Can AI data centers be mass-produced like cars? Runware aims to revolutionize the industry

Runware, an AI infrastructure company, is pushing the boundaries of data center manufacturing by aiming to mass-produce modular data centers like cars. The company, which previously launched the Sonic Inference Pod, now plans to build a factory that assembles these pods on a production line. Flaviu Radulescu, co-founder and CEO of Runware, stated that the company intends to scale production at its Romania-based factory to 30 megawatts per day by 2028, equivalent to nearly 1 gigawatt per month. This capacity would match the U.S. data center power addition for 2026. Radulescu explained that Runware will gradually increase production starting next year, with two factories already in operation. The company aims to deliver pods within 48 hours, including installation and setup, reducing build time from three weeks to one day. Customers can rent compute, supply their own GPUs, or use Runware’s standard design with customizable configurations. The company also highlighted its closed-loop liquid cooling system and plans to use renewable energy sources. Despite the ambitious goals, Runware faces challenges in sourcing components and managing supply chain logistics. However, the company has received $1 billion in committed interest for its Sonic pods, with over 20 megawatts already in the pipeline. Radulescu emphasized that Runware’s cost structure and in-house hardware team help avoid the pitfalls of previous manufacturing ventures like Arrival.