U.S. based cloud provider GMI Cloud has announced plans to build a $500 million artificial intelligence data center in Taiwan with support from Nvidia, signaling a major bet in Asia’s AI infrastructure. The data center is scheduled to be operational by March 2026 and will run on Nvidia’s advanced Blackwell GB300 processors.
The facility will host approximately 7,000 GPUs across 96 high-density racks, enabling it to process up to 2 million tokens per second. It will draw about 16 megawatts of power, underscoring the scale of the project and the energy demands associated with high-performance AI operations.
GMI Cloud CEO Alex Yeh emphasized that Taiwan’s data centers should be “strategic assets” for the island’s AI ambitions. Despite known power-supply challenges, he expressed confidence that infrastructure solutions can support the anticipated demand. He also noted that the company’s existing GPU utilization is “almost full,” signaling strong demand for compute capacity.
Early partners for the data center include cybersecurity firm Trend Micro, electronics company Wistron, and TECO, highlighting the facility’s importance to both local enterprise and global AI players. GMI Cloud is reportedly raising $400 million from Taiwanese banks for the project and anticipates additional funding later in 2025.
Industry observers see the investment as part of a broader trend: major cloud providers are increasingly prioritizing sovereign AI infrastructure. As countries look to secure strategic compute capacity, these “AI factories” are becoming central to national development strategies especially in regions vying to become technology hubs.