This week in AI, policy, models, robotics, and markets delivered a unified message: the race is moving toward distribution, embodiment, ownership, and economic returns. Jensen Huang backed an industry letter defending open-weight models, arguing American leadership requires an ecosystem where startups, universities, enterprises, and public institutions can operate advanced models. Moonshot released weights for Kimi K3, a mixture-of-experts model with native multimodality and a one-million-token context window, signaling open models as a parallel frontier paced by Chinese labs. Google DeepMind’s Gemini Robotics 2 extends AI from digital understanding to physical control, planning tasks, coordinating humanoid movement, and manipulating objects, suggesting the next race is turning reasoning into action. Markets provided a warning: Leopold Aschenbrenner’s Situational Awareness fund suffered a collapse and forced unwind after concentrated AI positions moved against it, showing a correct prediction can be financially fatal with misaligned concentration and leverage. Big Tech earnings offered a contrast. Microsoft was rewarded after Azure crossed $100 billion in annual revenue with expanding Copilot adoption. Amazon saw AWS growth accelerate alongside AI infrastructure investments. Meta received a harsher reaction as its infrastructure spending accelerated faster than market confidence, despite a strong core advertising business.