Deployment Case Study: Fermi America Hypergrid
The Fermi America HyperGrid AI Data Center Complex, located near Amarillo, Texas, is one of the most ambitious AI–energy projects in development worldwide. Designed as a fully integrated campus, HyperGrid will combine up to 11 GW of private energy generation with 18 million sq ft of AI data center space. The project represents a paradigm shift in siting and powering AI facilities, emphasizing energy autonomy, nuclear integration, and hyperscale growth.
Overview
- Developer: Fermi America, co-founded by former U.S. Energy Secretary Rick Perry
- Partners: Texas Tech University, Hyundai Engineering & Construction, Doosan Enerbility, Westinghouse
- Location: ~5,800 acres near Amarillo, Texas, adjacent to the DOE Pantex site
- AI Capacity: 18 million sq ft of compute floor space planned
- Energy Scale: Up to 11 GW (nuclear, natural gas, solar, wind, batteries)
- Timeline: 1 GW online by 2026 (gas + solar); full nuclear build-out by ~2032
Facility Specifications
| Dimension | Details |
|---|---|
| Site Size | ~5,800 acres |
| Compute Infrastructure | 18 million sq ft of AI data centers |
| Initial Power (2026) | ~1 GW (gas + solar, battery-backed) |
| Full Power (planned) | 11 GW (including 4 × Westinghouse AP1000 reactors) |
| Campus Role | Private “behind-the-meter” grid dedicated to AI workloads |
Energy Integration
- Nuclear: Four AP1000 reactors (1.1 GW each) planned, licensing under review by NRC
- Natural Gas: 600 MW turbines already secured to support Phase 1 operations
- Renewables: On-site solar and wind deployments to balance peak demand
- Storage: Battery energy storage systems (BESS) for grid stability and redundancy
- Design Model: HyperGrid operates as a private energy hub, reducing reliance on ERCOT and public utilities
Partners & Stakeholders
| Partner | Role |
|---|---|
| Texas Tech University | Academic partnership, research integration, workforce pipeline |
| Hyundai E&C | Engineering and construction support for nuclear components |
| Doosan Enerbility | SMR/nuclear component manufacturing and supply |
| Westinghouse | Supplier of AP1000 nuclear reactors |
| Fermi America | Developer and operator, driving the energy-first AI campus model |
Key Challenges
- Nuclear Licensing: Accelerated NRC review may still face political and technical hurdles.
- Construction Risk: AP1000 deployments (e.g. Vogtle) have historically faced delays and overruns.
- Financing: Capital requirements for an 11 GW project are massive, financing structure not fully disclosed.
- Integration: Coordinating nuclear, gas, solar, wind, and BESS in one campus is technically complex.
- Geopolitical Exposure: Reliance on international partners (Hyundai, Doosan) adds supply chain risks.
Strategic Importance
- Energy Autonomy: Demonstrates how AI campuses can bypass strained public grids.
- Scale Leadership: Potentially the world’s largest AI–energy complex by capacity.
- Policy Catalyst: May accelerate U.S. adoption of nuclear-powered AI campuses.
- Model: A reference point for “AI factories” that integrate energy, cooling, and compute from the ground up.
Future Outlook
- Phase 1 (2026): Initial 1 GW operations with gas + solar.
- 2030+: Ramp to multi-GW, nuclear-driven power backbone.
- Long Term: Blueprint for other U.S. and global AI–energy complexes.
FAQ
- Is this the largest AI data center project in the world? By planned capacity (11 GW), HyperGrid rivals or surpasses OpenAI–Oracle Stargate (TX) and Meta Hyperion (LA).
- Why is nuclear included? To provide baseload energy that renewables and gas alone cannot deliver reliably.
- When will it be operational? Target: 1 GW by 2026, full build-out by 2032.
- What makes this different from hyperscaler projects? HyperGrid is energy-first, with private nuclear integration, not just leased grid capacity.
- Will it connect to the Texas grid (ERCOT)? Designed primarily as a behind-the-meter energy island, though tie-ins are possible.