Runware Aims to Mass-Produce AI Data Centers Like Cars with New Factory

Image: TechCrunch · Source

Runware is building a factory to assemble modular AI data center pods on a production line, targeting 30 megawatts daily capacity by 2028 to accelerate data center deployment with faster shipping and installation.

Runware, an AI infrastructure company, is pioneering the mass production of modular AI data centers by developing a factory-based assembly line designed to build prefabricated data center pods. The startup’s approach is to manufacture these pods similarly to car production, aiming to efficiently meet the growing demand for AI compute facilities.

Earlier this year, Runware introduced the Sonic Inference Pod, a modular data center built entirely in a factory and shipped ready to install. Now, the company plans to scale this concept significantly at its Romania-based factory. By 2028, Runware targets a production capacity of 30 megawatts per day, or nearly 1 gigawatt per month, which would rival the total new data center power capacity expected in the U.S. annually.

Unlike some competitors such as Crusoe and Schneider Electric that also sell prefabricated pods, Runware emphasizes faster shipping times and actual factory-scale production rather than just infrastructure design. The company has two factories, with the newer facility expected to come online early next year, enabling pod build times to shrink from three weeks to a single day.

Customers can deploy the pods on their own sites or those arranged by Runware, with the pods brought online within 48 hours, covering setup and installation. The company also offers flexibility for customers to provide their own GPUs or rent compute through Runware. The pods feature a standard design but can be customized for specific hardware and requirements.

Runware’s CEO Flaviu Fadulescu highlighted the company’s strategy to avoid pitfalls seen by some startups attempting to reinvent manufacturing. They leverage a cost-effective workforce, widely available components, and in-house design of cooling and hardware systems. A former automotive plant manager has been brought in to help optimize production processes.

The company is mindful of energy concerns associated with data centers. Their pods employ closed-loop liquid cooling with air-cooled radiators, requiring no water connection and aiming for renewable power sources. They are silent during operation up to 86 degrees Fahrenheit and deployed on smaller sites than typical hyperscale data centers, reducing strain on local grids.

Runware has reportedly secured $1 billion in customer commitments representing over 20 megawatts of pods since launching sales, a testament to strong market interest. Following ramp-up in Romania, the startup plans to expand factory production capacity to the U.S. and global markets.

The company believes factory-built modular pods will become standard for large-scale AI deployments, enabling customers to build hyperscale data centers within months instead of years. This could signal a significant shift in how AI infrastructure is manufactured and deployed worldwide.