Who Flies, Who Commands, Who Decides

The Air Force treats "mission autonomy" and "flight autonomy" as two distinct layers of software, and the distinction explains why three companies can all claim a role on the same aircraft without competing on the same product. Flight autonomy software handles "basic or preplanned operations, like taking off, sticking to a flight path, or landing," per Air and Space Forces Magazine. Mission autonomy is the layer above it: software that lets a CCA "take complex action when given basic directions from a pilot in a manned aircraft." Shield AI's Hivemind sits in that second category. It was built, in the company's words, to reroute around no-fly zones, avoid or engage obstacles and respond to unexpected conditions without a human in the loop for each decision.

Shield AI was named a mission autonomy provider for CCA on 13 February 2026, after a competitive evaluation of Technology Maturity and Risk Reduction efforts, and its Hivemind software was integrated onto Anduril's own Fury airframe for system-level testing ahead of flight demonstrations. Shield AI describes Hivemind as compliant with the Air Force's Autonomy Government Reference Architecture and platform-agnostic, meaning the same software is designed to run on aircraft it did not build. That design choice lets Shield AI sit inside Anduril's own CCA programme as a supplier while the two companies separately compete for the single mission autonomy contract that follows.

CCA Increment 1's mission autonomy programme ran quietly for two years before it was named. Five companies were reported working on it as early as 2024, and by September 2025 Aviation Week had identified Collins Aerospace and Shield AI as the two selected to move forward, though no official confirmation followed. The Air Force confirmed both companies publicly on 12 and 13 February 2026, pairing Collins with General Atomics on the FQ-42 and Shield AI with Anduril on the FQ-44. On 17 June 2026, the Air Force widened the field: Anduril, Shield AI and Collins Aerospace were named as the three companies moving into the next phase of the CCA Increment 1 mission autonomy programme, with the service evaluating all three over the following year before selecting a single primary provider in 2027.

That same day carried the airframe decision. Anduril and General Atomics were awarded engineering, manufacturing and production contracts covering the first three lots of CCA Increment 1, arriving four months ahead of schedule and sufficient to field at least 150 aircraft by the end of the decade. The 14 September naming ceremony that opens this brief, Vengeance for the FQ-42 and Fury for the FQ-44, formalised that June decision rather than changing it.

The three companies carry valuations and revenue figures that price three different bets, and the figures are not directly comparable in the way a single "who's ahead" ranking implies. Shield AI's valuation rose from $5.3 billion in March 2025 to $12.7 billion in a March 2026 Series G of $1.5 billion in equity plus $500 million in fixed-return preferred financing, led by Advent International and JPMorganChase's Strategic Investment Group. Shield AI does not publicly disclose revenue. Anduril raised a $5 billion Series H in May 2026 at a $61 billion valuation, led by Thrive Capital and Andreessen Horowitz. The research firm Sacra estimates its 2025 revenue at $2.2 billion, up from an estimated $1 billion in 2024; Anduril itself does not disclose revenue. Palantir is the only one of the three with audited, regulator-filed numbers: $4.475 billion in FY2025 revenue, up 56% year over year, with $1.625 billion of GAAP net income at a 36% margin, and a market capitalisation of approximately $427 billion as of 18 September 2026.

Headline contract figures need the same care as valuations. Anduril's March 2026 Army enterprise contract carries a widely reported $20 billion figure, but Anduril's own president told reporters "there's no money attached to it, this is just a contract vehicle": it is a ten-year ordering ceiling consolidating more than 120 prior procurement actions, and its first actual task order, a Lattice-based counter-drone command and control award, was worth $87 million. The Army had set up a similar enterprise vehicle for Palantir the previous August, with a $10 billion ceiling. Neither ceiling is revenue; both describe how much a customer could spend, not what it has.

Maven's path to programme of record status underlines how separate its function is from the other two companies. On 9 March 2026, Deputy Secretary of War Steve Feinberg issued a letter designating Maven Smart System a Pentagon program of record, intended to equip warfighters with tools to "detect, deter, and dominate" adversaries across operational domains, per GovConWire's account. The letter also directed that oversight of the programme move from the National Geospatial-Intelligence Agency to the Department of War's Chief Digital and Artificial Intelligence Office within 30 days, with the US Army taking over future contracting actions with Palantir. Maven itself processes satellite, radar, drone, sensor and intelligence-report data to identify potential threats and targets for a human decision-maker; it is not named in any of the CCA programme reporting reviewed for this brief, and Palantir builds no aircraft for it to fly.