Thirsty AI

We are in the age of artificial intelligence that people a hundred years ago could not have imagined, and a hundred years later, it is everyone's pocket buddy. AI first captured the public's attention when it turned into chatbots, and people have been using it like a feast ever since, unaware of the fact that their responses are turning into training data for these systems.

For most people, AI is just a concept of chatbots, but in the real domain, these are large, enormous data centers, high-maintenance infrastructures that require billions of dollars to build. These data centers not only span acres of land and consume electricity but also require tons of water. Whenever you ask AI a pressing question, the servers run and generate an intense amount of heat. To prevent overheating, these centers need water. An immense amount of water to cool down their temperature.

The scale of this consumption is staggering. A 2024 report from the Lawrence Berkeley National Laboratory estimated that in 2023, data centers in the US consumed over 17 billion gallons of water for cooling alone. The irony is that Trump, in 2025, announced a dedication of 500 billion dollars to build AI infrastructure through the Stargate Project, which could make these figures double or even quadruple. The same report estimated that in 2023, U.S. data centers consumed an additional 211 billion gallons (800 billion liters) of water indirectly through the electricity that powers them. But that is just an estimate, because these big tech tycoons never reveal how much water their data centers actually consume. Estimates show that a single AI conversation of around fifty responses consumes roughly half a liter of water. Microsoft's global water consumption increased by thirty-four percent due to AI expansion. In 2023, Google reported consuming over 6 billion gallons of water (nearly 23 billion liters) to cool all its data centers, and Meta's data centers consumed 776 million gallons of water in the same year.

On one side, these big tycoons promise a big, bright future, but humans cannot have a bright future without water, which is the most essential and basic need. Humans can live without AI. In fact, we have lived without AI. Generations have lived without AI. But now the cost of Big Tech is going beyond human value and Earth's capacity. As water reserves deplete, there are direct repercussions for water-bankrupt regions, where about 3 billion people and more than half of global food production are concentrated.

And it is not finished yet. AI is still being trained and expanding, and large AI models like GPT-4 are estimated to have consumed around 700,000 liters of water during training. These are the projects of rich countries, and ultimately, they are the ones who will benefit. It is the same situation as when big countries polluted the environment beforehand, generated millions of dollars, and named themselves among thriving economies, but the negative effects were, and still are, faced by the Global South. Pakistan, which ranks among the most climate-vulnerable countries in the world, is no exception. It also lies among water-stressed countries, and as per the International Organisation for Migration (IOM), Pakistan is predicted to slip into absolute water scarcity by 2035, as its large population and inability to conserve fresh water are leading it toward despair.

Over this, countries including the US, UAE, and Saudi Arabia are building massive AI infrastructure while consuming freshwater at an industrial scale. The opportunity cost will be borne by underprivileged countries once again. Every year, nature gives us a water income in the form of rain, snow, and stored glaciers. Water is limited, as is the nature of every resource on Earth, but humanity is wasting it on machines rather than saving it for generations to come. We can indeed live without AI, but we will surely wither without water.