanned roles across defense, intelligence, DARPA, and open source initiatives — rattled off a handful. A contract worth $562 million was awarded this month for the MQ-25 Stingray, an autonomous refueling drone that extends the range of manned fighter jets flying off carriers.

The Navy has also been buying edge compute hardware from Armada, described loosely as shipping containers packed with servers meant for ship or remote deployment.

It brought in Gecko Robotics to handle inspection work that used to be done manually and dangerously; Fanelli said the move drew little pushback because almost nobody wanted that job in the first place.

Domino Data Lab is now running the Navy’s machine learning pipeline.

And in a case Fanelli seemed to particularly enjoy sharing, the Navy swapped a defense contractor’s years-delayed shipboard camera system for commercial cameras paired with software from Applied Intuition, cutting roughly four years off the timeline and expanding to more ships than expected.

Given the Strait of Hormuz has been a flashpoint this year — tanker strikes, a seized vessel, ceasefires that have broken down over shipping-route disputes — it seemed worth asking whether any of that commercial technology is showing up there now.

Fanelli was cautious.

“I often don’t know what’s classified and unclassified because I’m normally talking to people with clearances,” he said, adding that he’d have to check.

Interestingly, not many people sign off on a purchase like that, according to Fanelli. Buying decisions run through what he called a source selection committee, a small group rather than the sprawling review process outsiders might picture. He described the goal as keeping the process merit-based rather than adding layers of approval.

He didn’t get into why technologies typically fail once they’re inside the Navy — he’d arrived at his car and needed to figure out next steps for travel.

But when we talked last year, Fanelli explained that the most common reason promising technologies fail isn’t technical, but rather budgetary.

The Navy runs on long planning cycles, and a new tool only sticks around if it replaces or “turns off” something the Navy is already paying for.

Either way, before we parted ways, Fanelli shared an updated internal list of the technology areas the Navy is prioritizing, jointly issued by his office and the Portfolio Acquisition Executive for Mission Systems. It breaks down into five broad categories, for those of you either developing or funding defense tech.

The first is applied AI, covering machine learning and increasingly agentic software meant to turn raw data into decisions — sensor fusion, targeting support, autonomous behavior, and cyber operations built on software rather than hardware.

Quantum information science comes second, focused less on owning quantum hardware and more on applying quantum or quantum-adjacent techniques to navigation, secure communication, and cryptography.

Advanced networking is third, aimed at moving data securely across environments where connections are degraded or intermittent, whether that’s a ship at sea or a coalition partner’s network.

Fourth is electromagnetic spectrum operations, technology that senses and manages the spectrum itself, built to adapt in contested conditions rather than rely on fixed configurations.

The fifth category, digital engineering and interoperability, covers the plumbing: open APIs, model-based systems engineering, and zero-trust architecture meant to let different Navy systems talk to each other without custom integration work every time.

None is a funding commitment, the memo notes, and none of it ranks individual programs. It also adds as a reminder that its priorities “will change based on emergent, near-term, and long-term needs.”

In short, don’t take it as gospel. Instead, it’s meant to be read the way Fanelli talked about last year’s list — as a roadmap that companies and investors can plan around, so that decisions made now about where to put money have somewhere to go inside the Navy in a couple of years.

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Meta is refuting a journalist’s claim that its AI agent Muse read the user’s private messages without permission. Following an earlier report from Inc. columnis...