Mark Whitney is president of energy and environment at Amentum and a former principal deputy assistant secretary for the U.S. Department of Energy’s Office of Environmental Management.
Artificial intelligence is transforming business and society at lightning speed. But society is not ready for the electricity demands AI is creating. Data centers are projected to more than double their electricity use by 2030, according to the International Energy Agency, yet renewable energy and batteries alone cannot meet the 24/7, high-density demand of next-generation AI workloads.
The consequences are stark. Companies, government agencies and communities that fail to secure reliable, carbon-free power will face higher costs, slower growth and strategic disadvantages. Those that lock in energy supply now will gain a decisive edge by taking the following actions.
Nuclear energy is an immediate, national strategic imperative. Advanced nuclear technologies, including smaller, faster-to-build reactors, deliver reliable, scalable and zero-carbon power. Unlike wind or solar, nuclear operates 24/7, which is exactly what hyperscale AI data centers require.
The early movers are acting:
- Microsoft has entered a 20-year power purchase agreement with Constellation, based on the restart of Three Mile Island Unit 1.
- Amazon has secured access to nearly 2 GW of nuclear power via Talen Energy.
- Google is collaborating with Kairos Power on small modular reactors.
- Meta is issuing nuclear-focused energy requests.
The message is clear: secure dependable, carbon-free power now, or face capacity shortages and rising costs.
Additionally, smaller, faster-to-build nuclear reactors reduce construction time and costs compared to traditional plants. Their modular design allows factory-built components and scalable deployments. Co-locating AI data centers directly beside these reactors would further reduce transmission costs, free up grid capacity and ensure ultra-reliable power for AI workloads.
Current U.S. nuclear licensing and environmental review processes move far too slowly for AI-era needs. Recent federal initiatives, including 18-month reactor licensing timelines and multi-gigawatt pilot projects at U.S. Department of Energy sites, are a step forward, but predictable, fixed timelines are essential to scale nuclear quickly.
Meanwhile, China is building 30 new reactors with a 200-GW target by 2035. Energy security is tech security. Falling behind in nuclear risks falling behind in AI leadership.
Government and industry must work together
Building nuclear at the speed of AI requires expertise, workforce and regulatory knowledge. Enterprises that manage federal nuclear sites can help hyper-scalers design, build and operate next-generation nuclear-powered campuses. Joint ventures at federal sites, reactor upgrades and modular reactor deployments can deliver gigawatts of reliable power safely and efficiently.
The U.S. must act now to lead. In the AI era, electricity is no longer just a utility line item. It is the operating system of innovation, growth and national competitiveness. Companies and communities that secure nuclear-backed power will scale safely. Those relying solely on intermittent renewables risk outages, higher costs and missed opportunities. Nuclear is not optional. It is the clean and reliable backbone of AI.
By accelerating deployment, streamlining regulations and embracing public-private partnerships, America can power an AI revolution safely, sustainably and competitively. Fail to act and innovation will slow, costs will rise and the nation may cede its leadership to faster-moving competitors like China. The economic, strategic and technological stakes could not be higher.
