Meta announces its largest data center in Richland Parish, Louisiana (Hyperion)

A campus for AI training

Large language models are trained in data centers: buildings filled with servers, networking equipment and cooling systems, where many chips run for a long time on the same job. The size of such a building is usually described in two ways — floor area, and the electrical power it can deliver to the machines inside.

The announcement

In December 2024, Meta announced that it would build its largest data center to date in Richland Parish, Louisiana — its 23rd data center in the United States.^[1] The campus was described as 4 million square feet, representing an investment of more than $10 billion in the state and supporting more than 500 operational jobs.^[1]

Meta tied the site to open models rather than ordinary web traffic. The company said the campus would deliver more than two gigawatts of compute capacity for training future open-source large language models.^[2] A gigawatt is a billion watts; two gigawatts is the scale of electrical load at which a data center stops being only a construction project and becomes part of the grid's planning problem.

Construction and local investment

Meta said more than 5,000 construction workers were expected on site at peak construction.^[1] It also said it was investing more than $200 million in local infrastructure improvements, a figure that a later Meta update described as expanded to more than $300 million.^[1][2]

Electricity was part of the announcement. Together with its energy partner Entergy, Meta committed to bringing at least 1,500 megawatts of new renewable energy to the grid to match the data center's electricity use.^[1]

Hyperion

The Richland Parish campus is home to Hyperion, which Louisiana Economic Development describes as Meta's largest AI training cluster.^[3]

Expansion

In July 2026, Meta committed $50 billion to the Richland Parish project, expanding the campus to nearly 10 million square feet and 5 gigawatts of IT capacity.^[3] Set beside the original announcement — 4 million square feet, more than $10 billion, more than two gigawatts — the later numbers show how far the planned scale of AI infrastructure moved in a short period.

Why it matters

The project combines things that are usually discussed separately. For the state, it is a construction site, a set of local infrastructure works and a comparatively small number of permanent jobs — 500-plus operational positions against more than 5,000 people at peak construction.^[1] For Meta, it is where it plans to train future open-source models, and the power commitment with Entergy makes the model roadmap a question of generation capacity as well as chips.^[1][2] Louisiana Economic Development's description of Hyperion as Meta's largest AI training cluster places the site at the centre of that plan.^[3]