News

OpenAI Infrastructure budget increases to US$750 billion, with the first US$20 billion allocated to Georgia data center OpenAI announced on July 22, 2026, that it plans to invest US$750 billion in AI infrastructure construction by 2030. This budget is a significant increase of about 25% from the estimate earlier this year.

2 min read
OpenAI infrastructure budget increased to US$750 billion, with the first 20 billion allocated to the Georgia data center. OpenAI announced on July 22, 2026, that it plans to invest US$750 billion in AI infrastructure construction by 2030. This budget has increased significantly by about 25% from the estimate earlier this year. Against the background that the flagship data center project "Stargate" has stalled, this move signals that OpenAI is re-locking in long-term computing power advantages through more aggressive capital expenditures. As the first project of this huge investment plan, OpenAI plans to spend US$20 billion to build the 1,400-acre "Project Camellia" data center park in northwest Savannah, Georgia. The park has reached a cooperation with Georgia Power Company and is expected to gradually obtain at least 3.2 gigawatts of power supply from 2028 to 2032, and OpenAI will bear all new infrastructure and power service costs. In order to reduce the impact on the local power grid, OpenAI committed to reducing load by up to 1 gigawatt during peak power consumption periods. At the same time, the project received a 15-year 50% property tax exemption from Effingham County. Despite the large-scale electricity consumption of the project, neither party has disclosed a specific clean energy transition plan. Regulatory documents show that of the nearly 9.9 GW of new power generation capacity Georgia Power plans to add, more than half (approximately 5.8 GW) will rely on natural gas, with the rest supplemented by photovoltaics and energy storage. This has triggered market concerns that the expansion of computing power will increase fossil fuel dependence and carbon emissions. It is worth noting that OpenAI recently recruited Brett Mayo, an executive who led the construction of the xAI Memphis Colossus data center, to be responsible for the data center business, which may indicate that the first batch of GPU deployment nodes will be launched as early as 2028. At a critical period when generative AI has entered a critical period of high-end model training and large-scale application of inference, the implementation of OpenAI's huge infrastructure funds not only highlights the in-depth game between leading manufacturers in computing power and power resources, but also reflects the practical challenges faced by the construction of ultra-large-scale AI data centers in reshaping the regional power grid layout and promoting the transformation of the energy structure. via AI News (author: AI Base)