Armada Launches Orion, the 10MW Galleon, Bringing Distributed AI Deployments to the Sovereign AI Grid

Orion is Armada's most powerful modular data center yet — deployable in months, sovereign by design, and scalable from a single unit to hundreds of megawatts across distributed sites. Fossefall selecting Armada to accelerate Nordic AI factory development shows the modular model winning.

NEW YORK, Aug. 26, 2026 /PRNewswire/ -- Armada today launched Orion, the 10-megawatt Galleon and the most powerful member of its Galleon family of modular data centers. With Orion, the Galleon family now spans from rapidly deployable units to hundreds of megawatts across multiple sites — all built and delivered in months, sited directly at distributed, low-cost power, and owned and operated by the customer. Announced earlier this week, Fossefall selected Armada as one of its technology partners for future AI factories in Norway, Sweden and Finland, the latest proof that modular AI infrastructure is how capacity gets built now.

"AI infrastructure should be fast to deploy, easy to scale, and sovereign wherever it needs to live," said Dan Wright, Co-founder and CEO of Armada. "Orion raises the ceiling on what a single deployment can deliver — ten megawatts per unit, infinitely scalable, in months instead of years. And because our customers own and operate it, on their land and their power, it's infrastructure the cloud can never be: theirs."

Orion: Gigawatt-Class Ambition, One Deployment at a Time

Delivering 10MW of capacity per unit, Orion is designed to deploy inside existing structures and scale rapidly to hundreds of megawatts across distributed sites — bringing large-scale AI infrastructure within reach of a modular, incrementally financed build. It can be co-located with distributed, low-cost power that may be stranded or behind-the-meter without disturbing existing communities. Orion's architecture is engineered to be ready for the next generation of accelerated computing.

Every unit packages compute, networking, power, and cooling into a prefabricated six-module core, supporting up to 2,880 GPUs across 40 racks with the ability to achieve greater than N+1 redundancy. Orion is compatible with NVIDIA Blackwell with a Vera Rubin-ready architecture. Units will be deliverable in months from deposit.

Orion joins the full-stack Armada solution: Galleons across every form factor, all running Armada Bridge, the company's AI platform for distributed infrastructure. Together they turn a customer's land and power into running AI capacity — power distribution, cooling, compute, and the software to deploy and run models on it, including distributed inferencing and fine-tuning. The result is a complete, fully sovereign AI environment the customer operates, with no dependency on public cloud. Bridge provides multi-GPU orchestration across racks and sites, a model marketplace for deploying leading open and commercial models on sovereign infrastructure, and NVIDIA Token Factory enablement. A growing ecosystem of ISV and model partners delivers value-added services on top — including model compression and optimization to maximize throughput per megawatt, and model lifecycle management to keep deployed models current, governed, and performant in-country.

Fossefall: Modular AI Infrastructure in Action

The demand is already visible. On Tuesday, Fossefall announced that it had selected Armada to supply five Leviathan Galleons and be one of its technology partners for future AI factories in Norway, Sweden and Finland. It's part of its plan to build more than 1 gigawatt of Nordic-owned AI capacity by 2030. The deployment will run on infrastructure architected to be ready for future platforms including Vera Rubin. Fossefall will own and operate the GPU infrastructure inside the Leviathans — keeping the capacity, and the value it creates, under its control.

"We selected Armada for its best-in-class technology, proven execution capabilities, and ability to significantly accelerate time to market," said Øyvind Vesterdal, CEO of Fossefall. "We were also keen to work with a strategic partner manufacturing in Europe, to provide a resilient supply chain and reduce exposure to global disruptions. By refining clean energy into high-tech AI capacity, we will ensure national control over critical technology, create a billion-kroner expert industry, and produce significant local economic benefits."

Each Leviathan supports a maximum total IT load of 1.77MW and is designed to operate reliably across a wide range of environmental conditions. The site was chosen for its cold inland climate, which contributes to natural cooling and energy-efficient operations. Armada's closed-loop cooling system eliminates daily water consumption.

Fossefall selected a modular approach as the best fit for the project and local conditions: modular design reduces construction noise, site disruption, and build complexity, while allowing power infrastructure to be installed incrementally alongside each Leviathan. Fossefall will engage local firms for site preparation and installation, and, working closely with Armada, will train and support local employees to deliver this specialized work.

Sovereign AI Grid: The Future of AI Infrastructure Is Distributed

The AI buildout is colliding with reality. Energy is the bottleneck. Communities are pushing back on centralized gigawatt-scale data centers, with moratoriums spreading even in traditionally build-friendly regions. And data residency requirements are pushing workloads toward capacity that is owned and operated in-country. Gartner® predicts by 2030, neocloud providers will capture 20% of the $267 billion AI cloud market.¹

Armada believes the answer is distributed — and today's news shows the approach at work. Orion extends a family of form factors that now spans from rapidly deployable units to 100+ megawatt sites, all running the same full stack on Bridge. Fossefall has selected Armada as one of its technology partners to convert Nordic clean energy into sovereign AI capacity. WinDC, Armada's customer in Australia, is deploying Leviathans powered by renewable, curtailed energy. Deployment by deployment, capacity is being built where distributed, low-cost, and stranded power already exists — without waiting years on centralized construction. And every deployment is sovereign by design: owned and operated by the customer, on their land and their power, under their jurisdiction.

Those deployments are also assets. As NVIDIA argued this month in announcing financing platforms designed to mobilize more than $500 billion of third-party capital, AI compute has become investable infrastructure — revenue-producing, redeployable, and improving in performance over its life. Customer-owned, distributed capacity built on low-cost power sharpens that case: lower energy costs, faster time to revenue, and sovereign demand that doesn't move. Armada will make that argument today at the New York Stock Exchange, joining ICE, NativX, and Fossefall for a discussion on compute as an asset class.

That is what Armada is building: the Sovereign AI Grid. Distributed, sovereign, and rapidly deployable — it's how the AI race gets won without breaking the communities it's built in.

About Armada 
Armada is the hyperscaler for the edge, delivering modular AI infrastructure from first deployment to AI factory with speed, scale, and sovereignty. For more information, visit www.armada.ai.

Media contact: [email protected]

¹ Gartner, Sovereign AI: Infrastructure Strategies for a Multipolar World, Enrique Castera, January 6, 2026. Gartner is a trademark of Gartner, Inc. and/or its affiliates.

SOURCE Armada