Not covered in the article are two critical reasons:
1. They're breaking environmental laws in order to meet power demands. xAI has already been busted on this [1], but they keep finding willing accomplices in rural parts of the country to bypass public opposition or speedrun through regulatory exceptions [2].
2. Companies seem to be fudging their numbers when it comes to GPU capacity & current workloads [3], likely to inflate their IPO valuations. I know Ed Zitron is a divisive figure but I've not seen any journalist on the other side of the argument provide the volume of data that he has.
What is the advantage to not keeping them a secret? The populist movement against AI is growing rapidly, and is supported by bunk science which affirms people's pre-existing biases (like the idea that data centers suck up all the water in a community, or raise the ambient temperature by a single-double digit number of degrees F).
AI is a genuine source of economic growth. I can understand wanting to curtail it, but in return you are getting fewer jobs, less economic growth, more money to other countries who don't allow protesting or even complaining about data centers, etc.
The US has deindustrialized and since the rise of NIMBY politics hasn't built new industry in any scale at all. The first freeway revolts happened in 1959. In the meantime, basically everyone has agreed that industrial growth and manufacturing as an abstract concept is Good (TM) but definitely not in my backyard. For a long time these kinds of things were just being built in neighborhoods where the residents were too poor/politically disconnected to organize, but eventually that too became difficult.
Since automation became big in industrial processes, most industrial development has been labor-poor (few workers) but continues to be land intensive. That means while industry might generate tax revenue, it doesn't have a coalition of labor advocates willing to champion the capacity because not many jobs are created. So we find ourselves in 2026 with an inability to actually build new industrial capacity in any form. The anti-tech crowd is angry at the data centers, but the same exact thing is happening when it comes to permitting new power generation and transmission lines. In fact many of the concerns related to data center power usage could be allayed if we had more power generation but nobody wants power generation in their backyard.
For decades now the US has been dancing around the idea that there is no by-right way to build anything anymore, so building any new large structure becomes a collective action problem that just ends up failing. Even when things get built, costs are massive. This has even affected things that the US is ostensibly really good at making such as highways, as evident in the recent Texas highway expansions ongoing.
I think it's good to remind people that the industrial revolution was very close to never happening.
This is from Dud Dudley writing in 1665, whose own ventures to manufacture steel en masse before Abraham Darby succeeded.
> "I have been opposed by many adversaries, as by wood colliers, mine owners, and others who, being poor men, did, by misguided advice, throw down and destroy two of my furnaces and my works, and caused much of my pigs and bar iron to be carried away."
There were plenty of examples through history of "near-misses" where establishment land/wealth holders suppressed nascent steel industries. It was almost an accidental series of coincidences that the industrial revolution happened - the Glorious Revolution in England and Abraham Darby's secret financing network.
This is an excellent analysis. The NIMBYism is pretty self-evident I think but I never thought about the disconnect automation may have caused between industry and advocates for industry.
You're conflating AI and AI Data centres. What kind of "jobs" does AI data centre creates? More jobs for factory workers who assemble video cards to be shipped and installed in a rack?
How many technicians? And from where? Its unclear to me how often this stuff actually needs to be done. And how resilient the overall system is to failure. Is it imperative to swap out one point of failure immediately or can you let them batch up and send Joe from California out qaurterly?
Technicians are constantly behind in work. Many data centers are pulling 24/7/365 shifts to keep up with demand. Larger data centers have hundreds of full time employees.
Yes many servers are left in fail over states for long periods of time, but that can only be done because new capacity is actively being deployed to make up for that fail over. Modern data centers are far too big for a single person to be repairing things every once in a while. Stuff is breaking every hour of every day
Thanks for the info. So the technicians are there on site daily? dozens? hundreds?
Im not too surprised there is quite a bit to replace. Sure the failure rate should be low but the scale is massive. Its interesting to get an order of magnitude though. Every hour is more frequent than I thought.
I worked for a company that ran a pretty big, high-security data center--75,000 square feet serving thousands of customers. There was a team of 6-8 technicians and a couple dozen security guards. None of them were particularly great or high-paying jobs.
Regarding water usage, in general data centers do not use more water than other types of heavy manufacturing
> The Georgia data center is only using ~2% of the county’s water. For comparison, a pharmaceutical manufacturing plant is using ~4% of the county’s water. A construction plant for Rivian cars is using about the same amount of water as Meta’s data center. The data center is functioning like any other normal industry in the county.
Regarding data centers increasing ambient temp, the paper is simply measuring the surface temperature of the buildings, going against the claim that a data center, merely by its presence in a community, raises the ambient temperature by a few degrees or more
>Regarding water usage, in general data centers do not use more water than other types of heavy manufacturing
Are the claims really that "Data centers use more water than other types of heavy manufacturing"? I dont think so.
Even if thats true, that doesn't mean they cant have a disastrous effect on the local water supply. This isnt a good rebuttal.
Frankly I tend to think the anti-datacenter crowd is overreacting. But I don't think you've addressed the real criticisms being levied.
In some passing research I saw the datacenters do continuously consume water (its not a one time cost like some claims I've read). And smaller size ones may use water equivalent to around 1000 households, and larger ones may consume closer to the equivalent of around 20,000 households. Evidently the massive one in Utah will at least double the state's entire consumption of water.
Can all of these places handle it?
I dont know. But that's the question, not if other types of heavy manufacturing have higher demands. And frankly it's inevitable that at least some locations cannot handle it. Which doens't mean you should be anti-datacenter in general. It means you can't just blanket dismiss the water concern for all locations.
Data centers do use water, and a lot of it, but the claims being made are hyperbolic and not squared with reality.
One criticism I often see is that data centers somehow pollute the local water supply. Data centers use water in a closed loop, their impact on local water quality is negligible. Industrial manufacturing and even agriculture have a far greater deleterious effect.
> The EPA’s national assessments repeatedly identify agriculture as the leading source of impairment for rivers and streams due to nutrient and sediment runoff, with continued nitrogen and phosphorus problems that affect drinking water and coastal ecosystems.
The thing is, AI data centers bring in far more tax revenue than other water-guzzling domains (like golf courses), but use less water overall. Extreme panic over specifically their water use is disproportionate
> How much of this will be AI? Almost all this growth will be driven by AI, but because AI is only 20% of data center power use, its growth will have to be huge to triple total power usage. One forecast says AI energy use in America will be multiplied by 10 by 2030. Because water use is proportionate to energy use, we can multiply AI’s water use by 10 as well.
> So in 2030, AI in data centers specifically will be using 0.08% of America’s freshwater. This means it will rise to the level of 5% of America’s current water used on golf courses, or 5% of U.S. steel production, or be about 173 square miles of irrigated corn farms.
> The average American’s consumptive lifestyle freshwater footprint is 422 gallons per day. This means that in 2023, AI data centers used as much water as the lifestyles of 25,000 Americans, 0.007% of the population. By 2030, they might use as much as the lifestyles of 250,000 Americans, 0.07% of the population. Not nothing, but 250,000 people over 5 years is just 4% of America’s current rate of population growth. If you found out that immigration plus new births in America would increase by 4% of its current rate, would you first thought be “We can’t afford that, it’s way too much water”?
>One criticism I often see is that data centers somehow pollute the local water supply.
Yes, I think this one is completely misinformed. For example AOC held up some dirty water from a local resident's tap. Fine, that's bad. But it was a result of digging during the construction process and the fact that it was a data center was irrelevant. And the implication that it permanently ruined local's water supply was just wrong.
>The thing is, AI data centers bring in far more tax revenue than other water-guzzling domains (like golf courses), but use less water overall. Extreme panic over specifically their water use is disproportionate
I guess that's not surprising to hear. A lot of people against data centers are probably also against golf courses though. I think AI is valuable but a lot of opponents see it as a net negative. Not saying they are right - this is definitely a point against the anti datacenter crowd. But it is consistent from their perspective so I dont think this point will persuade them. Would need to attack the claim that AI is a net negative.
Modern data centers are effectively mines, but without even the upside of supporting a local economy.
They extract local resources (land, power, grid capacity, water, etc) and sell that as compute. As a rule, the mine operators are national or multi-national firms that have no presence with which to invest the extraction profits back into the community. The local resources are harvested, processed, and sold, and then the proceeds disperse into the books of these gargantuan, far-off operators.
The only way to recover some of the profits on those local resources where they're being harvested would be with taxes or similar ongoing development obligations, but the firms specifically predate on communities too politically weak to levy those against them.
If AI isn't just an speculative bubble with gross overinvestment, they may have important value as a national economic or security interest, but they're pretty terrible and lop-sided deal for existing communities. That's why they're kept quiet.
People's bewilderment about this stuff speaks more to how removed they are from their local civics.
Nearly everything in your town is built like this. The amount of people who come out of the woodwork to oppose coffee shops, housing development, new hospitals, bus stops, etc would astound you. Try attending a local city council meeting. Part of the reason civic infrastructure takes so long and costs so much is because of the enormous burdens of transparency.
Or the sheer number of things that can go wrong during zoning, development, etc. The best time to announce a new business is when construction is nearly done. And the cities themselves want the development to be secret because they don't want to be underbid by the town next door (did anybody actually like the transparency of Amazon's HQ2 process?)
The less NIMBYs are involved or aware of any infrastructure project, the better. Is that even a question? They ruined enough opportunities and wealth in the last decades.
Yes, it is a question. Obviously a balance must be struck but where the line is drawn will depend. The point about "ruining wealth opportunities" will likely rub a lot of people the wrong way even on a place like YCombinator forum
And what happened to Climate Change? Seems that they don't care about it anymore, they rather have their data centers which give them more control over society
When I’ve visited the big data centers, they had gates and security relatively far out from the buildings. How would you get close enough? It’s not like the World Trade Center in 1993 where you can just drive up next to the building. (It’s also not like the World Trade Center in 2026 where there is an anti-terrorism perimeter of steel-reinforced concrete bollards, but somewhere in the middle.)
Of course there are also data centers in the middle of cities but these are smaller.
Rather than going into detail, getting myself on yet another watchlist, and possibly inspiring someone to do something both criminal and counter-productive, I'll just quote Patton:
Fixed fortifications are a monument to the stupidity of man.
They’re not built in secret, it’s just no one cared until now. How come no one cared when AWS and the other big cloud providers were building out all that infrastructure? Some one should put together a map of all data centers (not just “ai” datacenters), current and planned, complete with power and water consumption. oh the humanity!
They're being built in as much secret as possible because everything is. Everything about land development is adversarial and the people who keep their mouths shut wind up taking fewer 4-7 figure screwings along the way than the honest people.
It sucks that this how it is but even just the most casual familiarity with all the rules and processes makes it obvious why this is the way it is.
You could find more than enough data points to write this same story about grocery stores or anything else.
We, as a society, set up rules to allow the people to have a voice in decisions via various means including public meetings and various review procedures. There is a cost to these rules, but the alternative you're describing is a system where rules and laws don't matter and anyone with the ear of a few influential people can do whatever they want.
Just as an example, there's a data center in the early stages about 10 miles from my house. The land developer spent a year shopping for a city council that would skip the hearings required by state law and finally found a small exurb willing to break the law. Now the developer is racing to break ground before the lawsuits and restraining orders hit. This isn't the way that a logical society should work.
But why shouldn't the possible neighbors of some unpleasant installation protest as much as allowed, and file lawsuits? Neighbors interests have value too, plus we in the USA have a constitutional right to free speech and petition for redress of grievances. The US also has an adversarial legal system. It pays off to file lawsuits if you can afford it.
My conclusion is that "NIMBY" isn't the problem, it's economic incentives and adversarial systems.
They aren't, and yet somehow you still find various groups who will fight them.
Seattle is currently dealing with this for a new WinCo -- which is low-priced and employee-owned, making it particularly unobjectionable -- on a site that used to be a Sam's Club, so it's not even really a new development: https://www.seattletimes.com/business/real-estate/winco-plan...
> North Seattle shoppers may need to wait even longer for a grocery store to fill a former Sam’s Club location left vacant since 2018
> Two years ago, discount grocery chain WinCo filed plans to remodel the building and reconfigure the parking lot on Aurora Avenue North. But the plans encountered opposition from a neighborhood group for their possible environmental impact.
> Last week, a hearing examiner overturned the city’s determination that the project would have no significant environmental impacts, casting doubts on the future of the project.
> ...
> For a while, everything seemed on track. The city conducted a State Environmental Policy Act process and found WinCo’s plans would have no significant environmental impact.
> Then, last fall, a coalition called Lake Washington Working Families appealed the decision. The group, which tried and failed to disrupt plans for a WinCo coming to Renton last spring, has no website and is not registered with the state — leading to online speculation about who exactly is behind the group. But Karl Anuta, a Portland-based lawyer representing the coalition, said it’s made up of King County residents.
> The coalition claimed the city’s environmental analysis of WinCo’s plans for the North Seattle site was inadequate and required further review. WinCo would have major traffic impacts, the appeal said, releasing pollutants into local bodies of water.
> In an interview Monday, Anuta, who primarily handles cases involving environmental law, said the group is not against having a WinCo store at the location but wants the city to seriously analyze the environmental impacts of such a large business.
> “The real issue for the Lake Washington Working Families was you’re going from eight years of nothing there to a larger facility with many impacts,” he said.
> “You can’t just permit stuff and expect the neighbors to deal with the consequences.”
This is why environmental regulations and processes are getting pushback -- not because people hate the environment, but because NIMBYs learned to weaponize these rules against almost any kind of development, even the kind of thing that the overwhelming majority of people in an area support.
> The group, which tried and failed to disrupt plans for a WinCo coming to Renton last spring, has no website and is not registered with the state — leading to online speculation about who exactly is behind the group. But Karl Anuta, a Portland-based lawyer representing the coalition, said it’s made up of King County residents.
I'm reading this as "WinCo competitors who live in King County" instead of NIMBYs. It seems real shady and of course they'd want to make it seem like it was just a group of good ol' regular folks.
But it’s not just ideological. The facts are, even with great strides in efficiency, datacenters use a TON of water and electricity. And without fail that results in increased prices for the local communities who see almost no benefit from these datacenters.
Ironically the very same people profiting most off of them are the ones saying they’re going to leave the country if they’re forced to pay anything resembling a fair tax rate. They’re always all about socializing costs and privatizing profits and the common folks are finally waking up to it.
This is false. Evaporative cooling systems consume significant amounts of water and do not reuse it. They can't anyway, recondensing the water would release all the heat they removed by evaporating it.
> Data centers use very little water, right down to none if they want. And state-of-the-art hyperscale data centers really are operated by AIs.
There are many, many reasons to oppose datacenters. Not the least is they're there to drive inequality to ever-greater heights and they're 21st century version of the toxic waste dump (put 'em where people are weak and marginalized).
But water use is a very simple argument, and sometimes you have to pound on those to get through to the general public that's not immediately affected.
Most people who believe that data centers use a lot of water aren't mislead about how much water data centers use, but they are unaware of how much water everything else uses. In the great scheme of things data centers simply do not evaporate that much water. They are too small and too few. As a comparative example, all the data centers in Arizona combined use less than 1/50th of the water evaporated by Arizona's own thermal power stations. Or, to choose another benchmark, the evapotranspiration of the rice crop in California is more than 250x the water used by data centers in California.
I think also to a degree, even if someone knows those other water use figures, it's easy to see an intrinsic value to the community from those sources. Many people do not see much value in pushing AI to such a degree where all this new compute is required, and others see a negative impact from this activity. It's much easier to argue against something you feel is wrong or bad than something that is arguably crucial for day-to-day life like electricity and staple crops.
You believe lies, 99% of data shows that data centers do not use appreciable amounts of water compared to almond farms, or gold courses, or bog standard lawns.
That source says data centers use a lot of water. Less than all almond farms combined, sure, but it doesn't support the parent argument. It's also 3 years out of date, and not relevant to protests against all the data centers that have not yet been completed.
You say that like environmentalists support those things. We don’t. We regularly criticize golf courses as a waste of water and land. We regularly call out the water waste and ecosystem impact of manicured, uniform lawns.
Who is "we"?? Some places do have water shortages. And regardless of whether the power goes out or not, more power consumption = higher prices + more pollution.
Much of the US is in a drought. While we can generate plenty of electricity, doing so with outdated fossils fuels and ‘natural gases’ introduces a lot of pollution that harms people and the environment. We need to be careful about making these decisions.
But that’s really the problem - “we” don’t get any say in these decisions. A bunch of corrupt politicians and rich oligarchs make these decisions that screw over the rest of us.
And yes, for the record, I’m not uniquely against “AI” data centers. I’m opposed to a lot of other environmentally harmful and wasteful developments. They don’t get hyped up like “AI” does, though.
Things would go over better with communities if they had assurances with regard to local impact and if those were violated they’d have recourse. Also would help if they provided some token compensation (like the Alaska fund). It wouldn’t be much but it’d be something and most people would probably take that.
This is give an inch take a mile type thinking. The arguments against data centers are based in emotion not reason, so it’s effectively a fools errand to try and placate them.
I'm as big of a skeptic on these AI companies and question deeply the long term value of data center buildout as a land use policy in the communities where they're being built. But:
> So let me ask the question directly: if AI data centers are such a tremendous benefit to communities, why are so many of them being built without meaningful community input?
Because of this (emphasis mine):
> They’re watching their utility bills climb, finding sick animals they can’t explain, and worrying about the long-term impacts on their health and property values
How are you supposed to have reasonable discussion about land use, economic impacts, zoning, etc when you're getting flooded with input from crockpots?
Can you say with absolute certainty that sick animals and long term impacts on their health are not caused by data centers? Certainly we have long list of examples where industrial activity contaminated areas with little oversight and it did kill animals and cause long-term impacts on health. I think you're saying that data centers do not pollute in a way that would cause sick animals or long-term impacts on health, but I don't think that can be stated confidently.
One reason is that politicians have vilified them for their own political means. Creating false narratives like they use huge amounts of water, when in fact the cooling is a closed loop system and use less water than a single busy restaurant. They are being used as yet another pawn piece to push inequity and climate change agendas. It’s been quite a successful strategy.
> Creating false narratives like they use huge amounts of water, when in fact the cooling is a closed loop system and use less water than a single busy restaurant
My understanding was that often multiple cooling methods are employed and the ratio of use shifts seasonally; however, that evaporative cooling was still primary method, especially in hotter climates like CA and AZ.
Can you help me understand what type of data center (size, location, etc) uses less water than a busy restaurant?
Microsoft CEO Nadella just recently said their Fairwater (315-acre facility in Wisconsin) only uses around the same amount of water as a single restaurant over the course of an entire year.
25 data center projects were canceled in 2025, up sharply from only 6 in 2024. 2026 numbers gonna be dramatically higher. 69 jurisdictions have enacted some form of moratorium, ban, or restriction on new data center construction. The friendly states such as Texas are just going to have all the data centers and thus reap all the financial and job benefits.
I think you're being single minded about the reason people don't want them. Not everyone is hungry for giant construction projects regardless of the "jobs gain"
Not covered in the article are two critical reasons:
1. They're breaking environmental laws in order to meet power demands. xAI has already been busted on this [1], but they keep finding willing accomplices in rural parts of the country to bypass public opposition or speedrun through regulatory exceptions [2].
2. Companies seem to be fudging their numbers when it comes to GPU capacity & current workloads [3], likely to inflate their IPO valuations. I know Ed Zitron is a divisive figure but I've not seen any journalist on the other side of the argument provide the volume of data that he has.
[1] https://www.selc.org/news/xai-built-an-illegal-power-plant-t...
[2] https://www.cnbc.com/2026/04/10/musks-xai-draws-more-opposit...
[3] https://www.wheresyoured.at/where-are-all-the-data-centers/
> I know Ed Zitron is a divisive figure
The main counterarguments to his claims are "but Sam/Dario/Satya/Jensen said X" and that we should treat them as gospel.
What is the advantage to not keeping them a secret? The populist movement against AI is growing rapidly, and is supported by bunk science which affirms people's pre-existing biases (like the idea that data centers suck up all the water in a community, or raise the ambient temperature by a single-double digit number of degrees F).
AI is a genuine source of economic growth. I can understand wanting to curtail it, but in return you are getting fewer jobs, less economic growth, more money to other countries who don't allow protesting or even complaining about data centers, etc.
The US has deindustrialized and since the rise of NIMBY politics hasn't built new industry in any scale at all. The first freeway revolts happened in 1959. In the meantime, basically everyone has agreed that industrial growth and manufacturing as an abstract concept is Good (TM) but definitely not in my backyard. For a long time these kinds of things were just being built in neighborhoods where the residents were too poor/politically disconnected to organize, but eventually that too became difficult.
Since automation became big in industrial processes, most industrial development has been labor-poor (few workers) but continues to be land intensive. That means while industry might generate tax revenue, it doesn't have a coalition of labor advocates willing to champion the capacity because not many jobs are created. So we find ourselves in 2026 with an inability to actually build new industrial capacity in any form. The anti-tech crowd is angry at the data centers, but the same exact thing is happening when it comes to permitting new power generation and transmission lines. In fact many of the concerns related to data center power usage could be allayed if we had more power generation but nobody wants power generation in their backyard.
For decades now the US has been dancing around the idea that there is no by-right way to build anything anymore, so building any new large structure becomes a collective action problem that just ends up failing. Even when things get built, costs are massive. This has even affected things that the US is ostensibly really good at making such as highways, as evident in the recent Texas highway expansions ongoing.
I think it's good to remind people that the industrial revolution was very close to never happening.
This is from Dud Dudley writing in 1665, whose own ventures to manufacture steel en masse before Abraham Darby succeeded.
> "I have been opposed by many adversaries, as by wood colliers, mine owners, and others who, being poor men, did, by misguided advice, throw down and destroy two of my furnaces and my works, and caused much of my pigs and bar iron to be carried away."
There were plenty of examples through history of "near-misses" where establishment land/wealth holders suppressed nascent steel industries. It was almost an accidental series of coincidences that the industrial revolution happened - the Glorious Revolution in England and Abraham Darby's secret financing network.
This is an excellent analysis. The NIMBYism is pretty self-evident I think but I never thought about the disconnect automation may have caused between industry and advocates for industry.
You're conflating AI and AI Data centres. What kind of "jobs" does AI data centre creates? More jobs for factory workers who assemble video cards to be shipped and installed in a rack?
I think OP is talking about the downstream implications of productivity growth. Which is still yet hypothetical.
Data centers require lots of technicians to install, maintain, and decommission servers. Who do you think installs those video cards?
How many technicians? And from where? Its unclear to me how often this stuff actually needs to be done. And how resilient the overall system is to failure. Is it imperative to swap out one point of failure immediately or can you let them batch up and send Joe from California out qaurterly?
Technicians are constantly behind in work. Many data centers are pulling 24/7/365 shifts to keep up with demand. Larger data centers have hundreds of full time employees.
Yes many servers are left in fail over states for long periods of time, but that can only be done because new capacity is actively being deployed to make up for that fail over. Modern data centers are far too big for a single person to be repairing things every once in a while. Stuff is breaking every hour of every day
Thanks for the info. So the technicians are there on site daily? dozens? hundreds?
Im not too surprised there is quite a bit to replace. Sure the failure rate should be low but the scale is massive. Its interesting to get an order of magnitude though. Every hour is more frequent than I thought.
I worked for a company that ran a pretty big, high-security data center--75,000 square feet serving thousands of customers. There was a team of 6-8 technicians and a couple dozen security guards. None of them were particularly great or high-paying jobs.
Data centers don't grow on trees.
Why are they bunk science? I’m not an environmental expert, but the research papers and policies I’ve read don’t seem to be egregiously wrong.
I’m asking genuinely, I’m open to changing my mind here.
Regarding water usage, in general data centers do not use more water than other types of heavy manufacturing
> The Georgia data center is only using ~2% of the county’s water. For comparison, a pharmaceutical manufacturing plant is using ~4% of the county’s water. A construction plant for Rivian cars is using about the same amount of water as Meta’s data center. The data center is functioning like any other normal industry in the county.
https://blog.andymasley.com/p/the-ai-water-issue-is-fake
Regarding data centers increasing ambient temp, the paper is simply measuring the surface temperature of the buildings, going against the claim that a data center, merely by its presence in a community, raises the ambient temperature by a few degrees or more
https://andymasley.com/writing/data-centers-heat-exhaust-is-...
I know both sources are from the same guy, but he cites many primary sources in his articles
>Regarding water usage, in general data centers do not use more water than other types of heavy manufacturing
Are the claims really that "Data centers use more water than other types of heavy manufacturing"? I dont think so.
Even if thats true, that doesn't mean they cant have a disastrous effect on the local water supply. This isnt a good rebuttal.
Frankly I tend to think the anti-datacenter crowd is overreacting. But I don't think you've addressed the real criticisms being levied.
In some passing research I saw the datacenters do continuously consume water (its not a one time cost like some claims I've read). And smaller size ones may use water equivalent to around 1000 households, and larger ones may consume closer to the equivalent of around 20,000 households. Evidently the massive one in Utah will at least double the state's entire consumption of water.
Can all of these places handle it?
I dont know. But that's the question, not if other types of heavy manufacturing have higher demands. And frankly it's inevitable that at least some locations cannot handle it. Which doens't mean you should be anti-datacenter in general. It means you can't just blanket dismiss the water concern for all locations.
Data centers do use water, and a lot of it, but the claims being made are hyperbolic and not squared with reality.
One criticism I often see is that data centers somehow pollute the local water supply. Data centers use water in a closed loop, their impact on local water quality is negligible. Industrial manufacturing and even agriculture have a far greater deleterious effect.
> The EPA’s national assessments repeatedly identify agriculture as the leading source of impairment for rivers and streams due to nutrient and sediment runoff, with continued nitrogen and phosphorus problems that affect drinking water and coastal ecosystems.
The thing is, AI data centers bring in far more tax revenue than other water-guzzling domains (like golf courses), but use less water overall. Extreme panic over specifically their water use is disproportionate
> How much of this will be AI? Almost all this growth will be driven by AI, but because AI is only 20% of data center power use, its growth will have to be huge to triple total power usage. One forecast says AI energy use in America will be multiplied by 10 by 2030. Because water use is proportionate to energy use, we can multiply AI’s water use by 10 as well.
> So in 2030, AI in data centers specifically will be using 0.08% of America’s freshwater. This means it will rise to the level of 5% of America’s current water used on golf courses, or 5% of U.S. steel production, or be about 173 square miles of irrigated corn farms.
> The average American’s consumptive lifestyle freshwater footprint is 422 gallons per day. This means that in 2023, AI data centers used as much water as the lifestyles of 25,000 Americans, 0.007% of the population. By 2030, they might use as much as the lifestyles of 250,000 Americans, 0.07% of the population. Not nothing, but 250,000 people over 5 years is just 4% of America’s current rate of population growth. If you found out that immigration plus new births in America would increase by 4% of its current rate, would you first thought be “We can’t afford that, it’s way too much water”?
>One criticism I often see is that data centers somehow pollute the local water supply.
Yes, I think this one is completely misinformed. For example AOC held up some dirty water from a local resident's tap. Fine, that's bad. But it was a result of digging during the construction process and the fact that it was a data center was irrelevant. And the implication that it permanently ruined local's water supply was just wrong.
>The thing is, AI data centers bring in far more tax revenue than other water-guzzling domains (like golf courses), but use less water overall. Extreme panic over specifically their water use is disproportionate
I guess that's not surprising to hear. A lot of people against data centers are probably also against golf courses though. I think AI is valuable but a lot of opponents see it as a net negative. Not saying they are right - this is definitely a point against the anti datacenter crowd. But it is consistent from their perspective so I dont think this point will persuade them. Would need to attack the claim that AI is a net negative.
Thanks for the reply and links, I’ll give it a read today.
Andy Massey writes about this stuff and is generally heavily disliked by the anti-data-center folks
* https://blog.andymasley.com/p/data-center-land-use-issues-ar...
* https://blog.andymasley.com/p/i-might-have-found-the-specifi...
>but in return you are getting fewer jobs,
Certainly not in any meaningful way.
Since electricity demand has outpaced supply they are increasing prices: https://www.bloomberg.com/graphics/2025-ai-data-centers-elec...
[dead]
Modern data centers are effectively mines, but without even the upside of supporting a local economy.
They extract local resources (land, power, grid capacity, water, etc) and sell that as compute. As a rule, the mine operators are national or multi-national firms that have no presence with which to invest the extraction profits back into the community. The local resources are harvested, processed, and sold, and then the proceeds disperse into the books of these gargantuan, far-off operators.
The only way to recover some of the profits on those local resources where they're being harvested would be with taxes or similar ongoing development obligations, but the firms specifically predate on communities too politically weak to levy those against them.
If AI isn't just an speculative bubble with gross overinvestment, they may have important value as a national economic or security interest, but they're pretty terrible and lop-sided deal for existing communities. That's why they're kept quiet.
People's bewilderment about this stuff speaks more to how removed they are from their local civics.
Nearly everything in your town is built like this. The amount of people who come out of the woodwork to oppose coffee shops, housing development, new hospitals, bus stops, etc would astound you. Try attending a local city council meeting. Part of the reason civic infrastructure takes so long and costs so much is because of the enormous burdens of transparency.
Or the sheer number of things that can go wrong during zoning, development, etc. The best time to announce a new business is when construction is nearly done. And the cities themselves want the development to be secret because they don't want to be underbid by the town next door (did anybody actually like the transparency of Amazon's HQ2 process?)
When I grew up in Singapore, the locations of all data centers are secret by default to prevent terrorist attacks.
I mean the framing is so silly.
"If bank vaults are so great, why would you not advertise their locations". It's a mystery, is what it is.
The less NIMBYs are involved or aware of any infrastructure project, the better. Is that even a question? They ruined enough opportunities and wealth in the last decades.
I don't recall anyone in my town being overly angry when fiber internet dug up their front yards.
But the datacenter they couldn't stop has made national news multiple times.
Also, what wealth? "Please allow me to exploit your land for little benefit to you" isn't a great selling point.
Yes, it is a question. Obviously a balance must be struck but where the line is drawn will depend. The point about "ruining wealth opportunities" will likely rub a lot of people the wrong way even on a place like YCombinator forum
some of these projects impact surrounding homes and businesses. NIMBYs should definitely be aware and involved
And what happened to Climate Change? Seems that they don't care about it anymore, they rather have their data centers which give them more control over society
To stop assholes from bombing them
When I’ve visited the big data centers, they had gates and security relatively far out from the buildings. How would you get close enough? It’s not like the World Trade Center in 1993 where you can just drive up next to the building. (It’s also not like the World Trade Center in 2026 where there is an anti-terrorism perimeter of steel-reinforced concrete bollards, but somewhere in the middle.)
Of course there are also data centers in the middle of cities but these are smaller.
Rather than going into detail, getting myself on yet another watchlist, and possibly inspiring someone to do something both criminal and counter-productive, I'll just quote Patton:
Trebuchet
We didn't have known datacenters bombed till now
I mean yes, Amazon servers in the middle east aren't doing great, but I guess it's not what you're talking about
It was included in what I was talking about. Why would you publicise something that just paints a big target.
Give me example where datacenter is bombed in the US?
Wish I could save comments like I can favorite threads.
There's a favourite button for comments if you click the timestamp.
The secrets are kept prior to breaking ground. After that everyone knows what it is.
They’re not built in secret, it’s just no one cared until now. How come no one cared when AWS and the other big cloud providers were building out all that infrastructure? Some one should put together a map of all data centers (not just “ai” datacenters), current and planned, complete with power and water consumption. oh the humanity!
They're being built in as much secret as possible because everything is. Everything about land development is adversarial and the people who keep their mouths shut wind up taking fewer 4-7 figure screwings along the way than the honest people.
It sucks that this how it is but even just the most casual familiarity with all the rules and processes makes it obvious why this is the way it is.
You could find more than enough data points to write this same story about grocery stores or anything else.
I enjoy reminding people the land acquisition history behind building Disney World.
We, as a society, set up rules to allow the people to have a voice in decisions via various means including public meetings and various review procedures. There is a cost to these rules, but the alternative you're describing is a system where rules and laws don't matter and anyone with the ear of a few influential people can do whatever they want.
Just as an example, there's a data center in the early stages about 10 miles from my house. The land developer spent a year shopping for a city council that would skip the hearings required by state law and finally found a small exurb willing to break the law. Now the developer is racing to break ground before the lawsuits and restraining orders hit. This isn't the way that a logical society should work.
Is "keeping your mouth shut" dishonesty?
But why shouldn't the possible neighbors of some unpleasant installation protest as much as allowed, and file lawsuits? Neighbors interests have value too, plus we in the USA have a constitutional right to free speech and petition for redress of grievances. The US also has an adversarial legal system. It pays off to file lawsuits if you can afford it.
My conclusion is that "NIMBY" isn't the problem, it's economic incentives and adversarial systems.
grocery stores? They are not comparable to data centers.
They aren't, and yet somehow you still find various groups who will fight them.
Seattle is currently dealing with this for a new WinCo -- which is low-priced and employee-owned, making it particularly unobjectionable -- on a site that used to be a Sam's Club, so it's not even really a new development: https://www.seattletimes.com/business/real-estate/winco-plan...
> North Seattle shoppers may need to wait even longer for a grocery store to fill a former Sam’s Club location left vacant since 2018
> Two years ago, discount grocery chain WinCo filed plans to remodel the building and reconfigure the parking lot on Aurora Avenue North. But the plans encountered opposition from a neighborhood group for their possible environmental impact.
> Last week, a hearing examiner overturned the city’s determination that the project would have no significant environmental impacts, casting doubts on the future of the project.
> ...
> For a while, everything seemed on track. The city conducted a State Environmental Policy Act process and found WinCo’s plans would have no significant environmental impact.
> Then, last fall, a coalition called Lake Washington Working Families appealed the decision. The group, which tried and failed to disrupt plans for a WinCo coming to Renton last spring, has no website and is not registered with the state — leading to online speculation about who exactly is behind the group. But Karl Anuta, a Portland-based lawyer representing the coalition, said it’s made up of King County residents.
> The coalition claimed the city’s environmental analysis of WinCo’s plans for the North Seattle site was inadequate and required further review. WinCo would have major traffic impacts, the appeal said, releasing pollutants into local bodies of water.
> In an interview Monday, Anuta, who primarily handles cases involving environmental law, said the group is not against having a WinCo store at the location but wants the city to seriously analyze the environmental impacts of such a large business.
> “The real issue for the Lake Washington Working Families was you’re going from eight years of nothing there to a larger facility with many impacts,” he said.
> “You can’t just permit stuff and expect the neighbors to deal with the consequences.”
This is why environmental regulations and processes are getting pushback -- not because people hate the environment, but because NIMBYs learned to weaponize these rules against almost any kind of development, even the kind of thing that the overwhelming majority of people in an area support.
> The group, which tried and failed to disrupt plans for a WinCo coming to Renton last spring, has no website and is not registered with the state — leading to online speculation about who exactly is behind the group. But Karl Anuta, a Portland-based lawyer representing the coalition, said it’s made up of King County residents.
I'm reading this as "WinCo competitors who live in King County" instead of NIMBYs. It seems real shady and of course they'd want to make it seem like it was just a group of good ol' regular folks.
You are not being curious or good faith here.
Uh, no? It was not built in secret before the AI hype.
In fact as late as Jan 2026 Google was proudly presenting their new data centers in Bangkok: https://www.googlecloudpresscorner.com/2026-01-21-Google-Clo...
Disclaimer: Google Cloud employee
This is the top comment:
“ And ask if AI is so great why hasn’t it designed resource efficient data centers?
And ask what the data centers are actually doing. Bitcoin mining, anyone?”
Says all you need to know about the competency level of this position, which is ideological and not fact-based
But it’s not just ideological. The facts are, even with great strides in efficiency, datacenters use a TON of water and electricity. And without fail that results in increased prices for the local communities who see almost no benefit from these datacenters.
Ironically the very same people profiting most off of them are the ones saying they’re going to leave the country if they’re forced to pay anything resembling a fair tax rate. They’re always all about socializing costs and privatizing profits and the common folks are finally waking up to it.
Data centers use very little water, right down to none if they want. And state-of-the-art hyperscale data centers really are operated by AIs.
This is false. Evaporative cooling systems consume significant amounts of water and do not reuse it. They can't anyway, recondensing the water would release all the heat they removed by evaporating it.
You don't have to evaporatively cool a data center. You can make a direct trade between energy efficiency and water consumption.
You don’t have to but in hotter climates, especially those with higher energy costs (ca), it’s a lot cheaper to cool evaporatively.
> Data centers use very little water, right down to none if they want. And state-of-the-art hyperscale data centers really are operated by AIs.
There are many, many reasons to oppose datacenters. Not the least is they're there to drive inequality to ever-greater heights and they're 21st century version of the toxic waste dump (put 'em where people are weak and marginalized).
But water use is a very simple argument, and sometimes you have to pound on those to get through to the general public that's not immediately affected.
Sorry but no, you don't get to lie just because your arguments don't resonate with the public. That makes you a bad person.
Source on that water claim? Everything I've seen suggests the opposite.
Most people who believe that data centers use a lot of water aren't mislead about how much water data centers use, but they are unaware of how much water everything else uses. In the great scheme of things data centers simply do not evaporate that much water. They are too small and too few. As a comparative example, all the data centers in Arizona combined use less than 1/50th of the water evaporated by Arizona's own thermal power stations. Or, to choose another benchmark, the evapotranspiration of the rice crop in California is more than 250x the water used by data centers in California.
I think also to a degree, even if someone knows those other water use figures, it's easy to see an intrinsic value to the community from those sources. Many people do not see much value in pushing AI to such a degree where all this new compute is required, and others see a negative impact from this activity. It's much easier to argue against something you feel is wrong or bad than something that is arguably crucial for day-to-day life like electricity and staple crops.
You believe lies, 99% of data shows that data centers do not use appreciable amounts of water compared to almond farms, or gold courses, or bog standard lawns.
https://tech.yahoo.com/science/articles/data-centers-less-wa...
That source says data centers use a lot of water. Less than all almond farms combined, sure, but it doesn't support the parent argument. It's also 3 years out of date, and not relevant to protests against all the data centers that have not yet been completed.
You say that like environmentalists support those things. We don’t. We regularly criticize golf courses as a waste of water and land. We regularly call out the water waste and ecosystem impact of manicured, uniform lawns.
compared to almond farms? You phrase that as if almond farms use a reasonable amount of water.
> datacenters use a TON of water and electricity
We have a shortage of neither.
Who is "we"?? Some places do have water shortages. And regardless of whether the power goes out or not, more power consumption = higher prices + more pollution.
We the people.
In places where there are significant water shortages, water hungry data centers are not being built.
Incorrect. Seems like you haven't really investigated this topic at all.
https://www.newsweek.com/map-data-centers-built-drought-hit-...
Much of the US is in a drought. While we can generate plenty of electricity, doing so with outdated fossils fuels and ‘natural gases’ introduces a lot of pollution that harms people and the environment. We need to be careful about making these decisions.
But that’s really the problem - “we” don’t get any say in these decisions. A bunch of corrupt politicians and rich oligarchs make these decisions that screw over the rest of us.
And yes, for the record, I’m not uniquely against “AI” data centers. I’m opposed to a lot of other environmentally harmful and wasteful developments. They don’t get hyped up like “AI” does, though.
Things would go over better with communities if they had assurances with regard to local impact and if those were violated they’d have recourse. Also would help if they provided some token compensation (like the Alaska fund). It wouldn’t be much but it’d be something and most people would probably take that.
This is give an inch take a mile type thinking. The arguments against data centers are based in emotion not reason, so it’s effectively a fools errand to try and placate them.
Post is from last week, and is just an intro to the map project, discussed here at length:
[dupe] https://news.ycombinator.com/item?id=48287952
I'm as big of a skeptic on these AI companies and question deeply the long term value of data center buildout as a land use policy in the communities where they're being built. But:
> So let me ask the question directly: if AI data centers are such a tremendous benefit to communities, why are so many of them being built without meaningful community input?
Because of this (emphasis mine):
> They’re watching their utility bills climb, finding sick animals they can’t explain, and worrying about the long-term impacts on their health and property values
How are you supposed to have reasonable discussion about land use, economic impacts, zoning, etc when you're getting flooded with input from crockpots?
Can you say with absolute certainty that sick animals and long term impacts on their health are not caused by data centers? Certainly we have long list of examples where industrial activity contaminated areas with little oversight and it did kill animals and cause long-term impacts on health. I think you're saying that data centers do not pollute in a way that would cause sick animals or long-term impacts on health, but I don't think that can be stated confidently.
The conflation of concerns and synechdochal arguments are a wonder to behold.
So many people flailing around to justify their feelings rather than thinking rationally.
One reason is that politicians have vilified them for their own political means. Creating false narratives like they use huge amounts of water, when in fact the cooling is a closed loop system and use less water than a single busy restaurant. They are being used as yet another pawn piece to push inequity and climate change agendas. It’s been quite a successful strategy.
> Creating false narratives like they use huge amounts of water, when in fact the cooling is a closed loop system and use less water than a single busy restaurant
My understanding was that often multiple cooling methods are employed and the ratio of use shifts seasonally; however, that evaporative cooling was still primary method, especially in hotter climates like CA and AZ.
Can you help me understand what type of data center (size, location, etc) uses less water than a busy restaurant?
Microsoft CEO Nadella just recently said their Fairwater (315-acre facility in Wisconsin) only uses around the same amount of water as a single restaurant over the course of an entire year.
That is not a reliable source at all.
Data Center, right now, have more negative impact. So, lets not dismiss that.
https://www.theguardian.com/environment/2026/feb/13/elon-mus...
So successful that data centers are everywhere? I'm not following the logic here.
25 data center projects were canceled in 2025, up sharply from only 6 in 2024. 2026 numbers gonna be dramatically higher. 69 jurisdictions have enacted some form of moratorium, ban, or restriction on new data center construction. The friendly states such as Texas are just going to have all the data centers and thus reap all the financial and job benefits.
I think you're being single minded about the reason people don't want them. Not everyone is hungry for giant construction projects regardless of the "jobs gain"
[dead]
There's a ton of expensive hardware that thieves might want. Why wouldn't you keep that quiet.