OpenAI's Georgia Plan: What $20 Billion Means for Local Power
OpenAI is planning a huge $20 billion data center, "Project Camellia," in Effingham County, Georgia, requiring an enormous 3.2 gigawatts of power. This project has sparked major concerns from environmental and consumer groups regarding its energy footprint and potential impact on local electricity customers. The debate highlights the broader environmental challenge of rapidly expanding AI technology, where the energy demand of powerful systems could significantly affect global climate goals and resource management. This escalating need for electricity and water underscores the urgency for sustainable solutions as AI continues to grow.
OpenAI's plan to build a massive $20 billion data center in Effingham County, Georgia, is bringing the energy demands of artificial intelligence into sharp focus. This project, known as Project Camellia, aims to draw 3.2 gigawatts of power, a staggering amount that has led to scrutiny from environmental and consumer advocates. They are pushing for more details about the power contract and better protections for Georgia's electricity users.
Environmental organizations, including the Southern Alliance for Clean Energy, Sierra Club, and Natural Resources Defense Council, have formally objected to the proposed agreement. They argue that key details about the project's infrastructure costs and potential effects on other power customers remain unclear. The groups stressed that a contract failing to cover all costs could leave everyday Georgia Power customers footing the bill. This is particularly concerning because 3.2 gigawatts is an exceptionally large demand; typical data centers use far less, while even large AI data centers rarely exceed 100 megawatts.
This project also brings to light the challenge of tracking ChatGPT's overall environmental impact. While the energy used for a single AI query might seem small, the sheer scale of billions of prompts daily means the total energy footprint is huge. As AI use expands, the overall demand for electricity can soar, even if efficiency per query improves. The International Energy Agency (IEA) estimates that global data centers consumed around 415 terawatt-hours in 2024, with demand expected to more than double by 2030. AI is a major driver, with electricity use from AI servers projected to grow approximately 30% annually through 2030.
Beyond electricity, AI data centers also require significant amounts of water for cooling. OpenAI states that Project Camellia will use a closed-loop cooling system to minimize water consumption. However, studies by the United Nations University estimate that global data centers could have a water footprint of about 9.3 trillion liters by 2030. This shows that AI’s environmental effects stretch far beyond just electricity use, encompassing critical resources like water.
OpenAI is also developing other large data centers, such as the 1-gigawatt Stargate project in Michigan and facilities in Norway that plan to use renewable power and efficient cooling. These efforts demonstrate a recognition of the need to address the environmental impacts of their expansion. However, the rapid growth of AI infrastructure underscores the urgency for greater transparency, efficiency, and a commitment to clean energy and sustainable water practices to mitigate the climate crisis.
Ultimately, the key question for ChatGPT's environmental impact isn't just about single prompts, but the combined effect of billions of interactions, increasingly powerful AI models, and the massive data centers required to run them. Lowering energy use per task is helpful, but the long-term environmental solution hinges on rapid improvements in efficiency, the widespread adoption of clean power, reduced water use, and clear reporting on the resources powering our AI future.