The United States is losing the artificial intelligence race not by failing to develop superior models, but by neglecting to ensure its technology becomes the global platform. While frontier capability is important, temporary benchmark leads do not guarantee durable dominance. The true national edge lies in fostering an ecosystem where the world builds upon American AI. Administration discussions on voluntary frameworks for testing advanced models exclude open-weight models, which can be freely adapted. This decision, while technically sound for closed models, lacks a strategy for open models.China is actively releasing open-weight models, allowing developers to customize and build products on them, converting diffusion into market power. This strategy is leading to a significant portion of AI applications globally, including potentially in the U.S., being built on Chinese foundations. The race is not just for individual models but for the underlying technology stack. China's policy is focused on strengthening its ecosystem through model adoption, creating a feedback loop that locks in global influence.American closed-model companies prioritize intellectual property and revenue, which may not align with maximizing national economic power. While ethical concerns regarding open-weight models are valid, unilateral restraint by the U.S. would not stop China from releasing them. Instead, it would cede leadership in this crucial area. The U.S. possesses the resources to lead in both closed and open AI markets, but requires a strategy to promote trusted American technology abroad. The future of AI leadership depends on whose technology stack becomes indispensable to developers and the global economy.