GMI Cloud raises $668 million to build more GPU capacity across the US, Taiwan, and Southeast Asia

GMI Cloud announced $668 million in new financing on Thursday, a combination of $223 million in Series B equity and a $445 million credit facility from CTBC Bank, as the Mountain View-based GPU cloud provider moves to expand capacity across the United States, Taiwan, and Southeast Asia ahead of surging demand for AI inference.

The Series B was led by ARCHIV, an AI and robotics investment firm, with NVIDIA participating alongside a cluster of Asia-Pacific investors including DSC Investment, Trend Micro, KB Investment, Kyobo Life, and KT Corporation. The Korea-heavy investor list reflects both the geographic expansion and a broader trend of Korean institutional capital flowing into AI infrastructure plays.

The numbers behind the raise are striking. GMI Cloud's contracted annual recurring revenue topped $600 million and grew nine-fold since the end of 2025, while live production ARR increased 4.5 times in the same period. The company's inference platform currently processes more than 4 trillion tokens per week, with major customers including Fireworks AI, Higgsfield, Nous Research, OpenRouter, Reflection, Cartesia, Trend Micro, and Utopai Studios.

Founded in 2021, GMI Cloud differentiates from hyperscaler cloud providers by focusing exclusively on on-demand GPU access for AI training and inference, using a proprietary Kubernetes-based Cluster Engine for automated resource allocation. The company has built out significant data center presence across Asia-Pacific — Taiwan, Thailand, and Malaysia — which gives it proximity to AI server supply chains and a cost advantage over US-only players.

CEO Alex Yeh said the capital will go toward GPU capacity expansion, growth of the inference services platform, and strategic hiring. “Our customers are scaling faster than ever, and they need infrastructure that keeps pace,” Yeh said in the announcement. The credit facility structure — two-thirds of the total — is notable: it suggests investors are backing hard infrastructure assets with debt rather than pure equity, a sign of the sector’s maturation.

As reported by SiliconANGLE, the round is one of the larger raises in the specialized GPU cloud segment in 2026, a year that has seen significant consolidation and capital concentration among the handful of providers with genuine scale.