Project Camellia marks a meaningful change in how OpenAI approaches the physical layer behind its artificial intelligence ambitions. The company is expanding its infrastructure strategy by leading the design and development of a major campus while continuing to work with infrastructure partners elsewhere. Announced July 22, the Effingham County project puts OpenAI in the lead on design and development while construction and operations arrangements remain under development. The campus will sit inside the Savannah Gateway Industrial Hub in coastal Georgia, turning an industrial tract into a potential anchor for a new regional computing market. OpenAI describes the project as a long-term AI infrastructure development and says it will contract with Georgia Power for 3.2 gigawatts delivered in phases from 2028 through 2032. The significance extends beyond the headline investment because Camellia combines planned compute capacity with a long-term power agreement, infrastructure development and local community commitments.

A $20 Billion Campus With a Longer Infrastructure Horizon

The $20 billion investment makes Project Camellia one of the largest private capital commitments associated with an AI infrastructure project in coastal Georgia. Effingham County says the project represents $20 billion of investment, while Bloomberg has reported that total spending could rise above $30 billion as development progresses toward full scale. The planned campus will occupy roughly 1,400 acres within the 2,600-acre Savannah Gateway Industrial Hub, positioning a previously industrial-focused site for one of the region’s largest technology developments. That scale gives the project a multiyear development profile, with OpenAI planning phased infrastructure delivery between 2028 and 2032. OpenAI says it is still working through infrastructure, phasing, design, financing and its operating model, signaling that the announced scope should not be treated as a final construction package.

Power is a central element of the Camellia design because OpenAI has entered a 25-year agreement with Georgia Power to support the planned facility. Reuters reported that Southern Company’s Georgia Power subsidiary signed a 25-year power supply agreement with OpenAI for the Effingham County project, with the full 3.2-gigawatt requirement scheduled for phased delivery. OpenAI says it will cover the infrastructure and electric-service costs required to serve the facility, while Georgia Power says the arrangement includes financial protections designed to shield its customers from those project costs. Meanwhile, OpenAI has committed to making up to 1,000 megawatts available through demand reduction during periods of high system demand. The arrangement demonstrates how a large AI facility can incorporate operational flexibility into its electricity requirements through demand response.

The Grid Contract Changes the Development Equation

The electricity arrangement illustrates how Project Camellia is tying the cost of required power infrastructure directly to the project rather than existing Georgia Power customers. OpenAI states, “Georgia families will not subsidize this project,” and says it will pay the full cost of the infrastructure and electric-service requirements. Georgia Power said OpenAI will pay the full infrastructure and electric-service costs and that the arrangement includes financial assurances designed to protect Georgia Power customers, consistent with rules approved by the Georgia Public Service Commission in January 2025. The arrangement makes the commercial structure around grid connection an important part of the project because OpenAI must cover the infrastructure and electric-service costs required to serve the facility. It also creates a precedent for developers to bring sophisticated load-management capabilities into campus design from the beginning.

Water is a major part of the project’s infrastructure and community commitments, with OpenAI planning a closed-loop cooling system designed to minimize ongoing water use. The company says the campus will use a closed-loop water system that recirculates water instead of continuously withdrawing and discharging it. OpenAI describes the system as operating like a car radiator and says ongoing water requirements should resemble those of an ordinary workplace with a similar employee population. The company states that the system is intended not to divert water from local communities, a material consideration as residents scrutinize the resource requirements of new industrial development. The approach places thermal management alongside power procurement as a front-end design decision instead of an efficiency feature added later. For infrastructure executives, that matters because cooling architecture increasingly influences site selection, utility planning, equipment configuration and the social license required to operate a high-density computing facility.

Community Benefits Become Part of the Infrastructure Model

Community commitments are another part of Camellia’s attempt to define what OpenAI calls a more accountable infrastructure model. OpenAI has pledged $80 million in community benefits over the life of the project, with spending directed toward priorities such as schools, public safety, health care, utilities, workforce training, housing, veterans’ services and small businesses. The company also plans to make up to $71 million in Codex credits available to eligible Georgia college, community college and technical school students. OpenAI says eligible students will receive $100 in credits through their ChatGPT accounts, connecting the project’s infrastructure investment with skills development for engineering, manufacturing, health care, education, entrepreneurship and skilled trades. The company expects the campus to generate hundreds of millions of dollars in state and local tax revenue and says it expects to become Effingham County’s largest taxpayer.

The social dimension became visible immediately after the announcement, with local residents raising questions about the project’s development process and potential effects. Local residents raised concerns about the development process and the scale of the proposed campus, while the July 23 public information open house drew a large crowd, according to local reporting. Residents questioned how the project had advanced before broader public awareness, while reporting from the Effingham Herald said transparency concerns dominated the gathering. OpenAI says it had already been meeting with officials, community leaders, schools and regional partners and planned the open house to gather further feedback. However, the company’s own statement acknowledges that major work remains on infrastructure, phasing, design, financing and operations. For contractors and equipment providers, the project’s development process remains an important consideration because OpenAI says infrastructure, phasing, design, permitting, financing and the operating model are still being worked through.

Effingham Could Become a New AI Infrastructure Market

Effingham County’s position on the data center map makes Camellia strategically distinct from OpenAI’s better-known infrastructure efforts. The project introduces a large-scale AI computing development into Effingham County through the existing Savannah Gateway Industrial Hub, a 2,600-acre industrial and logistics campus. OpenAI says its first campus in Abilene, Texas, is already operating at scale, while the company continues to pursue infrastructure through partnerships and projects in other U.S. locations. Camellia adds another layer by putting OpenAI directly into the design and development process for a major campus instead of relying solely on external capacity. Still, the project remains at an early stage, and OpenAI has said it will provide more information as the plans develop. That distinction matters for the supply chain because the project remains in development, with OpenAI continuing work on infrastructure, phasing, design, permitting, financing and its operating model.

The deeper market signal from Project Camellia is the convergence of compute strategy and infrastructure strategy inside the same corporate decision. Project Camellia underscores the growing importance of physical infrastructure to OpenAI’s effort to expand the computing capacity required for advanced AI. The project also remains dependent on the completion of infrastructure, permitting, financing, phasing and operating decisions before the planned campus reaches full development. Ultimately, the project’s value will be measured not only by how much capacity it creates, but by how effectively its power, cooling, construction and community systems operate together. Project Camellia therefore offers a preview of a broader infrastructure era in which AI developers increasingly compete not just for chips and capital, but for the ability to build durable physical platforms around them.