How Much Does Artificial Intelligence Drink?

Artificial intelligence is used in an ever-growing number of diverse fields, which means its water demand keeps rising. Water is used for cooling when ChatGPT performs complex computational processes. So, how much does artificial intelligence drink?

How much water does ChatGPT need?

How much water does ChatGPT consume?

Many people can no longer imagine their daily work without the help of artificial intelligence. ChatGPT is a truly powerful tool that, in the hands of a skilled user, can significantly ease the performance of professional duties. It can probably already be said that AI is changing the world, but does this come at no cost? It turns out that artificial intelligence is very thirsty. Computational processes consume large amounts of energy, which leads to the generation of high temperatures. Researchers from the University of California calculated that artificial intelligence, in this case ChatGPT, needs one glass of water to answer 25 queries. It is therefore easy to calculate that 100 queries require a litre of water. That doesn't seem like much, right? However, one must take into account that ChatGPT processes between 10 million and 1 billion queries a day. The amount of water required for devices to function properly is therefore very high.

The technology industry needs water

The technology industry and water

It is worth adding that it is not only artificial intelligence that is driving ever-greater water consumption. The entire technology industry needs more and more of it. Between 2021 and 2022, Microsoft increased the amount of water used to cool its data centres by 34%. In 2022, the giant announced that water for cooling the supercomputer at its data centre in Iowa would only be used when the outside temperature is above 85 degrees Fahrenheit. However, Fortune magazine uncovered surprising information. It turns out that Microsoft consumed approximately 11.5 million US gallons of water (about 43.53 million litres) in the month before the launch of the GPT-4 model, which accounted for 6% of total water consumption in Des Moines County. In response, local authorities and West Des Moines Water Works (the local water utility) announced that they would only consider further investments in Microsoft's data centres once the company presents and implements technologies that reduce water consumption.

Water consumption by the technology industry poses a significant challenge from the perspective of sustainable development. It is necessary to strive for efficient water resource management in production processes in order to minimise environmental impact. Innovations in water recycling and conservation strategies are key tools in this field.

Concepts from the article

How much does artificial intelligence drink and where does such enormous water demand come from?

AI — Artificial Intelligence. A set of methods and systems that perform tasks usually requiring human reasoning, learning, perception or content generation. In HEXX AI CMS it covers language models, document analysis and automations.

Why it matters: It takes over repetitive tasks — transcriptions, descriptions, translations, selection — and shortens the path from idea to publication. In an editorial office it means more time for work a machine cannot do: evaluation, contacts, decisions.

When it is used: For generating and editing content, document analysis, transcription, translation and automating the flow of materials.

What the risk of omitting it is: Without AI, the team performs bulk work manually (photo descriptions, language versions, summaries), and the competition publishes faster and more cheaply.

Summary

Answering just 25 ChatGPT queries requires a glass of water — according to research from the University of California. At a daily scale of even a billion queries, this translates into enormous resource consumption. Before the launch of GPT-4, Microsoft used 11.5 million gallons of water in Iowa, which forced a reaction from local authorities.

This is not just an AI problem — the entire technology industry is increasing its demand for water to cool data centres. Investment in recycling and conservation strategies is becoming crucial before local communities block the development of further supercomputers.