The race to develop and exploit the extraordinary capabilities of AI and other breakthrough technologies is accelerating at a dizzying pace. But while governments, businesses and citizens are scrambling to take advantage of the seemingly limitless ability of AI to transform almost every aspect of our lives, there’s another challenge looming on the horizon.
As economies in general, and tech companies in particular, are striving to transition to renewable energy sources and to reduce carbon footprint, the boom in AI-related data processing is producing a huge surge in demand for power. But, as the need for clean and secure electricity supplies soars, could nuclear be set to play a vital role in bridging the potential energy gap?
Powering AI will require 9% of US grid capacity by 2030
Powering the world’s rapidly expanding network of data centres has already had significant impacts on society and public policy. In Europe, major data centre clusters, around Dublin and Amsterdam for example, require so much electricity that further data centre expansion in those cities is on hold until new, additional sources of energy come on stream.
As recently as 2020, UK data centres used just over 1% of the nation’s electricity. By 2030 this figure is forecast to reach 7%. Demand is set to be even greater in the US, the global centre of AI innovation, with predictions that, by 2030, 9% of all grid capacity will be used to power AI technologies alone. It’s a monumental challenge that traditional energy utility organisations cannot meet alone.
SMRs will change the game for businesses transitioning to low carbon energy
New research published by Capgemini to coincide with the 2025 World Economic Forum in Davos reveals that 72% of business leaders say they will increase investment in climate technologies, including hydrogen, renewables, nuclear, batteries, and carbon capture, with nuclear energy in their top three climate technology investment priorities for 2025.
This direction of travel chimes with statements made during Davos by the International Energy Agency (IEA). The IEA heralds “a new era for nuclear energy, with new projects, policies and investments increasing, including in advances such as small modular reactors (SMRs)”.
According to IEA Director General Rafael Mariano Grossi: “one after another, technology companies looking for reliable low-carbon electricity to power AI and data centres are turning to nuclear energy, both in the form of traditional large reactors and SMRs.”
Around 60 new reactors are currently under construction in 15 countries around the world, with 20 more countries, including Ghana, Poland, and the Philippines, developing policies to enable construction of their first nuclear power plants. The US Energy Information Administration (EIA) estimates that, by 2025, global nuclear capacity could have increased by up to 250% compared to the end of 2023.
Clean, reliable, available – and safe
It’s easy to understand why nuclear is set to play an increasingly significant dual role in both powering the AI revolution and decarbonising industry. Its 99.999% guarantee of stable energy availability compares with just 30-40% from weather-dependent wind or solar generation.
Decades of continuous improvements in reactor design and operation make nuclear the second safest source of energy in the world after solar, according to the International Atomic Energy Authority (IAEA), although the Agency also points out that large scale solar power systems need 46 times as much land as nuclear to produce one unit of energy.
But it’s the potential to rapidly deploy SMRs that could have the most significant impact in preventing the looming energy gap, as AI-driven data processing requirements grow exponentially. It’s important to remember that most light-water SMRs are simply smaller versions of the large-scale GEN III+ technology with proven safety and operational records, with small generally defined as having a maximum output of 300 MWe. The underlying scientific and operational principles are not technologically new in themselves.
As the name suggests, SMR’s modular design enables major components to be constructed at speed in a factory environment, for bespoke assembly on site, located flexibly close to consumers. With a footprint the size of a sports stadium, they can easily be placed near the demand, such as data centres or industrial estates.
Reduced construction times, lower investment and running costs and the ability to add or reduce capacity as demand increases or decreases, are just some of SMRs’ obvious advantages. They’re ready-made replacements for fossil-fuel based generation, and as nuclear is less vulnerable to price fluctuations, owners and consumers of SMR generated power have more budget certainty and can plan more accurately for the long-term accordingly.
SMRs, specifically the advanced reactor designs, can also be adapted to supply heat for industrial applications, district heating systems and the production of hydrogen, and are increasingly regarded as catalysts for economic development and job creation.
Tech giants at the front of the queue
Many of the global tech giants are actively working on plans to develop their own SMR-based generating capabilities, to provide their own independent sources of safe, stable, low-carbon power, protected from the increasingly volatile open market.
It’s a race that’s not only vital that they win to ensure that we fuel the AI revolution, but by doing so we will accelerate the transition to a low carbon world economy.
