The Office of Naval Research has a new organizing idea, and it fits on a napkin: FEED at Speed. Behind the acronym is a bet that the hardest problem in naval technology is no longer inventing things — it is getting the things that work out of the lab and onto the water before they die in the gap between a promising prototype and a program of record.
On July 22, 2026, ONR announced its 2026 Science and Technology Strategy for Navy, Marine Corps Tech Future. Trade coverage caught up on July 27. The document restructures how the Navy's premier research office picks projects, justifies them, and measures success — and for anyone tracking unmanned and autonomous systems, the framing is hard to ignore: the exemplar capability trade coverage points to is an uncrewed vessel that actually reached the Fleet.
What 'FEED at Speed' Actually Means
FEED is five pillars, each a verb ONR wants applied to every investment decision:
- Focus — Invest in transformational, naval-unique problems that commercial markets will not solve on their own. ONR frames this as the "Only ONR" test: if industry or another agency is already going to build it, ONR should not be the one funding the science.
- Explain — Every effort must articulate a clear naval relevance and a credible transition pathway. No more research that is interesting but orphaned.
- Engage — Break down silos between the labs, the Fleet, and industry, including through recurring "Innovation Industry Days" meant to shorten the distance between a good idea and someone who can build it at scale.
- Deploy — Measure success by fielded capability, not by prototypes demonstrated. This is the pillar with teeth: a slick demo that never reaches a sailor now counts as a miss.
- Speed — Use artificial intelligence to accelerate scientific discovery and compress research timelines.
Dr. Rachel Riley, the Chief of Naval Research, framed the strategy as "a commitment to discipline, focus and renewed urgency," and told trade press the office intends to refine "every part of our business." The strategy is written to serve the Department of the Navy and, in the document's language, the "broader Department of War."
The Valley of Death Is the Real Target
Strip away the acronym and the strategy is aimed squarely at a problem the defense community has complained about for decades: the "Valley of Death," the stretch where a technology has been demonstrated but has no procurement money, no program office, and no path into the acquisition system. Prototypes go in; fielded systems rarely come out.
The 2026 strategy attacks that gap from two directions at once. The Explain pillar forces every program — not just the flashy ones — to show a transition plan up front, so that "what happens after the demo" is a design constraint rather than an afterthought. The Deploy pillar then changes the scoreboard, judging the office by capability that reaches the force rather than by the volume of prototypes it can parade. In other words: fewer science projects that end in a briefing slide, more that end in a hull, a payload, or an autonomy stack a unit can actually use.
Why This Lands in the Unmanned World
ONR could have illustrated the strategy with almost anything. The example that trade coverage foregrounds is unmanned: the Medium Displacement Unmanned Surface Vessel (MDUSV) Seahawk, delivered to the Fleet in 2026, held up as the model of a capability that actually made the crossing from experiment to operational asset.
That choice is telling. Uncrewed and autonomous systems are exactly the category most exposed to the Valley of Death. They tend to originate as science-and-technology efforts, they depend on software and autonomy that mature faster than traditional acquisition cycles, and they frequently lack an established program of record to inherit them. A ship program has decades of institutional gravity pulling it toward procurement; an autonomous surface vessel or an experimental autonomy behavior has far less. By naming a delivered unmanned vessel as the standard, the strategy's interpreters are effectively saying the transition problem it most wants to solve is the one that afflicts autonomy the hardest.
For the drone and UAS community, the read-through is the framework itself, not a specific air vehicle. The pillars are platform-agnostic. "Only ONR" focus, mandatory transition pathways, capability-not-prototype scoring, and AI-compressed timelines apply as cleanly to an autonomous aircraft, a collaborative-autonomy payload, or a counter-UAS sensor as they do to a surface vessel. If the strategy works as advertised, the beneficiaries are disproportionately the unmanned programs that have historically had the most trouble surviving the handoff.
The AI Bet — And the Caveat
The Speed pillar is where the strategy is most explicitly of-the-moment. ONR expects AI to accelerate scientific discovery and shrink the time between a research question and a usable answer. That is the mechanism the office is counting on to make "renewed urgency" more than a slogan.
Notably, the framing is not uncritical. Trade coverage stresses that ONR expects AI to speed discovery with appropriate validation and oversight — an acknowledgment that a research office cannot simply outsource judgment to a model and call the result science. For unmanned systems, where autonomy and machine learning are already core to the product rather than just the R&D process, that "validate and oversee" caveat is arguably the most consequential sentence in the whole strategy. It signals ONR intends to move fast on AI without treating validation as optional.
Reading the Strategy Skeptically
Strategy documents are cheap; transition is expensive. The Valley of Death has survived many previous frameworks that promised to close it, and every one of FEED's pillars describes an intention rather than a mechanism with a budget line attached. "Measure by fielded capability" is a genuinely different scoreboard — but only if leadership is willing to kill well-liked prototype efforts that cannot show a transition path, and only if the acquisition system on the other side of the gap actually has money to receive what ONR hands it.
What is different this time is the pairing. The discipline pillars (Focus, Explain, Deploy) are meant to stop wasting effort on research that was never going to transition, while the Speed pillar is meant to make the surviving efforts move faster. Whether that combination finally narrows the gap — or simply produces better-documented prototypes that still stall — is the question the next few years, and the next few unmanned deliveries, will answer.
Why It Matters
The Navy's top research office just rewrote its own definition of success to reward fielded capability over demonstrations, and it built AI-accelerated timelines and mandatory transition planning into the core of how it picks work. Trade coverage chose a delivered unmanned vessel as the model of what "good" looks like. For the UAS and autonomy sector, that is a signal about where naval S&T money and urgency are being pointed — and a reminder that in 2026 the decisive challenge is not building autonomous systems that work in a demo, but getting them across the Valley of Death and into the hands of the force. If FEED at Speed delivers even partly on its promise, the programs most likely to feel the difference are the uncrewed ones that have always struggled most to make that crossing.
