GMI Cloud Secures $668M Backed by Nvidia to Expand Global AI Infrastructure

The global race to construct specialized infrastructure for artificial intelligence reached another major milestone today as GMI Cloud, an AI-native cloud platform delivering high-performance GPU compute and inference services, announced a massive $668 million financing package. The funding consists of a $223 million Series B equity round alongside a $445 million credit facility arranged by CTBC Bank.

Strategic Backing from Nvidia and Asia-Pacific Tech Titans

The equity round was spearheaded by San Francisco-based AI and robotics investment firm ARCHIV, with direct strategic participation from AI semiconductor leader Nvidia. The round also drew prominent institutional support across the Asia-Pacific tech corridor, including South Korea's DSC Investment, KB Investment, Kyobo Life Insurance, KT Corporation, and cybersecurity pioneer Trend Micro.

This major capital infusion underscores the booming momentum behind so-called "neoclouds"—independent, GPU-first cloud providers engineered specifically to meet the compute-intensive demands of large language model (LLM) training and low-latency inference workloads.

Rapid Commercial Growth and Inference Demands

GMI Cloud's growth metrics reflect the intense enterprise appetite for dedicated AI computing capacity outside traditional hyperscalers. According to company disclosures:

• Contracted ARR Surge: Contracted Annual Recurring Revenue (ARR) has expanded by more than 9x compared to year-end 2025 levels.
• Live Production Capacity: Active production ARR has jumped over 4.5x over the same timeframe.
• Trillion-Token Throughput: The platform now processes approximately 4 trillion tokens per week across its distributed infrastructure.
• Ecosystem Adoption: Core developer and enterprise customers utilizing the infrastructure include Fireworks AI, OpenRouter, Reflection, Nous Research, Cartesia, and Utopai Studios.

Addressing Global Bottlenecks in GPU Provisioning

A persistent hurdle for AI engineering teams has been cluster lead times and capacity delivery. Unlike legacy cloud virtualization, modern multi-node GPU clusters require complex orchestration, high-speed interconnect fabric, and tight thermal management. Founder and CEO Alex Yeh highlighted that GMI Cloud's proprietary Kubernetes-based Cluster Engine automates resource allocation and environment provisioning, enabling engineers to spin up pre-configured GPU clusters in seconds while maintaining strict operational uptime.

The fresh financing will be allocated directly toward expanding GPU fleet capacity across data centers in the United States, Taiwan, and broader Asia-Pacific hubs, bolstering GMI Cloud's flagship Taiwan AI Factory initiative and sovereign AI deployments in Japan.

The Evolution of Neocloud Infrastructure

As enterprise adoption moves rapidly from experimental prototypes to full-scale autonomous agent deployments, specialized AI cloud platforms are carving out a distinct niche alongside Amazon Web Services, Microsoft Azure, and Google Cloud. By tailoring their hardware topology, networking layers, and commercial models directly to high-throughput inference, providers like GMI Cloud, CoreWeave, and Nebius are reshaping the cloud computing economics of the AI era.