SoftBank is planning to build a massive AI data center complex in Ohio capable of drawing up to 10 gigawatts of power, according to a Bloomberg report.
The facility would be developed on federally owned land at a former uranium enrichment site, with the first phase expected to deliver about 800 megawatts of capacity by early 2028 at a cost of $30 billion to $40 billion, making it one of the largest computing hubs globally.
To support the buildout, SoftBank is backing roughly $33 billion in natural gas power infrastructure, with turbines already sourced and expected to be deployed across the region through the end of the decade. The total planned generation capacity of about 9.2 gigawatts would rival some of the largest power projects in the US.
The scale reflects the surge in demand for AI infrastructure, as hyperscalers and governments race to secure compute and energy capacity. A 10-gigawatt data center would consume power comparable to millions of homes, putting pressure on grids already struggling to keep up with AI-driven demand.
The project is tied to a broader $550 billion US-Japan investment framework that includes energy and industrial infrastructure, and comes as policymakers push to secure domestic capacity in the global AI race.
SoftBank has not yet disclosed customers for the site, but said partners will be involved in sourcing chips and equipment. The company is working with local utilities to upgrade transmission infrastructure, with about $4.2 billion earmarked for grid expansion.
The proposal also highlights growing tension around AI energy use. Data center expansion has triggered backlash in parts of the US over rising electricity and water demand, even as governments prioritize building out capacity to compete with China in advanced technologies.
Disclosure: This article was edited by Estefano Gomez. For more information on how we create and review content, see our Editorial Policy.
