The AI-2040 project outlines a strategy where the US and China cooperate to delay transformative AI development. This plan relies on a layered defense stack involving compute centralization in neutral countries and strict military oversight. It proposes governing weights and hardware to ensure safety. Practitioners should evaluate these specific verification and incentive structures for feasibility.