Meta expands Louisiana AI data center to 5GW and $50 billion
Meta is expanding its Hyperion AI data center in Richland Parish, Louisiana, to five gigawatts of compute capacity and raising its regional investment to more than $50 billion. The company published the announcement on July 13, calling the project one of the world's largest AI infrastructure investments. Construction began in December 2024, when the original plan was valued at $10 billion.
The new scale puts Hyperion among the biggest data center projects ever attempted. Axios reports that its first section is expected to open in 2028. Meta says the completed site will support more than 1,000 permanent roles, while Louisiana officials expect about 7,500 construction jobs. Local companies have already received more than $1.6 billion in contracts, and Meta is adding over $1 billion for roads, water and wastewater infrastructure.
Power is the central issue. Meta says it will pay the full cost of the energy and related infrastructure used by the campus. Its latest agreement with utility Entergy Louisiana includes seven new natural-gas plants, three grid-scale batteries, nuclear uprates and purchased electricity. Axios says Meta's combined expansion plans will require ten new power plants with roughly 7.5 gigawatts of capacity, about six times New Orleans' peak summer demand.
That comparison explains why this is more than a construction story. Training and serving advanced AI models increasingly depends on access to enormous, reliable power supplies. Five gigawatts does not guarantee better AI, but it gives the company room to run far more chips and compete with the infrastructure programmes of OpenAI, Microsoft, Google, Amazon and xAI.
The local benefits are already visible. Richland Parish teachers received bonuses of up to $50,000 after tax revenues increased, and Meta is donating $5 million to Louisiana Delta Community College for data center training scholarships. At the same time, environmental groups and consumer advocates remain concerned about emissions, water use and whether residents will ultimately absorb part of the grid cost. Meta and Entergy say their agreements will save ratepayers more than $2 billion over twenty years, but that projection still needs to be tested.
For AI users and makers, no new model or feature arrives today. The significance is longer term: the physical cost of AI is becoming as strategic as model design. For businesses and policymakers, Hyperion shows that future AI competition will be shaped by electricity, financing, permits and community support as much as benchmark scores.