The era of inference is driving economic restructuring. NVIDIA CEO Jensen Huang has revised his computing power demand forecast for 2026-2027 from $500 billion to $1 trillion and introduced the concept of a 'Token Factory.' He unveiled the next-generation architecture codenamed Vera Rubin, which integrates with Groq to achieve 'decoupled inference,' boosting token generation speed by 350x. NVIDIA positions its open-source project OpenClaw as the Linux of the AI era and launched the NemoClaw AI agent platform, marking the software industry's shift toward an AaaS (AI-as-a-Service) model. Additionally, the physical AI sector is experiencing explosive growth, with autonomous driving technology entering large-scale commercialization and Disney's Olaf robot making a debut demonstration. Huang also disclosed the next-gen GPU architecture Feynman and announced NVIDIA's entry into space computing to build a 'digital twin' design system.