Armada raises $200M+ to build portable AI data centers for the 70% of the world without AI infrastructure

Summary

Armada, a full-stack edge infrastructure company founded roughly two and a half years ago, has raised over $200 million and is positioning itself as the hyperscaler for the 70% of the world that lacks AI infrastructure. Microsoft made a strategic investment in the company's latest round, and CEO Satya Nadella highlighted Armada publicly at Microsoft Ignite. The company is now going to market globally with Microsoft and SpaceX as anchor partners.

The Core Thesis

Conventional hyperscale data centers serve only about 30% of the globe, concentrated in major metro areas with reliable power and connectivity. Armada targets the remainder with rapidly deployable, modular data centers it calls Gallions, shipping-container-scale units that can be operational in weeks rather than the years required for traditional builds. The product line recently expanded with Leviathan, a megawatt-scale unit designed to co-locate with stranded energy sources and function as a portable AI factory.

Stranded Energy as a Supply Chain

A central element of Armada's strategy is monetizing stranded energy, particularly natural gas being flared at oil and gas sites across North Dakota, Texas, and the Middle East. Energy operators currently vent or burn this gas with no return; Armada proposes converting it via turbines into electricity to power on-site compute. The company estimates roughly 6 gigawatts of stranded energy exists domestically. This aligns with a recent Anthropic white paper calling for 50 gigawatts of new energy capacity for AI by 2028.

Target Customers and Deployments

Armada's customer concentration reflects two verticals: critical industry and defense. On the commercial side, the company is deployed with Saudi Aramco in a partnership that also involves Microsoft, with ambitions to scale across Aramco's global network of rigs and refineries. It is also active with major mining companies and large oil and gas operators. On the defense side, contracts with the U.S. Navy and broader DoD are public. A notable civil deployment is with the Alaska Department of Transportation, where Armada's platform reduced drone data processing latency for avalanche response from over a day to near real time.

The Microsoft Distribution Channel

The Microsoft relationship functions as both a capital and distribution partnership. Enterprise customers can procure Armada's products directly against existing Azure credits and Microsoft agreements, lowering friction significantly. Microsoft sellers earn incentives on Armada deals. The integration goes deeper operationally: Armada's Gallions can run Azure Stack hardware, meaning customers do not need to replatform applications already built for Azure. A three-way partnership with Microsoft and Halliburton is already active, extending production automation software used at connected energy sites out to remote field locations.

Starlink as Infrastructure Substrate

Armada treats Starlink, now live in over 140 countries since its public beta launch in November 2020, as foundational connectivity. Starlink began as a consumer and backup product for enterprises but is increasingly used as a primary link. Armada's architecture handles local data processing at the edge, then uses Starlink to transmit only metadata back to the cloud, reducing bandwidth dependency. Over a five-year horizon, Armada expects LEO satellite connectivity to reach fiber-equivalent performance globally. The company also has a drone data partnership with Skyo and integrates SDWAN devices from CradlePoint and Peplink, positioning itself as an orchestration layer across heterogeneous connectivity assets.

Scale and Trajectory

Armada prioritizes customers where it can see potential fleets of thousands to tens of thousands of connected assets. The business model mirrors a cloud provider, offering infrastructure, software, and AI as a managed service under what it calls the Armada Edge (AE) platform, which includes a fleet map view of all deployed Gallions and connected assets. With federated learning capabilities now in development, Armada is moving beyond inference at the edge toward localized model fine-tuning that can propagate improvements across distributed deployments.