Jensen Huang, Wu Zhao, and Li Yanhong presented differing perspectives on token economics in AI at recent industry events. At GTC 2026, NVIDIA's Jensen Huang introduced the concept of "Token Factory Economics," likening tokens to electricity with dynamic pricing based on demand and complexity. He emphasized that data centers are evolving into token production facilities, with NVIDIA's latest hardware capable of producing 700 million tokens per second. Meanwhile, Wu Zhao from Alibaba advocated for efficient token usage, comparing tokens to energy that should be used precisely and without waste. He criticized the trend of "token maxing," where companies inflate token consumption metrics without meaningful output. Li Yanhong of Baidu proposed a shift in focus from token consumption to Daily Active Agents (DAA), arguing that the true measure of AI success is the number of tasks completed by agents, not the volume of tokens consumed. This debate highlights the evolving landscape of token economics as AI continues to expand.