The United Nations Institute for Water, Environment and Health reports that global data center water consumption will double by 2030 due to the surge in artificial intelligence (AI). The report points out that the continuous expansion of data centers brings significant environmental costs.

Supercomputer Data Center (3)

Data shows that last year, global data centers consumed 448 terawatt-hours of electricity, equivalent to the total national electricity consumption of Saudi Arabia. Of this, the power required for AI computing accounted for one-fifth of the total electricity consumption. The water consumption of data centers reached 45 trillion liters, sufficient to meet the water needs of over 600 million people in sub-Saharan Africa. Meanwhile, data centers also emitted 189 million tons of carbon dioxide.

According to the lead author of the report, Kavi Madani, the public often views AI as a purely software product, but in reality, AI development relies on infrastructure support, including data centers, power facilities, cooling systems, and more. Therefore, it is expected that by 2030, global data center electricity consumption will reach 945 terawatt-hours, equivalent to Japan's national electricity consumption, with AI's electricity consumption proportion rising to 40%. Water consumption is expected to increase to 9.3 billion liters, and carbon dioxide emissions are projected to rise to 399 million tons.

The area occupied by data centers will also significantly expand, increasing from 6,900 square kilometers last year to 14,500 square kilometers by 2030. Although AI has the potential to improve energy efficiency by optimizing the grid and reducing resource waste, under the context of countries and companies racing to build computing infrastructure, overall water and electricity demand will still significantly increase.

Madani called for attention, stating that the current pursuit of development speed has overshadowed the principles of sustainable development. He said that although water and electricity resources will not be completely exhausted, without proper planning, data center construction could conflict with local resource supply and demand tensions. Therefore, he recommends scientific planning before infrastructure construction to achieve sustainable development.

Key Points:

🌍 Global data center electricity consumption will double by 2030, reaching 945 terawatt-hours.  

💧 Water consumption of data centers reached 4 trillion liters last year, expected to increase to 9.3 trillion liters by 2030.  

📉 The environmental cost of AI is becoming increasingly severe; ignoring it will exacerbate land resource shortages and e-waste issues.