OpenAI announced plans for $20 billion Georgia data center facility with 3.2 GW of power capacity, representing largest single datacenter capex commitment.
OpenAI plans to invest $20 billion to launch Project Camellia in Effingham County, Georgia, positioning the site as one of the largest AI infrastructure projects now moving toward development in the Southeast. The project represents a significant shift in strategy: it is the first known instance at scale where OpenAI has moved from primarily leasing compute capacity from cloud providers to taking a direct role in designing and building its own infrastructure.
The facility will be located in the Savannah Gateway Industrial Hub, an industrially zoned development area approximately 45 minutes outside Savannah. The 2,600-acre global logistics campus is developed by The Broe Group, according to Effingham County administrators. OpenAI expects the data center to generate less traffic than a warehouse project on the same site.
The project requires 3.2 gigawatts of electricity under a 25-year agreement with Georgia Power. Power will be delivered in phases between 2028 and 2032, reflecting a multi-year construction cycle tied as much to grid and utility work as to the data center buildings themselves. Under the arrangement, OpenAI will also provide up to 1,000 megawatts of flexible demand response, allowing power use to be reduced during periods of high demand. This configuration ties together electric transmission, generation, and grid-reliability planning.
As the lead designer and developer—a departure from its previous approach of renting chips from cloud providers such as Oracle and Amazon Web Services—OpenAI's greater control could reshape contractor demand, scheduling pressure, and procurement strategies. The company has hired Brent Mayo, who previously helped lead xAI's Colossus supercomputer build-out in Memphis, as head of data-center build and delivery. His role is expected to include keeping cloud-partner projects on schedule as well as overseeing the Georgia campus.
OpenAI said it is still selecting a partner to build and operate the facility, with significant work remaining on infrastructure, project phasing, design, financing, and the operating model. The company intends to prioritize local contractors and businesses while creating thousands of construction jobs and permanent on-site positions over the project's life. OpenAI also plans to use a closed-loop water system designed to keep ongoing water use low.
The Georgia project stands out against the backdrop of OpenAI's broader infrastructure commitments. According to Wall Street Journal reporting from July 22, 2026, OpenAI has raised its projected spending on computing power to around $750 billion through 2030, up from about $600 billion earlier this year. The company has contracted with Oracle for 6 gigawatts of data center capacity, expanded a multiyear Amazon Web Services deal to $138 billion over eight years, and previously committed additional cloud spending to Microsoft Azure.
The scale and phased delivery of projects like Project Camellia typically point to a long runway of site preparation, utility upgrades, power-distribution work, and staged facility construction. Data center projects of this scale depend not only on vertical construction but also on substations, interconnections, backup systems, electrical equipment procurement, and utility sequencing. The shift toward internal teams around execution could intensify competition for experienced mission-critical contractors, electrical trades, engineering talent, and suppliers of long-lead equipment.