US Space Force and Air Force test AI in joint C2 experiment

MASH is a concrete example of AI agents leaving the lab and being wired directly into real military command‑and‑control workflows. Rather than a single ‘killer app’, the exercise used a portfolio of LLM‑powered tools, agentic data platforms and ML models co‑designed with operators to generate and evaluate courses of action across air, space, cyber, maritime and ground domains. That’s exactly the kind of complex, high‑stakes environment that drives demand for more capable planning, reasoning and tool‑use models—and it gives the U.S. defense establishment strong incentives to push the frontier for operational advantage. ([orbitaltoday.com](https://orbitaltoday.com/2026/07/03/u-s-space-force-and-air-force-use-ai-together-in-first-complex-command-and-control-experiment/))

For the race to AGI, this kind of human‑machine teaming experiment matters because militaries can absorb vast quantities of compute and tolerate higher per‑token costs if it produces decision superiority. That in turn supports the economics of ever‑larger and more specialized models, with defense budgets effectively cross‑subsidizing fundamental capability improvements that later spill into civilian applications. It also tightens the link between frontier AI and strategic stability: as AI becomes embedded in battle networks, the pressure to secure privileged access, deny adversaries similar tools, and harden systems against manipulation will only grow.