Runware plans serial production of modular AI data centers at a factory in Romania
Runware plans serial production of modular Sonic Inference Pod data centers at a factory in Romania, aiming to reach a capacity of 30 MW per day by 2028. A second factory is to be launched at the beginning of next year, and pod installation is expected to take 48 hours.
Runware, an AI infrastructure provider, announced plans to build a factory in Romania that would manufacture modular Sonic Inference Pod data centers on an assembly line, similar to cars. According to the company, this should gradually increase production capacity to 30 megawatts per day by 2028, i.e. nearly 1 gigawatt per month. The company has been operating one pod-building factory for several years and this year started building a second one, which it plans to launch at the beginning of next year.
According to the company, installing and commissioning a finished pod at the customer's site should take 48 hours; the production of the pod itself is expected to shorten from three weeks to one day as production increases. The pods can be adapted to customer requirements and to new generations of GPU chips; cooling is handled by a closed liquid loop with air-cooled radiators, without the need for a water connection. The company stated that it is negotiating with energy suppliers about a network of sites with a capacity of 397 megawatts, with the possibility of expansion up to 1.2 gigawatts.
Similar prefabricated pods are already offered by Crusoe and Schneider Electric; Nvidia this year presented the Vera Rubin AI factories and DSX POD concepts. Runware states that, unlike these concepts, it is already actually operating the factory. So far it has raised over $60 million, including $50 million in a Series A round last December, and is planning a Series B round. According to the company, since it started offering pods to external customers a month ago, it has received binding interest worth $1 billion, corresponding to more than 20 megawatts.
Implementation of the plan is so far only an intention and requires extensive hiring, inventory, supply chain logistics, and capital. The article mentions, as a cautionary example, the startup Arrival, which unsuccessfully attempted mass production of electric vehicles. Runware continues to retain dependence on Nvidia for networking components, and imports some parts, such as pumps and fans, from Germany.
Why it matters
If Runware's plan were realized, it could shorten the time needed to expand AI computing capacity from years to months, since, according to the company, customers could plan the deployment of a data center directly around finished, assembly-line-produced pods. So far, however, this is an unconfirmed early-stage intention requiring extensive manufacturing, a supply chain, and capital, and the company itself points to the risks associated with failed mass-production attempts in other industries.
Relevant practical impact
What this means
For a business
For companies planning to deploy AI computing capacity, serial production of modular data centers could in the future shorten the time from order to launch from years to months, but this is an unfinished project whose large-scale operation has not yet been confirmed.
StrategyCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.