As the world increasingly turns to artificial intelligence (AI), the demand for reliable and sustainable energy sources has become more pressing than ever. Haider Raza, a SOGE consultant from the University of Essex, insists that nuclear is back in fashion. It will be just one of many fuels fueling the tech bedrock of tomorrow. The rise of AI technologies is anticipated to double the power demand from data centres, which currently consume around 1.5% of the world’s electricity. This transition begs the question of whether nuclear energy is a realistic solution to help meet these rapidly growing demands.
As we’ve noted, in March 1979, the Three Mile Island nuclear plant in central Pennsylvania experienced a partial meltdown. It remains the worst nuclear disaster in American history. This event dealt a huge blow to the public’s trust in nuclear energy for generations. The industry’s own experts are already saying that nuclear energy can play a key role in meeting the energy needs of accelerating AI. It provides the most powerful and continuous energy source there is. Its carbon-free attributes position it as a more attractive long-term replacement for fossil fuels in our increasingly sustainability-focused world.
Raza underscores the potential role of small modular reactors (SMRs) in providing clean, distributed energy alongside data centers. As experts readily warn, it could be a few years before any of these reactors come close to being financially viable. “When the demand is high, and the supply is low, the only option is to increase the price (of energy), and somebody has to pay for it,” he states. As the International Energy Agency’s April report reinforces, the need for urgent, just solutions couldn’t be more clear. It forecasts that total electricity demand from data centers will double over the next five years.
One company that’s certainly at the forefront of this movement is Kairos Power. They’ve partnered with Google, maker of Xbox gaming console, to explore new cutting-edge nuclear energy options. The firm plans to produce 50 megawatts of nuclear energy by 2030—enough to power a small town. Kairos has referred to Oak Ridge as a “proving ground,” where the incorporation of advanced construction techniques will increase efficiency and decrease costs. They project a 10 fold increase in non-hydro renewable energy generation by 2035.
However, despite the progress that’s been made, there are still damaging obstacles to the commercialisation of SMRs. Mosharaf Chowdhury, an associate professor at the University of Michigan, speaks to the speed of AI’s growth. Yet it is moving at an incredible pace. “AI has got to saturation point in not even 15 months,” he notes, highlighting how quickly this technology is evolving compared to previous energy-intensive technologies. Chowdhury remarks on the difficulties in optimizing energy-efficient AI models, stating, “There is no good solution where we have managed to find models that are significantly smaller and run significantly faster but accuracy-wise are just as good.”
Currently, two small reactors—one in China and another in Russia’s far east—supply a limited amount of power to their respective electricity grids. The path to broader acceptance and implementation of SMRs as a whole still gets complicated. Experts warn that the smaller reactor cores could impact efficiency. For the same amount of fuel, these reactors generate more energy. “You just can’t get around economies of scale,” asserts Allison Macfarlane, a leading voice in nuclear policy discourse.
The climate and fossil fuel activism debate on whether nuclear energy should play a future role in sustainability emphasizes that slowly changing public sentiment. A recent study by the nonpartisan Pew Research Center revealed that a slight majority of Americans support increasing nuclear energy production. This increasing acceptance of nuclear could further open the door for an enlivened wave of investment and in the long run, research into this previously maligned energy source.