OpenAI Targets Savannah For $20B Data Center
A year after OpenAI CEO Sam Altman expressed concerns about the possibility of an artificial intelligence bubble, the firm that created ChatGPT is planning to pump more than $20B into a new 3.2-gigawatt data center outside of Savannah.
OpenAI announced Wednesday that it has contracted with Georgia Power Co. for the necessary power, which will be delivered in stages between 2028 and 2032, the AI operator said in a press release. The move comes months after Georgia Power obtained permission from state regulators to expand the state’s power grid by 10 gigawatts by 2030.
OpenAI has already acquired the land for its data center project, called Project Camellia, in the Savannah Gateway Industrial Hub in Effingham County, according to The Wall Street Journal. OpenAI executives met with local leaders and school officials, as well as Georgia state officials, to drum up local support, OpenAI Vice President Sachin Katti told the publication.
Katti declined to say how much money the firm paid to Georgia Power for the energy allocation but called it a “meaningful amount,” the WSJ reported.
OpenAI also committed to pay the full cost of the infrastructure and electric costs to service its data center and said it would reduce power consumption during peak times of residential customer demand, according to the press release.
The company also said it would spend $80M in community benefits over the life of the project and provide $71M in credits to Georgia colleges and technical school students to use OpenAI’s Codex coding tool.
OpenAI increased its projected computing power spending from $600M earlier this year to $750M through 2030, after the firm secured new cloud-computing agreements with other companies, the WSJ reported, including an increase in its multiyear deal with Amazon Web Services to $138B over eight years. Those agreements have the company racing to lock in a surge in computing power to run its AI models, according to the WSJ.
Altman’s spending at OpenAI has been met with some resistance. In April, OpenAI Chief Financial Officer Sarah Friar told corporate leaders she was worried revenues weren’t rising fast enough for the company to be able to pay for future computing contracts, the WSJ reported. Friar also scaled back a previous claim from Altman that the firm planned to spend $1.4T on computing capacity, and instead told investors the company would likely spend $600M through 2030.
While data center demand is expected to jump overall over the next decade, AI workloads are expected to be a growing portion of the share, according to Goldman Sachs Research. AI consumes 13% of data center compute power but is projected to increase to 28% of the overall market by 2027.
But some experts warn that demand may actually fail to materialize, leaving institutional capital and investors with underutilized swaths of newly developed data centers around the globe.
Last August, Altman told The Verge the AI market was in bubble territory, CNBC reported, comparing it to the dot-com bubble that burst in March 2000.
“When bubbles happen, smart people get overexcited about a kernel of truth,” Altman was quoted at the time. “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes. Is AI the most important thing to happen in a very long time? My opinion is also yes.”
As OpenAI plots its initial public offering, technology journalist and AI industry critic Ed Zitron called OpenAI the “Lehman Brothers of the AI bubble,” saying the firm has been the single catalyst that has led to the AI investment mania in recent years, Business Insider reported.
"The only reason this has kept going so long is that OpenAI has yet to collapse. Its failure would be a watershed moment — the Lehman Brothers of the AI bubble, and an event that would define the end of one epoch, the start of another,” Zitron wrote last month in a newsletter post.
