Fermi America partners with South Korea’s Doosan for 11GW nuclear AI campus

Fermi America has signed a strategic partnership with South Korea’s Doosan Enerbility to advance nuclear power development at its massive 11-gigawatt HyperGrid AI campus in Amarillo, Texas. The memorandum of understanding covers both traditional large-scale nuclear plants and cutting-edge small modular reactor technologies for what promises to be one of the world’s largest AI infrastructure projects.

The HyperGrid campus represents a new scale of AI infrastructure, featuring 18 million square feet of data centers—roughly equivalent to 315 football fields of computing space. This facility will require up to 11 gigawatts of continuous power, enough electricity to power approximately 8 million homes, sourced from a hybrid mix of natural gas, renewable energy, and nuclear power.

Doosan Enerbility, headquartered in Changwon, South Korea, stands as the world’s largest supplier of nuclear plant components and brings significant expertise to the project. The company recently partnered with Korea Hydro & Nuclear Power in agreements with Amazon and X-energy to support over 5 gigawatts of nuclear capacity across the United States by 2039, demonstrating the growing intersection between nuclear power and tech infrastructure.

This partnership builds on Fermi America’s existing collaboration with Hyundai Engineering & Construction, signed earlier this month to co-develop nuclear-based hybrid energy systems in Texas. Under that agreement, the companies will jointly advance project planning, conduct feasibility studies, and execute both Front-End Engineering Design (the detailed planning phase) and Engineering, Procurement, and Construction phases.

The project timeline moves aggressively forward. Fermi America recently submitted its Combined Operating License Application to build AP1000 nuclear reactors—advanced pressurized water reactors designed by Westinghouse—with the submission accepted for regulatory review in record time according to the company. Construction on the nuclear complex is expected to begin in 2026, with the first reactor projected to become operational by 2032.

Why nuclear power is becoming essential for AI infrastructure

The surge in artificial intelligence applications has created unprecedented demand for continuous, reliable electricity that traditional power sources struggle to meet efficiently. Nuclear power addresses several critical requirements that make it increasingly attractive for AI infrastructure projects.

1. Uninterrupted power supply

Unlike solar panels that depend on sunlight or wind turbines that require consistent breezes, nuclear power plants operate continuously regardless of weather conditions. They provide steady electricity 24 hours a day, 365 days a year—a crucial requirement for AI systems that cannot afford power interruptions during training or inference operations. Even brief power outages can cause significant data loss and computational setbacks in AI workloads.

2. Exceptional energy density

Nuclear power delivers extraordinary efficiency compared to other energy sources. A single uranium fuel pellet the size of a fingertip contains as much energy as a ton of coal or 149 gallons of oil. This exceptional energy density means nuclear plants can generate massive amounts of electricity using relatively small physical footprints—a critical advantage as AI campuses require enormous power loads in concentrated areas.

3. Minimal carbon emissions

Nuclear power produces virtually zero greenhouse gas emissions during operation, making it attractive for technology companies committed to carbon neutrality goals. As governments and corporations intensify their net-zero commitments, nuclear power offers a pathway to meet dramatically increasing energy demands without expanding carbon footprints—particularly important as AI workloads are projected to consume exponentially more electricity in coming years.

4. Advanced small modular reactor technology

Next-generation nuclear technologies are making atomic power more flexible and deployable than traditional large-scale plants. Small Modular Reactors represent a breakthrough approach, offering factory-built units that can be installed closer to data centers and tech campuses. These systems provide localized, dedicated power while being faster and less expensive to deploy than conventional nuclear facilities, making them particularly suitable for AI infrastructure projects.

Strategic implications for the AI industry

The Fermi America project signals a broader shift in how technology companies approach power infrastructure for AI operations. Traditional data centers typically require between 10 to 50 megawatts of power, while AI-focused facilities demand significantly more—often 100 megawatts or higher for advanced machine learning operations.

This power requirement creates a strategic bottleneck for AI development. Companies cannot simply plug massive AI systems into existing electrical grids without substantial infrastructure investments. The nuclear approach offers a solution that provides both the scale and reliability required for next-generation AI applications while addressing environmental concerns that increasingly influence corporate and regulatory decisions.

The involvement of major international players like Doosan Enerbility and Hyundai Engineering & Construction also demonstrates how AI infrastructure development is becoming a global industrial priority, requiring partnerships that span continents and combine expertise from both technology and energy sectors.

As artificial intelligence applications become more sophisticated and computationally intensive, projects like Fermi America’s HyperGrid campus may represent the future of AI infrastructure—massive, purpose-built facilities powered by advanced nuclear technology designed specifically to support the enormous and continuous power demands of artificial intelligence systems.