Equivalent to a Nation's Annual Electricity Consumption! AI Power Usage Like ChatGPT to Surge in 4 Years
-
October 12, Fast Technology News - With the emergence of generative AI like ChatGPT, companies and institutions are rapidly launching their own large-scale AI models.
This technology undoubtedly brings numerous conveniences and previously untapped opportunities to humanity, such as making it easier for students to solve complex problems and advancing medical research, among others.
However, these conveniences come at a significant cost. Take Google as an example: its AI training consumes 10% to 15% of the company's total electricity usage, amounting to approximately 2.3 billion kilowatt-hours annually, equivalent to the yearly electricity consumption of all households in a U.S. state capital.
Even after training is completed, these AI systems still require substantial computational power to operate, resulting in high energy consumption. OpenAI spends up to $700,000 daily on running ChatGPT.
A recent international study suggests that if interest in AI technology continues to grow and chip supply remains sufficient, generative AI (including ChatGPT, Bing Chat, etc.) will lead to massive energy consumption, potentially surpassing the electricity demands of some countries.
The study indicates that energy consumption related to artificial intelligence is likely to increase significantly in the coming years. By 2027, the energy consumed by generative AI could power a country the size of the Netherlands for an entire year, equivalent to approximately 85-134 terawatt-hours (TWh).
However, with continuous technological advancements, more efficient and environmentally friendly generative AI solutions are expected in the future, which should help address the electricity consumption issue.