Why would military purposes preclude multiple parties having artificial intelligence? It seems you are assuming that if anyone achieves superintelligent machines, they will have a decisive enough advantage to prevent anyone else from having the technology. But if they are achieved incrementally, that need not be so.