AI push from US tech giants leads to skyrocketing water consumption figures
It’s probably no surprise that some tech giants are falling short when it comes to sustainability. Harmful mining practices, a huge carbon footprint and massive energy consumption pose threats to green goals.
Now there is new research from the Financial Times has also found that water usage at tech giants has reached unsustainable levels in some locations.
In these centres, water is mainly used to cool powerful equipment. Air mechanisms are used to remove heat. As a result, up to nine litres of water evaporate for every kWh of energy consumed.
A precious resource
The report notes that about two-thirds of the world’s population experiences severe water shortages for at least one month each year, and that number is only set to increase. That equates to 2.7 billion people experiencing water stress each year, with 2.2 million children dying from water-related diseases.
However, that won’t stop tech companies from using at least 7 billion gallons of water in their data center warehouses in Virginia alone by 2023.
Microsoft admitted that 42% of the water it consumed globally in 2023 would come from “water-stressed areas.” Google also claimed that 15% of its freshwater withdrawals would come from “highly water-stressed” areas.
The AI Factor
The environmental impact of artificial intelligence has yet to be fully assessed, but the sector is caught between enormous challenges. greenhouse gas emissions and dangerous levels of water consumption. While water cooling has been shown to reduce emissions, it comes at the cost of one of our most valuable resources.
Servers running AI algorithms generate more heat than classic computer systems, requiring 6 times more kilowatts per rack – with a 5-50 prompt conversation with ChatGPT using about 500 ml of water only.
Amazon recently said it is “committed to being a good water steward” and that data centers will be “water positive” by 2030. But environmentalists warn that AI will “explode” demand for computing infrastructure.
