Runware Plans Assembly-Line Factories for Modular AI Data Center Pods
Runware plans to build a factory that assembles modular data center pods on a production line, much as automakers assemble vehicles. The company's targeted production pace is nearly 1 gigawatt of data center capacity per month, which it expects to reach by scaling production at its Romania-based factory to 30 megawatts per day by 2028, according to TechCrunch.
Runware already operates one factory, which it has used for a few years to build pods. It started building a second factory earlier in 2026. That facility is expected to be finished within months and go online early next year.
Build time is the figure the company is pushing down. Runware says a pod once took three weeks to build and will soon take one day as production ramps. Its stated goal is to bring pods online within 48 hours for installation and set-up, a window that does not include build time. The Sonic Inference Pod, which the company unveiled, is a modular data center built in a factory and shipped ready to install.
The pods have a published specification. In an August 2026 blog post, Runware describes its Sonic Pods as modular 1MW inference data centers, with 1,200 GPUs packed into a 20-foot container and closed-loop liquid cooling. Inference is the stage where a trained model produces output for a user. Closed-loop cooling recirculates the same liquid rather than discarding it after each pass.
Runware's website adds that its orchestration layer and the modular pods together form what it calls the Sonic Inference Engine. The company states that its modular AI data centers run at half the cost, and that it offers inference at up to 90% below market rates. It also states that its platform has served more than 10 billion inference requests across image, video, audio, language and 3D, that more than 1 million developers build production AI features on it, and that applications running on its infrastructure reach more than 300 million end users. The platform, the website says, maintained 99.99% uptime over the last 90 days.
On the commercial side, Runware offers serverless GPUs billed by the second, with workloads scaling down to zero when idle. Listed pay-as-you-go pricing for RTX PRO 6000 GPUs is $1.99 per GPU-hour, and reserved serverless capacity is listed as low as $0.63 per GPU-hour. Dedicated bare-metal clusters on Sonic Pods start at 72 GPUs, and the company says it offers more than 400,000 models on one endpoint.
Runware has raised more than $60 million to date and plans to raise a Series B, according to TechCrunch.
In my view, the central question is whether the manufacturing cadence holds. Production targets for a factory still under construction are forward-looking by nature. Moving from a three-week build to a one-day build is a ramp the company has described, and the coming year will have to deliver it.
The broader context here is that most of the delay in standing up AI capacity has come from site preparation, power and installation, rather than from the servers themselves. If a pod can be built, tested and shipped as a finished unit, the gap between committing capital and running inference could shrink. That would matter to anyone whose workloads are constrained by how quickly capacity can be placed.
