OpenAI Eyes $500 Billion Data Centre on Federal Land as AI Compute Race Heats Up

OpenAI is in advanced discussions to lease a proposed 10-gigawatt data centre campus on federal land in southern Ohio, a facility that could cost at least $500 billion to build, making it one of the most ambitious AI infrastructure projects ever announced.

The deal, reported by The Information and citing two people with direct knowledge of the discussions, would see the facility developed by SB Energy, a unit of SoftBank, on Department of Energy land. OpenAI would control the equipment under a 20-year lease, with payments beginning once operations commence. The first phase is expected to be operational by 2028.

Nvidia is understood to be a key partner in the arrangement, expected to supply hardware for the facility and provide a financial guarantee for both OpenAI's lease and SB Energy's project financing. Neither OpenAI nor Nvidia had commented publicly at the time of writing.

The scale of the proposed campus, ten gigawatts, is extraordinary by any measure. For context, the entire US power grid currently generates around 1,100 gigawatts. A single facility of this size would represent a meaningful fraction of the national energy demand required just to run artificial intelligence workloads.

The Ohio deal is part of a broader wave of hyperscale AI infrastructure investment sweeping the technology industry. Earlier this year, OpenAI paused its proposed Stargate data centre project in the United Kingdom, citing regulatory hurdles and elevated energy costs. Apollo and Blackstone are separately financing a $35 billion AI capacity expansion for Anthropic using Broadcom chips.

The trajectory of AI infrastructure spending is becoming one of the defining stories in enterprise technology. For organisations across the Gulf region and beyond that are building AI strategies today, the question of compute access, who controls it, where it is located, and at what cost, is increasingly central to long-term planning. The UAE's confirmed role as host of the 2028 Artificial Intelligence Summit underlines how directly the region's leadership is positioning the Gulf as an active participant in global AI infrastructure governance, not merely a consumer of Western compute capacity. For enterprise technology leaders evaluating which AI platforms to build on, the compute dependency question connects directly to platform selection: the GCC-specific comparison of GPT-5, Claude, and Gemini examines how infrastructure ownership and data residency factor into that decision for Gulf organisations.

The facility would be among the largest data centres ever constructed, and its development timeline will depend heavily on regulatory approvals, grid infrastructure expansion, and the readiness of Starship, SpaceX's fully reusable rocket system, which underpins related orbital compute ambitions across the industry.