Shield AI says its Hivemind software followed directions from an F-22 pilot and coordinated multiple uncrewed aircraft in tests last month, according to Fortune. The report also highlights a growing gap between the speed of defence technology and the Pentagon’s lengthy acquisition process.
Shield AI Watch analysis
What happened
Fortune reports that Shield AI chief executive Gary Steele described the tests at an event with Tyler Sweatt, head of defence software company Second Front Systems. Steele said Shield expects to get its autonomous fighter aircraft from conception to first flight in less than two years. That is a company projection, not a reported flight result.
The contrast is striking: Fortune cites a July Government Accountability Office report that put the average time to deliver a capability across major Pentagon weapon programmes at more than 12 years. The article also reports that defence-tech investment reached $37.9 billion across 269 deals in the third quarter of 2026, citing PitchBook, and says Shield AI announced on 5 November that it was raising $2.5 billion to support its acquisition of radar company Echodyne in a transaction valued at $5.3 billion.
Why it matters
AI-enabled aircraft are moving through tests while the systems that buy and field military technology can take years to catch up. That does not mean a successful software test is the same as a deployable capability: procurement, integration and further testing still stand between a demonstration and use in the field.
The money is moving quickly, too. Fortune reports that investors put record sums into defence technology in the quarter, while startup leaders warned that slow military approvals could delay returns. The sector’s challenge is not just building new technology, but getting it adopted without mistaking a brisk development timeline for proof of operational readiness.
Our read
This is a useful glimpse of the pace problem in military AI: software may be tested on a much shorter cycle than conventional weapons programmes, but the hard question is what happens after the demonstration. Shield AI’s reported test and projected timeline are notable; neither settles how its systems will perform in operational conditions or how quickly the military could adopt them. The difference between a promising test and a fielded capability is where the paperwork, and quite a lot else, lives.
What to watch
- Whether Shield AI publishes more detail on the tests and the role Hivemind played.
- Whether its autonomous aircraft meets the company’s stated timeline for first flight.
- Whether Pentagon acquisition reforms shorten delivery times, or private investment simply piles up behind the same bottleneck.
Discussion spark: Should defence buyers accept faster, iterative AI-system development even when it means changing established testing and procurement processes, or should military adoption remain deliberately slow?
Sources and evidence
- Source update (18 November 2026, 01:11 UTC)
Independent WittyWires tracker for public updates about Shield AI. Not affiliated with or endorsed by Shield AI; this is not an official account.