The rapid expansion of the AI industry is posing unprecedented challenges to global energy and water resources. This morning, the head of the National Energy Administration revealed during the introduction of the "14th Five-Year Plan" for the new energy system that generating a 5-second high-definition video with AI consumes the same amount of electricity as charging 10 mobile phones.

Since early February this year, global energy markets have experienced intense fluctuations, with many countries facing varying degrees of energy shortages and supply tensions. The continuous rise in AI computing power has undoubtedly exacerbated this situation.

It's not just about electricity consumption — 10 questions consume a bottle of mineral water

The high water consumption of the AI industry is also alarming. At the 2026 Summer Davos Forum, Wu Xuchu, vice chairman of KPMG China, pointed out that having a 5-minute conversation with AI consumes approximately 500 milliliters of cooling water. According to experts from the National Geographic Society, asking AI 10 questions consumes about the same amount of water as a bottle of mineral water.

It is worth noting that the water used by AI is not limited to the cooling towers of data centers. According to a report by the World Economic Forum, the water consumption in chip manufacturing and power generation far exceeds that of data center operations themselves. Looking at the entire AI industry chain, from chip production, power supply to computing power maintenance, the annual freshwater consumption reaches 23 billion cubic meters, accounting for 3.7% of the total industrial freshwater use globally. Although this proportion may seem low, freshwater has the special characteristic of being difficult to transport over long distances, making its scarcity much higher than electricity, which can be easily adjusted across regions.

40% of data centers are located in water-scarce areas, exacerbating the crisis through layout conflicts

More concerning is the serious mismatch between industry layout and water resource capacity. Data shows that about 40% of data centers and nearly one-third of chip factories are located in water-scarce areas. When companies choose locations for their industries, they often prioritize electricity prices and land costs, placing water resource capacity in a secondary position.