At COMPUTEX 2025, NVIDIA CEO Jensen Huang emphasized the transformative impact of AI, likening it to essential infrastructure like electricity and the internet.
He described AI data centers as “AI factories,” producing valuable outputs termed “tokens.” Jensen Huang also highlighted the widespread adoption of NVIDIA’s CUDA-X platform across industries, the company’s initiatives in 6G development using AI, and advancements in accelerating quantum supercomputing.
NVIDIA and Foxconn Partner to Build AI Supercomputer in Taiwan
NVIDIA and Foxconn are collaborating with Taiwan’s government to develop an AI factory supercomputer powered by 10,000 NVIDIA Blackwell GPUs, aimed at advancing AI research and industry innovation across Taiwan. Managed by Foxconn’s Big Innovation Company as an NVIDIA Cloud Partner, the facility will provide AI computing resources to researchers, startups, and enterprises, fostering the growth of AI applications in sectors like semiconductors and robotics.
TSMC plans to leverage the AI factory for accelerated semiconductor R&D, achieving substantial performance gains over previous systems. Meanwhile, the Taiwan National Science and Technology Council aims to establish a robust AI ecosystem, integrating advanced infrastructure to propel smart city initiatives and AI-driven industries.
NVIDIA Empowers AI Development with DGX Spark and DGX Station Systems
NVIDIA has partnered with Taiwan’s top system manufacturers, including Acer, GIGABYTE, and MSI, to expand the availability of DGX Spark and DGX Station — AI supercomputers designed for desktops. Built on the NVIDIA Grace Blackwell platform, these systems provide robust AI compute power in a compact form factor, enabling developers, researchers, and enterprises to prototype, fine-tune, and deploy AI models with unprecedented efficiency.
DGX Spark, equipped with the GB10 Grace Blackwell Superchip, delivers 1 petaflop of AI compute and 128GB of unified memory, making it ideal for generative AI and data science workloads. Meanwhile, DGX Station, powered by the GB300 Ultra Superchip, offers 20 petaflops of AI performance and 784GB of memory, serving as a centralized compute node or multi-instance system.
Both systems are preconfigured with NVIDIA AI software and NIM microservices, facilitating seamless deployment to DGX Cloud or other accelerated infrastructures, driving AI innovation across industries.
NVIDIA Accelerates Enterprise AI Transformation with RTX PRO Servers
NVIDIA has launched RTX PRO Servers and the NVIDIA Enterprise AI Factory validated design to expedite the shift from CPU-centric IT infrastructure to AI-driven data centers. Powered by RTX PRO 6000 Blackwell GPUs, these servers deliver exceptional performance and energy efficiency across enterprise workloads, from AI and engineering simulations to digital AI agents.
The new infrastructure design integrates RTX PRO Servers, Spectrum-X Ethernet networking, BlueField DPUs, and NVIDIA AI Enterprise software, enabling partners to build AI factories that accelerate product development and business operations. Leading enterprises like Foxconn, Cadence, and Lilly are adopting NVIDIA’s AI Factory blueprint to optimize manufacturing, smart vehicles, and healthcare systems using AI-powered infrastructure.
Foxconn plans to deploy the AI Factory model to revolutionize semiconductor and electronics production with AI, aiming to create smarter, more efficient manufacturing processes. NVIDIA’s strategy underscores its commitment to advancing enterprise AI adoption, positioning itself as a catalyst for AI-driven business transformation.
NVIDIA Accelerates Humanoid Robot Development with Isaac GR00T Updates
NVIDIA has introduced NVIDIA Isaac GR00T N1.5, a customizable foundation model for humanoid reasoning, and GR00T-Dreams, a synthetic motion data generation blueprint, to accelerate humanoid robot training. With NVIDIA Blackwell systems powering these advancements, leading robotics developers like Agility Robotics, Boston Dynamics, and NEURA Robotics are adopting the Isaac platform for enhanced humanoid capabilities.
GR00T-Dreams, unveiled at Computex, leverages Cosmos Predict world foundation models (WFMs) to generate synthetic neural trajectories, enabling robots to learn new tasks using a single image input. This data generation blueprint complements the previously released GR00T-Mimic, which augments existing datasets using NVIDIA Omniverse and Cosmos platforms.
The updated GR00T N1.5 model, developed in just 36 hours using GR00T-Dreams, enhances robots’ ability to adapt to new environments and follow user instructions for complex tasks. Early adopters, including AeiRobot and Foxlink, are implementing these models to streamline industrial workflows and household automation, positioning NVIDIA as a leader in the physical AI revolution.
NVIDIA Advances Quantum-AI Research with G-QuAT and ABCI-Q Supercomputer
NVIDIA has launched the Global Research and Development Center for Business by Quantum-AI Technology (G-QuAT), housing ABCI-Q — the world’s largest quantum-focused supercomputer. Developed in collaboration with Japan’s AIST, ABCI-Q integrates 2,020 NVIDIA H100 GPUs connected through the Quantum-2 InfiniBand networking platform, marking a breakthrough in hybrid quantum-GPU computing.
ABCI-Q leverages NVIDIA’s CUDA-Q, an open-source platform for orchestrating large-scale quantum computing applications, enabling researchers to tackle complex challenges in sectors like healthcare, energy, and finance. The system combines superconducting qubits from Fujitsu, neutral atom processors from QuEra, and photonic processors from OptQC to execute hybrid quantum-AI workloads across diverse quantum modalities.
This initiative supports NVIDIA’s strategy to accelerate quantum computing development, focusing on quantum error correction and application development. By integrating advanced quantum hardware with AI supercomputing, NVIDIA aims to expedite practical quantum computing use cases, positioning itself as a leader in the next generation of computational innovation.
NVIDIA DGX Cloud Lepton: Revolutionizing AI Compute Access
NVIDIA has introduced DGX Cloud Lepton, a groundbreaking AI platform that connects developers with tens of thousands of GPUs through a global marketplace supported by major cloud providers like CoreWeave, Foxconn, and Softbank Corp. This platform addresses the growing demand for AI by offering both on-demand and long-term GPU compute capacity in specific regions, ensuring compliance with data sovereignty regulations and optimal latency for critical workloads.
Key benefits include improved productivity through a unified development and deployment experience, frictionless multi-cloud deployment, and predictable enterprise-grade performance. DGX Cloud Lepton also integrates with NVIDIA’s advanced software stack, streamlining AI application development.
Additionally, NVIDIA unveiled Exemplar Clouds to help partners enhance security and performance using NVIDIA’s benchmarking tools and operational frameworks. Yotta Data Services, the first Asia-Pacific partner, exemplifies this initiative, setting a new standard for AI cloud performance.
InfotechLead.com News Desk
