OpenAI plans to spend $30B+ on a massive new data center in Georgia, securing 3.2GW of energy, with several hundred MWs set to come online starting in 2028
OpenAI’s Georgia plan extends a rapid shift from generalized cloud demand to named, multigigawatt capacity commitments. It follows OpenAI’s disclosure that it had contracted for 10GW of US AI compute capacity, including more than 3GW added in a single 90-day period.
The project also gives physical form to an expanding capital plan: related coverage says OpenAI lifted its projected cloud spending through 2030 to about $750B. A phased 2028 start means this is a longer-dated buildout rather than near-term capacity.
First-order effects
OpenAI adds a planned 3.2GW Georgia power allocation to its US compute pipeline, with several hundred megawatts targeted to begin coming online in 2028.
A project of this scale concentrates demand for grid interconnection, data-center construction and AI hardware around a single deployment schedule, making execution against the 2028 ramp consequential for OpenAI’s available compute.
It reinforces the pressure on other major AI builders and cloud suppliers to secure power and site capacity earlier, rather than treating compute as capacity that can be added quickly.
Third-order effects
If similar commitments continue, AI capacity planning will increasingly resemble utility-scale infrastructure development: power availability and construction timelines will shape model deployment as much as cloud contracts do.
The pattern raises compute-execution risk: larger announced capacity pipelines can improve supply visibility, but their value depends on power delivery and build schedules materializing as planned.
The trend:AI infrastructure is becoming a long-duration, power-constrained capital buildout in which access to gigawatt-scale electricity is a core competitive input.
Bloomberg's take on OpenAI infra spend is different than WSJ. Much more nuanced. “The startup is currently looking for partners to help finance and develop the data center, Katti said, and may opt to do some of the operation itself.” “OpenAI plans to spend tens of billions of
We're developing Project Camellia, a long-term AI infrastructure project in Effingham County, Georgia. There is significant work ahead. Here is the approach guiding that work: - Georgia families will not subsidize this project. OpenAI will pay all infrastructure and
NEW: Lot's of scoop-y details in my latest story for @WSJ - OAI raised its projected spending on compute to ~ $750B through 2030 - OAI is planning a 3.2 GW data center in Georgia - Sam Altman hired Brent Mayo, one of the architects of Elon Musk's data-center build-out 👀 [image]
👀 WSJ: OpenAI “said Wednesday it would invest $20 billion to kick off a data center called Project Camellia, in Effingham County, Ga. Sachin Katti, OpenAI's vice president of compute strategy, said the company has contracted with utility Georgia Power to receive 3.2 gigawatts
Ruh-roh for certain companies “The project represents the first site in which OpenAI is the lead designer and developer. At its other sites, OpenAI rents chips from cloud providers such as Oracle and Amazon Web Services.”
Very proud to have led this project with our incredible team at OpenAI and can't wait to see the community benefit we can deliver to Effingham County. …
Three separate AI infrastructure announcements. One single day. 6.2 gigawatts of AI infrastructure. wtf - OpenAI: 3.2 GW for Project Camellia (~$20B initial investment, ~$750B projected compute spend through 2030). - SpaceXAI: reportedly planning another Texas AI campus as [image…