엔비디아 창립자 겸 CEO 젠슨 황(Jensen Huang)이 20일 대만 타이베이에서 열린 컴퓨텍스(COMPUTEX) 2025 기조연설에서 AI 인프라 산업에 대한 새로운 구상을 발표하며, “엔비디아는 이제 전 세계적인 AI 인프라 공급자다. 모든 기업은 자사 비즈니스를 운영하고 제품에 인텔리전스를 적용하기 위해 AI 팩토리를 구축하거나 임대하게 될 것이다. 물리 AI와 범용 로보틱스는 차세대 수조 달러 규모의 산업이며, 엔비디아는 이러한 범용 로보틱스 구축을 위한 AI 인프라를 제공한다”고 밝혔다.

▲Nvidia CEO Jensen Huang is giving a keynote speech at Computex 2025.
Unveiling massive investments and technological innovations for an AI-based future
Emphasizes strengthening cooperation with Taiwanese companies such as TSMC and Foxconn
“NVIDIA is now the global AI infrastructure provider. Every company will build or lease an AI factory to run their business and apply intelligence to their products. Physical AI and general-purpose robotics are the next trillion-dollar industries, and NVIDIA provides the AI infrastructure to build these general-purpose robotics.”
NVIDIA founder and CEO Jensen Huang announced new vision for the AI infrastructure industry in his keynote speech at COMPUTEX 2025 in Taipei, Taiwan on the 20th.
He emphasized that AI will become an essential infrastructure like electricity and the Internet, and that building AI factories is essential for this.
Jensen Huang explained in his keynote speech, “AI is not just a technology, it is a core infrastructure that will change the entire industry. Now, data centers should be called AI factories, not just a collection of servers. Here, energy is invested to create valuable data, which we call tokens.”
At this event, NVIDIA unveiled various solutions that support AI computing. In particular, NVLink Fusion maximizes the scalability and performance of data centers, enabling companies to build optimal AI infrastructure.
Additionally, DGX Spark and DGX Station are designed to give AI researchers the freedom to prototype, fine-tune, and perform inference from the desktop to the data center.
RTX PRO Server is an essential component for building AI factories, enabling companies to operate on-premise AI infrastructure.
Jensen Huang emphasized that “AI is developing at an accelerating pace, and now enterprises need more computing power.”
Jensen Huang emphasized that Taiwan plays a key role in building a global AI infrastructure. He announced that Nvidia is working with Foxconn and the Taiwanese government to build an AI factory supercomputer, which will support researchers and startups.
Additionally, Nvidia said that $5 trillion worth of global factories are using AI technology to increase productivity.
In particular, major manufacturers such as TSMC, Foxconn, and Gigabyte are developing next-generation industrial AI solutions using NVIDIA Omniverse.
NVIDIA is advancing agentic AI and physical AI for the future of AI. The Isaac GR00T-Dreams and GR00T N1.5 announced at this event are expected to help enhance the intelligence of humanoid robots.
Jensen Huang said, “AI is evolving beyond simply analyzing data to understanding the world and taking action. The next step is “general purpose robotics,” he said.
At the end of his keynote, Jensen Huang thanked NVIDIA's ecosystem partners and emphasized, "We are creating a completely new industry encompassing AI factories, AI agents, and robotics. This opportunity is once-in-a-lifetime and very special."