Runware plans to assemble AI data center pods on a production line, treating them like cars rather than custom-built structures. The company, which unveiled the Sonic Inference Pod months ago, told TechCrunch it intends to scale output at its Romania-based factory to 30 megawatts of capacity per day by 2028. That pace equates to nearly 1 gigawatt per month. At that rate, Runware would produce roughly as much power in a single year as the entire United States added to its data center capacity this year, approximately 10 gigawatts.
While firms such as Crusoe and Schneider Electric already sell prefabricated units, Runware claims it can deliver them faster. Nvidia also announced the Vera Rubin AI factories and DSX POD earlier this year to help companies build their own infrastructure. The distinction here is that Runware has constructed a factory to do the work, rather than simply designing tools for others to build one.
Flaviu Radulescu, co-founder and CEO, said production will ramp up gradually starting next year. The company currently operates two facilities: one used for several years to build the pods, and a second site under construction that should finish within months and go online early next year. Customers may provide their own site for assembly, or Runware can locate one for them.
The goal is to bring the pods online within 48 hours. This timeframe covers installation and setup, not the manufacturing of the unit itself. A pod previously took three weeks to build but will take one day once production scales. Runware will supply spare parts if components fail. Clients can rent compute from the provider, supply their own GPUs, and pay a monthly fee for the pods and operations. Power and internet are charged separately.
Pods can be customised to meet specific requirements and the latest GPU chips. Chip delivery times often lag, which works in Runware’s favour. Radulescu noted that their pod delivery times are shorter than current GPU lead times, giving them the window to adapt infrastructure to the hardware ordered. “For new GPU generations, our in-house hardware team can design, build, and test components itself, allowing us to adapt quickly,” he said.
These remain plans until execution proves them. The factory requires significant staff, inventory, supply chain logistics, and capital. The electric vehicle startup Arrival, once valued at over $13 billion, struggled to mass-produce cars cost-effectively compared with traditional plants. Runware, which has raised more than $60 million to date, plans to raise a Series B. It is currently finishing the new factory using the $50 million Series A raised last December.
“Our cost structure is very different,” Radulescu said, explaining how the company seeks to avoid Arrival’s fate. “We do have headcount and materials cost,” he continued. “But our factory staff is cost-effective relative to the value of deploying compute for AI.” Designing own systems and buying widely available, affordable parts helps keep costs low. “We’ve brought in a highly experienced plant manager from the automotive industry to work alongside our chief of hardware and help us get that execution right,” he added.
Sourcing components remains a challenge. Although Runware designs and builds its own cooling and pod systems with standard parts, some items such as pumps and fans come from Germany, and the company still depends on Nvidia for networking. Still, Runware claims it has received $1 billion in committed interest for its Sonic pods since offering them to outside customers a month ago. “Representing over 20 megawatts, which we’re now working to deliver,” Radulescu said.
Data centers in the U.S. and abroad have faced criticism over energy consumption and grid impact. When pressed, Radulescu stated his company’s pods require only power and an internet connection. They use a closed-loop liquid cooling system with air-cooled radiators to release heat, meaning they do not need a water connection. “We’re working with power providers on a site pipeline covering 397 megawatts, with potential to expand to 1.2 gigawatts,” he said, referring to planned sites. He added that the company deploys modular data centers on sites much smaller than those used by hyperscalers like Amazon, Microsoft, and Google.
The sites typically already have power, so they should not further strain the grid or disrupt the community. The pods run completely silent up to temperatures of 86 degrees Fahrenheit. “We use energy from renewable sources whenever possible,” he continued. “We are actively working on expanding our network of energy providers and sites. We’re aiming to have renewables as our main energy source.”
Next, Runware will bring the new factory online early next year, then expand to the U.S. and other regions worldwide. Modular systems are the future and will soon become “the standard for large AI deployment,” Radulescu said.
“Customers should be able to plan a hyperscale data center around factory-built pods from the beginning and have it working in months instead of years,” he added.
What it means
For people making things, the shift from custom builds to factory lines means faster access to compute. Instead of waiting weeks for a site to be prepared and built, a pod can be installed and running in two days. This speed allows teams to deploy models without being held back by infrastructure delays. The ability to swap in new GPU generations quickly also means hardware upgrades will happen faster than before.
