FQ-42 Vengeance and FQ-44 Fury: the Air Force split the airframe from the brain

Six contractors. Two airframes. One unanswered question about who supplies the brain. On 17 June 2026 the US Air Force moved its Collaborative Combat Aircraft programme into full-scale manufacturing and, in the same release, declined to name a software winner. Over 150 aircraft by the end of the decade. At least 500 autonomous aircraft in service by 2032. Both airframes already hold production contracts. The autonomy decision does not exist yet.

That structure has a consequence for anyone underwriting defence hardware. The margin will not sit with whoever ships the first 150 tails. It will sit with whoever wins a software marketplace that has barely started running.

Two airframes that are not the same aircraft

The hardware award went to two companies on the same day. General Atomics Aeronautical Systems received the engineering, manufacturing and production contract for the FQ-42A. Anduril received the same for the FQ-44A. Both awards landed four months ahead of the original schedule, which is the detail procurement offices care about most: it means neither programme needed rescuing before it could be built.

Then came the names. On 14 September 2026, at the Air & Space Forces Association conference in National Harbor, Secretary of the Air Force Troy Meink gave the two aircraft popular names. FQ-42A became Vengeance. FQ-44A became Fury. The same speech carried a fleet-level number attached to them: at least 500 autonomous aircraft in service by 2032, with the CCAs expected to take on missions that crewed fighters fly today.

Naming an aircraft is procurement theatre, usually. Here it did real work. A service that intends to buy 150 of something has stopped calling it a demonstrator.

The two airframes come from genuinely different design lineages, which is the point of running a split award. Anduril's FQ-44 is a jet-powered uncrewed fighter developed from the Fury design, and the company says it flew multiple aircraft across dozens of sorties from different airfields, switched between two autonomy software suites mid-flight, and carried inert air-to-air munitions on test flights. General Atomics describes the FQ-42A as a purpose-built uncrewed fighter whose modular design is meant to take mission autonomy software quickly, and it says the aircraft went from contract award to first flight in fifteen months. The company also traces the airframe to the XQ-67A lineage, which is a different answer to the same question.

Programme record as published by the Department of the Air Force, the two primes, and Inside Defense. Cost per CCA tail was described as classified by programme official Chris Helfrich.

What the table cannot show is the price. Air Force officials have kept cost per tail classified, and neither prime has published a unit figure. For an investor, that absence is more informative than any number would be. A programme that publishes its unit cost invites unit-cost competition. A programme that hides it is buying something else, and the Air Force has been unusually direct that affordability drove the accelerated software awards.

The software market the Air Force refused to consolidate

Here is the unusual part. Alongside the airframe awards, the Air Force issued mission autonomy production contracts to six companies, and gave all six a six-year baseline vehicle. Anduril, General Atomics, Lockheed Martin, Northrop Grumman, RTX Collins Aerospace, Shield AI. Each one gets a framework for continuous competition and rapid software development. Nobody owns the programme's software outright.

Three of the six were then pulled forward into the first of two six-month competitive phases: Anduril with Lattice, Shield AI with Hivemind, and RTX Collins Aerospace with Sidekick. They were selected on schedule and cost proposals, per the service. After the first six months the Air Force evaluates progress, runs a second competitive period, and selects a single primary mission autonomy provider for Increment 1. That decision is planned for summer 2027.

Mission autonomy is the cornerstone of the CCA concept, and leveraging a competitive, multi-vendor environment ensures we capture the latest technology.— Troy Meink, Secretary of the Air Force

That quote is the acquisition strategy in one sentence. The aircraft is treated as a platform. The autonomy is treated as a market with a buyer who can switch suppliers. Most defence programmes buy an integrated system from a prime and then spend the rest of their lives paying for the integration. This one deliberately separated the two halves on 17 June 2026.

There is a second clock running. Trade reporting indicates Increment 2 work is already running, and the follow-on increment is expected to draw on what Increment 1 teaches. A two-phase competition ending in 2027 leaves roughly one cycle of real flight data before the follow-on requirement is set. Vendors that treated the baseline contract as a research grant will be behind vendors that treated it as a delivery problem.

The Navy is running a different aircraft under the same name

CCA is not one programme with one air vehicle. In early September 2026 the Naval Air Systems Command issued a request for information for its own Increment 1, carrier-based prototypes. Published requirements call for a weaponised, extended-range unmanned aircraft that can catapult from a Gerald R. Ford-class or Nimitz-class deck and recover with arrested landings, inside the deck, elevator and hangar bay dimensional limits those ships impose.

The interesting part of that RFI is the standards list. NAVAIR asked for Modular Open Systems Approach and Autonomy Government Reference Architecture compliance, plus Open Mission Systems and Universal Control Interface protocols for onboard mission software. Those are the same open-interface requirements the Air Force has been pushing across its autonomy programmes. The name is shared. The aircraft is not.

A carrier-based uncrewed fighter that launches off a moving deck, recovers onto a moving deck, and flies on a navigation and datalink budget shaped by a forward-deployed ship is a different engineering problem from an airbase asset with a runway. Sharing a programme name does not make the two vehicles interchangeable, and it does not mean the same autonomy stack flies both.

For suppliers, that is a second addressable market arriving early. The Navy's Increment 1 is still a request for information, and industry responses come before a contract. But the interface requirements are written down now, and the teams building against them will be the teams bidding later.

What the numbers will and will not tell you

Three numbers anchor this programme: six baseline autonomy vendors, three accelerated awards, one primary provider due by summer 2027. Around them sit two fleet goals — over 150 CCAs by the end of the decade, at least 500 autonomous aircraft by 2032 — and one deliberate blank, the classified cost per tail.

The blank matters more than it looks. Unit cost is the variable that decides whether affordable mass procurement stays affordable once attrition rates meet real aircraft, and no external observer can currently model it. Every published estimate of CCA economics is therefore an estimate of something else: production rate, learning curve, or sustainment assumptions. Buyers should treat any confident per-aircraft number in circulation as unsourced until the service releases it.

Anduril has described its own schedule as the fastest path from prototype to production for a fighter aircraft in more than fifty years: prototype award in April 2024, ground testing in April 2025, first flight in October 2025, production contract in June 2026. Two years from a standing start to a manufacturing line. General Atomics points to a different metric, fifteen months from contract award to first flight. Both claims are company-sourced, and both are checkable against the Air Force's own flight-test record, which credits independent evaluation at Edwards Air Force Base and operational assessment by the Experimental Operations Unit at Nellis.

What to watch next

There is a precedent worth remembering from the other end of the autonomy market. In September we covered a five-year defence IDIQ for drop-in maritime autonomy kits, where the supplier's pitch was software that fits hardware somebody else already owns. That contract shape is now visible on the Air Force side too, at a much larger price tag and with a public competition attached.

Sources