Shield AI’s Hivemind flies two FQ-44A Fury drones in simulation

On Oct. 26, two U.S. Air Force FQ-44A Fury Collaborative Combat Aircraft (CCAs), built by Anduril, flew in formation at the Nevada Test and Training Range. Shield AI’s Hivemind autonomy software controlled the formation and its beyond-visual-range counter-air tactics against a virtual threat.

Shield AI disclosed the sortie Nov. 9 and called it the first time two CCAs had flown together as a coordinated team.

No human pilot controlled either aircraft in the air. A ground operator supplied high-level direction, Air & Space Forces Magazine reported, while Hivemind handled the formation flight and tactical response. That division put the software in charge of aircraft-level coordination while a person retained mission-level intent.

“The difference here was, it was a full-on tactical air-to-air engagement,” said Todd Wesley, Shield AI’s vice president of Hivemind engineering.

What did the formation flight demonstrate?

The engagement used no live weapons. Shield AI plans to add more aircraft and more complex engagements before live-munitions tests, although it gave no date for those tests. Until then, the Oct. 26 result remains a simulated demonstration of coordinated flight, not evidence of live combat capability.

The Oct. 8 test at Edwards Air Force Base in California came first. An F-22 pilot acted as an airborne “quarterback,” issuing high-level directives that Hivemind converted into maneuvers and simulated combat tasks on a single FQ-44A.

Wesley said each flight feeds data into the next software release. The process, he said, is nearing the point where the Air Force could fly a sortie, update the software overnight and fly again the next day.

The Oct. 8 flight asked Hivemind to translate one pilot’s intent; the Oct. 26 sortie required two aircraft to preserve formation and share a counter-air task. That is the added software problem.

How many drones could one fighter control?

Air Force officials have discussed one fighter controlling two to five CCAs and, more recently, as many as eight. Those arrangements depend on autonomy software coordinating formations with little direction.

Those are control concepts discussed by Air Force officials, not flight-test results. The reported test included neither the discussed eight-aircraft configuration nor live weapons.

What changes at Increment 1?

The program’s production path now has three dated markers:

• June: The Air Force awarded the first CCA Increment 1 production contracts for Anduril’s FQ-44A and General Atomics’ FQ-42A.
• Summer 2027: The Air Force plans to select a single primary mission-autonomy software provider for Increment 1. Anduril, RTX’s Collins Aerospace and Shield AI are in the contest.
• Fiscal 2027 request: The program seeks $2.37 billion, including $996.5 million for procurement.

The $2.37 billion request covers the overall CCA program. It is not a stated Shield AI contract value, booked revenue or a guaranteed order for a specified number of FQ-44As.

A single-provider award would concentrate Increment 1 mission autonomy with one of the three competitors, linking Shield AI’s software cadence to the same program that selected the FQ-44A and FQ-42A airframes.

What do the market figures show?

Kratos, whose MQ-58 Valkyrie is another drone wingman, recently put that aircraft’s cost at $6 million to $8 million each, according to MLQ.ai coverage. The estimate applies to the MQ-58; the source gives no unit price for the FQ-44A or FQ-42A.

The MQ-58 is not among the two aircraft named in the June Increment 1 production contracts, so its $6 million-to-$8 million estimate cannot stand as a proxy for either CCA’s price.

Shield AI also agreed last week to buy radar maker Echodyne for $5.3 billion, according to the same coverage. The Oct. 26 report identifies only Hivemind software and does not say Echodyne hardware flew on either aircraft, so the $5.3 billion deal should not be read as hardware content in this test.

Shield AI still faces three measurable gaps: no date for live-munitions tests, no disclosed unit price for the FQ-44A, and no reported flight with the discussed eight-aircraft formation. Its next challenge is to turn a two-aircraft simulated engagement into repeatable releases, larger formations and eventually live-munitions tests.

The Air Force’s planned choice of one Increment 1 autonomy provider by summer 2027 will decide whether Hivemind takes that slot or leaves it to Anduril or Collins Aerospace.