Microsoft issued a statement on July 24, 2026, titled "Open Weights and American AI Leadership," declaring that "the US should take the lead in developing open models to secure AI leadership." Major technology companies including Google, OpenAI, NVIDIA, and AMD endorsed the statement. The article notes that AI models are categorized as "closed models" or "open models," with the United States leading in closed models while Chinese companies advance in open models. For example, "Kimi K3," which appeared on July 17, 2026, outperformed GPT-5.6 Sol and Claude Fable 5 in some tests and is scheduled to be made an open model on July 27, 2026. A ranking by third-party organization "Artificial Analysis" shows Chinese open models achieving performance comparable to American closed models, while American open models have more modest performance. Microsoft argues that "American AI leadership will be defined not by 'one cutting-edge model,' but by 'a powerful, open ecosystem that permeates every sector,'" and that open models offer advantages such as "enabling the development and use of AI without paying exorbitant fees" and "enabling AI use that is not tied to a single provider." While acknowledging risks like being "difficult to control and track," Microsoft argues that "instead of banning open models, it is important to allow defenders to access models with equivalent performance."