Summary of Huang Renxun's GTC Keynote Address

1. From chip companies to trillion-dollar "token factories", the economic restructuring of the reasoning era. Huang Renxun announced that the demand for computing has entered a "millionfold growth" stage, predicting that the computing power demand from now until 2027 will double from $500 billion to $1 trillion. He defined the concept of a "token factory", pointing out that future data centers will no longer be storage centers, but factories that produce intelligent tokens. NVIDIA emphasized that through extreme "collaborative design", its unit token cost has reached a world-class and unshakable level, forcing global CEOs to manage "token production rates" like managing assets. 2. Vera Rubin architecture integrates with Groq, decoupling reasoning to achieve 350x acceleration. The speech unveiled the next-generation architecture codenamed Vera Rubin. The system adopts a fully liquid-cooled design and introduces the Vera CPU optimized for intelligent tasks. The most groundbreaking breakthrough is the integration of Groq team's deterministic flow processor technology, achieving "decoupled reasoning" through Dynamo software: using Rubin's massive memory to process complex contexts and using Groq chips to rapidly generate tokens. This combination has achieved an astonishing 350-fold leap in token generation speed for gigawatt-level factories. 3. OpenClaw: The "Linux moment" of the AI era and the intelligent agent operating system. Huang Renxun defines the open-source project OpenClaw as the Linux of the AI era. This is an operating system-level framework that allows AI intelligent agents to autonomously call tools, execute code, and manage file systems. To address enterprise security pain points, NVIDIA has launched the NeMo Claw reference design, integrating the OpenShell privacy barrier, marking the software industry's comprehensive transformation from SaaS to AaaS. 4. The "ChatGPT moment" of physical AI: from autonomous driving to Disney Olaf. The speech showcased the explosion of physical AI, announcing that autonomous driving has entered a large-scale commercial stage. The NVIDIA RoboTaxi Ready platform has added partners such as BYD, Hyundai, and Geely, covering 18 million new vehicles annually, and has reached a deployment agreement with Uber. The climax was when the Disney robot Olaf appeared, demonstrating the physical adaptability evolved in the virtual world through Omniverse and Newton solvers. 5. Feynman architecture and space deployment: vertical integration towards the stars and the sea. Looking to the future, Huang Renxun revealed the next generation GPU architecture after Rubin, Feynman, which will be equipped with the LP 40 processor and Rosa CPU, supporting co-packaged optical technology. In addition, NVIDIA announced its entry into space computing and the development of the Vera Rubin Space-1 computer. Through the Omniverse DSX platform, the company is building a "digital twin" global design system, achieving vertical integration from the hardware layer to cosmic-level infrastructure.