- AI’s growing energy consumption, projected to exceed 1,000 TWh by 2026, highlights the need for reliable, low-carbon energy sources like nuclear power.
- Nuclear energy can support AI’s energy demands while AI can enhance nuclear power through predictive analysis, optimization, and safety improvements.
- Tech giants like Microsoft, Amazon, and Google are investing in nuclear energy, recognizing its potential to meet the escalating electricity needs driven by AI.
Artificial intelligence (AI) has been a hot topic for years, gaining more and more momentum each day. Companies are catching up on the hype, fearing they’re missing out on opportunities to utilize this opportunity for company growth. You can see and hear it everywhere, from the news to your home, school, and workplace. We are all surrounded by it in our everyday life. But what does AI have to do with nuclear energy? Well, they are strongly associated with each other. The two may end up in a mutually beneficial relationship, helping each other overcome its shortcomings.
As many people may know, AI consumes a lot of energy compared to the average household. AI has been growing exponentially, along with it is the increase in energy consumption. Additional sources of energy are required to support the use of AI and nuclear energy is one of the options. Nuclear energy is proven to be the cleanest source of energy in the world, emitting a minimal amount of carbon emissions compared to its counterparts. While nuclear energy is helpful to AI, nuclear energy can also be beneficial to AI. They can be used in tasks such as data analysis, theoretical modeling, and experiment design. These AI applications expedite foundational research, such as the evaluation and compilation of nuclear and atomic data, while also driving forward technological innovation by conducting analysis and asserting risks.
How much energy is AI actually using?
According to data from the International Energy Agency, power consumption from AI and cryptocurrency data centers is projected to more than double, reaching more than 1,000 terawatt-hours (TWh) by 2026. In 2022, these data centers collectively consumed an estimated 460 terawatt-hours (TWh) of electricity, which is two percent of all global electricity usage. In just two years, this consumption is anticipated to exceed 1,000 TWh per year. “This demand is roughly equivalent to the electricity consumption of Japan,” said the report.
We have yet to fully comprehend just how immense energy demands AI will entail. Training AI, especially, demands a significant amount of energy, surpassing the electricity consumed by traditional data center operations. ChatGPT specifically has taken our technological climate by storm. For instance, training a large language model such as GPT-3 is estimated to require nearly 1,300 megawatt hours (MWh) of electricity, equivalent to the annual consumption of approximately 130 US households. Besides the training, ChatGPT alone consumes approximately 500,000 kilowatt-hours (kWh) of electricity per day. It is equivalent to what 17,241 average US households use.
The future of AI is nuclear energy
As the world is transitioning to Net-Zero, an alternative energy source of reliable and low-carbon energy is a must for companies. The increase in demand for energy consumption underscores the critical necessity for expanded green energy solutions worldwide. Mark Zuckerberg, CEO of Meta, highlights the pressing concern: without additional sources of affordable, dependable, and environmentally sustainable power, the global infrastructure may struggle to meet the escalating electricity demands driven by the AI revolution.
Tech firms have been eyeing nuclear energy as an alternative. Microsoft and Amazon, for instance, have announced plans to integrate nuclear power into their data centers in the Eastern US. Last summer, Microsoft partnered with Constellation, a leading operator of nuclear power plants, to incorporate nuclear-generated electricity into its Virginia facilities. Before that, Microsoft signed a deal with Helion Energy, a nuclear fusion company, in which they will be providing electricity to Microsoft in five years. Earlier this year, Amazon Web Services (AWS) purchased Talen Energy’s 960 MW data center campus in Pennsylvania, which receives power from the nearby 2.5GW Susquehanna nuclear power station. Additionally, Google participated in a $250 million fundraising round for the fusion startup TAE Technologies.
Ai benefiting nuclear energy
AI can be utilized for physics-based predictive analysis across various aspects including design, manufacturing, construction optimization, operational effectiveness, iteration of new reactor designs, model-based fault detection, and advanced control systems. Without the help of AI, the progress of developing and constructing nuclear power plants would be much slower. These applications have the potential to lower operation and maintenance expenses for nuclear power plants and enhance safety through intelligent data-driven or physics-based model-driven analyses.
AI plays a crucial role in advancing fusion research by leveraging its capability to tackle intricate problems. Through modeling and simulations, AI helps in experiments and scientific breakthroughs, particularly in the fusion domain. Integrating AI with digital simulations of actual nuclear facilities enables the industry to optimize complex procedures and enhance reactor design, performance, and safety. These optimizations not only boost operational efficiency but also cut down on maintenance costs.
Machine learning, which allows AI to learn from extensive data analysis, automates tasks, enhances reliability, and minimizes errors. Additionally, AI’s analytical and predictive capabilities allow them to monitor power plant processes and detect anomalies swiftly. AI algorithms have also enhanced engineers’ ability to forecast fuel requirements and optimize fuel configuration for maximum power generation while ensuring operational safety and efficiency—otherwise time-consuming and costly.
Nelly Ngoy Kubelwa, a nuclear engineer specializing in innovative technology at the IAEA, indicated that “AI, together with other technologies, like digital twins, could decisively boost the efficiency of nuclear power production.”
What’s next?
In the age of digitalization, AI offers solutions that can lead the nuclear power industry towards a sustainable future. It’s intriguing that a symbiotic relationship could develop between nuclear technology and AI. The advanced analytics and automation capabilities of AI have the potential to spur innovations that tackle nuclear challenges, including safety enhancements and operational optimization. Along with, using nuclear energy to power the data center and training of AI. This mutual interaction suggests a win-win scenario where both technologies complement each other, leveraging strengths to overcome respective weaknesses. Until more renewable power sources are fully established, nuclear power remains incredibly important.
“We need nuclear power to get to a low-carbon future,” said Ahmed Abdulla, assistant mechanical and aerospace engineering professor at Carleton University.
