Taiwan's GMI Cloud raises $663M to expand on-demand GPU cloud worldwide

The Series B was led by ARCHIV, an investment firm focused on robotics and artificial intelligence, with participation from Nvidia Corp., Redwood, DSC Investment, KT Corp., Kyobo Life, KB Investment and Trend Micro Inc. The credit facility was provided by ChinaTrust Commercial Bank, which the company described as one of Taiwan's largest financial institutions.

GMI Cloud is one of a group of specialized "neocloud" providers that have emerged in recent years to serve demand for GPUs and other resources used to run AI workloads. The best-known of these is CoreWeave Inc., which went public last year, while Europe has Nebius Group N.V., the former Yandex N.V. that was renamed after its Russian search business was sold.

Like those rivals, GMI Cloud sells an AI infrastructure-as-a-service platform built for resource-intensive training and inference. It offers on-demand access to Nvidia's H100 and H200 chips and to newer Vera Rubin GPUs. The platform runs on a proprietary Kubernetes-based system called Cluster Engine, which automates resource allocation and instance provisioning, allowing developers to spin up GPU clusters from preconfigured environments in seconds rather than configuring them manually, as they would on traditional cloud platforms.

The company operates dozens of data centers in Taiwan, Thailand and Malaysia, a footprint that contrasts with the mostly U.S.- and Europe-centered capacity of CoreWeave and Nebius. It is also building a presence in the United States and launched a sovereign AI initiative in Japan earlier this year.

Founder and Chief Executive Alex Yeh said demand for AI compute has become global, with U.S. hyperscalers and AI startups seeking more GPU capacity in Asia, where they have large customer bases, alongside thousands of Asia-Pacific enterprises running their own AI projects. "Our customers are scaling faster than ever and they need infrastructure that keeps pace," Yeh said. "AI is driving a new renaissance and reliable compute is its foundation. Our goal is to build that foundation across continents, with an ecosystem of products on top of it."

Yeh pointed to Taiwan's role as the world's largest manufacturing base for the AI server systems used in data centers, saying proximity to suppliers gives GMI Cloud a secure supply chain and the ability to procure systems quickly for new sites. "In AI infrastructure, a delivery date is a promise," he said. "Customers plan launches, hiring and revenue around those dates. Our place in Taiwan's supply chain is how we keep that promise, cluster after cluster globally."

GMI Cloud said the annual recurring revenue it has under contract is set to grow tenfold by the end of the year from its level in December 2025, while revenue already being realized has risen fivefold over the past year. Its inference platform processes an average of more than 4 trillion tokens per week, according to the company, which lists Higgsfield Inc., OpenRouter, Nous Research Inc., Fireworks AI Inc. and Trend Micro among its enterprise customers.

Editor's Summary

GMI Cloud, a Taiwan-based neocloud, raised $663 million through a $223 million Series B led by ARCHIV with Nvidia among the investors and a $440 million credit facility from ChinaTrust Commercial Bank. The company plans to use the capital to expand its on-demand GPU platform, which spans data centers in Taiwan, Thailand and Malaysia, and to extend its footprint in the United States and Japan. Its growth comes as contracted annual recurring revenue is projected to rise tenfold by year-end, intensifying competition with CoreWeave and Nebius in the AI compute market.