Former U.S. government official and current OpenAI Head of Strategy, Ball: Kimi's performance in proxy coding conversations is already comparable to the world's top-tier open models expected in Q1 2026. However, in practical use, this model has an extremely high token consumption rate, and its actual inference costs do not show significant competitive advantages. Ball believes that the reason Chinese tech companies choose to open up such high-performance model weights is 75% due to a lack of full strategic awareness regarding Artificial General Intelligence (AGI). The remaining 25% stems from the domestic shortage of computational power for large-scale inference, making China's open-weight strategy objectively a secondary byproduct of U.S. export controls. Since few in the international market are willing to pay high subscription fees for sub-frontier models from non-U.S. origins, open-sourcing has become the core strategy for lagging companies to secure ecosystem positioning. This accelerated release of capabilities is triggering commercial contrasts on the supply side of large models. The surge in open-source models inherently has a decelerating effect, as it commodifies frontier capabilities and hinders further capital expenditure (Capex) by global tech giants. Ball emphasizes that a future dominated by open-source models is pushing AI from being a competitive market product to a "public good" provided uniformly by nations as "digital public infrastructure," essentially moving toward digital nationalization at the technological level. As the U.S. 美国 responds to this trend, Ball predicts the Trump administration will adopt an asymmetrical regulatory intervention strategy. The U.S. government is unlikely to directly "ban open-source" through executive orders but may issue soft legal notices via regulatory bodies like the Federal Reserve (Fed), such as claiming that "Chinese large models may contain potential backdoors," thereby creating panic and compliance risks within global regulated markets. This tactic aims to manufacture uncertainty, forcing highly regulated overseas mainstream enterprises to proactively sever toolchain access to Chinese open-source models, thereby completing a defensive segmentation of computational infrastructure on both physical and legal fronts.