Runware Plans Factory for Mass Production of Modular AI Data Centers

Mass Production of AI Data Centers

Runware, an AI infrastructure company, announced its intention to establish a factory for the mass production of modular data centers. The company aims to assemble these units on a production line, similar to car manufacturing. This follows their earlier introduction of the Sonic Inference Pod, a factory-built, ready-to-install modular data center.

Scaling Production Capacity

Runware plans to scale production at its Romania-based factory to 30 megawatts of data center capacity per day by 2028. This equates to nearly 1 gigawatt per month. This production rate would match the projected annual data center power additions for the entire U.S. in 2024, which is approximately 10 gigawatts. The company already operates one factory and is building a second, expected to be operational early next year.

Deployment and Customization

Runware's modular data centers are designed for rapid deployment, with installation and setup taking 48 hours. The company states that the build time for a single pod will decrease from three weeks to one day as production scales. Customers can provide their own sites or use Runware's services to find one. The pods offer customization options for customer requirements and GPU chips, and Runware provides spare parts. Customers can rent compute from Runware or supply their own GPUs, paying a monthly fee for the pods and operations, with power and internet charged separately.

Industry Context

The move reflects an industry-wide push for faster and more efficient data center deployment. Other companies, such as Crusoe and Schneider Electric, also offer prefabricated data center solutions. Nvidia has also introduced initiatives like the Vera Rubin AI factories and the DSX POD to accelerate AI data center construction. Runware differentiates itself by focusing on factory-line assembly for its modular units.