OpenAI’s push to build a $20 billion, 3.2-gigawatt data center in rural Georgia has run into stiff local opposition, now capped by a third lawsuit that alleges fraud, civil conspiracy, and civil RICO violations. The fight over “Project Camellia” points to a growing friction in the AI buildout: hyperscale projects that rely on secrecy are clashing with residents who want straight answers on power use, water, and tax breaks.
Quick Take
- The catalyst: A proposed 1,440-acre, 4.4 million-square-foot OpenAI data center in Effingham County, Georgia, is now facing its third major legal challenge from residents and conservation groups.
- The allegations: The latest 44-page complaint accuses local officials and development authorities of civil RICO violations, fraud, and conspiracy to secretly orchestrate nearly $900 million in tax incentives.
- The energy footprint: Project Camellia’s projected 3.2-gigawatt power demand represents a massive shock to the local grid, potentially requiring extensive new natural gas generation and sparking heavy climate pushback.
- Why it matters for AI: As LLM parameters and compute needs scale exponentially, the physical rollout of AI is increasingly throttled not by GPU supply chains, but by local zoning, energy availability, and public backlash over opaque, NDA-driven negotiations.
What happened
Under the code names “Project Gateway” and “Project Camellia,” OpenAI affiliate Octans GA LLC and local development authorities quietly assembled plans for a huge campus in Effingham County, about 30 miles from Savannah. The $20 billion project moved forward through nondisclosure agreements and a 15-year payment-in-lieu-of-taxes deal that would cut the property-tax bill by roughly half, or nearly $900 million.
That secrecy has now produced three separate lawsuits. Earlier filings challenged zoning changes that opened industrial land to data centers (Ordinance 2026-352) and accused officials of violating Georgia’s Open Meetings and Open Records Acts. The newest 44-page complaint goes further, charging fraud, civil conspiracy, and civil RICO violations. It claims county commissioners and the Effingham County Industrial Development Authority worked together to steer the deal past public review. Site-plan votes have been delayed while disputes over cooling water, environmental buffers, and the project’s overall scale remain unresolved.
Why it matters now
Effingham County is a clear example of the physical limits now facing frontier AI. Training and running the largest models demands enormous infrastructure, yet the old hyperscaler approach—NDAs, shell companies, and quick land grabs—is running into resistance as the footprint grows.
Project Camellia’s 3.2-gigawatt load would require new natural-gas plants to keep the lights on, directly conflicting with state climate goals. Environmental groups calculate the added emissions could match those of millions of gasoline cars, turning a local zoning fight into a wider argument over the climate cost of scaling AI.
For the industry, the Georgia cases mark a shift. If communities can use open-government laws and RICO claims to slow or stop $20 billion campuses, the pace of next-generation model deployment will hit real-world constraints. Intelligence cannot outrun the grid or local consent.
Who is most affected
| Stakeholder | Impact | Why |
|---|---|---|
| OpenAI / Hyperscalers | Deployment delays | Cascading litigation and public scrutiny threaten the timeline of a critical 3.2 GW compute cluster. |
| Local Utilities (Georgia Power) | Extreme grid stress | Fulfilling a multi-gigawatt load requires rapid, likely fossil-fuel-based, grid expansion and transmission overhauls. |
| Effingham Residents | Tax and resource strain | Facing potential noise, water usage, and the financing of a $900M tax abatement package negotiated without initial public input. |
| Local Governments | Legal liability | The use of NDAs and closed-door meetings to secure AI investments is now a target for RICO and fraud lawsuits. |
About the analysis
Researched and written with AI assistance under the editorial responsibility of the i10X News team. The title image is an AI-generated illustration. Read our AI Use Policy. Spotted an error? Request a correction.
This is an independent, research-based i10x analysis built for AI founders, infrastructure investors, and policy watchers tracking the physical scaling of intelligence. It synthesizes local investigative reporting, environmental policy analysis, and legal filings surrounding the Project Camellia dispute to map the real-world friction of AI deployment.
i10x Perspective
The era of stealth hyperscale deployment is over. As AI energy demands cross the gigawatt threshold, data centers are no longer invisible backend utilities; they are massive, community-altering industrial projects. The legal siege in Georgia proves that while Silicon Valley can iterate LLMs at lightspeed, it cannot fast-track the physical reality of land, water, and local democracy. We’ve seen this pattern before with other large infrastructure builds: once the scale becomes visible, transparency becomes non-negotiable. To win the AI infrastructure race, companies will have to abandon the NDA playbook and start negotiating directly—and transparently—with the communities powering their compute.
