Time to go nuclear? Inside the battle to power AI

Essential brief

Key facts

Global data center power demand is projected to exceed 200 gigawatts by 2030, driven largely by AI workloads.

The U.S. faces a potential 80-gigawatt shortfall in data center power capacity by 2030.

Renewable energy will cover much of the increased demand but may not provide consistent supply on its own.

Small modular nuclear reactors (SMRs) could supply about 10% of AI’s power needs, offering a scalable and low-carbon option.

Regulatory challenges and public perception remain significant barriers to widespread SMR deployment.

Highlights

Global data center power demand is projected to exceed 200 gigawatts by 2030, driven largely by AI workloads.

The U.S. faces a potential 80-gigawatt shortfall in data center power capacity by 2030.

Renewable energy will cover much of the increased demand but may not provide consistent supply on its own.

Small modular nuclear reactors (SMRs) could supply about 10% of AI’s power needs, offering a scalable and low-carbon option.

As artificial intelligence (AI) technologies become increasingly integrated into everyday life, the energy demands of the data centers powering these systems are soaring.

Current estimates suggest that global data center power consumption will more than double by 2030, reaching over 200 gigawatts.

In the United States alone, demand could rise to between 100 and 130 gigawatts, creating a significant shortfall of approximately 80 gigawatts compared to existing capacity.

This growing energy appetite has sparked a debate on how best to sustainably power the AI revolution.

Renewable energy sources such as wind and solar are expected to cover a substantial portion of this demand, but their intermittent nature poses challenges for consistent supply.

Nuclear energy, particularly through small modular reactors (SMRs), emerges as a promising solution to fill around 10% of AI’s power requirements.

SMRs offer advantages including scalability, lower upfront costs, and enhanced safety features compared to traditional nuclear plants.

However, the deployment of SMRs faces regulatory hurdles and public skepticism, which could delay their contribution to the energy mix.

Balancing the urgent need for reliable, carbon-free power with environmental and social considerations is central to this energy transition.

As AI applications proliferate, ensuring that their infrastructure is powered sustainably will be critical to minimizing the sector’s environmental footprint.

The interplay between renewables and nuclear power will likely define the future landscape of AI energy supply, influencing policy decisions and investment strategies worldwide.