The US Department of Energy announces a partnership with SoftBank and its affiliate SB Energy to develop a 10 GW data center with its own power supply in Ohio
The U.S. Department of Energy on Friday announced a public-private partnership to develop a major data center with its own power supply …
Context & Ripple Effects
This partnership formalizes an Ohio buildout that had previously surfaced as SoftBank's plan for a consortium-backed power plant sized for AI data centers. It also follows OpenAI and SoftBank's joint investment in SB Energy's data-center growth, putting SB Energy at the center of the group's infrastructure strategy.
The project extends a broader shift from acquiring data-center capacity to assembling power and compute campuses together. SoftBank's earlier Lordstown plant-remodeling plan connected Ohio manufacturing assets to OpenAI-oriented data-center equipment, while OpenAI and Oracle had already outlined a 5GW US capacity expansion.
First-order effects
- SoftBank and SB Energy gain a federal public-private partner for a proposed 10GW Ohio data-center project whose power supply is developed alongside the campus.
- The DOE is directly tied to a large AI-infrastructure development, making this a power-and-compute project rather than a conventional standalone data-center announcement.
Second-order effects
- A self-powered campus reduces dependence on securing equivalent external grid capacity, raising the competitive bar for developers pursuing large AI workloads in power-constrained markets.
- The Ohio plan gives SoftBank and SB Energy a more integrated proposition to prospective tenants: capacity paired with dedicated power, rather than data-center space alone.
Third-order effects
- If such projects advance, AI data-center competition will increasingly hinge on control of powered capacity and the ability to coordinate energy development, financing, and tenant demand.
- Public-sector participation may make large compute campuses a more explicit part of industrial and energy policy, while concentrating execution risk in a smaller set of developers able to deliver both power and facilities.
The trend: AI infrastructure is being built as utility-scale, power-integrated capacity rather than as conventional leased data-center real estate.