Effingham County OpenAI Data Center's Water Needs Under Scrutiny
OpenAI's proposed data center in Effingham County, Georgia, intends to use a closed-loop dry cooling system to minimize onsite water use. However, experts warn that this system will significantly increase the data center's electricity demand, leading to substantial indirect water consumption at power plants. Residents and scientists are questioning the overall water footprint and the system's effectiveness in Coastal Georgia's humid climate.
Effingham County, Georgia, is facing scrutiny over the water demands of a proposed OpenAI data center after its initial application lacked clear information on water supply. The Savannah area's existing water struggles, particularly with the Floridan aquifer being a designated "red zone" for new withdrawals, have heightened local concerns regarding the project's environmental impact.
OpenAI spokesman Aaron McLear stated that the data center will utilize a closed-loop dry cooling system, which rejects heat to outdoor air and limits ongoing water use after an initial fill, avoiding the evaporative cooling towers common in many facilities. However, Effingham County Manager Tim Callanan estimated the initial fill could require around 25 million gallons. Experts like Akanksha Menon, a Georgia Tech assistant professor, and Michelle Addington, a retired architecture professor, questioned the system's effectiveness in Coastal Georgia's hot, humid climate, suggesting it might not dissipate full heat loads or could be highly inefficient.
A significant concern highlighted by experts is the indirect water consumption associated with the data center's electricity needs. Professor Kelly T. Sanders from USC explained that dry cooling significantly increases electricity demand, which in turn requires substantial water for power generation, especially given Georgia's energy mix (natural gas, nuclear, coal). Tonya Bonitatibus, the Savannah Riverkeeper, noted that energy is the state's largest water user. Sanders calculated that the 3.2 gigawatt data center could indirectly consume between 4 billion and 8 billion gallons of water annually from power plants.
OpenAI maintains the system will be designed for Southeast Georgia's extreme weather conditions but could not provide specific design temperatures or the name of the design firm, leaving many questions about the project's total water footprint unanswered. The trade-offs between onsite and offsite water usage remain a central point of discussion.