Jung Yeon-tae’s Professional GPT Revolution 8 | NVIDIA’s Vision for the AI Industrial Revolution

NVIDIA's recent GTC event in Berlin detailed a shift in the AI industry, moving beyond large language model-based generative AI toward "physical AI" capable of interacting with the real world. CEO Jensen Huang suggested that combining Europe's manufacturing strengths in precision machinery and automobiles with AI could generate new industrial competitiveness. This evolution means robotics technology is advancing to perceive surroundings, manipulate objects, and perform actual tasks, in addition to understanding human language.

A notable development discussed was the progression of humanoid robot design, moving from a focus on individual hardware to integrated AI development platforms. NVIDIA is expanding a framework that connects AI models for robotics, computing hardware, simulation, training data, and development environments. This approach aims to provide a more cohesive system for building and deploying advanced robotic capabilities, enabling AI to perform real-world tasks alongside humans.

For AI data centers, competition is increasingly influenced by deployment speed and operational efficiency, not just semiconductor performance. NVIDIA addressed this by introducing DSX POD, a modular system engineered for integrating and validating major data center components in factories before on-site installation. The company stated this system could reduce deployment time at a prepared site from around six months to as little as 14 days, though this is a company-stated performance target under specific conditions, not a verified average for all data centers.