The panel discusses the release of Kimi K3, an open-weight AI model from Chinese lab Moonshots AI, which has achieved top performance across multiple benchmarks, surpassing US frontier models despite export controls. This marks an 'AI Sputnik Moment' for the US, proving that frontier AI can be replicated at a fraction of the cost. The conversation explores how open-source models are disrupting the industry, with US export controls backfiring and spurring more efficient Chinese models. The panel notes that intelligence cannot be constrained, and the focus is shifting from benchmarks to practical use cases. They also discuss the rise of small, efficient models that can run on smartphones, the implications of AI hyperforecasters matching human prediction accuracy, and the urgent need for safety regulations on autonomous humanoid robots. The episode concludes with a debate on open versus closed source models and a call for companies to urgently evaluate Kimi K3 on their own data.