Meta Hyperion: Compressing a 5-GW AI Campus Into a Regional Infrastructure Program

Meta's Richland Parish project illustrates the new meaning of rapid AI-data-center construction. In December 2024, the project was publicly described as a $10 billion, more-than-4-million-square-foot data center. By July 2026, Meta and Louisiana Economic Development described an expanded plan of more than $50 billion, nearly 10 million square feet, and 5 GW of compute capacity housing Meta's Hyperion AI training cluster.

The project is not a finished 5-GW facility. It is a construction and regional-infrastructure program whose scale expanded rapidly after the initial announcement. The relevant case-study question is therefore not how quickly a company can pour concrete. It is how site preparation, contractors, transmission, power supply, water and wastewater, workforce, capital, and community systems can be assembled quickly enough to support an AI campus that changes its own requirements while it is being built.

What the public record establishes

Entergy's December 2024 release described Meta's planned Richland Parish data center as more than 4 million square feet and a $10 billion investment on the Franklin Farm mega site. The release identifies a more-than-1,400-acre certified site, proximity to utility infrastructure and transportation, and a utility plan that included generation, substations, and high-voltage transmission.

In July 2026, Meta stated that it was expanding the AI-optimized Richland Parish data center to 5 GW of compute capacity, calling it the largest in Meta's fleet and the home of Hyperion, its largest multi-gigawatt AI training cluster. Meta stated that the project would support more than 7,500 peak-construction roles and 1,000 operational roles once complete. Louisiana Economic Development described the same expansion as more than $50 billion, nearly 10 million square feet, and one of the largest data centers in history.

The delivery apparatus is also public. Meta identifies DPR Construction, Turner Construction Company, and Mortenson Construction as general-contracting partners. It states that it had contracted more than $1.6 billion with Louisiana businesses and invested more than $1 billion in local infrastructure improvements, including roads, water, and wastewater systems. At the corporate level, Meta reported $72.22 billion in 2025 capital expenditures and forecast 2026 capex of $115 billion to $135 billion, driven principally by infrastructure investment.

System architecture: delivery is a regional coordination problem

Evidence test: proven, claimed, and unresolved

LegacyGrid AI assessment

The strongest lesson from Hyperion is that speed is not a single construction metric. A 5-GW AI-campus plan is delivered only if multiple external systems advance at the same time: grid studies and transmission, generation and substation procurement, civil work, water and wastewater, fiber, equipment supply, skilled labor, finance, permitting, and public accountability. The public record documents an unusually rapid expansion in stated scope. It does not yet show an operating 5-GW campus or establish that every supporting system will arrive on the same schedule.

The 2024 Entergy plan alone included more than 2 GW of planned generation, multiple substations, nearly 100 miles of 500-kV transmission, and additional 230-kV lines. The later 5-GW compute target is substantially larger than that initial generation figure, so an accurate analysis must resist treating early power announcements as the full final supply architecture. The appropriate reading is that the power system, like the campus, must continue to evolve.

A multi-gigawatt AI campus is a regional infrastructure program. Its actual critical path may sit in a turbine queue, transformer factory, transmission right of way, water permit, interconnection study, skilled-labor constraint, or optical-cable supply chain.

Decision lessons

1. Preserve optionality when the end-state is likely to expand. Owners should assess whether easements, switchyard footprints, rights of way, water and wastewater capacity, laydown areas, and backbone fiber can accommodate a changed scale without a full site redesign.

2. Treat every external dependency as a critical path. Schedule governance should track power, water, site, labor, procurement, and permitting dependencies as one integrated program rather than assigning them to disconnected teams.

3. Make community commitments measurable and recurring. Announcements should be the starting point for measurements of local contracts, workforce outcomes, safety, water withdrawals and restoration, public-infrastructure completion, rate effects, and reliability results—not the conclusion.

What this case does not show

The public record does not establish a finished or energized 5-GW facility, final compute load, a final power-resource mix, actual cooling-water use, construction completion dates, final cost allocation, or realized ratepayer and reliability outcomes. The original illustration on this page is a conceptual editorial visual; it is not an as-built campus depiction, final site plan, or Meta construction document.

Primary-source register

1. Entergy Louisiana — “Entergy Louisiana to power Meta's data center in Richland Parish”, December 2024.
2. Meta Data Centers — “Deepening our investment in Richland Parish, Louisiana”, July 13, 2026.
3. Louisiana Economic Development — “Meta commits more than $50 billion for North Louisiana project”, July 13, 2026.
4. Meta Investor Relations — “Meta Reports Fourth Quarter and Full Year 2025 Results”, January 2026.

Professional Practice Boundary: LegacyGrid AI provides advisory, planning, assessment, and systems-engineering analysis. This case study is not final design, permitting, construction, commissioning, legal, financial, or professional-engineering advice. Project facts, targets, and regulatory status may change after publication.