In a highly anticipated keynote at the Taipei Music Center, NVIDIA founder and CEO Jensen Huang opened COMPUTEX 2025 to a packed house of over 4,000 attendees. His message was clear and powerful: we are entering a new era where artificial intelligence becomes the backbone of modern infrastructure. AI is now infrastructure, and this infrastructure, just like the internet, just like electricity, needs factories,” Huang declared. “These factories are essentially what we build today.
Jensen Huang challenged traditional concepts of data centers, reimagining them as AI factories powerful hubs where energy is transformed into intelligence. They’re not data centers of the past,” he explained. “These AI data centers, if you will, are improperly described. They are, in fact, AI factories. You apply energy to it, and it produces something incredibly valuable, and these things are called tokens.
With this bold redefinition, Huang painted a compelling vision of a world transformed by AI—impacting every country, every industry, and every company. As NVIDIA continues to push the boundaries of innovation, the stage is set for a global shift in how we think about computing, productivity, and the very infrastructure of our digital age.
NVIDIA Expands CUDA-X, Powers the Future of AI and Robotics
NVIDIA CEO Jensen Huang showcased the explosive growth and adoption of CUDA-X, the company’s comprehensive AI platform. Standing before a massive wall of partner logos, Huang highlighted how industries worldwide are leveraging CUDA-X for everything from next-gen wireless technology to quantum supercomputing. The larger the install base, the more developers want to create libraries, the more libraries, the more amazing things are done,” Huang said. “Better applications, more benefits to users.
NVIDIA Building 6G on AI
Jensen Huang emphasized how NVIDIA and its partners are pioneering the development of 6G using AI, setting the stage for faster, smarter global connectivity. He also revealed the company’s latest strides in accelerating quantum supercomputing, bringing science fiction closer to reality.
Jensen Huang also outlined the rapid evolution of AI itself, describing a progression from agentic AI systems that can understand, think, and act to physical AI that interacts with and understands the world, ultimately leading to general robotics.
This accelerating advancement, he explained, is fueling massive demand for high-performance computing. In response, NVIDIA is delivering powerful new technologies such as the Grace Blackwell NVL72 systems and advanced networking solutions, while enabling major AI infrastructure deployments from CoreWeave, Oracle, Microsoft, xAI, and others around the world.
These are gigantic factory investments, Huang said with a grin. The reason why people build factories is because you know you know the answer. The more you buy, the more you make.
Building AI for Taiwan
NVIDIA CEO Jensen Huang underscored Taiwan’s vital role in the global technology ecosystem, stating, “It all starts in Taiwan.” But beyond being a global AI powerhouse, Taiwan is also becoming a hub for homegrown innovation—thanks to a deepening collaboration between NVIDIA, Foxconn Hon Hai Technology Group, and the Taiwan government. Huang announced plans to build a state-of-the-art AI factory supercomputer in Taiwan, powered by NVIDIA Blackwell infrastructure. This powerful system will serve as a national resource for researchers, startups, and industries across Taiwan—including key partners like TSMC. “Having a world-class AI infrastructure here in Taiwan is really important, Jensen Huang emphasized. The initiative not only strengthens Taiwan’s position in the global AI landscape but also ensures it has the local tools and computing power to drive its own breakthroughs in science, technology, and industry.
NVIDIA Unveils NVLink Fusion
NVIDIA CEO Jensen Huang introduced NVLink Fusion, a groundbreaking architecture designed to give hyperscalers the freedom to build semi-custom compute solutions using NVIDIA’s powerful NVLink interconnect. This innovative technology is engineered to eliminate traditional data center bottlenecks, enabling unprecedented levels of scalability and flexibility for AI infrastructure. With NVLink Fusion, partners can design systems that are finely tuned to the unique demands of their AI workloads delivering greater performance, efficiency, and customization. “This incredible body of work now becomes flexible and open for anybody to integrate into, Jensen Huang said, emphasizing NVIDIA’s commitment to openness and collaboration as the AI era continues to accelerate.
Blackwell Everywhere
At the heart of the modern AI revolution is NVIDIA Blackwell—the powerful engine driving innovation across every layer of the AI ecosystem. Jensen Huang highlighted how Blackwell delivers “one architecture” that seamlessly spans the full spectrum of AI applications. From cloud AI to enterprise AI, from personal AI to edge AI, Blackwell provides a unified foundation that simplifies development, enhances performance, and accelerates deployment at every scale. With this cohesive approach, NVIDIA is empowering developers and organizations to build the future of AI on a platform that’s flexible, scalable, and ready for everything from data centers to devices.
DGX Spark and DGX Station Redefine Developer Supercomputing
NVIDIA is bringing the power of AI supercomputing directly to developers with the DGX Spark now in full production and set to be available within just a few weeks. Designed as a personal AI supercomputer, DGX Spark is being launched in collaboration with top-tier partners including ASUS, Dell, Gigabyte, Lenovo, and MSI, making it widely accessible for developers looking to build and train advanced models locally.
DGX Station, a powerhouse system delivering up to 20 petaflops of performance all from a standard wall socket. its capabilities, noting that it can run a 1 trillion-parameter model, essentially giving developers their own personal DGX supercomputer. Together, these systems mark a major step toward democratizing AI development, putting incredible performance in the hands of individuals and small teams.
NVIDIA RTX PRO Servers
NVIDIA CEO Jensen Huang unveiled the NVIDIA RTX PRO Servers, a new line of enterprise-grade systems purpose-built for the next generation of agentic AI. Now in volume production, these servers are part of the newly validated NVIDIA Enterprise AI Factory design, providing businesses with a powerful, scalable platform for AI, design, engineering, and enterprise workloads. With universal acceleration capabilities, RTX PRO Servers enable NVIDIA partners to build and operate their own on-premises AI factories, bringing advanced AI performance and flexibility directly into enterprise environments. This new offering underscores NVIDIA’s commitment to delivering the tools organizations need to harness AI wherever they operate whether in the cloud or on-site.
NVIDIA AI Data Platform
As AI continues to reshape the computing landscape, it also demands a new approach to storage. At COMPUTEX 2025, NVIDIA CEO Jensen Huang emphasized that “the compute platform is different, so the storage platform for modern AI is different.” To meet these evolving needs, Huang introduced the latest advancements in intelligent storage infrastructure, powered by the NVIDIA AI Data Platform reference design. This next-generation platform integrates seamlessly with NVIDIA RTX 6000 PRO Blackwell Server Edition GPUs, enabling high-performance, scalable storage solutions optimized for modern AI workloads. Huang also spotlighted NVIDIA’s growing ecosystem of partners who are building these cutting-edge systems, designed to keep pace with the massive data demands of today’s AI-driven innovation.
Physical AI
NVIDIA is pushing the boundaries of robotics with its vision for Physical AI—intelligent agents that go beyond digital interactions to perceive, understand, and act in the real world. “Agents are essentially digital robots,” said NVIDIA CEO Jensen Huang. “They can perceive, understand, and plan.” To accelerate the development of physical robots, Huang emphasized the importance of training in highly accurate simulated environments. To that end, NVIDIA has partnered with DeepMind and Disney to create Newton, the world’s most advanced physics training engine for robotics. Newton allows developers to simulate realistic physical environments with unmatched precision, enabling robots to learn complex tasks before they ever touch the real world. This marks a significant leap toward making intelligent, capable robotics a practical reality.
NVIDIA Unveils Isaac GR00T Tools
Jensen Huang introduced groundbreaking tools to accelerate the development of humanoid robots, marking a major leap forward in the field of robotics. Central to this advancement is the Isaac GR00T-Dreams blueprint, designed to generate high-quality synthetic training data essential for teaching robots to understand and interact with the world. Complementing this is the Isaac GR00T N1.5 Humanoid Robot Foundation Model, a powerful AI model that will serve as the core intelligence for future humanoid machines. Together, these innovations form a comprehensive platform to train, simulate, and deploy humanoid robots more efficiently than ever before—bringing the dream of intelligent, human-like robotics closer to reality.
Powering the Factories of the Future with NVIDIA Omniverse
As industries around the world invest over $5 trillion in next-generation factories, optimizing their design and operation has become mission-critical. At COMPUTEX 2025, NVIDIA CEO Jensen Huang highlighted how industrial physical AI is transforming manufacturing by combining simulation, automation, and AI. Leading the charge are Taiwan’s top manufacturers including TSMC, Foxconn, Wistron, Pegatron, Delta Electronics, Quanta, GIGABYTE, and others who are leveraging the NVIDIA Omniverse platform to create digital twins of their production environments. These virtual replicas enable real-time collaboration, advanced simulations, and continuous optimization, helping manufacturers boost efficiency, reduce costs, and accelerate innovation. With Omniverse at the core, these companies are shaping the future of semiconductor and electronics manufacturing through intelligent, AI-driven industrial design.
A New Chapter in Taiwan’s AI Journey
Closing his COMPUTEX 2025 keynote with a dramatic flourish, NVIDIA CEO Jensen Huang unveiled NVIDIA Constellation the company’s brand new Taiwan office, revealed through a cinematic video that showed NVIDIA’s Santa Clara headquarters launching into space and landing in Taiwan. The announcement highlights the company’s deepening commitment to its growing workforce in Taiwan, a region at the heart of the global technology ecosystem.
Huang expressed profound appreciation for Taiwan’s leading role in advancing AI and technology. “The work Taiwanese companies are doing has changed the world,” he said, extending gratitude to NVIDIA’s ecosystem partners. He described the opportunities ahead as “extraordinary” and “once in a lifetime,” as the industry stands at the threshold of a new era. “We are in fact creating a whole new industry to support AI factories, AI agents, and robotics with one architecture,” Huang concluded, reinforcing NVIDIA’s vision for a unified, AI-driven future.
