
AI
Next to animal agriculture, all the world’s data centers use practically nothing.
1.4 millionabout 1.4 million prompts to match the water in one cow-flesh burger
Going by OpenAI’s figure for water usage in a single prompt[1] and the University of Michigan study of how much water it takes to produce a quarter-pound cow-flesh patty[2], you would need about 1.4 million prompts to equal the water in one cow-flesh burger[1][2]. Plant burgers with the same taste and flavor already exist, and they use 99% less water, 93% less land, and 90% fewer greenhouse gases[2], with no animal cruelty. Also, farm animals produce about 40× more greenhouse gases than all data centers on the planet combined[12][13]. So, if you really cared about the environment, you would stop harming animals.
What the evidence says
The main points first. Every claim is tied to a source below.

Our World in Data · Poore & Nemecek (2018)
Farm animals use about 200 times more freshwater than AI is projected to use
Wisser and colleagues estimated in 2024 that animal agri-food systems consume about 228 km³ of freshwater a year[1], and growing feed makes up more than 90% of those withdrawals[1]. A September 2025 analysis by Morgan Stanley projected that water use linked to AI data centers will reach about 1,068 billion liters a year by 2028, roughly 1 km³[2]. Divide 228 by 1.07 and you get about 213.
Lawrence Berkeley National Laboratory’s 2024 report gets numbers of a similar size for U.S. data centers alone[3], and even after you add the water used to generate their electricity, farm animals use far more[3]. CBS News, using LBNL and University of Twente numbers, reported that Americans use more water to grow food, especially meat, than all U.S. data centers combined[3][4].
Read full article →OpenAI query water · University of Michigan Beyond Burger LCA
One cow-flesh burger uses as much water as about 1.4 million AI prompts
Going by OpenAI’s figure for the water in a single prompt[1] and the University of Michigan study of how much water a quarter-pound cow-flesh patty takes[3], you would need about 1.4 million prompts to equal the water in one cow-flesh burger[1][3].
OpenAI put an average ChatGPT query at about 0.32 milliliters of water[1], and the Michigan study put that patty at 437.7 liters[3]. Divide one by the other and you get about 1.4 million[1][3]. You would have to use a chatbot nonstop for years to send that many prompts, while tens of billions of farm animals are fed and watered every day, whether or not anyone opens a chat window.
Depending on the sources and how you calculate it, the figure ranges from about 8,800 to 5.5 million prompts[1][2][4]. The low end comes from Li et al. 2023, which put a GPT-3 reply at about 50 milliliters[2]. The high end uses OpenAI’s 0.32 milliliters[1] and Mekonnen and Hoekstra’s estimate of about 1,750 liters (about 460 gallons) for a quarter pound of cow flesh[4].
The 437.7-liter figure counts only blue water, meaning water taken from rivers, lakes, and wells[3]. It leaves out green water, the rain that falls on pasture and feed crops, so the full amount is higher[4]. Many people wrongly assume figures like this are mostly rain.
Read full article →The bottle-per-email water figure is misleading
In 2024, the Washington Post reported that ChatGPT used about 519 milliliters of water to write a 100-word email, a little more than a bottle, and the number spread widely[1]. But that estimate modeled GPT-4 writing a full email, which is neither an average prompt nor the figure OpenAI later published, about 0.32 milliliters[3].
The research paper behind those headlines, Li et al. 2023, put a medium-length GPT-3 reply at about 10 to 50 milliliters, not a bottle each[2]. The headlines turned that into a bottle per email.
OpenAI’s published average[3] and the Michigan burger study give about 1.4 million prompts per burger[3], and other studies give higher and lower numbers around it[2].
Read full article →University of Michigan LCA · cited by Beyond Meat, A Burger With Benefits
Heller & Keoleian, University of Michigan, CSS18-10
Vegan burgers already use about 99% less water than cow-flesh burgers
The University of Michigan study that Beyond Meat cites compared a quarter-pound Beyond Burger with a quarter-pound U.S. cow-flesh burger[1][2]. The plant patty used 99% less water, 93% less land, 90% fewer greenhouse gases, and 46% less energy[1][2]. Measured as scarcity-weighted water (Table ES1), the Beyond Burger used 1.1 liter-eq and the cow-flesh burger 218.4 liter-eq[2]. Measured as the actual volume of water consumed (Tables 7 and 8), it was 3.27 liters against 437.7 liters (about 116 gallons), still more than 99% less[2].
Once you count the rain that falls on pasture and feed crops, Mekonnen and Hoekstra put cow flesh at about 15,400 liters per kilogram globally, or about 460 gallons for a quarter pound[4]. You don’t have to give up burgers to cut that water use, because the vegan version is already in stores.
Read full article →
Our World in Data · How is the world’s land used?
Data centers take up a tiny fraction of the land used for farm animals
According to Our World in Data, drawing on FAO and on Poore and Nemecek, about 44% of the world’s habitable land is already farmed, about 48 million km²[1][2]. Grazing and feed for farm animals take up about 80% of that farmland[1][2], an area roughly the size of the Americas[1]. If the whole world ate vegan, farmland would shrink from about 4.1 billion hectares to about 1 billion[1][2].
Industry counts put the number of data centers worldwide at about 10,000[3]. Even if every one of them took up 200 acres, the total would be about 8,000 km², a fraction of one percent of the land used for farm animals[1][3]. Some new data centers are going up on former farm fields, but that’s a local planning fight, nothing like the land used for cows.
Read full article →
Our World in Data · Poore & Nemecek (2018)
Farm animals produce about 40 times more greenhouse gases than all data centers
The International Energy Agency estimated that the world’s data centers used about 415 TWh of electricity in 2024, about 1.5% of global electricity, and that generating that power released about 180 million metric tons of CO₂, roughly 0.5% of the world’s CO₂[1]. In its 2035 base case, that rises to about 300 million metric tons, still under 1.5% of energy-sector emissions[1].
The UN’s Food and Agriculture Organization puts farm animals at about 14.5% of all human greenhouse gas emissions, or about 7.1 billion metric tons of CO₂-equivalent, mostly from methane, manure, and land cleared to grow feed[3]. Divide 7.1 billion by 180 million and you get about 40[1][3]. Poore and Nemecek put the whole food system at 26% of global greenhouse gases[4], and Xu and colleagues found that animal foods, including their feed, make up about 57% of food-system emissions[5]. Data centers are a real and growing drain on electricity, but breeding tens of billions of animals is a far bigger climate problem.
Read full article →
Our World in Data / FAO slaughter statistics
Animal agriculture kills about 88 billion land animals a year, and AI doesn’t slaughter any
Faunalytics, using FAO slaughter data, counts about 87.9 billion land animals killed for food in 2024[1], or about 241 million a day[1]. About 78.5 billion of them were chickens[1]. Add about 130 billion farmed fish a year[3] and an estimated 1.1 to 2.2 trillion wild-caught fish[4], and the food system kills more animals than anything else on the planet. AI doesn’t raise, crate, or stun animals to make its product. Building data centers and power lines can harm wildlife, the way building any factory can, but that’s accidental harm, which is different from breeding tens of billions of animals in order to kill them.
Read full article →Data centers can strain local water supplies, but farm animals use far more water
A single hyperscale data center can use millions of gallons of water a day[2], and about 40% of U.S. data centers are in areas with high water stress[4]. For the towns next door, that’s a real problem, but it doesn’t make AI a bigger water user than farm animals. Farm animals push some of the world’s rivers past their freshwater limits, and most of that water goes to irrigating feed crops spread across many river basins. Both things are true: don’t build a cooling tower on a failing aquifer, and don’t pretend chatbots are the main drain on water. The main drain is animal farming.
Read full article →Articles
Farm animals have a far bigger water, land and climate footprint than AI
Farm animals use about 200 times as much freshwater as AI data centers are projected to use, and they have about 40 times the climate impact of every data center on Earth. Next to animal farming, AI’s footprint is small.
Read full article →
One cow-flesh burger uses as much water as about 1.4 million AI prompts
Going by OpenAI’s own figure for the water used by an average ChatGPT prompt and a University of Michigan study of a quarter-pound cow-flesh patty, you would need about 1.4 million prompts to use as much water as one burger. Plant burgers that do the same job are already on the shelf.
Read full article →
Sources
Papers and datasets. Each card keeps its caveats, including small samples and single-animal cases.
[1]The Verge / OpenAI · Review
Sam Altman says an average ChatGPT query uses about 0.32 milliliters of water
OpenAI’s Sam Altman put the water used by an average ChatGPT query at about 0.32 milliliters. By the University of Michigan’s count, a 1/4 lb cow-flesh patty consumes 437.7 liters of water, as much as about 1.4 million prompts. Going by Mekonnen and Hoekstra’s figure of about 1,750 liters for the same quarter pound, it’s about 5.5 million.
[2]University of Michigan CSS, 2018 · Review
A Beyond Burger uses 99% less water and 93% less land than a cow-flesh burger
Heller and Keoleian at the University of Michigan found that a 1/4 lb Beyond Burger uses 99% less water, 93% less land, and 46% less energy than a 1/4 lb U.S. cow-flesh burger, and produces 90% fewer greenhouse gases. The cow-flesh burger consumes 437.7 liters of water and the Beyond Burger 3.27 liters, or 218.4 versus 1.1 liter-equivalents once weighted for water scarcity.
[3]Beyond Meat, 2018 · Review
Beyond Meat’s summary of the University of Michigan burger study
Beyond Meat’s summary of the University of Michigan life-cycle assessment says the Beyond Burger uses 99% less water, 93% less land, and 46% less energy than a 1/4 lb U.S. cow-flesh burger, and produces 90% fewer greenhouse gases.
[4]Ecosystems, 2012 · Review
A quarter-pound cow-flesh patty takes about 1,750 liters of water
Mekonnen and Hoekstra put the global average water footprint of cow flesh at about 15,400 liters per kilogram, counting green, blue, and grey water, so a quarter-pound patty takes about 1,750 liters, or around 460 gallons. At OpenAI’s figure of 0.32 ml per prompt, that’s about 5.5 million prompts, or tens of thousands to hundreds of thousands at Li and colleagues’ 10–50 ml.
[5]Water, 2024 · Observational
Farm-animal food systems use about 228 cubic kilometers of freshwater a year
Wisser and colleagues found that farm-animal food systems consume about 228 cubic kilometers of freshwater a year, more than 90% of it to grow feed.
[6]FoodFacts · Review
Even in 2028, AI data centers may use about 200 times less water than farm animals
FoodFacts, summarizing Morgan Stanley and other reported figures, puts AI data-center water use near 1 cubic kilometer in 2028, roughly 200 times less than the 228 cubic kilometers used for farm animals.
[7]Lawrence Berkeley National Laboratory, 2024 · Review
U.S. data centers used about 66 billion liters of water on site in 2023
Lawrence Berkeley National Laboratory found that U.S. data centers used about 66 billion liters of water on site in 2023, and projected about 150–275 billion liters for 2028. Even more is used indirectly at the power plants that supply them, and farm animals use more still.
[8]arXiv, 2023 · Review
GPT-3 uses about 500 ml of water for every 10–50 replies
In “Making AI Less Thirsty,” Li and colleagues estimated that GPT-3 uses about a 500 ml bottle of water for every 10–50 medium-length replies, or about 10–50 ml each, depending on when and where it runs. At that rate, the 437.7 liters consumed for a 1/4 lb cow-flesh patty in the University of Michigan study equal about 8,800 to 44,000 prompts, the low end of the range.
[9]CBS News · Review
Americans use more water growing food, especially meat, than all U.S. data centers
CBS News reports that Americans use more water to grow food, especially meat, than all U.S. data centers use combined.
[10]Faunalytics / FAO · Observational
About 87.9 billion land animals were killed for food in 2024
Faunalytics, using FAO data, counts about 87.9 billion land animals killed for food in 2024. No animals are slaughtered to run AI.
[11]Science, 2018 · Meta-analysis
Avoiding meat and dairy is the biggest way to cut your food’s footprint
Poore and Nemecek found that avoiding meat and dairy is the single biggest way to cut the land, water, and climate impact of a meal. Food causes about 26% of the world’s greenhouse gas emissions.
[12]IEA, 2025 · Review
Data centers used about 1.5% of the world’s electricity in 2024
The International Energy Agency found that data centers used about 415 TWh in 2024, about 1.5% of the world’s electricity, and generating that power released about 180 million metric tons of CO₂, roughly 0.5% of the world’s CO₂. The FAO puts farm animals at about 14.5% of all human-caused greenhouse gases, about 40 times as much.
[13]FAO, 2013 · Review
Farm animals cause about 14.5% of human-caused greenhouse gas emissions
The FAO found that farm-animal supply chains accounted for about 14.5% of human-caused greenhouse gas emissions, about 7.1 billion metric tons of CO₂-equivalent, from methane, manure, and land converted to grow feed.