Shield AI’s Hivemind on Anduril’s Fury: The CCA Program Now Has a Split-Stack Architecture

The Air Force’s decision to pair Shield AI’s Hivemind autonomy software with Anduril’s Fury (YFQ-44A) aircraft for CCA flight demonstrations signals that the service is deliberately decoupling the autonomy layer from the airframe — a structural choice that reshapes how both companies compete and how the program’s eventual downselect will be scored.

This is not a routine teaming arrangement. Shield AI, which closed a $2B raise at a $12.7B valuation in March 2026, has built Hivemind specifically as a platform-agnostic autonomy stack — it has previously flown on F-16s and MQ-25s without GPS or datalinks. Anduril, rated DOMINANT in our coverage with $6.3B in total funding and YFQ-44A production now confirmed started at Arsenal-1 in Pickaway County, Ohio, brings the airframe and manufacturing capacity. The Air Force selecting Hivemind to run on Fury rather than Anduril’s own Lattice platform is the critical data point: it means the USAF is actively testing whether best-of-breed autonomy software can be disaggregated from the vehicle integrator, rather than accepting a single vendor’s vertically integrated stack. Lattice is fielded across Space Force surveillance networks, Army Robotic Combat Vehicle frameworks (DIU selection, April 2024), and now underpins an $87M JIATF-401 counter-drone C2 contract — but it was not chosen here.

For Anduril, this cuts two ways. The Fury airframe advancing to flight demonstration is a concrete milestone toward the USAF downselect that could yield a multi-billion-dollar program of record — and production has already begun at Arsenal-1, ahead of the originally planned Q2 2026 target. But Hivemind’s selection over Lattice for this demonstration introduces a competitive variable: if the Air Force standardizes on Hivemind as its CCA autonomy layer, Anduril’s software moat narrows on the air domain even as its manufacturing position strengthens. Northrop Grumman’s Project Lotus, revealed in October 2025 as a turbofan-powered CCA Increment 2 candidate, adds further pressure — the downselect field remains contested, and the Fury/Hivemind pairing is a demonstration, not a contract award.

BOTTOM LINE

Procurement officers and program analysts tracking the CCA downselect should treat the Fury/Hivemind pairing as evidence that the USAF is structurally open to mixed-vendor autonomy architectures — meaning neither Anduril’s Lattice nor Shield AI’s Hivemind has a locked position, and both companies’ CCA revenue cases remain contingent on downselect outcomes that are not yet determined.

Confidence: MODERATE — The teaming and demonstration selection are confirmed, but the downstream implications for the downselect scoring criteria and program-of-record conversion are not yet public, and Anduril’s private financials remain unaudited.

Source: https://twitter.com/shieldaitech/status/2022346426974347463

Product Portfolio — Anduril

Signal Activity — Anduril

Deal History — Anduril

Competitive Positioning — Anduril