Have you read that Xinhua article? It said how much water AI uses for each answer, and the result was quite shocking.

I did. The article said data center cooling and power plant operations both require water, and a single query can use over ten milliliters.

Ten-plus milliliters doesn’t sound like much, but if you ask a thousand questions a day, that’s ten liters—about a whole bottle of mineral water.

What’s even scarier is that many data centers are built in water-scarce areas, making local residents’ drinking water even tighter.

So it’s not just a global total issue, but a local supply risk. One large data center could drain residents’ wells.

AI companies say they achieve water neutrality through wetland restoration or groundwater recharge, but often not in the same location.

That sounds like ‘balanced on paper, unbalanced in reality.’ What can ordinary people do?

For example, residents in Pima County, Arizona, rejected a high water-consuming data center project—an example of successful public participation.

This shows that protecting water requires hearing all sides: not only company reports, but also real local conditions and community concerns.

Right. We should enjoy the convenience of AI while advocating water conservation and sustainable development, and maintain rationality and responsibility toward technology.