Amazon has joined forces with X-energy, Korea Hydro & Nuclear Power (KHNP), and Doosan Enerbility in a bid to bring advanced small modular reactors (SMRs) online across the U.S., a move tailored to the surging electricity needs of AI-era data centers. The partners plan more than 5 gigawatts of new nuclear capacity by 2039 and say they will pursue up to $50 billion in public and private financing to make it happen.
The agreement stretches well beyond a single site. It covers engineering and design, supply-chain buildout, construction planning, investment strategy and long-term operations—plus potential joint deployments outside the U.S. It also leans on Korean industrial expertise: KHNP brings decades of operating experience, and Doosan, a major provider of nuclear components, will lock in manufacturing capacity for U.S.-based X-energy’s reactor modules.
More Power Needed
The need for the additional power is compelling. By 2030, U.S. data centers could consume 214 to 675 terawatt-hours annually—a 2.6x jump from 2023—driven by AI training and inference. SMRs, as a source of nuclear power, offer predictable, carbon-free power at (or adjacent to) data centers, lowering transmission losses while delivering constant capacity. Driving all of this is Amazon’s data center expansion strategy: the company is backing a multi-project approach rather than a one-off pilot.
At the center of the plan is X-energy’s Xe-100, a high-temperature, gas-cooled SMR design that produces 80 MW of electricity or 200 MW of high-grade steam. The reactor uses helium as a coolant and TRISO-X fuel, which are uranium kernels sealed in multiple ceramic layers to enhance containment. X-energy claims that the design has a 60-year operating life.
Two U.S. deployments are already in progress. In Texas, a four-reactor Xe-100 project at Dow’s Seadrift site is moving through federal review. The Nuclear Regulatory Commission accepted the construction permit application in May 2025 for an 18-month review. In Washington state, Energy Northwest is preparing a site that would start with four Xe-100 units and could scale to 12, a campus that Amazon is helping advance.
An Operating Model for the Future
Together, those projects illustrate the model the alliance is trying to replicate: standardize a design, stand up a supply chain and repeat at multiple locations.
Amazon is investing heavily in nuclear. The company’s Climate Pledge Fund led a $700 million round for X-energy to accelerate commercialization, and Amazon has said growing AI demand requires more “always-on” carbon-free sources to complement its large wind and solar portfolio. Over the past year, the company has directed more than $1 billion into nuclear projects and technologies, framing these investments as essential to keep cloud growth aligned with decarbonization goals.
Still, nuclear projects, whether modular or large-scale, face predictable hurdles: financing structure, local permitting, and a limited workforce for specialized trades. However, if this alliance can execute, the template could spread: hyperscalers underwriting predictable, on-site (or near-site) nuclear to anchor AI campuses, standardized SMR blocks replicated across regions, and a supply chain tuned for replication. For an industry scrambling to square AI growth with grid limits and climate commitments, that would be more than a power contract—it would be an operating model.
