
NVIDIA’s GTC keynote centers on the transition from traditional computing to an era of "AI factories" driven by generative AI and agentic systems. The presentation highlights the critical shift toward inference, where massive token generation now defines computing demand and revenue potential. Key technological advancements include the Vera Rubin architecture, which leverages extreme co-design and NVLink 72 to achieve a 35-fold increase in performance per watt. The integration of Grok’s deterministic data flow processors further optimizes low-latency token generation for agentic workflows. Additionally, the introduction of OpenCLAW provides an open-source framework for personal agents, mirroring the foundational impact of Linux and HTML. By vertically integrating hardware, software, and simulation platforms like Omniverse DSX, the infrastructure enables the rapid deployment of physical AI and robotics, fundamentally reinventing enterprise IT into a multi-trillion-dollar industry powered by autonomous, reasoning agents.
Sign in to continue reading, translating and more.
Continue