GMI Cloud raises $223M Series B for GPU cloud capacity

GMI Cloud raises $223M Series B for GPU cloud capacity

30 September 2026: GMI Cloud, a provider of GPU infrastructure for training and running AI models, has closed a $223 million Series B led by ARCHIV, a San Francisco investment firm focused on AI and robotics. NVIDIA took part. The round came with a separate $445 million credit facility led by CTBC, which the company counts together with the equity as $668 million of new funding.

Not yet independently verified. Revenue and token figures are the company's own. The credit facility is debt and is reported only on the company's blog. We will update this when it can be confirmed, and remove this note.

The company announced the financing on 30 September and published a longer account on its blog on 6 October. Asia-Pacific investors also joined the Series B: DSC Investment, Trend Micro, KB Investment, Kyobo Life and KT Corporation, among others. Crunchbase News lists GMI Cloud as based in Mountain View, California, and included the round in its weekly list of the largest US financings.

GMI says the proceeds go to capacity in the United States, Taiwan and the rest of Asia-Pacific, building on a Taiwan AI Factory announced in 2025 and a Japan sovereign AI initiative announced this year. Money will also go to its inference services and hiring.

The company supplied its own growth figures. It says contracted annual recurring revenue is now more than nine times its year-end 2025 level, and live revenue in production has grown more than 4.5 times over the same period. Its inference platform handles about 4 trillion tokens a week, and in September it passed 1 trillion tokens served through OpenRouter. It names Fireworks, Higgsfield, Nous Research, OpenRouter, Reflection, Cartesia, Trend Micro and Utopai Studios as customers.

Alex Yeh, founder and chief executive, said: "Our customers are scaling faster than ever, and they need infrastructure that keeps pace." Chenyu Zhao, co-founder of customer Fireworks, is quoted calling GMI Cloud "one of our strongest and most reliable providers" across NVIDIA GB200 and GB300 systems.

This year GMI became an NVIDIA Exemplar Cloud on GB300 NVL72 systems for training. It also launched GMI Agentbox for agent infrastructure and GMI Router, which it says optimises performance and token use, as well as a startup programme in March and a developer programme in May.

Company profile on TrustList: GMI Cloud

Sources

• GMI Cloud: We Raised Over $660 Million to Accelerate Global AI Infrastructure Expansion, 6 October 2026
• Crunchbase News: The Week's 10 Biggest Funding Rounds, 3 October 2026