Almost every single citation here is another article/blog post. Even on citations for actual statistics. The fact is that 90% of data centers are open loop systems and DO in fact use an inordinate amount of water. Closed loop data centers are not a great solution either as they require much more investment up front and require the infrastructure to handle the toxic sludge that they have to excrete every month when they "bleed the lines"
I think this article hand waves and side steps what I see as two notable issues.
The main one is aquifer depletion. The consumptive vs withdrawal argument mostly holds water, but consumptive is a sliding scale. Evaporative cooling for data centers is solidly at the far ‘truly consumptive’ end of that scale because the consumed water will not reenter the local watershed. That’s problematic because aquifers are very slow to refill. So this is genuinely a concern in water stressed areas.
The other is the weak growth model. I suspect we’re only going to see faster and faster growth of data centers in coming years, making the consumption there more exponential than linear. Meanwhile the majority of the other consumptive consumers are strongly tied to demographics and population growth is slowing everywhere. For example agricultural water use in the US has held steady or even declined in recent years.
In fairness, part of that agricultural decline in use is from advancement in technology and methodology and we’ll likely see the same with data centers, but those numbers are unpredictable.
On the whole I agree that the concern over data centers in terms of water (and electricity) usage is overblown to an extent, but I think we do need to pay closer attention to the points that actually matter when looking at the situation.
As a Euro/American, it's kind of shocking to me that the US is fast re-inventing communism (and fascism) from first principles: apparently, industries and technologies are 'good' or 'bad' based on their alignment with the ideology of a vocal 'elite'. Last time this happened (70s) we put nuclear energy on a shelf and ended up with global warming.
The West has invented a perfect mechanism for controlling the use of resources for the common good: money. If water is a scarce resource, charge a lot of money for it. If AI (or whomever) can pay for it, so be it, other sources of water (desalination perhaps?) will be invented.
I am saying this as a liberal and an AI-sceptic: Ideology should never outweigh outcomes.
It is interesting, and this conversation has been happening buried within other threads on HN - it would be good to have the discussion as its own topic.
Sadly it's flagged, so for now I guess it's still too contentious.
The AI water usage is a real problem because AI datacenters are effectively unregulated.
Yes, a car factory would be expected to use water, which is why we have environmental regulations in place.
"Lagging behind in AI" is treated as a national emergency which warrants forgetting about regulations.
This isn't some sort of quirk or accident; it's very much created by people writing headlines lie "The AI water issue is fake".
The following are symptomatic and typical:
- 800 data center approved in spite of vehement protests of the locals [1]
- Data center blatantly violates water regulations. When it comes to light, administration refuses to do anything about it [2]
- Data center gets constructed in historically disadvantaged / redlined areas disproportionately and adversely affecting Black people. Air quality drops. Nothing is done about it [3]
The AI water issue is indeed a red herring, because it is a part of a larger issue of the AI industry having free reign over anything that's deemed necessary for its existence.
The impact on people's lives and health is real. To say that the issue is "fake" is disingenuous at best, disinformation at worst.
I'd say it's heavily biased but tries hard to not look like it, in a way similar to what tobacco and fossil fuel corporations have been doing for decades.
The author is financed by some 'effective altruism' cultists operating a supposedly philantropic fund.
For a less partisan article on the same topic, this one might be worthwhile to you:
The problems with 'data centers for AI' don't stop at water consumption, which is absolutely a real risk in some cases due to the relevant corporations having a knack for finding areas with really small or stressed aquifers and building there.
It's also somewhat common to skirt or just straight up abuse legal protections against corporate tyranny, e.g. by running noisy or polluting turbines.
Then the idea itself to eradicate the Internet and put everyone behind a SaaS database intermediary should be quite unpalatable to any sane and freedom loving person. Especially since these databases obviously make quite a lot of people insane or otherwise mentally disabled, including prominent figures in the movement to push this through.
>The developer, Quality Technology Services, owed nearly $150,000 for using more than 29 million gallons of unaccounted-for water.
>The company said its water consumption was so high last year because of temporary construction-related activities, such as concrete work, dust control and site preparation.
I do find it interesting that the framing is "a data center" used a bunch of water if it really is "manufacturing concrete uses a bunch of water."
Different source and event, same misleading headline as the one mentioned in the parent article.
I think that largely depends on how you engage with the internet. To continue the metaphor, the internet has an absurdly high noise floor, but it’s easily filtered noise. So if you do that filtering you’re actually left with a pretty high SNR.
I think it's harmful & a turn off to frame what is a big huge general human issue as political, as party line stuff.
There's a ton of reasons that are very very sensible to be very pissed off about AI. People very rarely benefit from human support staff being turned into chatbot AI. The unyielding limited mechanized systems we are forced to interact with suck.
And the billionaires themselves are so loudly advertising that the purpose of these things is to replace people. To obsolete so much labor. When it's so clear they suck, in so many ways. That people are upset is not just "the party line".
I've gone on about AI (even though I personally love it & think it's a miracle, having help in life I never thought I would get). But I don't want to fixate on AI. The sociological effect here, of how people constantly hunt for basically parallel constructions for what they want to believe: that's such a major factor in understanding people and societal forces. I think you're really smashing your ability to think to bits to chalk this up as "the party line". It's such a deeper more interesting force than than.
It again just goes to ai, which I think is not by point, but man what a moment of people seeking their existing biases: the "AI generated Monet" painting that actually is in fact a real Monet (Water Lilies). People went to town!
https://bsky.app/profile/segyges.bsky.social/post/3mlsgc53ry...
This effect deeply deeply deeply predates the internet. I think perhaps yes there is a little more biting sarcasm & anger & negative energies that have infected us, by being exposed to so much attention-driven systems where negative energies rule! But this chasing your existing beliefs has been around probably since before the dawn of man. I'd put money in it.
"AI's water usage is being approved without proper planning, because of the arguably fake sense of urgency around it."
Other industries that use significant water have significant regulations already . AI has been desperately trying to avoid ANY regulation (unless it forces folk to use AI.)
I don't personally think water usage is the biggest issue with how AI is being rolled out, but it's one that is easier to engage the public on then copyright, or societal context collapse. :)
I'm pretty convinced the only reason people care is because it's been memeified, and to be fair, the running gag of someone chugging a bottle of water before giving a lacklustre response to an easy question is pretty funny. But this sudden care for water consumption despite the public's general apathy towards it with regards to literally everything else (mining, livestock, textiles, energy generation, Coca Cola, etc), just seems manufactured.
Near 100% of the complaints I see online and in person about AI water usage are based on the idea that it’s consuming out of control amounts of water.
It’s frequently brought up as a reason not to use AI. The public perception right now is that doing anything with AI causes a lot of water to be “used” and that this is a very bad thing.
I can't imagine what possessed the author to write this much text on the topic. But yes, that was always my impression. My pet conspiracy theory - which is completely unsubstantiated - is that if you're pushing for data centers in rural communities, you actually want people to get fixated on water usage, because you can then talk to the county commissioners and shoot down these criticisms so easily.
Mega-scale AI data centers have other externalities. They're often touted as a way for rural counties to become the hubs of the digital era, but they don't employ many people, don't generate a whole lot of tax revenue, and basically just leverage cheap electricity at the expense of local residents. So it's a sham in that respect. You're not gonna have X, Google, Amazon, or Meta reinvigorate your community. You're just gonna have ratepayers subsidize some inference via higher electricity bills.
I have no doubt that someone will chime in saying with an "actually..." that electricity is fungible and therefore, it doesn't matter where the datacenter is built. If it were so, they wouldn't be getting built in places like Wyoming or eastern Washington, and electricity prices in these markets would be the same as they are in the SF Bay Area. In practice, though, there's plenty of factors that make the US electricity markets a lot more local.
The electricity rates in the SF Bay Area are astronomical primarily because PG&E is trying to pay off the fire damage and deaths they have been causing regularly since 2015.
In my opinion, the company is a colossal disaster. But more importantly, CPUC is complicit, bordering on incompetent [1]. Our regulatory commission is not in the business of protecting utility consumers.
Using these electricity rates as an example of where rates are going because datacenters are being built in those areas is wild! The rates have essentially nothing to do with datacenters.
The part in the article that is actually of interest to me focuses on the tax revenue section. Can someone versed in this explain?
My initial reaction is b.s. Companies building these data centers, in many instances, get tax breaks to start building. On top if it they get different breaks on cost of electricity or materials. And on top of it, we all know that corporates pay less taxes than individuals already. And last but not least, data centers don't require a lot of staff, so there is no "trickle down".
> Importantly, I am not saying that it’s impossible for data centers to ever cause any problems with water. They require careful planning in the same ways other large industrial buildings do.
(Emphasis added)
I think that part of why AI data centers seem to have more stringent local objections even in areas that are generally pretty friendly to industrial development than do other industries that use a lot of water (or a lot of electricity) may be that there often isn't enough time to do that careful planning.
If you wanted to build say a new paper manufacturing plant there would be local objections because paper manufacturing a lot of both water and electricity. But nobody urgently needs a new paper manufacturing plant. There is plenty of time to address the objections and figure out something that will make most of the locals happy.
With AI data centers there is a (possibly bogus) need for a lot in a short timeframe, and so a lot of pressure on officials to approve them without the careful planning such a project should have.
The table of contents opens links in a new tab. If they didn't, they would require a full page reload, because they don't use fragments. This is seemingly how substack is designed.
1. No accounting for other countries or externalities such as large corporation leveraging to get what it wants at the expense of community members.
2. > Put another way, almost all (80%) the reported water used by AI occurs during the generation of electricity
Well that sure seems like a problem to me cuz that’s water that now needs to be used for generation that wasn’t prior and it’s a significant amount. And that impacts my water and electricity prices now.
Edit for 3: > Consumptive use can harm total access to freshwater, but freshwater sources are also regularly being replenished.
Yeah I don’t think so when Mexico City is sinking because its aquifer was depleted.
All in all he’s got some interesting points but I think he’s hyper focused on numbers and ignoring broader things.
The entire global semiconductor industry, source of vast benefits, only uses about 50000 acre-feet of water per year, which is essentially nothing. As a point of comparison this is half a percent of what the paper industry consumes.
Where is the source for that? I am only familiar with one site, the STMicro one in the Alps, which already used 4000 acre-feet / 5 million m³ per year in 2023 [1], and it's been at least doubled since then. This is a huge chunk of the total water consumption of the region, and there are NIMBY demonstrations frequently because of that [2]. It's also surprisingly polluting. Whether you think the value is worth or not is a different story (I worked there, so guess).
The figures in that report are consistent with what I said. This is because the usage we're talking about is evaporation. This should not be confused with withdrawals. Semiconductor industry mainly withdraws water and then discharges it as effluent. Evaporation is a minor component of their withdrawals.
Listen, I really like LLMs and diffusion models and machine learning and all this stuff, and I want to see it happen in a just and sustainable way. "AI" doesn't necessitate extreme waste. If anything, reasonable policy constraints would push "AI" to be even better.
I feel like the way many companies implement AI right now is very very wasteful. For context I'm looking into adding some AI elements to my SaaS app and I'm looking at running on-device TinyBERT intent classifiers then have my API take it from there (still experimenting with this).
I feel like this is a pretty sustainable way to implement AI in an application, meanwhile I see most companies just implement with OpenAI API + some custom prompts on top.
Granted I've had to do this for some of my clients and it's a pretty easy way to implement AI, though I always have the sinking feeling that we could achieve the same thing in a way more efficent manner and a bit more effort.
> If anything, reasonable policy constraints would push "AI" to be even better.
Like what, though? I'm not opposed to AI regulation at all, but the very last thing I expect it to fix is the resource constraints around GPGPU compute.
Can you engage on the content? I hate AI and would love to have this as a numerate way to complain about it, but my current perception is that the water problem is nowhere near as large as the concern.
I don't engage with the "firehose of falsehood" types of articles as it's a losing battle. This guy's is a high school teacher whose side-gig is churning out pro-AI content for validation.
I simply disregard the entire article since I know the author's MO, and I inform other HN members of it.
I despise armchair libertarians and Substack "intellectualism" as much as the next guy, but your comment doesn't offer any pushback to the claim being made.
I don't engage with the "firehose of falsehood" types of articles as it's a losing battle. This guy's is a high school teacher whose side-gig is churning out pro-AI content for validation.
I simply disregard the entire article since I know the author's MO, and I inform other HN members of it.
So basically you're asking everyone to just trust you, a random commentator that provides no evidence or sourcing, over an article written with extensive sourcing, details, and explanations?
You might be the correct person here, but you're not going to convince anyone like this.
I don't see the point of measuring how much water we use if we're not poluting it. It's water. It will evaporate and rain down again. It's renewable resource.
By that logic water can never be wasted, it all stays on Earth and eventually comes down somewhere.
Of course water use above replenishment rates is bad, it doesn't magically rain down in the same spot and all the underground water tables get full again. They deplete, meaning existing consumers have to dig deeper or just go without water. Even ancient peoples knew that if you take too much water from a well, it will dry out.
I imagine you would see the point in measuring how much water data centres use when one opens near you, and you can't flush your toilet any more.
You can drain aquifers faster than they get re-filled. Fresh water is renewable, but it renews relatively slowly in many places.
Months in some places. But in the arid locations, it can take thousands of years for water to go from surface to aquifer.
A lot of these data centers and chip fabs are built in arid places because labor is cheap, taxes are low, and land is cheap. The reason for those three things is that there's not enough freaking water in the first place.
Slurping it up to run digital addiction mills and predatory advertising falls somewhere on a spectrum ranging from just plain stupid to abhorrently immortal.
I agree but then using a volume of water as a measure without provided context is just fearmongering.
Draining oasis in a desert might have much higher impact than one of the thousands of lakes in canada but still, it's a renewable resource. Most of the places suitable for datacenters have plenty of it anyway as datacenters are more suited for colder climates which usually have plenty of water.
My interpretation of their point was that more demand for water with a fixed supply of water will only make our water problems worse. Not that data centers are causing this problem on their own.
Water resources are isolated, so moving purified water from one basin to another via evaporation does matter. But the OP is right that the specific cases of data centers usually don't matter.
It will evaporate and rain down again somewhere else. It's lost locally. In a place with a limited local water supply, that still will cause plenty of local pain.
Almost every single citation here is another article/blog post. Even on citations for actual statistics. The fact is that 90% of data centers are open loop systems and DO in fact use an inordinate amount of water. Closed loop data centers are not a great solution either as they require much more investment up front and require the infrastructure to handle the toxic sludge that they have to excrete every month when they "bleed the lines"
I think this article hand waves and side steps what I see as two notable issues.
The main one is aquifer depletion. The consumptive vs withdrawal argument mostly holds water, but consumptive is a sliding scale. Evaporative cooling for data centers is solidly at the far ‘truly consumptive’ end of that scale because the consumed water will not reenter the local watershed. That’s problematic because aquifers are very slow to refill. So this is genuinely a concern in water stressed areas.
The other is the weak growth model. I suspect we’re only going to see faster and faster growth of data centers in coming years, making the consumption there more exponential than linear. Meanwhile the majority of the other consumptive consumers are strongly tied to demographics and population growth is slowing everywhere. For example agricultural water use in the US has held steady or even declined in recent years.
In fairness, part of that agricultural decline in use is from advancement in technology and methodology and we’ll likely see the same with data centers, but those numbers are unpredictable.
On the whole I agree that the concern over data centers in terms of water (and electricity) usage is overblown to an extent, but I think we do need to pay closer attention to the points that actually matter when looking at the situation.
I posted this because I thought it was interesting. I've seen a lot of comments about AI water usage.
This article feels a bit biased and I would love to see HN's take.
As a Euro/American, it's kind of shocking to me that the US is fast re-inventing communism (and fascism) from first principles: apparently, industries and technologies are 'good' or 'bad' based on their alignment with the ideology of a vocal 'elite'. Last time this happened (70s) we put nuclear energy on a shelf and ended up with global warming.
The West has invented a perfect mechanism for controlling the use of resources for the common good: money. If water is a scarce resource, charge a lot of money for it. If AI (or whomever) can pay for it, so be it, other sources of water (desalination perhaps?) will be invented.
I am saying this as a liberal and an AI-sceptic: Ideology should never outweigh outcomes.
It is interesting, and this conversation has been happening buried within other threads on HN - it would be good to have the discussion as its own topic.
Sadly it's flagged, so for now I guess it's still too contentious.
To say it's a bit biased is an understatement.
The AI water usage is a real problem because AI datacenters are effectively unregulated.
Yes, a car factory would be expected to use water, which is why we have environmental regulations in place.
"Lagging behind in AI" is treated as a national emergency which warrants forgetting about regulations.
This isn't some sort of quirk or accident; it's very much created by people writing headlines lie "The AI water issue is fake".
The following are symptomatic and typical:
- 800 data center approved in spite of vehement protests of the locals [1]
- Data center blatantly violates water regulations. When it comes to light, administration refuses to do anything about it [2]
- Data center gets constructed in historically disadvantaged / redlined areas disproportionately and adversely affecting Black people. Air quality drops. Nothing is done about it [3]
The AI water issue is indeed a red herring, because it is a part of a larger issue of the AI industry having free reign over anything that's deemed necessary for its existence.
The impact on people's lives and health is real. To say that the issue is "fake" is disingenuous at best, disinformation at worst.
[1] https://www.yahoo.com/news/articles/800-acre-data-center-app...
[2] https://www.tomshardware.com/tech-industry/georgia-data-cent...
[3] https://islandpress.medium.com/data-centers-threaten-black-c...
Is regulation the word you're looking for? It seems like you just need to get pricing right not regulation.
Yes, the word I'm looking for is "regulation".
Environmental pricing doesn't make much sense now, does it?
Another word I'm looking for is externality, as in: negative externality.
Actually, I'm not looking for words at all. I know these words well, and use them with intent.
Do you?
I'd say it's heavily biased but tries hard to not look like it, in a way similar to what tobacco and fossil fuel corporations have been doing for decades.
The author is financed by some 'effective altruism' cultists operating a supposedly philantropic fund.
For a less partisan article on the same topic, this one might be worthwhile to you:
https://www.lincolninst.edu/publications/land-lines-magazine...
The problems with 'data centers for AI' don't stop at water consumption, which is absolutely a real risk in some cases due to the relevant corporations having a knack for finding areas with really small or stressed aquifers and building there.
It's also somewhat common to skirt or just straight up abuse legal protections against corporate tyranny, e.g. by running noisy or polluting turbines.
Then the idea itself to eradicate the Internet and put everyone behind a SaaS database intermediary should be quite unpalatable to any sane and freedom loving person. Especially since these databases obviously make quite a lot of people insane or otherwise mentally disabled, including prominent figures in the movement to push this through.
The internet has disconnected us from reality so much that a lot of people simply don't care. If it supports the party line they are for it.
https://www.politico.com/news/2026/05/08/georgia-data-center...
>The developer, Quality Technology Services, owed nearly $150,000 for using more than 29 million gallons of unaccounted-for water.
>The company said its water consumption was so high last year because of temporary construction-related activities, such as concrete work, dust control and site preparation.
I do find it interesting that the framing is "a data center" used a bunch of water if it really is "manufacturing concrete uses a bunch of water."
Different source and event, same misleading headline as the one mentioned in the parent article.
Contra Sir Tim Berners-Lee, the signal-to-noise ratio on the internet has been terrible.
One seriously wonders if GenX didn't manage to hit a sweet spot for technological availability ahead of everything just going to seed.
I think that largely depends on how you engage with the internet. To continue the metaphor, the internet has an absurdly high noise floor, but it’s easily filtered noise. So if you do that filtering you’re actually left with a pretty high SNR.
Theres major issues that people got really good at figuring out how to make really outsized spikey signals.
I think it's harmful & a turn off to frame what is a big huge general human issue as political, as party line stuff.
There's a ton of reasons that are very very sensible to be very pissed off about AI. People very rarely benefit from human support staff being turned into chatbot AI. The unyielding limited mechanized systems we are forced to interact with suck.
And the billionaires themselves are so loudly advertising that the purpose of these things is to replace people. To obsolete so much labor. When it's so clear they suck, in so many ways. That people are upset is not just "the party line".
I've gone on about AI (even though I personally love it & think it's a miracle, having help in life I never thought I would get). But I don't want to fixate on AI. The sociological effect here, of how people constantly hunt for basically parallel constructions for what they want to believe: that's such a major factor in understanding people and societal forces. I think you're really smashing your ability to think to bits to chalk this up as "the party line". It's such a deeper more interesting force than than.
It again just goes to ai, which I think is not by point, but man what a moment of people seeking their existing biases: the "AI generated Monet" painting that actually is in fact a real Monet (Water Lilies). People went to town! https://bsky.app/profile/segyges.bsky.social/post/3mlsgc53ry...
This effect deeply deeply deeply predates the internet. I think perhaps yes there is a little more biting sarcasm & anger & negative energies that have infected us, by being exposed to so much attention-driven systems where negative energies rule! But this chasing your existing beliefs has been around probably since before the dawn of man. I'd put money in it.
https://andymasley.com/writing/
was 50$ bounty, now 300$ - that's an unusually brazen stance.
clicking a link to in the Contents section opens a new page instead of jumping to an anchor. wonder if he's getting paid by the click.
Maybe I missed it in the article, but it's less:
"AI uses more water than other things"
and more:
"AI's water usage is being approved without proper planning, because of the arguably fake sense of urgency around it."
Other industries that use significant water have significant regulations already . AI has been desperately trying to avoid ANY regulation (unless it forces folk to use AI.)
I don't personally think water usage is the biggest issue with how AI is being rolled out, but it's one that is easier to engage the public on then copyright, or societal context collapse. :)
I'm pretty convinced the only reason people care is because it's been memeified, and to be fair, the running gag of someone chugging a bottle of water before giving a lacklustre response to an easy question is pretty funny. But this sudden care for water consumption despite the public's general apathy towards it with regards to literally everything else (mining, livestock, textiles, energy generation, Coca Cola, etc), just seems manufactured.
Near 100% of the complaints I see online and in person about AI water usage are based on the idea that it’s consuming out of control amounts of water.
It’s frequently brought up as a reason not to use AI. The public perception right now is that doing anything with AI causes a lot of water to be “used” and that this is a very bad thing.
I can't imagine what possessed the author to write this much text on the topic. But yes, that was always my impression. My pet conspiracy theory - which is completely unsubstantiated - is that if you're pushing for data centers in rural communities, you actually want people to get fixated on water usage, because you can then talk to the county commissioners and shoot down these criticisms so easily.
Mega-scale AI data centers have other externalities. They're often touted as a way for rural counties to become the hubs of the digital era, but they don't employ many people, don't generate a whole lot of tax revenue, and basically just leverage cheap electricity at the expense of local residents. So it's a sham in that respect. You're not gonna have X, Google, Amazon, or Meta reinvigorate your community. You're just gonna have ratepayers subsidize some inference via higher electricity bills.
I have no doubt that someone will chime in saying with an "actually..." that electricity is fungible and therefore, it doesn't matter where the datacenter is built. If it were so, they wouldn't be getting built in places like Wyoming or eastern Washington, and electricity prices in these markets would be the same as they are in the SF Bay Area. In practice, though, there's plenty of factors that make the US electricity markets a lot more local.
The electricity rates in the SF Bay Area are astronomical primarily because PG&E is trying to pay off the fire damage and deaths they have been causing regularly since 2015.
In my opinion, the company is a colossal disaster. But more importantly, CPUC is complicit, bordering on incompetent [1]. Our regulatory commission is not in the business of protecting utility consumers.
Using these electricity rates as an example of where rates are going because datacenters are being built in those areas is wild! The rates have essentially nothing to do with datacenters.
[1]: https://www.turn.org/why-are-my-energy-bills-so-high
The part in the article that is actually of interest to me focuses on the tax revenue section. Can someone versed in this explain?
My initial reaction is b.s. Companies building these data centers, in many instances, get tax breaks to start building. On top if it they get different breaks on cost of electricity or materials. And on top of it, we all know that corporates pay less taxes than individuals already. And last but not least, data centers don't require a lot of staff, so there is no "trickle down".
Curious to understand this better.
> Importantly, I am not saying that it’s impossible for data centers to ever cause any problems with water. They require careful planning in the same ways other large industrial buildings do.
(Emphasis added)
I think that part of why AI data centers seem to have more stringent local objections even in areas that are generally pretty friendly to industrial development than do other industries that use a lot of water (or a lot of electricity) may be that there often isn't enough time to do that careful planning.
If you wanted to build say a new paper manufacturing plant there would be local objections because paper manufacturing a lot of both water and electricity. But nobody urgently needs a new paper manufacturing plant. There is plenty of time to address the objections and figure out something that will make most of the locals happy.
With AI data centers there is a (possibly bogus) need for a lot in a short timeframe, and so a lot of pressure on officials to approve them without the careful planning such a project should have.
The table of contents opens links in a new tab. If they didn't, they would require a full page reload, because they don't use fragments. This is seemingly how substack is designed.
Two things jumped out to me.
1. No accounting for other countries or externalities such as large corporation leveraging to get what it wants at the expense of community members.
2. > Put another way, almost all (80%) the reported water used by AI occurs during the generation of electricity
Well that sure seems like a problem to me cuz that’s water that now needs to be used for generation that wasn’t prior and it’s a significant amount. And that impacts my water and electricity prices now.
Edit for 3: > Consumptive use can harm total access to freshwater, but freshwater sources are also regularly being replenished.
Yeah I don’t think so when Mexico City is sinking because its aquifer was depleted.
All in all he’s got some interesting points but I think he’s hyper focused on numbers and ignoring broader things.
Plus semiconductor manufacturing et al which are also heavy users of water.
The entire global semiconductor industry, source of vast benefits, only uses about 50000 acre-feet of water per year, which is essentially nothing. As a point of comparison this is half a percent of what the paper industry consumes.
Where is the source for that? I am only familiar with one site, the STMicro one in the Alps, which already used 4000 acre-feet / 5 million m³ per year in 2023 [1], and it's been at least doubled since then. This is a huge chunk of the total water consumption of the region, and there are NIMBY demonstrations frequently because of that [2]. It's also surprisingly polluting. Whether you think the value is worth or not is a different story (I worked there, so guess).
[1] https://www.st.com/content/dam/aboutus/sustainability/report...
[2] https://stopmicro38.noblogs.org/post/2026/03/17/rando-pas-de...
50000 for the entire industry is bullshit, even if you limit it to the US.
The figures in that report are consistent with what I said. This is because the usage we're talking about is evaporation. This should not be confused with withdrawals. Semiconductor industry mainly withdraws water and then discharges it as effluent. Evaporation is a minor component of their withdrawals.
No, the report is already not counting whatever can be reused. It's the point of it.
What is the "AI water issue"
The blog post never defines it
Listen, I really like LLMs and diffusion models and machine learning and all this stuff, and I want to see it happen in a just and sustainable way. "AI" doesn't necessitate extreme waste. If anything, reasonable policy constraints would push "AI" to be even better.
I feel like the way many companies implement AI right now is very very wasteful. For context I'm looking into adding some AI elements to my SaaS app and I'm looking at running on-device TinyBERT intent classifiers then have my API take it from there (still experimenting with this).
I feel like this is a pretty sustainable way to implement AI in an application, meanwhile I see most companies just implement with OpenAI API + some custom prompts on top.
Granted I've had to do this for some of my clients and it's a pretty easy way to implement AI, though I always have the sinking feeling that we could achieve the same thing in a way more efficent manner and a bit more effort.
> If anything, reasonable policy constraints would push "AI" to be even better.
Like what, though? I'm not opposed to AI regulation at all, but the very last thing I expect it to fix is the resource constraints around GPGPU compute.
there fucking isn't "extreme waste". that's the point.
You know this is from 2025 because of the large number of emdashes in the text and in the comment liked by the "author".
[flagged]
[dead]
[dead]
[flagged]
Can you engage on the content? I hate AI and would love to have this as a numerate way to complain about it, but my current perception is that the water problem is nowhere near as large as the concern.
I don't engage with the "firehose of falsehood" types of articles as it's a losing battle. This guy's is a high school teacher whose side-gig is churning out pro-AI content for validation. I simply disregard the entire article since I know the author's MO, and I inform other HN members of it.
You get $300/falsehood[0]. Can you engage with one drop of the firehose, perhaps?
edit to add: Wait, is he actually a self-proclaimed libertarian? Is there a source for that?
[0]https://andymasley.com/writing/
To quote your original comment before the edit:
> "Andy Masley is nothing but a libertarian AI shill writing on substack and his thoughts should not be taken into consideration on this topic."
Andy Masley is nothing but a <person I disagree with> and his thoughts should not be taken into consideration on this topic.
You do you buddy
I despise armchair libertarians and Substack "intellectualism" as much as the next guy, but your comment doesn't offer any pushback to the claim being made.
I don't engage with the "firehose of falsehood" types of articles as it's a losing battle. This guy's is a high school teacher whose side-gig is churning out pro-AI content for validation.
I simply disregard the entire article since I know the author's MO, and I inform other HN members of it.
So basically you're asking everyone to just trust you, a random commentator that provides no evidence or sourcing, over an article written with extensive sourcing, details, and explanations?
You might be the correct person here, but you're not going to convince anyone like this.
I don't see the point of measuring how much water we use if we're not poluting it. It's water. It will evaporate and rain down again. It's renewable resource.
By that logic water can never be wasted, it all stays on Earth and eventually comes down somewhere.
Of course water use above replenishment rates is bad, it doesn't magically rain down in the same spot and all the underground water tables get full again. They deplete, meaning existing consumers have to dig deeper or just go without water. Even ancient peoples knew that if you take too much water from a well, it will dry out.
I imagine you would see the point in measuring how much water data centres use when one opens near you, and you can't flush your toilet any more.
You can drain aquifers faster than they get re-filled. Fresh water is renewable, but it renews relatively slowly in many places.
You can drain aquifers faster than they get re-filled. Fresh water is renewable, but it renews relatively slowly in many places.
Months in some places. But in the arid locations, it can take thousands of years for water to go from surface to aquifer.
A lot of these data centers and chip fabs are built in arid places because labor is cheap, taxes are low, and land is cheap. The reason for those three things is that there's not enough freaking water in the first place.
Slurping it up to run digital addiction mills and predatory advertising falls somewhere on a spectrum ranging from just plain stupid to abhorrently immortal.
I agree but then using a volume of water as a measure without provided context is just fearmongering.
Draining oasis in a desert might have much higher impact than one of the thousands of lakes in canada but still, it's a renewable resource. Most of the places suitable for datacenters have plenty of it anyway as datacenters are more suited for colder climates which usually have plenty of water.
Because water access is constrained geographically. You have to pipe it in if you use it all up. The US is seeing the worst spring drought on record: https://time.com/article/2026/05/10/drought-US-farmers-crops...
That's not the fault of datacenters though. Not even a local heavy industry.
My interpretation of their point was that more demand for water with a fixed supply of water will only make our water problems worse. Not that data centers are causing this problem on their own.
At a global scale yes, on the local scale perhaps not (or not on an acceptable timeframe)
Water resources are isolated, so moving purified water from one basin to another via evaporation does matter. But the OP is right that the specific cases of data centers usually don't matter.
It will evaporate and rain down again somewhere else. It's lost locally. In a place with a limited local water supply, that still will cause plenty of local pain.
Yeah but building a datacenter in colder wetter climate will hopefully bring the water somewhere else, usually to somewhere dryer.
At least that's my logic
This assumes that they do actually build in colder wetter climates. Most of the time they dont.