Runware plans assembly-line production for AI data center pods
AI infrastructure company Runware plans to scale production of its Sonic Inference Pod modular data centers through a Romania-based factory model. Co-founder and CEO Flaviu Fadulescu told TechCrunch that the company is targeting 30 megawatts of data-center capacity produced per day by 2028, or nearly 1 gigawatt per month.
The company has two factories: one that has built pods for several years and a second whose construction began earlier this year. Runware expects the newer site to be completed within months and to begin operating early next year. It says production will increase gradually from next year.
From modular pod to production line
Runware introduced the Sonic Inference Pod only months ago as a factory-built unit that can be shipped to a customer site ready for installation. Its stated aim is to manufacture pods on an assembly line, borrowing a production approach associated with the automotive sector.
Fadulescu said customers may provide a site or ask Runware to locate one. The company aims to bring a pod online within 48 hours, a figure covering installation and setup rather than manufacturing. It says a pod that once required three weeks to build will soon take one day as production ramps up.
The pods use a standard design but can be configured for customer requirements and newer GPU generations. Runware says its in-house hardware team can design, build and test components for new chips. The company argues that pod delivery can be faster than GPU lead times, allowing it to adapt infrastructure while customers await ordered hardware.
Commercial model and execution risks
Customers can either rent compute from Runware or provide their own GPUs and pay a monthly fee for pods and operations. Power and internet are billed separately, while Runware says it will provide spare parts if components fail. It has raised more than $60 million, including a $50 million Series A last December, and plans to raise a Series B.
Runware says it has received $1 billion in committed interest since opening the Sonic pods to outside customers a month ago. Fadulescu said this represents more than 20 megawatts that the company is working to deliver. Those expressions of interest do not remove the operational challenge of building a high-volume manufacturing operation.
The company will need staffing, inventory, supply-chain logistics and capital to execute its plan. It designs its cooling and other systems around standard components, but pumps and fans are sourced from Germany, while networking still depends on Nvidia. Runware has also hired an experienced automotive plant manager to work with its chief of hardware.
Power, cooling and site requirements
Runware says its pods require power and an internet connection. They use closed-loop liquid cooling with air-cooled radiators, which the company says avoids the need for a water connection. It also says the pods operate silently at temperatures up to 86 degrees Fahrenheit.
Fadulescu said Runware is working with power providers on a site pipeline covering 397 megawatts, with potential expansion to 1.2 gigawatts. The company says its sites are typically smaller than hyperscale facilities and already have power, and it aims to use renewable energy whenever possible. For businesses assessing modular AI capacity, the relevant decision is not only pod lead time but whether power, connectivity, GPUs and operating arrangements are ready at the selected site.