The podcast explores the escalating conflict over Chinese open-weight AI models and their potential restriction in the US. The White House is considering measures like adding firms to the entity list and requiring security guarantees, while the abrupt resignation of a key AI official highlights policy confusion. A controversial tweet from OpenAI strategist Dean Ball, who argued that open-weight models are 'decelerationist' and suggested creating regulatory risk around Chinese models, sparked backlash from figures like Epic CEO Tim Sweeney, who accused OpenAI of seeking a monopoly. Critics like David Sacks and Aaron Levie countered that gatekeeping is ineffective and that accelerating US AI progress is the real solution. The debate also examines whether Chinese models constitute 'dumping,' noting that China lacks the compute power to serve these models globally, as shown by Moonshot's Kimi K3 launch overwhelming their servers. The episode underscores that open-weight models are not free due to high inference costs, and the unresolved tension between regulating closed models and leaving open models unregulated will shape future AI costs, architecture, and market incentives.