Meta is sinking $10 billion into Louisiana to build its wildest AI aspirations, setting the template for the grid buildout
Briefly

A massive construction project has transformed over 2,000 acres of former farmland in Richland Parish, northeastern Louisiana, into the site for Meta's largest data center complex. The $10 billion development comprises nine buildings and more than 4 million square feet of server space and sits adjacent to the Haynesville Shale gas field to supply substantial new gas-fired power. Named Hyperion, the project could scale from over 2 gigawatts to as much as 5 gigawatts of computing capacity, potentially using energy comparable to millions of homes to train open-source large language models, amid aggressive hiring and strategic investments in AI firms.
On a quiet patch of former farmland in northeastern Louisiana, a fleet of excavators has leveled more than 2,000 acres of reddish clay earth. This is rural Richland Parish, once a floodplain tangled with meandering bayous and wild canebrake where black bears still wander and a quarter of the 20,000 residents live below the poverty line. Enter Meta-the sixth-largest company in the world by market cap. The tech giant is keen on making Richland home to its wildest AI aspirations- courtesy of a tremendous amount of new gas-fired power. The region has ample land and sits adjacent to Louisiana's huge Haynesville Shale gas field.
In December, construction began on Meta's biggest-yet data center: a $10 billion complex of nine buildings, housing bank upon bank of servers that will take up over 4 million square feet, an area larger than Disneyland. Meta chairman and CEO Mark Zuckerberg isn't stopping there. He dubbed the project "Hyperion" in July-a data center "supercluster" that eventually could use the energy equivalent of 4 million homes and become the world's biggest data center project. Zuckerberg said Hyperion would cover a "significant part of the footprint of Manhattan." The project entails more than 2 gigawatts of computing capacity-Zuckerberg said it could eventually expand to 5 gigawatts-programmed to train open-source large language models.
Read at Fortune
[
|
]