Textron and Merlin Want the Cargo Plane to Fly Itself

On the morning of 17 July, a Cessna 208B Grand Caravan lined up on Runway 27 at Oshkosh, Wisconsin. Nobody was flying it.

Air traffic control asked once, half-skeptical. "3 Charlie Bravo, are you the autonomous Caravan?"

"Affirm."

Merlin Pilot, autonomy software sitting where the second pilot would normally sit, flew the approach and put the aircraft on the runway at 10:02 a.m. Central. The safety pilot watched from the right seat. His account afterward was almost boring: the software landed the way he would have.

Boring is the point. A demo proves the code works on a good day. A routine landing in front of 600,000 people proves something harder — that autonomy can be trusted in public, on a certified airframe, with a human watching and choosing not to intervene.

One company builds the airframe. The other ships the pilot as software. The concept is the clearest test yet of whether autonomy sells as a kit rather than as a new aircraft.

A Gap Between a Drone and a C-130

Military logistics runs on a barbell. Small quadcopters and one-way attack drones handle the last mile. Large turboprops and jets like the C-130 move pallets by the tonne across continents.

The middle is thin.

Not every resupply run needs a four-engine airlifter, and the aircraft that fly the middle distance are shrinking. The two companies frame the SkyCourier UX as the missing rung: a medium-payload, fixed-wing aircraft that can take off and land with no crew, fly a short runway, and carry enough to sustain a dispersed unit for a day rather than an hour.

The wording in the companies' announcement is careful. "Military planners are increasingly focused on how to sustain forces across distributed and contested operational environments," said Bob Gibbs, vice president of special missions sales at Textron Aviation. When routes are constrained or unavailable, the argument goes, the aircraft that survives the trip is the one that needs no runway infrastructure and no pilot to risk.

That framing is where the money is. Distributed operations assume small units spread across a wide area, each needing food, fuel, batteries, and spare parts on a schedule nobody can guarantee. A tactical drone drops a few kilograms. A C-130 needs a prepared strip and a crew of three. The SkyCourier sits between them.

Certification is what makes the SkyCourier interesting to a software company. The software vendor does not have to certify a new aircraft. It has to convince regulators that its software can fly one that already has a type certificate, a maintenance program, and thousands of flight hours behind it.

Capability comparison. Textron Aviation and Merlin materials, 2026

What Merlin Actually Sells

Merlin is not an aircraft company. Founded in Boston in 2017, it writes software that translates pilot skills into certified autonomy — holding a flight path, avoiding obstacles, separating from traffic, and talking to air traffic control.

The company describes the product as a pilot, not an autopilot. That distinction matters to regulators. An autopilot follows a commanded path. A pilot is expected to understand the mission, adapt when the plan breaks, and make a decision when the radio goes quiet. Its pitch is that the software manages the whole flight, on any aircraft it is installed in.

2026 turned that pitch into a track record. The company began trading publicly in March. In April it raised $80 million through a private investment in public equity, or PIPE. In June it completed a critical design review for autonomy on the C-130J with United States Special Operations Command. In July it landed the Caravan at Oshkosh, then signed an agreement with Israel Aerospace Industries (IAI) to push the same software into commercial cargo aircraft. In September it appeared on Textron's stand.

Military planners are increasingly focused on how to sustain forces across distributed and contested operational environments.— Bob Gibbs, vice president of special missions sales, Textron Aviation

MERLIN — FROM CARAVAN TO SKYCOURIER
─────────────────────────────────────────────────────────────
  2017 ──── Mar 2026 ──── Apr 2026 ──── Jun 2026 ──── ◉ Sep 2026
  🔬        🏭            💰            🧪            🏭
  Founded   Public        $80M PIPE     C-130J         SkyCourier
  Boston    listing       raised        design review  UX concept

That sequence is the actual asset. The company has accumulated the two things autonomy companies struggle to buy: flight hours on certified airframes, and a public currency to fund the certification work that has no shortcut.

Contested Logistics Is the Test Case

Autonomy has an easier time in the air than on the ground. Airspace is already instrumented, air traffic control already speaks in procedures, and a fixed-wing aircraft at altitude faces fewer surprises than a truck on an unmarked road.

That is why the first credible autonomy customers are logistics commands, not shooters. A resupply mission has a destination, a payload, and a schedule. It does not require a machine to decide who to engage. It requires a machine to fly a route it has rehearsed, and to abort when the route closes.

The Oshkosh landing is the proof point that matters here. The approach was flown in a normal airspace environment, with smoke reducing visibility, and the software handled it to touchdown. The company later described the pilot's role as oversight rather than control.

The military version adds a harder requirement. A contested resupply aircraft has to fly when satellite navigation is degraded, when the runway is a stretch of road, and when the crew is not there to take over. The value proposition is that the software becomes more capable than a fatigued pilot precisely in the conditions where a pilot is most needed.

Whether commanders believe that is a procurement question. The SkyCourier UX is a concept, not a contract. The two companies have shown a floor model and a thesis. They have not announced an order, a price, or a certification timeline.

The Retrofit Math

New military aircraft programs take a decade and cost billions before the first unit flies. Autonomy kits promise a different curve. The airframe already exists, the factory already runs, and the engineering risk moves from metal to software.

The airframer has the factory. The SkyCourier is in production, supported by a commercial freight order and a special-missions catalog that already includes an in-flight operable door for airdrop. Every variant shares the same wing, the same gear, and the same avionics baseline. Adding an autonomy option to an existing type is cheaper than designing a clean-sheet cargo drone large enough to matter.

The funding tells the same story from the other side. The company raised $80 million in April to fund program execution, on top of earlier venture rounds, and went public before it had a production contract. That is a bet that certification milestones are worth more than revenue in the early years of a category.

For an investor, the retrofit model has an obvious attraction and an obvious flaw. The attraction is that every certified airframe in the world becomes a potential customer without a new factory. The flaw is that the software has to be certified aircraft by aircraft, and each certification is slow, expensive, and specific.

Certification Is the Bottleneck

Flight software in civil aviation is held to a standard that would make most technology companies quit. Every line of safety-critical code needs a traceable requirement, a documented test, and a reviewer who signs their name to it. Adding autonomy does not replace that process. It adds to it.

Aviation safety rules were written for a crew. Every function a pilot performs — see, decide, act — has to be proven safe before it can be delegated to software, and each proof is specific to the aircraft and the mission. Its advantage is time in the air on airframes the regulator already knows.

Certification is also where the liability sits. If a pilotless SkyCourier lands short, the question of who is responsible — the airframer, the software vendor, the operator, or the commander who ordered the flight — has no settled answer. Insurance markets dislike that ambiguity. Procurement lawyers dislike it more.

The counterargument is that autonomy rarely arrives as a single decision. It arrives as an incremental permission: first a co-pilot that handles cruise, then one that lands, then one that flies with a human on the ground. Each step expands the envelope without requiring regulators to accept a fully pilotless future in one ruling. Merlin's roadmap follows that shape.

If that path holds, the SkyCourier UX is less a product than a placeholder for a permission that does not yet exist. The concept tells the customer what the aircraft would do the day the rules allow it. The rules are the schedule.

The Other Autonomy Bets

The two companies were not alone at AFA. Swarm Aero used the same conference to preview Gamera, a concept for massed long-range strike aircraft it hopes to fly in 2027 and to position for a future Air Force modular aircraft program. The two unveilings framed opposite bets on what autonomy is for: one removes the pilot from an aircraft that already exists, the other imagines an aircraft that only makes sense without one.

The gap between those bets is where the investment risk sits. Removing a pilot from a certified turboprop is a certification problem with a fixed, if long, path. Designing a new autonomous aircraft is a clean-sheet problem with no baseline — faster to imagine, slower to certify, and dependent on a procurement program that may not survive a budget cycle.

In between sit the ground and maritime analogues. As we wrote in September, Sea Machines won a five-year contract to supply drop-in autonomy kits for vessels — the same retrofit logic, on water instead of in the air. Ground vehicles followed the pattern earlier, with autonomy packages sold onto platforms that already exist. Aviation is simply the last domain to get the kit.

That is why the category has a name now: optionally crewed. The idea is an aircraft certified to fly with a pilot or without one, so the same airframe serves a commercial freight route today and an autonomous resupply mission later. The operator buys one aircraft and two ways to fly it. For an airframer, that is a larger addressable market with no new type certificate.

The retrofit model is quietly becoming the default path to autonomy for capital-intensive hardware. It is cheaper for the customer, faster to field, and easier to insure because the base platform has a service history. Its weakness is that value accrues to whoever owns the software layer, and airframers have historically treated software as a feature bundled with metal rather than a business with its own margin.

Pilot supply sharpens the argument. Airlines and air forces both face a shortage of experienced crews, and the shortage is worst in the unglamorous segments — night freight, remote strips, repetitive resupply runs. Those are exactly the missions where removing the pilot is least controversial and most valuable, because the alternative is not a heroic aviator but a delayed pallet.

That is the tension inside the Textron-Merlin concept. The airframer can offer autonomy as one more option on a catalog it already sells. The software maker needs autonomy to be the thing customers ask for by name. A concept vehicle serves both stories at once, which is why neither company has put a price, a launch customer, or a timeline against it.

For now the SkyCourier UX is a claim about a market that does not exist yet: a middle rung of military logistics, flown by software, bought as a kit. The claim is plausible. The hard work — certification, a launch customer, and a liability framework — is still ahead.

What to Track

The deeper bet is not about one turboprop. It is about whether autonomy follows the pattern of avionics — a certified component that any airframer can buy and any operator can trust — or the pattern of the engine, where the maker of the platform captures the value.

Textron and Merlin have placed opposing bets on the same aircraft. Textron wants autonomy to sell more airframes. Merlin wants the airframe to sell more autonomy. Both can be right for a while. Eventually one of them owns the customer relationship, and the SkyCourier UX is where that argument starts.