Runware Takes a Manufacturing Approach to AI Data Centres
AI infrastructure startup Runware is exploring a new way to build data centres by manufacturing them on assembly lines rather than constructing them entirely at customer locations.
The company plans to produce modular data centres in factories, using a manufacturing model similar to automobile production. Its objective is to reduce deployment time and help businesses access computing infrastructure faster as demand for artificial intelligence continues to grow.
The initiative builds on Runware's Sonic Inference Pod, a modular data centre designed to be manufactured in a factory, transported to a customer site and installed with minimal construction work.
According to CEO and co-founder Flaviu Fadulescu, the company aims to bring these pods online within 48 hours of installation and setup.
Building Data Centres on an Assembly Line
Runware currently operates two factories in Romania. One has been used to manufacture its modular data centre pods, while the second is expected to become operational in early 2027.
The company plans to gradually increase manufacturing capacity, targeting production of 30 megawatts of data centre capacity per day by 2028. This would represent nearly one gigawatt of capacity per month.
The manufacturing approach is designed to standardise production, improve efficiency and reduce the time required to deploy computing infrastructure.
Runware says a pod that previously took approximately three weeks to manufacture could eventually be produced in just one day as production scales.
Customers can provide their own installation sites or work with Runware to identify suitable locations. The company also offers options for customers to supply their own GPUs or rent computing capacity directly.
Sonic Inference Pods Designed for Faster Deployment
Runware's Sonic Inference Pods are modular computing facilities designed to support AI workloads.
Unlike conventional data centres that require extensive construction, these units are manufactured in advance and transported to locations with suitable power and internet connectivity.
The pods feature a closed-loop liquid cooling system with air-cooled radiators to release heat. This design reduces the need for a direct water connection.
The company also says its standardised architecture allows it to adapt the pods to different customer requirements and newer GPU generations.
Runware's strategy is to separate infrastructure manufacturing from the time-consuming process of building a traditional data centre at every location.
Targeting 30 Megawatts of Daily Production by 2028
Runware has set an ambitious manufacturing target. By 2028, it wants its Romania-based factory to produce 30 megawatts of data centre capacity every day.
The company says it has received approximately $1 billion in committed customer interest for its Sonic pods since making them available to external customers.
That interest represents more than 20 megawatts of capacity that the company is working to deliver. However, customer interest and commitments should not be confused with completed deployments or confirmed revenue.
Runware has raised more than $60 million to date, including a $50 million Series A round announced in December 2025. It plans to raise a Series B to support its next stage of expansion.
Manufacturing Challenges Could Slow Expansion
Despite the ambitious plans, scaling modular data centre production will involve significant operational challenges.
The company will need sufficient inventory, skilled factory workers, reliable component suppliers and efficient logistics to meet its manufacturing targets.
Some components, including pumps and fans, are sourced from Germany, while the company continues to depend on Nvidia for networking equipment.
Runware has also brought in an experienced plant manager from the automotive industry to help manage manufacturing operations.
The experience of electric vehicle startup Arrival, which struggled to scale cost-effective vehicle production, illustrates the difficulties of turning an assembly-line concept into a sustainable manufacturing business.
Runware will need to demonstrate that its model can deliver consistent quality and competitive costs as production increases.
Addressing AI Infrastructure and Energy Requirements
The rapid growth of AI is increasing demand for data centre capacity, creating pressure on electricity infrastructure and other resources.
Runware argues that modular facilities can be deployed at smaller sites that already have access to power, potentially reducing the need for large-scale construction projects.
Its cooling system is designed to operate without a direct water connection, while the company says it aims to use renewable energy wherever possible.
Runware is working with power providers on a site pipeline covering 397 megawatts, with potential to expand to 1.2 gigawatts.
However, the environmental impact will depend on the energy sources, local infrastructure and operating efficiency of individual deployments.
The Future of AI Data Centre Manufacturing
Runware's manufacturing strategy reflects a broader shift in how technology companies are approaching AI infrastructure.
As demand for computing power increases, businesses are exploring modular systems that can be manufactured in advance, transported and deployed more quickly than traditional facilities.
The company's target of producing 30 megawatts per day by 2028 is ambitious, and execution will determine whether its manufacturing model can scale successfully.
If Runware delivers on its plans, factory-built data centre pods could become another option for companies looking to expand AI computing capacity without waiting years for conventional infrastructure projects.
