How much water does AI consume?

Summary

---

More or Less Podcast: "How much water does AI consume?"

BBC Radio 4 | Host: Charlotte MacDonald | Date: March 28, 2026
Episode Focus:
A deep dive into the startling claims about the amount of fresh water consumed by Artificial Intelligence (AI)—and an investigation into what the numbers really mean for the environment and for the future.

---

Episode Overview

This episode explores how much water AI truly uses globally and untangles some commonly misreported statistics about AI’s water consumption. With the rapid expansion of AI and large language models, public concern has grown over the “invisible” resources required to power and cool AI data centers. The presenters scrutinize alarming figures reported in the media, trace their origins and accuracy, and ask: should we be worried about AI’s thirst for water?

---

Key Discussion Points & Insights

1. The Origins of Alarming Water Consumption Figures

• Initial Claim:
  • Karen Howe in "Empire of AI" writes: AI could consume 1.1 trillion to 1.7 trillion gallons of fresh water yearly by 2027—“half the water annually consumed in the UK” (04:10).
• Conversion: This equals 4.2 to 6.6 trillion liters.
• Issue Highlighted:
  • The distinction between water CONSUMED (not returned to the system) vs water WITHDRAWN (temporarily used, some returned afterward).

2. Debunking the Numbers: Consumption vs. Withdrawal

• Nathan Gower explains:
  • The 4 to 6 trillion liter figure incorrectly refers to withdrawal, not consumption.
  • “So Howe got her figures from a paper…those figures…aren’t for water consumption. They’re…for…water withdrawal.” (03:40)
• Correction:
  • Real consumption is only 10% of that: 380 to 600 billion liters per year (04:23).

Notable Quote:

“So the author's stuck the wrong label on the 4 to 6 trillion litre figure. It should be for withdrawal, not consumption.”
— Charlotte MacDonald [04:12]

3. Digging Deeper: Flawed Research Chain

• Underlying Source Issue:
  • The original paper used an estimate from researcher Alex de Vries Gao.
  • That estimate was not for all AI servers in use, only for those built in 2027.
• Effect:
  • Massive underestimation of total electricity use (05:32), leading to flawed water figures.

Notable Quote:

“It’s a bit of a mess.”
— Nathan Gower [05:52]

4. Estimating Actual AI Water Consumption

• Alex de Vries Gao’s Methodology:
  • Traced number of AI chips produced, server modules, then actual servers.
  • Looked at utilization rates and water used per unit of power (06:13).

Notable Quote:

“I tried to look at the biggest buyers…like Microsoft, Google…how do these companies’ data centers perform in terms of water intensity…used those numbers to translate my power demand estimate into…water consumption estimate.”
— Alex de Vries Gao [06:13]

• Key Figure:
  • By late 2025, AI systems were consuming about 750 billion liters of water per year (07:14).
  • That’s more than global bottled water consumption (446 billion liters).

Memorable Explanation:

“This is exceeding the level of global bottled water consumption…It could cause a lot of problems if this consumption is concentrated in a single location…But we just don’t know at this time.”
— Alex de Vries Gao [07:30]

5. Where Does the Water Go?

• 90% of AI’s water consumption happens off-site at power stations (from rivers/lakes).
• 10% happens on-site in data centres, often using drinking water (08:26).
• Water Consumption vs. Withdrawal Debate:
  • Some experts argue water withdrawal, even if mostly returned, can still impact ecosystems and local supplies.

6. Predicting the Future

• Bottleneck:
  • Can tech companies find the power (and water) needed for relentless AI growth?
• Unknowns:
  • Future supply chains make precise projections speculative.
  • Local impacts may be severe, but global numbers don’t give the full picture (09:53).

Notable Quote:

“This is still going to be adding on top…But again, I can’t say anything about whether it’s going to be possible to find a home for all this equipment.”
— Alex de Vries Gao [09:06]

---

Timestamps of Important Segments

• 01:16 — Intro, framing the environmental question around AI’s “invisible” resource use.
• 03:05 — Defining water consumption vs. withdrawal.
• 04:23 — Actual water consumption figures clarified.
• 05:40 — How academic errors propagate through media.
• 06:13 — Alex de Vries Gao describes his methodology.
• 07:14 — Staggering “750 billion liter” figure revealed.
• 08:26 — On-site vs. off-site water use explained.
• 09:06 — Challenges in making future predictions.

---

Notable Quotes & Memorable Moments

• Charlotte MacDonald [04:12]:

  “So the author's stuck the wrong label on the 4 to 6 trillion litre figure. It should be for withdrawal, not consumption.”

• Nathan Gower [05:52]:

  “It’s a bit of a mess.”

• Alex de Vries Gao [07:30]:

  “This is exceeding the level of global bottled water consumption... It could cause a lot of problems if this consumption is concentrated in a single location... But we just don’t know at this time.”

• Alex de Vries Gao [09:06]:

  “I can make statements based on…hardware, which…means…the cumulative power demand of AI systems is still going to be rising. This is still going to be adding on top... But again, I can’t say anything about whether it’s going to be possible to find a home for all this equipment.”

---

Summary Table

| Concept/Statistic | Flawed Figure | Correct Figure | Notable Detail | |:------------------------------------|:----------------------------|:----------------------|:-------------------------------------------------| | Global AI Water “Consumption” (2027)| 4.2–6.6 trillion liters | 380–600 billion liters| Error: Figure was for withdrawal, not consumption| | Actual 2025 AI Water Use | N/A | 750 billion liters | Exceeds global bottled water consumption | | % On-site (data center) | N/A | 10% | Mostly uses drinking water | | % Off-site (power stations) | N/A | 90% | Uses rivers & lakes; location matters |

---

Conclusion

This episode neatly unravels how a startling statistic—AI will soon “consume half the UK’s water”—got inflated through a series of academic and journalistic misinterpretations. The true figure is dramatically lower (but still significant), centering debate on what metrics (consumption vs withdrawal) truly matter for the environment. The episode closes with uncertainty about AI’s future water demands and location-specific risks, reminding listeners that “huge” global numbers are only part of the story.

---

Transcript

[Advertisement Voice]

This BBC podcast is supported by ads outside the uk.

Today's episode is sponsored by Smart Travel, a podcast from NerdWallet. You know that one friend who always finds the best travel deals, picks the right credit cards and somehow ends up in first class for the price of coach? Smart Travel is like that friend, but in podcast form. NerdWallet's travel journalists sort out what's worth paying for. From lounge access to travel advisors to hotel loyalty programs, make your travel dollars work harder. Follow Smart Travel on your favorite podcast app.

The best B2B marketing gets wasted on the wrong people. So when you want to reach the right professionals, use LinkedIn ads. LinkedIn has grown to a network of over 1 billion professionals, including 130 million decision makers. And that's where it stands apart from other ad buyers. You can target your buyers by job title, industry, company role, seniority, skills, company revenue so you can stop wasting budget on the wrong audience. It's. It's why LinkedIn Ads generates the highest B2B return on ad spend of major ad networks. Spend $250 on your first campaign on LinkedIn Ads and get $250 credit for the next one. Just go to LinkedIn.com Broadcast that's LinkedIn.com Broadcast. Terms and conditions apply.

[Charlotte MacDonald]

Hello and thanks for downloading the more or less podcast. We're the programme that looks at the numbers in the news and in life and an AI water consumption. Hi, I'm Charlotte MacDonald. When you sit in front of an AI chatbot and start typing away, the responses appear on your screen like magic information, apparently springing out of fresh air. The truth is of course, very different when we send our queries off. These are dealt with by a vast network of data centers which whir into action to come up with the answers. These data centers contain servers which themselves contain processing chips. Those run on electricity from the power grid and as they operate, they generate lots of heat and need to be cooled to prevent overheating. Both electricity generation and cooling data centers use water. With AI expanding rapidly, some people are worried about how AI's water use might escalate in the future. One striking figure came in a book called Empire of AI, written by the US journalist Karen Howe. According to her, surging AI Demand could consume 1.1 trillion to 1.7 trillion gallons of fresh water globally a year by 2027, or half the water annually consumed in the UK. That's between 4.2 and 6.6 gallons trillion litres of fresh water, which sounds like a big number, but should we Trust it. Nathan Gower has been looking into this one. Hi, Nathan.

[Nathan Gower]

Hi, Charlotte.

[Charlotte MacDonald]

Nathan. Four to six trillion litres of fresh water Sounds like a lot.

[Nathan Gower]

It definitely is, but there's all sorts of problems here. So Howe's claim is about the amount of water that could be consumed. This has a specific meaning. It's water that's taken out of a system or source but not returned. This happens through evaporation when it's used for cooling, whether that's on site at data centers or off site at the power plants where electricity is generated. They also evaporate water when they cool the generators.

[Charlotte MacDonald]

Okay, so this claim seems to be saying that trillions of litres of water are going to get consumed. You know, they're taken out of the water system and then not returned.

[Nathan Gower]

Yes, but the claim is just not right. Howe got her figures from a paper published by academics from the University of California, Riverside. Except Those figures of 4 to 6 trillion liters aren't for water consumption. They're for something different called water withdrawal. That is the total amount of water that gets taken out of a water system or source. Now, some of this water will be returned, but some of it won't, will be consumed. So basically the consumption figure is a subset of the wider withdrawal figure.

[Charlotte MacDonald]

Okay, so this author's stuck the wrong label on the 4 to 6 trillion litre figure. It should be for withdrawal, not consumption. So what is the actual consumption figure?

[Nathan Gower]

So the paper says between 380 and 600 billion litres could be consumed in 2027. That's about 10% of the of the original 4 to 6 trillion figure. This mistake was pointed out by an American substacker, Andy Masley, and Karen Howe has since issued a correction.

[Charlotte MacDonald]

Right, so that's settled now.

[Nathan Gower]

It would be really great if it was. So Karen Howe read those figures in the academic paper and misinterpreted them, mistaking the larger figure for withdrawal as the figure for consumption. But I've discovered that those figures in the paper, they were actually wrong in the first place.

[Charlotte MacDonald]

How so?

[Nathan Gower]

The paper says it takes an estimate for global AI electricity use in 2027 and extrapolates from that to a figure for global AI water use in the same year. But that electricity estimate comes from work by another researcher, Alex de Vries. Gao. I spoke to Alex and it turns out that he wasn't estimating total global AI electricity use in 2027. Instead, he. He made an estimate for global AI electricity use by those AI servers that might be produced in 2027 alone.

[Charlotte MacDonald]

Right. So it's not counting all the AI servers and data centers built in previous years, most of which will presumably still be running.

[Nathan Gower]

Exactly. So they've basically underestimated electricity consumption and that throws off the water estimates. Then on top of all this, an author comes along and misinterprets these flawed figures.

[Charlotte MacDonald]

This sounds like a mess.

[Nathan Gower]

It's a bit of a mess. So let's try and salvage something that researcher Alex Dufries GAO has also made estimates for global AI water consumption. To do that, you need to know how much electricity global AI systems might demand. Now, the tech giants don't publish these numbers, so Alex came up with a novel strategy.

[Alex de Vries Gao]

Just to get to the power demand, you actually need to take a deep dive into the supply chain of AI hardware. What I did was looking at how many AI chips could have been produced in the past years, how many AI server modules could have been made with that, and ultimately how many AI servers have been made with that. When you get to that point, you still need some assumptions to figure out, like how much are those servers going to be utilized. What I did there was I tried to look at the biggest buyers of AI server equipment, which is large tech companies like Microsoft, Google, et cetera. I kind of examined, like, how do these companies, data centers typically perform in terms of water intensity? And I used those numbers ultimately to translate my power demand estimate into a indirect water consumption estimate for AI server hardware. And then on top of that, you still need to include the water that's actually being consumed in the data center itself, the direct water consumption.

[Nathan Gower]

Using this method, alex estimated that AI systems at the end of 2025 were consuming water at a rate of 750 billion litres per year. Now, I know what you're going to ask Charlotte. Is 750 billion litres a big number? Well, here's Alex.

[Alex de Vries Gao]

This is absolutely a big number. This is exceeding the level of global bottled water consumption, which is at about 446 billion liters. That in itself is a significant amount of consumption. But is it really a problem? It could certainly be. It could cause a lot of problems if this consumption is concentrated in a single location where water scarcity is already a potential problem, but we just don't know at this time. So that really makes it hard to make that translation, is this a problem or not? But at the same time, it also makes it impossible to say this is not a problem at all. It's a pretty huge number. It's going to have an effect on local freshwater supply for sure. We just don't really know where and we just don't really know how much this is going to hurt in which locations.

[Nathan Gower]

For Alex's estimate for water consumption, only about 10% of that is happening on site at data centers which typically use drinking water from local supplies. The other 90% is happening off site at power stations which use water from sources like rivers and lakes. Also, Alex has only made estimates for water consumption. He thinks that's the most important metric when thinking about overall water scarcity. But I've spoken to another researcher who thinks that water withdrawal matters equally, if not more, and argues that even if water is eventually returned to a system or source, increased demand can still have important consequences.

[Charlotte MacDonald]

What about future predictions?

[Nathan Gower]

Here's Alex again.

[Alex de Vries Gao]

One of the big bottlenecks that is starting to appear. Can these tech companies even find sufficient power to power their data centers? Like they're getting a lot of equip. Where are they going to be able to actually use all that equipment? I can make statements based on, let's say, the current supply chain capacity for producing AI hardware, which I know this year is probably going to be at least similar to the previous year, 2025, which means that the cumulative power demand of AI systems is still going to be rising. This is still going to be adding on top of the production of the power three years. But again, I can't say anything about whether it's going to be possible to find a home for all this equipment.

[Charlotte MacDonald]

Well, thank you Nathan, and thanks to Alex Devries Gao, as well as Professor Xiao Le Ren who also helped with this episode. That's all we have time for this week, but if you have any more questions or comments, please email us on moreorlessbc.co.uk we'll be back next week and until then, goodbye.

[Tyler Redick]

Tyler redick here from 2311 Racing. Game night's fun until someone spends five minutes lining up one shot. Chalk. Breathe. Rechock. Still aiming. While they figure it out, I fire up Champa Casino. I can spin anywhere, anytime. And there's always a new social casino game every week. Spins happen way faster than that shot. Play [e-mail] let's chum them. Sponsored by Chumba Casino. No purchase necessary VGW Group voidware prohibited by Law21.

[Nathan Gower]

Terms and conditions apply.

[Advertisement Voice]

The best B2B marketing gets wasted on the wrong people. So when you want to reach the right professionals, use LinkedIn ads. LinkedIn has grown to a network of over 1 billion professionals, including 130 million decision makers. And that's where it stands. Apart from other ad buys. You can target your buyers by job title, industry, company role, seniority skills, company revenue so you can stop wasting budget on the wrong audience. It's why LinkedIn Ads generates the highest B2B return on ad spend of major ad networks. Spend $250 on your first campaign on LinkedIn Ads and get $250 credit for the next one. Just go to LinkedIn.com Broadcast that's LinkedIn.com Broadcast Terms and Conditions apply.