The news
GMI Cloud, a provider of on-demand graphics processing unit (GPU) capacity for artificial intelligence, announced on September 30, 2026, that it raised a $223 million Series B round led by ARCHIV, according to Pulse 2 and SiliconANGLE. GMI Cloud funding also includes a $445 million credit facility led by CTBC (ChinaTrust Commercial Bank), for $668 million in total, Pulse 2 reported. SiliconANGLE put the credit line at $440 million.
Nvidia (NVDA) took part in the equity round alongside DSC Investment, KT Corp., Kyobo Life, KB Investment and Trend Micro, according to both reports; SiliconANGLE also listed Redwood among the investors. Pulse 2 described ARCHIV as a San Francisco-based investment firm, and SiliconANGLE said it focuses on AI and robotics. Valuation was not disclosed in the coverage we reviewed.
GMI Cloud sells AI infrastructure as a service. Customers can rent Nvidia H100, H200 and Vera Rubin chips, and the company runs a Kubernetes-based Cluster Engine, its own software for allocating computing resources automatically. Kubernetes is a widely used open-source system for managing software containers across many servers.
The company operates data centers in Taiwan, Thailand and Malaysia and is expanding into the US and Japan, SiliconANGLE reported. Pulse 2, citing the company, described plans to add GPU capacity across the US, Taiwan and the Asia-Pacific region with an emphasis on Nvidia GB200 and GB300 NVL72 rack systems.
On growth, SiliconANGLE reported that GMI Cloud expects annual recurring revenue (ARR) under contract to grow 10-fold by the end of 2026, and that realized ARR rose 5-fold over the past year. Its inference platform, which runs trained models for customers, processes about 4 trillion tokens a week. Customers named include Higgsfield, OpenRouter, Nous Research, Fireworks AI and Trend Micro.
Founder and Chief Executive Alex Yeh tied the raise to reliable delivery, saying customers plan launches, hiring and revenue around when their infrastructure arrives, and pointing to the company's place in Taiwan's supply chain, as reported by SiliconANGLE and Pulse 2.
The numbers
- Series B equity
- $223 million
- Credit facility
- $445 million
- Total new financing
- $668 million
- Weekly tokens processed
- 4+ trillion
- Realized ARR growth, past year (SiliconANGLE)
- 5x
Why CEOs should care
For companies buying AI compute, each well-funded neocloud is another place to source GPUs beyond Amazon, Microsoft and Google. Procurement teams negotiating training or inference capacity should get quotes from specialist providers like GMI Cloud and compare price per GPU-hour, contract length, and how quickly new chips such as Vera Rubin actually become available.
CFOs and boards should look closely at how these providers are financed. GMI Cloud's package leans heavily on debt: the credit facility is roughly twice the size of the equity round. Debt-funded GPU fleets depend on steady rental demand to service loans, so buyers signing multi-year commitments should ask about the provider's financial health, data center ownership and what happens to capacity if the lender's terms change.
Technology and security leaders should also check where workloads will physically run. With sites in Taiwan, Thailand and Malaysia and new ones planned for the US and Japan, data residency, latency and export-control rules for advanced Nvidia chips may differ by region. Ask which location your capacity will sit in and which compliance certifications apply.
The bigger picture
GMI Cloud is part of a group of so-called neoclouds, specialist GPU rental firms that have raised large sums of equity and debt to buy Nvidia hardware. Nvidia investing in its own customers is a recurring pattern in this market. Strategic investors from Korea, Japan and Taiwan in this round, including KT, Kyobo Life and Trend Micro, also point to Asian interest in building AI capacity outside the US hyperscalers.
What’s next
Watch for GMI Cloud to announce specific US and Japan sites and timelines for Vera Rubin capacity. SiliconANGLE reported the credit line at $440 million and the total at $663 million, slightly below the $445 million and $668 million figures in other coverage of the announcement.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
Companies in this story
