Why Saving Water is the Next Big Challenge for Tech Giants

news
Wednesday, 21 January 2026 at 10:48
Generative AI
As the world races to build better AI, a new problem is rising: the industry is getting incredibly thirsty. A new study from Xylem and Global Water Intelligence shows that AI will drive up global water use by 30 trillion litres by 2050. That is an almost impossible amount to imagine. It is roughly three-quarters of the water held by the massive Three Gorges Dam.
93b7024d-79c1-4c70-8cab-ccf6b51f54ab

Where the Water Goes

While many blame data centres, they’re only a small part of the problem. The real "thirst" is hidden behind the scenes. More than half of that water is actually used just to generate the massive amount of electricity needed to keep AI running. Another huge chunk goes into making the computer chips themselves. By 2050, the total water needed for the AI industry is expected to jump by 129%.

Building in the Wrong Places

The location of these tech hubs is another big worry. Right now, about 40% of data centres are sitting in places that are already running low on water. Even worse, many new projects are being planned for these same dry areas. If this keeps up, tech growth could end up competing with local towns for the water they need to survive.

There is Still Hope

Not everything's doomed. If we move fast, we can turn this around. Folks who know say recycle our water and fix those damn leaky pipes. Every drop counts now, so why not reuse 'em all and keep our gadgets humming without sucking the planet dry? It's just about getting smart—squeeze every bit out of what we've got to hang onto the tech we dig.

Looking to 2050

We can still afford the time, but the clock is ticking. As AI moves into our homes and lives, the effects of that on the natural world are getting too massive to ignore. In such a scenario, the challenge for the next generation of leaders is deeply personal: they need to show the way to power our digital dreams without giving away the very water that will sustain us all.
loading

Loading