In a republished column, Xi Chunying argues that closed models may determine the ceiling of AI capability, open-weight models the pace of diffusion, and industrial applications where capital ultimately settles. The article uses Kimi K3 as a case study, citing its technical report for claims about model scale and capabilities, then explores three scenarios: closed-model oligopoly, long-term coexistence, and open-weight market dominance. Its forecasts about investment flows and profit migration are analytical scenarios rather than independently measured outcomes.