Taipei 17 Nov 2025: In what is shaping up to be one of the most significant moves of the current global AI infrastructure race, U.S.-based GPU-as-a-service specialist GMI Cloud today announced it will invest USD 500 million in a new AI “factory” facility in Taiwan, built in collaboration with NVIDIA.
Enterprises in Asia will have a compelling regional alternative to U.S. or European-based infrastructure, with lower latency, local support, and possibly more favourable regulatory positioning.
The facility is set to become operational by March 2026, the project is positioned to be a cornerstone of Asia’s AI ecosystem, addressing demand for high-capacity compute, regional data sovereignty, and the next phase of generative-AI scale infrastructure.
What the facility will be
- The facility will house roughly 7,000 NVIDIA Blackwell-architecture GPUs across 96 high-density racks.
- It will draw about 16 megawatts of power and be capable of processing nearly 2 million tokens per second.
- It is explicitly branded as an “AI Factory”, a term increasingly used to denote large-scale compute campuses optimised for training, fine-tuning and inference at industrial scale.
- Early anchor customers are already named: among them ,Trend Micro (cybersecurity), Wistron (electronics manufacturing), VAST Data (storage infra), TECO Electric & Machinery (industrial solutions) and NVIDIA itself.
“AI factories are where intelligence is produced, turning data into insight and innovation for the future,” said Raymond Teh, Senior Vice President of Asia Pacific at NVIDIA. “GMI Cloud’s AI factory infrastructure will help continue the region’s leadership in AI infrastructure and innovation.”
This announcement of a new AI factory comes at the intersection of several major trends: AI compute demand, Large language models, multimodal AI, and real-time inference pipelines. Enterprises are driving demand for massive GPU clusters. GMI Cloud says its GPU utilisation is “almost full”.
NVIDIA continues to position itself not just as a GPU vendor, but as a platform company enabling global “AI factories”. Its Blackwell architecture, NVLink, quantum InfiniBand networking and BlueField DPUs are all called out in the blueprint for this facility.
The GMI Cloud / NVIDIA Taiwan AI-Factory represents a bold bet large-scale, regionally anchored, next-generation AI-compute infrastructure built for the new wave of generative and multimodal AI workloads. If executed successfully, it may become a blueprint for how Asia builds sovereign AI capability. For investors, technologists and regional policymakers alike, this is a landmark announcement signalling the next phase of the AI infrastructure boom.
Also Read: U.S. Invests $8.9 Billion in Intel: A Bold Move to Reinvent Semiconductor Supremacy
