In recent years, concerns about artificial intelligence (AI) have expanded beyond its societal implications to include its environmental impacts, particularly water consumption. Headlines have called out the large amounts of water required for cooling AI data centres. However, when compared against the water usage of cattle farming, AI’s demands are a drop in the bucket, paling in comparison. Understanding the big picture is crucial to refocusing efforts on the far greater environmental challenge: livestock farming.In recent years, concerns about artificial intelligence (AI) have expanded beyond its societal implications to include its environmental impacts, particularly water consumption. Headlines have called out the large amounts of water required for cooling AI data centres. However, when compared against the water usage of cattle farming, AI’s demands are a drop in the bucket, paling in comparison. Understanding the big picture is crucial to refocusing efforts on the far greater environmental challenge: livestock farming.

AI systems, such as ChatGPT, rely on massive data centres that require significant cooling to function effectively. Globally, approximately 11,000 data centres serve millions of users, consuming water primarily for cooling and electricity generation. 

Water use across data centres can vary significantly, anywhere from 68,000 litres in small facilities to 2.1 million in the largest ones – equivalent to the daily consumption of about 4,200 people. With more and larger data centres comes a greater demand for water; Google’s data centres consumed 20% more water in 2022 compared to 2021, while Microsoft’s water usage increased by 34% in that same year. 

Around half of data centre demand is for AI, and that is rising rapidly. This is because for the past few years, the primary method companies have used to create smarter, more capable AI models is to increase the amount of data and computing power used to build it, often by 10–100 times. This brings with it large increases in water use.

These figures are undoubtedly concerning. However, we should be cautious in sounding the alarm bells on AI for two reasons: firstly, there is significant progress in reducing AI’s water impact, and secondly, it may distract from far greater sources of wasted water.

Despite these numbers, the water usage associated with AI remains significantly lower than that of coal, biodiesel, and nuclear energy. Where Google’s AI can require half a million gallons (1.89 million litres) per day, nuclear cooling systems can demand up to half a million gallons of water per minute.

Moreover, efforts to mitigate AI’s water footprint are already underway. Companies like Google have pledged to use recycled water for cooling in some facilities and have engaged with local communities to minimise their impact on regional water resources. Google and Microsoft are both prototyping “zero water solutions”; closed loop systems that capture and re-use evaporated water. Additionally, new cooling designs could save up to 125 million litres annually.

So while AI companies are building bigger models and serving them to more people than ever before, they are also working on efficiency improvements to partially offset these increases. As resource-hungry as AI data centres appear to be, the dark reality is that for decades, another industry has dominated our resources, particularly, fresh water use. ChatGPT’s water requirements pale in comparison to animal agriculture, specifically, beef and dairy. 

In stark contrast to AI, animal agriculture is a major driver of global water consumption and environmental degradation. Livestock accounts for over a quarter of all of humanity’s water use. Much of this water is used to grow crops like corn and soy, which serve as feed for billions of farmed animals. Feed can account for a staggering 41% of all the water used in agriculture.

Beef and dairy are particularly water hungry. Producing a kilogram of beef alone requires a staggering 15,400 litres of water. Globally, the daily water usage for beef production amounts to approximately 2.49 trillion litres, with leather and other cow products adding up to 2.53 trillion litres per day. This means that together, beef, leather, and by-products demand over 5 trillion litres of water. This is roughly the equivalent of 2 million Olympic sized swimming pools per day. Beyond direct water use, manure run-off and agricultural fertilisers contribute to water pollution, leading to harmful algae growth and biodiversity loss in freshwater ecosystems.

The environmental footprint of dairy production dwarfs many industrial activities, including AI’s water consumption. In fact, dairy farming tops all other agricultural sectors by billions of litres. 

Beef has deservedly drawn scrutiny over the past decade, but it’s time dairy faced the same. Although beef uses more water per kilo, we produce far more dairy – 552 million metric tons vs.  60 million for beef and veal.

Industrial dairy’s toll on water systems is staggering. A Food & Water Watch report highlights how major basin states (such as Arizona, California and Colorado) are strained by the sector. Alfalfa, a key feed crop for dairy cows, consumes 2.2 trillion gallons of water annually across the seven basin states. This fresh water is vital for residents, who have now had to face severe droughts. Meanwhile, the 2.5 million dairy cows on mega-dairies in these states require an additional 218 million gallons of water per day for drinking and washing alone.

To make matters worse, industrial dairy farming not only depletes water, but also degrades the quality of ecosystems. Improper manure and fertiliser management leads to contaminated local water resources such as wetlands and forests. In both scale and impact, industrial cattle farming (both beef and dairy) is one of the most water-intensive and water-destructive systems on the planet.

Of course, everything we do requires environmental resources, so its not enough to argue that something has a high environmental impact. The costs must be weighed against the benefits. The most water-intensive industry in the world could still be worthwhile if it provides substantial benefits and there is no good alternative. Can such a case be made for the beef and dairy industries?

1. Impact on human health: Far from being essential for human health, beef, and dairy consumption increase the risk of cancer, heart disease, and diabetes, putting a heavy burden on public healthcare systems. Moreover, intensive animal farms are a leading cause of new diseases.
2. Impact on other environmental outcomes: Alongside water, beef, and dairy are some of the most carbon and land-intensive foods in the human diet. Grass-fed beef and dairy are a key driver of deforestation worldwide. If cattle were able to form their own country, they would rank 3rd behind China and the United States among the world’s largest greenhouse gas emitters. 
3. Impact on food security: While meat and dairy is food that many people eat regularly, that does not make it important to food security. In fact, the modern meat industry undermines food security, by increasing dependence on imports, costing taxpayers money that they could spend on healthier foods, and consuming large amounts of food that could be fed to humans. 
4. Impact on the economy: AI and tech creates well paying, future-proof jobs, and AI tools unlock new levels of productivity for the entire economy. On the other hand, jobs in animal agriculture are declining over time, and many are low paid and back-breaking. 

While water and energy use from AI is positioned to reduce overtime, resource use from animal agriculture is not. While we can make some efficiency gains in dairy (better genetics and animal welfare for example) most of these gains have already been made. There is ultimately very little potential to make dairy more water efficient than it currently is. After all, milk is mostly water, meaning that producing vast amounts of milk will always require using vast amounts of water.

On the other hand, this presents a bold opportunity; transitioning away from water-intensive beef and dairy offers a significant opportunity to solve global water scarcity. One solution is a plant-forward diet. Plants can provide roughly two-thirds of the global protein supply with just a quarter of agricultural land. Moreover, switching to plant-based meat could reduce water use by up to 99%, while cultivated meat could lead to a 66% reduction. Plant based milk alternatives are already mainstream, found in supermarkets and cafés near you.

It is encouraging to see incremental gains in AI’s water efficiency. However, these improvements are minor compared to the transformative impact that reforms in beef and dairy could achieve.

As the world grapples with environmental challenges, prioritising efforts to tackle the largest contributors to water scarcity is essential. While AI’s water use is on a concerning upward trajectory, it is minuscule compared to the trillions of litres used daily by animal agriculture.  Criticising AI for its water use without acknowledging livestock farming’s vastly greater impact misdirects public concern and resources. To ensure our water security for decades to come, replacing beef and dairy in our diets is essential.