GMI Cloud, a Mountain View-based NVIDIA Cloud partner and GPU-as-a-service provider, has announced plans to open a $500 million AI factory in Taiwan to serve as a central hub for the region’s AI infrastructure. Designed to help enterprises train and deploy large-scale AI models, the new data center is designed to expand the region’s computing capacity.
According to the announcement on Monday, the AI factory will be powered by 7,000 NVIDIA Blackwell Ultra GPUs installed across 96 GB300 NVL72 high-density racks, a configuration expected to deliver processing speeds of up to 2 million tokens per second. Built on NVIDIA’s Blackwell architecture, the system will incorporate NVIDIA NVLink, NVIDIA Quantum InfiniBand, NVIDIA Spectrum-X Ethernet networking and NVIDIA BlueField DPUs to create a fully integrated, high-performance supercomputing environment.
GMI Cloud CEO Alex Yeh described the factory as “the blueprint for the heart of Asia’s AI future,” emphasizing its role in helping turn advanced AI ambitions into practical, scalable solutions. By pairing GMI Cloud’s infrastructure solutions and regional expertise with NVIDIA’s architecture, the initiative is meant to support next-generation AI development across the region.
