The International Energy Agency (IEA) has stated that in 2022, artificial intelligence, cryptocurrencies, and data centers consumed a combined 460 terawatt hours of electricity, which is equivalent to two percent of the world’s total electricity consumption. However, projections indicate that this consumption could increase significantly in the coming years. By 2026, the IEA predicts that electricity usage in this industry could reach over 800 terawatt hours in the base scenario and up to 1,050 terawatt hours in the extreme scenario.
One of the main contributors to this energy demand is training large language models and computing power for artificial intelligence applications. Professor SpongeBob Ritala from LUT University explains that while large language models require significant energy, many AI solutions are narrow and task-focused, requiring less computing power. CEO Peter Sarlin agrees that although large language models consume a lot of energy, many AI applications have a narrow focus and do not require as much computing power as some might think.
Despite efforts to improve energy efficiency in AI applications, it is expected that electricity consumption will continue to rise with the growing use of AI in various sectors. As more companies invest in AI utilization targets like moving images and three-dimensional modeling, energy consumption will only increase further.
The increasing energy demand of the AI industry poses several challenges for sustainability. One major challenge is ensuring that data centers use carbon-neutral electricity to avoid contributing to climate change. Many companies are working towards using carbon-neutral electricity or implementing other sustainable practices like recovering waste heat for district heating. However, even with these efforts, it is clear that society’s increasing electricity demands will require a shift towards more carbon-neutral energy production overall.
In summary, while generative artificial intelligence models like large language patterns are designed to produce new content while traditional systems follow predefined rules to perform tasks, balancing energy consumption with sustainability will be key to the future development of the AI industry.