According to a new analysis, the electricity consumption of artificial intelligence is expected to reach nearly half of the total electricity consumption of global data centers by the end of 2025. This means that AI energy consumption will soon surpass Bitcoin mining. The analysis was conducted by Alex de Vries-Gao, a PhD student at the Environmental Institute of Vrije Universiteit Amsterdam. He has previously tracked the electricity consumption and environmental impact of cryptocurrencies, and published relevant data on his website, Digiconomist.

Wires and cables

Currently, AI already accounts for up to one-fifth of data center electricity consumption. Although this figure is difficult to determine precisely because major tech companies do not specifically disclose the energy consumed by their AI models, de Vries-Gao predicted through the supply chain of specialized computing chips used for AI. Despite improvements in efficiency, researchers found that the power demand of AI is still growing rapidly, and more attention is urgently needed.

De Vries-Gao originally thought that after alternative cryptocurrencies like Ethereum transitioned to low-energy technologies, his research work might gradually come to an end. However, with the emergence of large-scale AI technologies like "ChatGPT," he realized that this is another typically energy-intensive technology, especially in competitive markets. "I thought, oh my goodness, here we go again," he said.

In assessing the energy consumption of AI, de Vries-Gao pointed out that the rapid development of AI in modern society is attracting increasing attention. As this field expands, understanding its energy requirements becomes increasingly important. Although technology continues to advance, how to balance energy consumption and sustainable development will be an important issue that must be addressed in the future.

This research highlights not only the rapid development of AI technology in modern society but also the energy challenges it brings. As more enterprises and institutions adopt AI, how to effectively manage and optimize its energy consumption will have far-reaching impacts on the environment and economy.

Key points:

🌍 Artificial intelligence is expected to account for nearly half of global data center electricity consumption by the end of 2025.  

💡 Currently, AI already accounts for up to one-fifth of data center electricity consumption.  

⚡ Despite efficiency improvements, the power demand of AI is still growing rapidly and requires attention.