AI's growing demand is causing a significant increase in data centre energy consumption, with predictions showing a more than double increase from 2022 to 2026.,The expansion of data centres in Asia, particularly in China, is raising concerns about energy usage and water consumption.,Efforts to mitigate environmental impact include advancements in chip hardware efficiency and exploring renewable energy sources.
The AI Energy Conundrum
Artificial Intelligence (AI) is transforming the world in ways we never imagined. However, this transformation comes with a hefty price tag: a colossal consumption of energy. Power-hungry data centres across the globe are working round the clock to support AI's insatiable appetite for energy. The International Energy Agency (IEA) forecasts that global data centre electricity demand will more than double from 2022 to 2026, with AI playing a significant role in this surge.
Data Centres: The Backbone of AI
Data centres are massive facilities filled with thousands of computer servers running continuously. In the era of 5G and cloud-based storage, they have become indispensable, supporting everything from financial transactions to social media and government operations. They require a constant and reliable energy supply, accounting for more than 1% of global electricity use.
AI's rapid growth has dramatically accelerated data centre construction and usage. Training AI models is extremely energy-intensive, consuming power at a much higher rate than traditional data centre activities. For instance, a ChatGPT query uses ten times more energy than a standard Google query. For more on how people are using this technology, see How People Really Use AI in 2025.
AI vs. Climate Goals: The Asian Perspective
This energy usage surge threatens to undermine climate pledges made by major tech companies. In Asia, data centres are growing rapidly, particularly in China. The Hong Kong-based think tank China Water Risk estimates that data centres in China consume 1.3 billion cubic meters of water per year, nearly double the volume used by the city of Tianjin for households and services. This growth is part of a broader trend of AI Boom Fuels Asian Market Surge.
Potential Solutions
AI companies are working on solutions to reduce energy consumption. Efficiency improvements in chip hardware, such as NVIDIA's new line of GPUs with 25 times lower energy consumption, could significantly reduce energy usage. Additionally, AI technology could help combat climate change by addressing inefficiencies. This push for efficiency is also seen in other areas, such as Nvidia Jetson AGX Thor sets a new pace for robotics and physical AI.
Industry leaders are also investing in renewable energy sources like wind, solar, and nuclear power. However, many data centres still rely on fossil fuels, especially during the transition period.
Regulation on the Horizon
Regulation may be imminent. The European Commission is moving towards regulating data centre sustainability across the EU. In the US, the House Committee on Energy and Commerce held a hearing about AI's energy usage in June. Transparency is key to understanding AI's environmental impact and implementing effective regulations. This is a global concern, as countries like Taiwan’s AI Law Is Quietly Redefining What “Responsible Innovation” Means. For more detailed information on data center energy consumption and its environmental impact, the International Energy Agency provides comprehensive reports here.
Comment and Share:
What are your thoughts on AI's growing energy consumption? Do you believe that advancements in hardware efficiency and renewable energy sources can mitigate its environmental impact? Share your thoughts below and don't forget to Subscribe to our newsletter for updates on AI and AGI developments at AI in Asia.
Latest Comments (2)
the doubling of energy demand by 2026 feels conservative given the current pace of model training. I mean a ChatGPT query uses 10x more energy than a Google query - that alone is a huge jump. Most companies aren't even optimising models for power efficiency yet.
Even back then the China growth was huge. Here in Manchester, we're seeing much more distributed models for AI inference emerging, hopefully taking some heat off the big server farms.
Leave a Comment