AI infrastructure company Runware said on 30 October 2026 that it is building a factory that “will build 10 gigawatts of new AI compute capacity every year”, producing modular data centers “on an assembly line similar to car manufacturing”, according to its announcement. TechCrunch reported the plan the same day.
Runware says it will offer factory-produced pods for Nvidia Vera Rubin deployments at “$1 million per usable megawatt per year”, and cites Cushman & Wakefield’s figure of $17.6 million per megawatt for average upfront greenfield development costs. By our arithmetic, 17.6 years of pod-only fees would equal that upfront cost; the two figures measure different things.
Short explainer · Runware’s Factory Targets 10GW of AI Data Centers a Year
A vertical news Short narrated from this piece’s headline and opening line.
Business Pill · Grid Cost Allocation
A one-minute explainer of grid cost allocation: who pays for a new power line as electricity demand grows. It teaches the general idea only and says nothing about Runware or this story.
The key insight: Runware says it will offer factory-built pods for Vera Rubin deployments at $1 million per usable megawatt per year, against the $17.6 million per megawatt average upfront development cost it cites; 17.6 years of those fees equal that upfront figure, by our arithmetic.
Update (30 Oct): what Runware’s CEO told TechCrunch. Co-founder and CEO Flaviu Radulescu told TechCrunch that Runware plans to scale production at its Romania-based factory to 30 megawatts of data-center capacity per day by 2028, which TechCrunch describes as nearly 1 gigawatt per month. He also said Runware already has two factories: one it has used for a few years to build the pods, and a second it started building earlier this year that is expected to go online early next year. By our arithmetic, 30 megawatts a day for 365 days is about 11 gigawatts a year, close to the 10-gigawatt annual rate Runware’s announcement targets for the end of 2027. These are company plans, not delivered capacity.
Runware says the manufacturing facility, located in Europe, “will begin production in the first quarter of 2027 and reach that annual production rate by year-end”. It says the production line is designed for “the equivalent of up to 30 megawatts of compute facilities per day”.
The company says the factory will deliver its Sonic Pods in two formats: “a 1-megawatt data center in a 20-foot shipping container and a 2-megawatt data center in a 40-foot shipping container”. Runware says the line will accommodate Nvidia reference architectures for Vera Rubin.
Founder and CEO Flaviu Radulescu said in the announcement: “Demand for AI is outpacing the industry’s ability to build data centers”. TechCrunch spells his surname Fadulescu; we use the company’s spelling.
Per megawatt, as stated in Runware’s 30 October 2026 announcement: its pod-only service rate of $1 million per usable megawatt per year, excluding hardware and electricity, and the $17.6 million average upfront greenfield development cost it cites from Cushman & Wakefield, excluding chips. An annual fee and an upfront cost; 17.6 years of fees equal the upfront figure, by our arithmetic.
The Price of a Megawatt
Runware says the $1 million rate is “based on usable IT capacity and excludes compute hardware and electricity”. Cushman & Wakefield’s $17.6 million figure, as Runware cites it, covers modern facilities in the United States and Canada and excludes chips and GPUs.
Dividing $17.6 million by $1 million a year gives 17.6 years, by our arithmetic. That compares an annual service fee with an upfront build cost, so it shows the scale of the gap, not a like-for-like saving.
Runware lists GPU capacity on Sonic Pods “from $0.63 per GPU-hour for NVIDIA RTX PRO 6000, $3.62 for GB300 NVL72 and $5.07 for Vera Rubin NVL72”.
Counting the Capacity
Thirty megawatts a day for 365 days is about 10.95 gigawatts, by our arithmetic, consistent with the 10 gigawatts a year Runware states. Runware says its current site pipeline “covers 397 megawatts, with expansion to 1.2 gigawatts in the next 6 months”, about three times larger, also by our arithmetic.
Runware says “41 Sonic Pods are being assembled by EOY” at its existing assembly operation in Europe. TechCrunch reports Runware’s claim of $1 billion in committed interest for its pods, which Radulescu told TechCrunch represents “over 20 megawatts, which we’re now working to deliver”.
Runware’s gigawatts describe a factory production rate for data-center enclosures, not power generation, so they are not additive with those figures.
The Structural Read
As we read the announcement, Runware is selling data-center capacity as an annual service rather than a building, with the rate excluding compute hardware and electricity.
As we read the figures, the stated 30 megawatts a day works out to about 10.95 gigawatts a year by our arithmetic, consistent with Runware’s 10 gigawatt target, which Runware says it aims to reach by the end of 2027.
As we read TechCrunch’s report, the claims rest on a company that has raised more than $60 million to date and says it has $1 billion of committed interest representing over 20 megawatts.
Flaviu Radulescu, founder and CEO of Runware, in the company’s announcement, 30 October 2026
“Demand for AI is outpacing the industry’s ability to build data centers”
Three Implications
FOR RUNWARE Runware says the factory will begin production in the first quarter of 2027 and that individual pods are designed to be installed and commissioned in under 48 hours at a prepared site.
FOR BUYERS Runware says customers can bring their own site or deploy at a Runware location, and supply their own GPUs or have Runware provide them.
FOR THE AI BUILDOUT Our prior coverage reports AM Intelligence seeking up to $3 billion in debt for an Nvidia build and Chevron’s roughly 2.67-gigawatt Kilby gas plant; factory output and power-plant gigawatts measure different things.
The framework says “The structural bottleneck isn’t computing, it’s electrons.”
As we read it, Runware is betting that factory-built pods can be placed wherever suitable power already exists. The framework was published in December 2025 and does not discuss Runware or modular data centers.
What Is Not Established
The announcement does not name the factory’s country; TechCrunch reports Runware’s factory as Romania-based. Runware does not disclose the factory’s cost or named customers. TechCrunch reports that Runware, which has raised more than $60 million to date, is finishing the factory with the $50 million Series A it raised last December and plans to raise a Series B. The 10-gigawatt rate is a target for the end of 2027, and the $1 billion of committed interest is Runware’s claim as reported by TechCrunch.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
Runware says its European factory will build modular AI data centers on an assembly line from the first quarter of 2027, aiming for 10 gigawatts a year by year-end, and will offer pods for Vera Rubin deployments at $1 million per usable megawatt per year, against the $17.6 million per megawatt upfront cost it cites.
A note on sourcing. Runware facts come from its 30 October 2026 announcement; the Romania location and the $1 billion committed-interest claim come from TechCrunch. Figures labelled our arithmetic are our own calculations.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
