Japan is building its own defense AI, without waiting on the Americans

On September 4, 2026, Aviation Week's Aerospace Daily & Defense Report noted, among a dozen items that week, a partnership between Japan's NEC and Mitsubishi Heavy Industries (MHI) "to strengthen Japan's defense industrial base and develop sovereign data and AI capabilities." The announcement ran to a single sentence, with no disclosed budget or timeline — but it sits inside a broader pattern the same issue documented without naming it as such.

That same week, Hermeus tapped Anduril to provide the mission autonomy software for its Quarterhorse aircraft, Shield AI integrated GE Aerospace's AVEN thrust-vectoring nozzle into its X-Bat combat drone, and the US Navy advanced its collaborative combat aircraft (CCA) surrogate program. Three different programs, one common denominator: the software layer that now defines a combat aircraft — autonomy, sensor fusion, swarm coordination — is concentrated in a handful of American companies, Anduril and Shield AI foremost, in a market where Palantir holds an equivalent position on the data and command side.

It is this concentration, more than rising Chinese capability alone, that is pushing an ally like Japan to bring a capability in-house rather than simply buy it. Tokyo committed, as early as its December 2022 National Security Strategy, to raising defense spending to 2% of GDP by fiscal 2027; the NEC-MHI deal is one of the first visible translations of that effort into military AI specifically, rather than into further purchases of already-American platforms.

For the aerospace and defense clients Cardan-AI advises, the signal to watch is not this single deal but whether it replicates: if Europe, South Korea, or Gulf states follow the same "AI sovereignty" logic instead of buying off the shelf, the Western supply chain for autonomous defense systems could fragment into several non-interoperable software stacks — a trade-off Cardan-AI's full analysis unpacks below.