In January 2026, the Department of Defense released its AI Acceleration Strategy with a blunt goal: make the U.S. military “AI-first.” The strategy’s most concrete mechanism is seven “pace-setting projects” — outcome-oriented vehicles meant to rapidly build AI infrastructure, data, models, policies, and talent. Here’s what each one is.
Warfighting
Swarm Forge — a competitive mechanism pitting U.S. forces against elite tech innovators on human-machine teaming.
Agent Network — development and experimentation with AI agents for battle management and decision support.
Ender’s Foundry — AI-enabled simulation, with tight feedback loops between simulation development and simulation operations.
Intelligence
Open Arsenal — accelerating the pipeline from technical intelligence (TechINT) to fielded capability.
Project Grant — moving deterrence from “static postures” to “dynamic pressure with interpretable results.”
Enterprise
GenAI.mil — the big one: democratizing AI experimentation across the department at all classification levels. Launched December 2025, built in 74 days, now near 2 million users across Gemini, ChatGPT, and Grok, certified for CUI at IL5.
The strategy behind the projects matters as much as the projects: incentivize internal experimentation, kill bureaucratic obstacles to model integration, and concentrate investment where the U.S. holds asymmetric advantages — computing, model innovation, operational data. A May 2026 move cleared seven providers for classified networks, making multi-vendor competition explicit policy and single-vendor dependency an officially recognized risk.
For anyone selling AI to DoD — or assessing it — these seven projects are where the requirements, the money, and the scrutiny are going.