Nvidia founder and CEO Jensen Huang opened Computex on Monday with a keynote address that delivered his vision for an AI-powered future.
His opener for the annual computing event, underway in Taipei, Taiwan, included announcements covering AI infrastructure, quantum computing, humanoid robotics and boosts to the AI ecosystem.
Nvidia, Foxconn, Taiwan Government Plan Massive AI Supercomputer
Nvidia and Foxconn Hon Hai Technology Group have announced an extension of their partnership with the Taiwan government to build an advanced AI factory supercomputer.
The facility would feature 10,000 Nvidia Blackwell GPUs and operated by Foxconn’s subsidiary, Big Innovation Company, an Nvidia Cloud Partner.
The supercomputer aims to support researchers, startups and industrial sectors. Taiwan’s National Science and Technology Council plans to use it to provide cloud computing resources to the local technology ecosystem. Semiconductor multinational TMSC researchers plan to use it to accelerate R&D with performance improvements orders of magnitude faster than previous systems.
The initiative aligns with Taiwan’s vision to create an AI-focused industrial ecosystem and transform into a “smart AI island,” supporting innovation across smart cities, electric vehicles and manufacturing sectors.
Nvidia Targets Humanoid Robotics Acceleration with Advanced AI Platforms and Foundation Models
Nvidia has unveiled a suite of technologies aimed at accelerating the development of humanoid robots, including the Nvidia Isaac GR00T N1.5 foundation model update, GR00T-Dreams synthetic data generation blueprint and Blackwell systems for robot development. Together, these aim to address critical challenges in robotics development by generating vast amounts of synthetic motion data and improving robots’ ability to adapt to new environments and tasks.
The GR00T-Dreams blueprint enables developers to generate entirely new synthetic motion data from a single image, reducing development time. Nvidia Research demonstrated this by creating the GR00T N1.5 update in 36 hours instead of months. Robotics companies, including Agility Robotics, Boston Dynamics, Foxlink and Neura Robotics, are already adopting these technologies.
Nvidia is also expanding its ecosystem with new simulation frameworks like Cosmos Reason and Cosmos Predict 2, along with universal Blackwell systems from major manufacturers, including Cisco, Dell and Lenovo, to support every stage of robot development.
“Physical AI and robotics will bring about the next industrial revolution,” said Huang.
“From AI brains for robots to simulated worlds to practice in or AI supercomputers for training foundation models, Nvidia provides building blocks for every stage of the robotics development journey.”
Nvidia Launches World’s Largest Quantum Research Supercomputer in Japan
Nvidia also announced the launch of the Global Research and Development Center for Business by Quantum-AI Technology, or G-QuAT, home to ABCI-Q, the world’s largest research supercomputer dedicated to quantum computing.
Developed by Japan’s National Institute of Advanced Industrial Science and Technology, the system features 2,020 Nvidia H100 GPUs interconnected by Nvidia’s Quantum-2 InfiniBand networking platform. It is integrated with Nvidia CUDA-Q, an open-source hybrid computing platform for quantum applications.
The ABCI-Q supercomputer integrates AI supercomputing capabilities with multiple quantum processors from Fujitsu, QuEra and OptQC. The quantum-GPU computing infrastructure aims to accelerate research in quantum error correction and applications development across healthcare, energy and finance sectors.
“Seamlessly coupling quantum hardware with AI supercomputing will accelerate realizing the promise of quantum computing for all,” said Nvidia senior director of computer-aided engineering, quantum and CUDA-X, Tim Costa.
“Nvidia’s collaboration with AIST will catalyze progress in areas like quantum error correction and applications development, crucial for building useful, accelerated quantum supercomputers.”
Other Nvidia Computex Announcements
-
Nvidia Launches NVLink Fusion for Custom AI Infrastructure Development
Nvidia’s new NVLink Fusion silicon aims to help industries build semi-custom AI infrastructure using Nvidia’s vast partner ecosystem. MediaTek, Marvell, Fujitsu and Qualcomm are among early adopters. The technology allows for the integration of custom silicon with Nvidia GPUs in rack-scale architectures, delivering up to 1.8 TB/s bandwidth per GPU—14x faster than PCIe Gen5.
Nvidia said it aims to address the “tectonic shift” in data center architecture as AI becomes integrated into every computing platform, providing hyperscalers with a path to scale out AI factories to millions of GPUs using any ASIC and Nvidia’s end-to-end networking platform.
-
Taiwan Manufacturers to Build Nvidia’s Personal AI Supercomputers
Taiwanese systems manufacturers Acer, Gigabyte and MSI will produce Nvidia DGX Spark and DGX Station systems, extending the availability of personal AI supercomputers that deliver up to 1 petaflop and 20 petaflops of AI compute, respectively.
These desktop systems, powered by Nvidia Grace Blackwell technology, enable developers, researchers and data scientists to prototype, fine-tune and run inference on AI models locally while maintaining the same software architecture as industrial-strength AI factories.
-
Nvidia Accelerates Enterprise IT Transition to AI Factories with RTX PRO Servers
Nvidia said it is supporting the IT infrastructure transition to enterprise AI factories with new RTX PRO Servers built with RTX PRO 6000 Blackwell Server Edition GPUs and a validated Enterprise AI Factory design.
Global partners, including Cisco, Dell, HPE and Lenovo, plan to offer full-stack solutions based on this design, which combines RTX PRO Servers with Nvidia networking, storage and AI Enterprise software. The acceleration platforms aim to enable enterprises to run different workloads, from multimodal AI inference and physical AI to design and scientific computing.
-
Nvidia Launches DGX Cloud Lepton Marketplace for Global AI Development
Nvidia has announced the DGX Cloud Lepton platform, a compute marketplace connecting AI developers with tens of thousands of GPUs from global cloud providers, including CoreWeave, Foxconn and SoftBank.
The platform targets the challenge of securing reliable GPU resources by unifying access to cloud AI services across the Nvidia compute ecosystem, while enabling developers to tap into region-specific GPU capacity for both on-demand and long-term computing needs.