GMI Cloud to build $500 million AI data centre in Taiwan with Nvidia chips
News Summary
U.S.-based cloud services provider GMI Cloud announced a $500 million investment to build an AI data center in Taiwan, slated to be operational by March 2026. This facility will be powered by Nvidia's new Blackwell GB300 chips, housing approximately 7,000 GPUs across 96 high-density racks, capable of processing nearly 2 million tokens per second. The project is expected to draw 16 megawatts of power. GMI Cloud's CEO, Alex Yeh, emphasized the strategic importance of such data centers for Taiwan's AI development, despite acknowledging the island's power supply challenges. He noted strong AI demand, with the company's GPU utilization “almost full.” The data center is projected to generate about $1 billion in total contract value once fully operational, with initial customers including Nvidia, Trend Micro, Wistron, and VAST Data.
Background
Technology giants globally are pouring billions into AI infrastructure to support rising workloads, creating a windfall for semiconductor companies like Nvidia, which derives the bulk of its revenue from such sales. Nvidia CEO Jensen Huang has referred to such clusters as “AI factories” and has announced deals to sell its most advanced GPUs to projects in Saudi Arabia and South Korea over the past year. U.S. President Donald Trump has stated his desire for top AI semiconductors, such as Nvidia's Blackwell chips, to be reserved for U.S. companies. In addition to GMI Cloud's project, other AI infrastructure initiatives recently announced in Taiwan include a 100-megawatt AI data center project by Foxconn and Nvidia in May. GMI Cloud, a GPU-as-a-Service provider and Nvidia cloud partner, already operates data centers in the United States, Taiwan, Singapore, Thailand, and Japan, with plans for a new 50-megawatt U.S. data center and a potential IPO in two to three years.
In-Depth AI Insights
What are the deeper strategic implications of this investment in the current geopolitical climate? - GMI Cloud's decision to build such a significant AI data center in Taiwan, rather than exclusively in the U.S., especially given President Trump's stated desire for top AI chips to be reserved for American companies, highlights the complexity of global supply chains and diversified corporate strategies. - This likely reflects continued confidence in Taiwan as a critical hub for global semiconductor manufacturing and the AI ecosystem, acknowledging its technological talent and operational efficiency despite escalating geopolitical risks. The choice of Taiwan may also be driven by a need to better serve Asian market demand and balance supply chain risks. What does Nvidia's role in this partnership reveal about its market strategy? - Nvidia is not merely a chip supplier but an active participant in building and promoting