Skip to content
AI Usage3 min read

Georgia Power Cleared for 3.2 GW OpenAI Data Center Deal

Georgia Power receives regulatory approval to supply up to 3,200 MW of electricity for OpenAI's $20B Project Camellia data center campus in Effingham County.

AB

Author

AUG Bot

Published

Power grid transmission infrastructure serving a hyperscale AI data center campus in Georgia

Georgia Power Cleared for 3.2 GW OpenAI Data Center Deal

State regulators approve 3,200 MW supply contract for Project Camellia in Effingham County with 1,000 MW grid demand-response commitment

State regulators in Georgia have approved a major power contract between Georgia Power and OpenAI for "Project Camellia," a 3.2-gigawatt data center complex in Effingham County. The deal will deliver up to 3,200 megawatts of electricity between 2028 and 2032 to support OpenAI's expanding artificial intelligence workloads.

Key details

Under the approved agreement, Georgia Power will scale power delivery in phases to supply the $20 billion facility, which requires an electrical load roughly equivalent to the capacity of three nuclear reactors at Plant Vogtle. OpenAI has committed to funding all infrastructure and electrical service upgrades needed to connect the facility to the regional power grid.

To safeguard grid stability during peak energy demand, OpenAI agreed to a flexible demand-response mechanism allowing Georgia Power to reduce the campus's power draw by up to 1,000 megawatts. Georgia Power stated that the influx of revenue from large customers like OpenAI is expected to generate $950 million annually starting in 2029, projecting a $180 annual bill reduction for typical residential customers using 1,000 kilowatt-hours per month. OpenAI also announced plans to use a closed-loop cooling system to minimize water consumption and provide annual independent audits of its environmental commitments.

Why this matters

The 3.2-gigawatt deal represents one of the largest single-facility power supply contracts ever executed for artificial intelligence infrastructure. By incorporating a 1,000-megawatt demand-response obligation and requiring full upfront infrastructure funding from the developer, the agreement establishes a template for managing utility-scale AI energy demand without straining grid capacity or shifting costs onto local consumers.

Context

The approval follows growing regulatory and public scrutiny over AI data center expansion across the United States. In Georgia and neighboring states, rapidly rising electricity demand from hyperscale facilities has spurred debates over utility generation planning and potential rate increases for residential consumers. The inclusion of flexible demand management and closed-loop water systems reflects shifting industry standards as tech companies face stricter municipal and state oversight.

Risks and open questions

While the contract includes ratepayer protection mechanisms and demand-response capabilities, integrating a 3.2-gigawatt load presents ongoing grid reliability and environmental management challenges. Questions remain regarding the specific fuel mix Georgia Power will deploy to supply the additional capacity between 2028 and 2032, as well as whether real-world water recycling in closed-loop systems will match projected conservation benchmarks.

What happens next

Georgia Power and OpenAI will proceed with preliminary engineering and site infrastructure development in Effingham County ahead of the initial power delivery phase scheduled for 2028. State regulators and Public Service Commission staff will monitor project milestones and review OpenAI's annual independent compliance audits as construction progresses.


Source: CBS News Published on AI Usage Global, author: AUG Bot

Older post
Related

Read more

More posts that expand on the topics, companies, and AI trends covered in this story.

Digital representation of power transmission lines and data center grid queue management in the UK
AI Usage

UK Grid Faces 73 GW Phantom Data Center Backlog Under Ofgem Rules

UK energy regulator Ofgem proposes strict financial hurdles to purge 73 GW of speculative 'phantom' data center applications clogging the power grid.

Digital representation of carbon removal technology and AI data center energy infrastructure
AI Usage

Climeworks Urges Carbon Removal Inclusion in Rapid AI Infrastructure Expansion

A new whitepaper from Climeworks Solutions outlines strategic pathways for data center developers and hyperscalers to integrate carbon removal into AI infrastructure planning.

Digital representation of a high-density AI data center campus and compute server racks
AI Usage

Anthropic Signs $45 Billion Compute Deal with Nscale

Anthropic agrees to a $45 billion, six-year compute lease with Nscale for Nvidia Vera Rubin server capacity from a West Virginia data center.