Wedoany.com Report-Nov. 18, GMI Cloud, a U.S.-based cloud services provider, announced on Monday that it will build a $500 million artificial intelligence data centre in Taiwan with support from Nvidia. The company stated that the facility is scheduled to begin operations by March 2026 and will run on Nvidia’s Blackwell GB300 chips. According to the plan, the data centre will include around 7,000 GPUs installed across 96 high-density racks, offering processing capacity of nearly 2 million tokens per second. The project will require about 16 megawatts of power.
NVIDIA logo is seen near computer motherboard in this illustration taken January 8, 2024.
GMI Cloud Founder and CEO Alex Yeh said Taiwan requires additional data centres as “strategic assets” to support the island’s AI development. He noted that power-supply issues can be addressed and added that demand for AI services has remained strong, with the company’s GPU utilisation “almost full”. Yeh said: “You want to promote local ecosystems - you have to build the data centre first, you have to build the AI cluster first.”
The announcement comes during a period of accelerated global investment in AI infrastructure. Technology companies worldwide are allocating significant resources to support growing workloads, boosting demand for high-performance semiconductors produced by firms such as Nvidia. The company derives a substantial share of its revenue from supplying GPUs for AI clusters. Nvidia CEO Jensen Huang has previously described these large-scale computing setups as “AI factories”, and over the past year Nvidia has announced agreements to deliver advanced GPUs to projects in Saudi Arabia and South Korea.
Recent AI infrastructure developments in Taiwan include a 100-megawatt AI data centre project announced by Foxconn and Nvidia in May. These projects reflect growing expansion in the region as companies increase their capacity for cloud services, data processing, and AI model training.
GMI Cloud is a GPU-as-a-Service provider and one of Nvidia’s cloud partners. It currently operates data centres in the United States, Taiwan, Singapore, Thailand, and Japan. In addition to the new Taiwan facility, the company plans to construct a 50-megawatt data centre in the United States. GMI Cloud is also considering an initial public offering in two to three years.
Yeh said the partnership project with Nvidia is expected to generate around $1 billion in total contract value once the site becomes fully operational. Initial customers for the Taiwan AI facility include Nvidia, cyber-security company Trend Micro, electronics manufacturer Wistron, Chunghwa System Integration, data-infrastructure provider VAST Data, and industrial solutions firm TECO.
Overall, the planned data centre marks a major expansion of GMI Cloud’s regional infrastructure and reflects continued global growth in investments supporting AI computing needs.
