Funding · October 7, 2026
Also published in Français, 한국어, Polski, ไทย
GMI Cloud Secures 668 Million USD in Series B and Credit Facility
GMI Cloud has secured a total of 668 million USD in funding.
BRIDGE reported that the total amount consists of 223 million USD raised through equity in a Series B round and a 445 million USD credit facility led by CTBC. The Series B round was led by ARCHIV, an investment firm specializing in artificial intelligence and robotics, with participation from NVIDIA. Other investors from the Asia-Pacific region include KB Investment, Kyobo Life, KT Corporation, DSC Investment, and Trend Micro.
Founded in 2021, the company provides a single cloud platform that offers foundations for artificial intelligence agents, inference, and graphics processing unit clusters. The company intends to use the new funds to expand its graphics processing unit supply capacity in the United States, Taiwan, and other Asia-Pacific regions. This expansion follows the announcement of the Taiwan AI Factory, the first facility in Asia announced in 2025, and sovereign artificial intelligence initiatives in Japan announced this year. The funding will also support strategic hiring and the growth of inference services.
GMI Cloud aims to serve artificial intelligence companies and hyperscalers in the United States that require computing power in both the United States and Asia, while also supporting Asia-Pacific companies that want to operate production artificial intelligence locally to meet compliance and data requirements. The company leverages its connections with Taiwan, where artificial intelligence server manufacturing is concentrated, to maintain close relationships with manufacturers and ensure predictable timelines from ordering to operation.
According to the company, contract-based annual recurring revenue has exceeded 600 million USD, representing more than a nine-fold increase by the end of 2025. Production-based annual recurring revenue grew more than 4.5 times during the same period. The company also stated that its inference platform processes approximately 4 trillion tokens per week.
Major customers include Cartesia, Fireworks, Higgsfield, Nous Research, OpenRouter, Reflection, Trend Micro, and Utopai Studios. Chenyu Zhao, a co-founder of Fireworks, stated that GMI Cloud is one of the most reliable and powerful providers of GB200 NVL72 and GB300 NVL72 systems from NVIDIA.