In our recent commentary on generative artificial intelligence (AI), which can be found here, we discussed the growing energy demands of data centres. Generative AI is an artificial intelligence algorithm that is capable of generating different types of content, including text, images, videos or other data, for example in response to a user query. Examples of generative AI include Gemini, ChatGPT, AlphaCode, Midjourney and DALL-E. Data centres house the processing infrastructure used to run and train generative AI algorithms and analysts estimate that the power demand of these centres will grow by 160% by 2030.
In view of the rising power demands of power centres, tech giants are considering different sources of energy. For example, a coalition of Amazon, Google and Meta publicly endorsed a vision of tripling nuclear power capacity by 2050. The risks and safety concerns relating to nuclear power generation have led to a decreased use of this energy source and phasing-out of nuclear reactors in the past. However, the energy demands of AI applications and climate goals have led to renewed interest in nuclear power. Especially in view of nuclear power’s arguably lower carbon impact than conventional energy sources, and reduced power generation volatility compared to renewable energy sources.
While the use of AI algorithms and the almost zero carbon dioxide emission of nuclear power generation may make the use of nuclear power more attractive, a number of fundamental questions concerning issues, such as waste management, the long-term impact of nuclear power generation on the planet and our responsibilities to future generations, remain.
Yet tech giants may be able to do more than simply endorse increasing nuclear power capacity. Their AI could be used to address the noted concerns. The AI algorithms being developed may provide a turning point in the use of nuclear power. Such algorithms have the potential to enhance the safety of nuclear reactors and their infrastructure while reducing their downtime. For example, AI-based systems could be used to monitor reactor performance, identify anomalies and predict potential failures before they occur. These systems could also optimise reactor operations, dynamically adjust parameters to maximize efficiency, and detect and respond to cyber threats, to name a few possible uses of AI in nuclear power infrastructure.
Innovation is already happening in AI and energy. A measure of such innovation is patent filings. European Patent Office data shows that despite a levelling out in AI applications, growth in AI related to green technology remains strong with a 35% increase year-on-year in European patent publications in 2023. To parallel the well-known science fiction movie. The Matrix, “machines” (read generative AI) could be used to “generate” (read make safer) the very power they need to get stronger (read innovation in green AI).
The energy landscape is undergoing a seismic shift. Last week a coalition of tech giants — Amazon, Google and Meta — publicly endorsed a bold vision: tripling global nuclear energy capacity by 2050.