The US Navy and the Defense Innovation Unit are putting $50 million behind Shield AI’s X-BAT, a fighter-sized jet drone built to take off and land on its tail from a trailer or a warship deck. The award, announced August 13, is the first publicly announced investment in an aircraft that has never flown. The Air Force, which Shield AI is separately courting with autonomy software for its Collaborative Combat Aircraft program, is deliberately keeping its wallet closed.
A destroyer never had a runway to lose, while the Air Force has spent two years war-gaming what happens when Chinese missiles crater its Pacific runways in a war’s opening hours. Both services see the appeal of a jet that needs neither. Only one is paying to find out if it works. DroneXL has covered Shield AI’s vertical landings since the V-BAT era, and that record says the last ten feet are where this concept gets expensive.
The Navy Wants Strike Aircraft That Fit On A Destroyer #
The Department of the Navy Rapid Capabilities Office, its Portfolio Acquisition Executive for Aviation, and the Defense Innovation Unit announced the $50 million on August 13 under RIMES, a program launched in February 2026 to give surface warships a long-range strike aircraft that needs no runway or flight deck.
The award expands risk-reduction work the Navy was already running with Shield AI, per the Navy’s announcement. RIMES grew out of an arithmetic problem: a warship’s vertical-launch cells hold a finite number of missiles and cannot be reloaded at sea fast enough to sustain a long fight. The February solicitation asks for a drone that delivers 1,000-pound (454-kilogram) class munitions, the same class an F/A-18 carries, across a one-way range of at least 1,400 nautical miles (1,611 miles / 2,593 kilometers), from ships without large flight decks. The Arleigh Burke-class destroyer, the Navy’s most numerous surface combatant, can land a helicopter and nothing fixed-wing.
X-BAT is the second aircraft RIMES has backed, after the program picked Mach Industries and Whisper Aero’s hybrid-electric Atlas in June. Shield AI’s entry is the louder bet: a cranked-arrow jet the size of a fighter powered by the same General Electric F110 engine that powers an F-16, with enough thrust to climb off its tail and, in theory, settle back down on it. Shield AI cofounder Brandon Tseng framed the concept at its October 2025 unveiling: “airpower without runways is the holy grail of deterrence.”
Landing A Fighter On Its Tail Is The Hard Part #
Vertical landing is where X-BAT’s design meets physics. The aircraft must descend tail-first on the thrust of its F110 engine and stay stable while its own wings catch the wind like a sail, a maneuver rockets have proven and fighter-sized drones have not.
Retired Gen. Mike Holmes, a former head of Air Combat Command, told Air & Space Forces Magazine the rocket-landing technology is understood, if not routine. Heather Penney of the Mitchell Institute for Aerospace Studies was blunter about the aerodynamics: those wings act as a sail on the way down, and “it’ll be an amazing technical feat” if X-BAT can land reliably in winds above a couple of knots.
The propeller-driven V-BAT lands the same way at a fraction of the weight, and its recovery record cuts both ways. It aced the Coast Guard’s cutter trials in 2025 and flew more than 130 sorties over Ukraine through early 2026. It also severed a Romanian naval officer’s fingers during a landing this June, and a 2024 landing mishap in front of Navy officers in Fort Stockton, Texas ended with a serious injury, Forbes reported. X-BAT weighs far more and comes down on a jet’s thrust instead of a propeller’s.
The Air Force Has Bigger Bills To Pay #
The Air Force is not investing in X-BAT. Its Collaborative Combat Aircraft finalists all use runways. Analysts say the service, already paying for the F-47, B-21, and Sentinel, would rather let the Navy fund the experiment and buy the result later if it works.
“I’m not surprised that the Navy moved first,” Holmes said. Destroyers and amphibious ships already carry fuel and munitions, so X-BAT’s logistics come nearly free at sea. The Air Force version of the concept, popping jets up from hidden trailers scattered across the Philippines, would need fuel and weapons trucked to every one of those sites.
Penney’s advice to the service was to let the Navy and DIU pay, then buy the aircraft off the shelf if it proves the business case. The Air Force already has two runway-dependent CCA prototypes in flight test and the F-47 and B-21 programs consuming its budget. Shield AI puts X-BAT’s price at about $27 million, close to the Air Force’s CCA target, though that figure belongs to an aircraft that has not flown yet (price targets are never more confident than before first flight). Penney also cautioned that full mission autonomy is harder than most people realize, which puts the load on Shield AI’s Hivemind, the software that just taught Taiwan’s attack drones to fly themselves.
DroneXL’s Take #
The Navy structured this the way experimental aviation should be funded: the service with the clearest use case pays to find out, and everyone else buys the answer. My worry is narrower than the concept. It is the last ten feet of every X-BAT landing.
The concession first: the V-BAT’s vertical recovery has passed real tests, including Coast Guard trials that went about as well as trials can go. But the V-BAT is a drone small enough that crews assist its landings by hand, and it has still drawn blood twice since 2024. Scale the same maneuver up to a fighter-sized jet balancing on an afterburning F110, on a moving deck, in sea wind, and Penney’s two-knot line stops sounding like snark and starts sounding like the test plan.
What I like about this structure is that nobody is pretending otherwise. DIU money exists to buy answers to exactly this kind of question cheaply, before a program of record inherits the risk. If X-BAT lands the way SpaceX boosters land, the Navy gets a reusable strike arm for 70-plus destroyers at $27 million apiece, if the price holds. If it cannot, $50 million is a cheap way to learn that, and Mach’s slower, gentler Atlas is standing right there.
Shield AI promised a first flight before the end of 2026 when it unveiled X-BAT last October. That date is now the whole story. A jet that flies in the next four months keeps the schedule credibility the price target depends on. A slip would not kill the program, but it would tell you how far the render still is from the rocket.
Sources: Air & Space Forces Magazine, US Navy, Defense Innovation Unit DroneXL uses automated tools to support research and source retrieval. All reporting and editorial perspectives are by Haye Kesteloo.