The Eastern Herald reported on July 24, 2026 that OpenAI plans to build Project Camellia, a $20 billion, 1,400-acre AI data center campus near Savannah, Georgia that will draw at least 3.2 gigawatts from Georgia Power. The article says this project is part of an expanded $750 billion AI infrastructure spending plan through 2030 and is backed by a 15-year 50 percent property tax break approved by Effingham County.
This article aggregates reporting from 2 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Project Camellia is what it looks like when an AI lab starts to behave like an integrated utility. A single 3.2‑gigawatt campus, backed by a local tax deal and custom grid build‑out from Georgia Power, effectively gives OpenAI its own dedicated power plant city. That scale, combined with the reported $750 billion multi‑year infrastructure budget, would make OpenAI one of the largest single drivers of gas‑fired generation and transmission upgrades in the Southeast.
For the race to AGI, this is a direct play to remove compute as the binding constraint. A campus of this size, operating over a four‑to‑five‑year window, can support multiple generations of frontier training runs, heavy agent workloads and large‑scale evaluation infrastructure under one operational roof. It also reduces dependence on Microsoft, Oracle or other cloud partners for the absolute top end of workload density. If Stargate, the earlier OpenAI–Microsoft–Oracle mega‑project, is facing headwinds, Camellia looks like the hedge: own the land, own the power contracts, own the chips.
The flip side is exposure. OpenAI is now bound to the politics of rate cases, gas plant permitting, local water and land‑use battles, and a national debate about whether hyperscale AI justifies that much new fossil generation. How regulators and communities respond will tell us a lot about how fast future AGI‑class infrastructure can be built in democratic economies.


