NVIDIA Redefines Itself as AI Infrastructure Architect at Computex 2025
TAIPEI — NVIDIA CEO Jensen Huang delivered a landmark keynote at Computex 2025, Asia’s premier tech expo, positioning the AI chip giant as a full-stack “AI infrastructure company” poised to redefine industrial and consumer innovation. The two-hour address outlined how NVIDIA’s ecosystem—spanning hardware, software, and robotics—is accelerating the transition to an “AI factory” era.
The presentation revolved around four pillars: advancements in consumer RTX GPUs, next-gen data center solutions like the GB300 Blackwell Ultra AI platform, industrial metaverse tools via Omniverse, and breakthroughs in autonomous systems. Huang emphasized NVIDIA’s newly unveiled Blackwell architecture, designed to supercharge AI training, inference, and “agentic” AI development. “Applying more computational power during inference isn’t just about speed—it’s about unlocking precision at scale,” he noted.
The Dawn of a Trillion-Dollar Robotics Revolution
Huang declared physical AI and robotics the “next industrial revolution,” with humanoid systems representing a trillion-dollar frontier. NVIDIA’s Isaac robotics platform—powered by its Jetson Thor processors—is central to this vision. The company also introduced Isaac GR00T N1.5, a customizable foundation model for humanoid robots, alongside simulation tools like GR00T-Dreams and the Cosmos Predict world model to streamline development.
“From AI brains for robots to digital twins for training, we’re building every layer of this ecosystem,” Huang asserted. Industry heavyweights like Agility Robotics, Boston Dynamics, and Xiaomi-backed RoboTech are already adopting NVIDIA’s stack, signaling a race to commercialize multipurpose humanoids.
Market Forces: From Labs to Factory Floors
Morgan Stanley’s recent “Humanoid Robotics 100” report underscores the sector’s momentum, forecasting a $60 trillion global market by 2050. China alone could see 59 million humanoids deployed, per the analysis. This aligns with NVIDIA’s bets on “physical AI” driving demand for its Blackwell systems and simulations.
Meanwhile, companies like WIMI Hologram (WIMI) are leveraging AI vision and natural language processing to enhance robotic adaptability. By refining motion control algorithms, WIMI aims to bridge the gap between lab prototypes and real-world industrial applications.
The Great Pivot: Tokens, Agents, and Robots
Huang’s repeated emphasis on “agentic AI,” “models,” and “robots” signals NVIDIA’s strategic trifecta. As generative AI matures, the company is betting that the next frontier lies in embodied intelligence—systems that translate digital smarts into physical action. With rivals racing to claim their stake, NVIDIA’s infrastructure-first approach could determine who owns the architecture of tomorrow’s smart factories and cities.
One thing is clear: the neural engines powering this transformation are being built today—and NVIDIA intends to supply every gear in the machine.
Original article, Author: Tobias. If you wish to reprint this article, please indicate the source:https://aicnbc.com/971.html
