OpenAI is bringing a new AI data center to Effingham County, Georgia, under a project it is calling Project Camellia. The company announced the buildout in a blog post that pairs the site with commitments to responsible energy, community investment, local hiring, and public access to Codex, its coding-focused AI product. The move deepens OpenAI's push to lock down US compute capacity outside the traditional hyperscale hubs of Northern Virginia, Ohio, and Phoenix.
Effingham County sits just inland from Savannah, a stretch of rural Georgia that has emerged as an industrial magnet on the back of the Hyundai Metaplant and adjacent supplier factories. Project Camellia adds an AI-specific anchor to that industrial base, and it puts OpenAI directly into a region already competing for grid capacity, water rights, and skilled trades workers with automotive megaprojects.
OpenAI framed the announcement around four commitments: energy that the company describes as responsibly sourced, direct investment in the surrounding community, local jobs at the site, and offering residents access to Codex. The company did not publish capacity figures, power draw, or a construction timeline in the announcement.
Key facts
- 01OpenAI announced Project Camellia, a new AI infrastructure site in Effingham County, Georgia.
- 02The project comes with commitments on responsible energy sourcing, community investment, and local job creation.
- 03OpenAI is offering the local community access to Codex, its coding-focused AI system.
- 04Project Camellia extends OpenAI's expanding US data center footprint into the rural Southeast.
The Codex piece is the unusual part. Rather than the standard package of workforce-training grants and STEM scholarships that accompany most data center announcements, OpenAI is dangling access to its production coding model as part of the community offer. The framing positions the data center not just as a taxpayer and employer but as a distribution point for AI tools in a county that would otherwise be a late adopter.
Georgia has become one of the most contested data center markets in the country. Georgia Power, the state's dominant utility, has repeatedly revised its load forecasts upward as AI operators queue up for interconnection, and the state legislature has spent the past two sessions debating how to allocate power, tax incentives, and water permits among competing hyperscale requests. Project Camellia lands in the middle of that debate.
OpenAI's infrastructure spending has accelerated sharply through 2025 and into this year. The company has signed multi-year compute deals with Microsoft, Oracle, and CoreWeave, committed to the multi-site Stargate buildout with SoftBank and Oracle, and is now layering directly-branded projects like Camellia on top of that partner capacity. The pattern is a company that no longer trusts leased capacity alone to keep pace with its training and inference demand.
The Effingham County announcement follows recent BloombergNEF projections that data centers will consume 20% of US electricity by 2035, a forecast that has forced utilities, regulators, and rural counties to negotiate power terms that were unheard of two years ago. Southeast utilities in particular have been the target of AI operator interest because of relatively lower industrial power rates and available land near existing transmission.
Local reception in Effingham County has not yet been detailed in the announcement, and OpenAI did not disclose the specific site, the utility contract, or whether the facility will use closed-loop cooling. Those details typically emerge during the county zoning and permitting process, which is where similar projects across Georgia have run into their sharpest scrutiny over water use and grid impact.
OpenAI also did not name a construction partner or a target energization date for Camellia. The company said only that it intends to build with the community rather than around it, language meant to preempt the friction that has accompanied AI data center siting in Memphis, central Ohio, and the Dublin metropolitan area over the past 18 months.
The open questions on Camellia are the ones OpenAI did not answer: how many megawatts, drawing from which generation mix, cooled how, and delivering how many permanent jobs versus construction jobs. Until those numbers land in a filing or a county council packet, the announcement is a statement of intent rather than a project of measurable scale.
For OpenAI, the strategic logic is straightforward. Every self-branded site the company can point to strengthens its position with regulators and enterprise customers who want to know that OpenAI controls its own compute destiny rather than renting it entirely from Microsoft or Oracle. Project Camellia is small in the context of Stargate but useful as a proof point that OpenAI can execute directly in a rural US market. The bigger test comes when the utility interconnection terms and the community-benefit agreement are made public — that is when the pledges in today's blog post get priced.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
