AI threatens to raid the water reserves of Europe's driest regions
(politico.eu)52 points by molteanu 12 hours ago
52 points by molteanu 12 hours ago
TBH, if I had to choose, I'd rather have chocolate than "AI data centers".
These comparisons to existing uses of water besides cooling "AI data centers" ignore the fact that water use for "AI data centers" is additive. It is a _new_ use of water not seen before. As such, the comparisons to existing uses are unconvincing.
A bizarre "justification" such as "[existing use, e.g., chocolate] is permitted to deplete water supplies therefore [new use, e.g., AI data centers] should be permitted to deplete water supplies by similar amounts" does not scale. The supply of water is not unlimited. Not all new uses can be accomodated without affecting other uses.
This strange "reasoning" something like "so-and-so already does it" may appear to be based on some AI cheerleader notion of "fairness" but in reality it is based on a disregard for the limited nature of the resource, as if there is an unlimited supply. There is more money behind "AI" than chocolate. Truthfully, in the event of shortage, "fair" allocation where every use is accomodated is not realistic.
People want chocolate. There are hundreds of years of history to prove it.^1 But do people want "AI". It is being forced on them. If they accept it, if they essentially have no choice, then is that the same as "wanting".
1. The history of "AI" is absurdly short by comparison. It includes long periods of years where interest in "AI" was lost. Perhaps an AI cheerleader can dig up something about "Chocolate Winters".
> These comparisons to existing uses of water besides cooling "AI data centers" ignore the fact that water use for "AI data centers" is additive. It is a _new_ use of water not seen before. As such, the comparisons to existing uses are unconvincing.
No, I'm not ignoring that. I'm saying two things:
1) When put next to existing water usage, data center water usage stops looking as crazy large as is implied in TFA. All data centers for all compute purposes in the EU used as much water in 2024 as the creation of a single food product used in 2019. That's important perspective to have when we're talking about additive water usage, because it hints out what percentage increase we're talking about [0].
2) I don't accept the presumption that existing usage has preeminence now and forever by simple virtue of being older. We should be able to look at all water usage that currently stands and decide which usages are worth keeping and which are worth adding.
You're welcome to disagree that data centers (AI and otherwise, because non-AI workloads are included in the numbers) are worth more than chocolate, but it should be as part of a fair comparison between them on the merits, with no grandfathering chocolate and agriculture in just because the chocolate companies and farmers got there before anyone was paying attention to monitoring water usage.
[0] If you can find a percentage increase from data centers figure that would be even better, but I can't find absolute numbers for total water usage in the EU to construct such a figure.
"Amazon does not disclose its total water footprint."
I agree that the raw numbers without context can be misleading but couldn't it be argued that the water usage for chocolate is way too high and that we're vastly overproducing? Adding on more and more usage and saying it's only a percentage of what we already waste isn't a good long-term strategy.
Sure, but then it's silly to throw a fit about data centers which pretty clearly provide a greater benefit to Europe's interests than chocolate does. It's just another example of how the rhetoric gives the established water abusers a free pass while scrutinizing every cubic meter used for tech.
Be very careful about considering this a "grassroots campaign" against "evil foreign megacorps".
This is lobbying effort directed at policy makers and the public.
As others have pointed out: There is an very simple solution to "solve" water waste/allocation: Just put a price on it.
BUT one big interest group, namely agriculture/farmers, absolutely don't want that, because they historically could pretty much use water for free (and/or underpaid massively)-- any rational discussion about water use/price/allocation is undesirable to them, because it is likely to make the situation worse for them comparatively.
This is also why the whole discussion centers round emotional arguments against allowing industrial water use at all, instead of arguing that small/local farmers should get a better price on it.
Putting a price on it does not solve it because the entities that compete for the water with the data centers are out-matched financially. How is a city that has a budget that is constantly dealing with budget cuts supposed to pay more for water than a multi-billion dollar company? Taxing it does not solve it either, because there are so many incentives to writing loopholes into the tax code.
They don't even have to use water - there are alternatives. The solve is changing behavior of the leaders in this greedy industry.
> Putting a price on it does not solve it because the entities that compete for the water with the data centers are out-matched financially.
I don't think this argument works at all, because bigger datacenter operator does NOT mean "tolerates higher OPEX".
> How is a city [...] supposed to pay more for water than a multi-billion dollar company?
I do not understand this; the city would not compete with the datacenter operator for water-- the farmers would, and both of those would be paying the municipality for the water (ideally), not vice-versa. Residential users already pay much more for water (typically, compared to farming/industry) so any renegotiation is unlikely to affect them much.
Yep. And at the same time, it's the residents that are asked to conserve. Meanwhile, we let data centers off the hook when it comes to conservation?
When I drive around my town, I can't drive more than a few blocks without seeing a sign about Stage 3 water restrictions. For residents. Just doesn't seem fair.
Can someone help me understand how cooling data centers "uses" water? The water presumably isn't destroyed it's just put back hotter than before right? I don't imagine it comes out as steam like a nuclear plant but I suppose it could.
I asked a friend who works on datacenter this so it's secondhand information but - apparently it's cheaper to use mist cooling where you essentially spray cold water on hot parts and it evaporates off instead of the normal closed loop system that you'd see in a home water cooling for example. And while it's not destroyed, it means you need a constant supply of clean water, and the evaporated water doesn't necessarily go back to the body of water you got it from.
Kind of crazy we let them blow through our limited clean water supplies just because it's cheaper
In addition to the other concerns of evaporative cooling, there is one issue with a simple water cooled loop that just runs river water through a heat exchanger and then dumps it again. You can only raise a river's temperature so much before it becomes a problem for the ecosystem. This issue has in the past forced power plant shutdowns, so it's not that simple.
https://www.ans.org/news/article-6268/french-nuclear-plant-l...
> The water presumably isn't destroyed it's just put back hotter than before right?
I'd assume the water isn't put back into circulation after passing through a data center, so if anything it might be cooled and then reused in the same data center, best case scenario.
The results is the same, farmers who already have to fight to get enough water in Aragon now have to fight more and compete against companies like Amazon and other foreign investors for the water. Aragon isn't exactly wetlands, so hard to not feel the local government is making the wrong choice here.
As other have stated, evaporative cooling is most of the usage, AC dries the air and requires adding moisture to the air to prevent problems with static electricity, this drives a much smaller portion of the usage.
Note that evaporative cooling also produces significant amounts of brine, as it is similar to distillation, producing a brine that is difficult to treat and often not useful for other purposes.
With AC can't you just well re-introduce it? The cold side could be almost closed system as nothing demands you to remove water in such location. For occupied spaces lower humidity generally feels better so moisture is removed. But for data centre it could be re-introduced?
You still have to transfer the heat to something.
Note that it is quite common for the refrigerant or chilled water to be a closed loop, but still have an evaporative cooling tower etc.
People are trying to use geothermal etc...
Evaporative cooling efficiency in dry climates is hard to beat for capital, energy and space.
Rain can deposit water in seas or lands away from country's border. Anyhow its always efficient and easy to collect water from a running river on its way to sea.
Data centres can be designed to harvest the steam / evaporated water and reuse it but thats expensive too and they might not choose to do so.
Apparently, it does not have to be this way. According to https://youtu.be/GhIJs4zbH0o?t=895 Stargate is designed with a "closed loop system" that will be filled up just once.
In the video he paints the picture of asking for "a million gallons of water one time" VS asking for "a million gallons of water per hour", that cannot possible be a faithful comparison.
For these new planned data centers, realistically, what would the "one-time" volume be VS how much they would need per hour/day? And not some straw-man argument like what the guy in the video said.
It's a fairly shallow info pop piece that feels more like a commercial. The only point with regards to this discussion (and assuming it is not all just a lie, which I did not investigate, so it might just be): You can have a closed system that does not "use up" water.
> You can have a closed system that does not "use up" water.
But those "a million gallons of water" (or whatever it consumes/uses) have to come from somewhere, and cannot go somewhere else at the same time as it goes to the data center, so from the perspective of the Aragon farmers, isn't the water "used up"?
Similar to other industries, the water isn't lost per se, but it's deferred. It takes a considerable amount of time for water to do the whole loop and end back where it started, so the more you defer, the less you have available for consumption at any given time.
Exactly this. This is a major issue in my region right now, central Texas. Data centers are going up in some of the most drought stricken areas of Texas, and since it is hot and dry, they want to save money on cooling by misting. This uses precious fresh water that would have gone to support the overwhelming growth of local cities. So now everyone who lives in the area is paying more for water service because the cities are competing for water with the data centers. Guess who has more lawyers and money?
I still don't understand why more attention is not being paid to oil immersion cooling! https://par.nsf.gov/servlets/purl/10101126
It's corporate greed and an unwillingness to face the external consequences of demand for compute. Just because misting is cheaper doesn't make it automatically better.
When water changes phases from liquid into a gas it sucks a lot of energy into it. This is how sweat works to cool you. The water comes in, cools things through conduction (think liquid cooling in a computer) and then they evaporate it and it sucks a bunch of heat. It is then in the air and becomes rain eventually.
killing any healthy living parts in the water and contaminating the water with difficult molecules, of various degrees of toxicity. As others point out, the flow and duration of the flow of water is part of any ecosystem. There are already living things around water, so it also depends on how much living things are valued compared to industrial operations.
Per the article, all current European data centers used 62 million cubic meters of water for 2024. That is 3.4% of only Spain's existing desalination capacity (≈1.8 billion m³/yr).
Seems like this is solvable:
1. Keep rolling out the closed-loop cooling improvements now appearing in new DC designs.
2. Add more desal capacity where it’s cheap (sunny coastlines + renewables) to cover the residual demand.
Sources: - https://aedyr.com/plantas-desaladoras-agua-salobre-espana/
If the data centre is coastal then desalination might be an extra step. A salt water heat exchanger might do the job.
I hope this cheap populism doesn't win out. As someone who actually lives in Spain I want there to be good jobs for my children in the future.
Without tech companies and data centres we will just be a theme park for tourists with the poorly paid, precarious hospitality jobs that go along with that.
> Without tech companies [...]
Maybe. Tech companies might have some good jobs for a while. But they will continue to transfer their profits to their parent companies in the US (and/or some tax shelter somewhere).
> and data centres
Data centres employ very few people. Pretty much exclusively during construction. Once operational, very little actual labour is required. Mostly maintenance.
> we will just be a theme park for tourists with the poorly paid, precarious hospitality jobs that go along with that.
No more poorly paid or precarious than a data centre job for Big Tech, I would imagine.
I agree with the sentiment that Spain (and much of Europe, really) needs a big economic boost beyond tourism. But AI fueled data centres ain't it.
Yes, but thats is not what the article advocates for (allocation of water between farming/residential/industry).
It wants to keep industrial water use away without even having a discussion about water allocation/price. This is because farmers (all around the world tbh) are getting an insanely good deal right now (on water), and any public discussion of water price is only ever gonna make things worse for them.
> Without tech companies and data centres we will just be a theme park for tourists with the poorly paid
Since when was Aragon ever a tourist hotspot? As far as I know (as another "actual" Spain resident), tourists flock to the coastal areas and the islands (and for some reason, Madrid), not to the inner-mainland like Teruel and Zaragoza.
People go to Teruel to see Albarracín (IMHO one of the most beautiful villages in Spain -- 100% worth the trip) and to make side trips to places like Valderrobres, Calaceite, Rubielos de Mora, Puertomingalvo, Cantavieja, and Mirambel. Tends to be more popular with non-Spanish tourists.
I agree, the inner mainland is beautiful, and so is Teruel, but that's kind of besides the point. Are you disagreeing with me that the tourist hotspots are the coasts + islands? In my experience, it's more of us who already live in Spain who visit various inner mainland areas, not so much tourists who are staying for a week or two.
I meant for the country as a whole.
But Zaragoza is quite touristic, although probably mostly domestic tourism.
How would "tech companies and data centres" alleviate the situation? Tech companies typically employ highly specialized staff that are often not even local to the community. The result of inviting tech companies to your country can be seen in Ireland (/Dublin).
Data centres are even worse - they need only a handful of staff members while draining incredible amounts of energy and water.
You live in Spain - why not advocate for boosting the energy sector, better grid infrastructure, more renewables, etc.? This would harness a tangible, sustainable strength of your country within the EU and, considering the blackout last month, is definitely something to work on before any tech company or datacentre can settle in Spain ...
Also happening in Mexico
https://www.context.news/ai/thirsty-data-centres-spring-up-i...
> the company [microsoft] plans to be water positive by 2030 which means that they “put more water into the local basins where we operate than we withdraw”
> AWS has the same target, while Google has pledged to “replenish 120 percent of the freshwater volume we consume, on average, across our offices and data centers by 2030.”
How is that supposed to work?
The cynic in me can’t help thinking of an high-energy or production-externalities-imported system, but I’d be glad to ear about a sustainable local water creation.
If we would really have a chance to ask an official, whomever that may be, either from government or the tech companies, the (scripted) answer would probably go like this:
By 2030, AI will make revolutionary advancements in water management which will reduce our total water consumption, reduce waste, improve the wastewater treatment efficiency by 15x, so that, overall, the industry is not consuming but producing water.
These AI water articles are all terrible. Data centers could use no water but more power. The evaporative cooling is to save energy. In fact it would still save energy to desalinate water and use that for evaporative cooling. But none of that really matters since data centers don't use much water in comparison to farms and other industries.
Articles basically pander to the anti-AI crowd now -- knowing this group will run and share stories like this on all the socials -- and it just isn't useful.
Firstly, they're talking about data centres, not "AI". AI is just the boogeyman now and 100% of usage suddenly is imagined to be ChatGPT exchanges. In reality it is mostly servicing sites like this, ad networks, backing up your iCloud photos, running your bank, etc.
Secondly, price and regulate a resource appropriately and this wouldn't happen. The only data centres that run evaporative cooling do so because it's the least expensive option, and because it's allowed. In every normal place they run a closed circuit and the water usage is basically a rounding error.
Further, articles like this never give a context. 94 million m2 (the 2030 forecast for every data centres across Europe combined, not just "AI") sounds super large. Unfathomably large. Paris uses double this. Of course Paris is a massive city, but then think of every other large city across Europe, every farm, etc. It ends up being a small slice, for something that is very important in people's lives.
> In reality it is mostly servicing sites like this, ad networks, backing up your iCloud photos, running your bank, etc.
In reality, GPUs slurp a lot of energy, and it would be missing the current situation if you don't think new data centers will include a lot more GPUs than they did just 5 years ago. Not saying the new data centers will mostly be GPUs, but current context surely changed the calculations of what a "modern" data center should contain.
> price and regulate a resource appropriately and this wouldn't happen [...] In every normal place they run a closed circuit
Pricing and regulation doesn't suddenly mean foreign investors won't try to get an already scarce resource allocated to them. And regardless of how closed the circuit is, once the water goes to the data center, it cannot also go to other uses, so the farmer ends up with less water.
Here's a random example of one, current flow 2.16: https://mapy.com/en/zakladni?pid=107312153&newest=1&yaw=1.53...
I feel like this water-AI datacenter idea is the worst kind of whataboutism luddite thinking.
To begin with:
- it's almost certain that there is an over capacity of compute being built, and some kind of bubble.
- money is being wasted being thrown after non-viable ideas
*but*
This is a technology that will fundamentally change the way humans think and do things. There will be plenty of amazing new discoveries that will benefit all of humanity that will come out of all of this. -probably the most purely virtuous will be bio-medical related (alphafold etc.)
Yes, the capitalistic waste of resources is a shame, but any comment saying that the money should/can be used in a more cooperative/communal way are completely disconnected from reality.
The relatively high-waste wheel of capital has been spun up and because moore's law seems to generally apply- if not to model training and inference itself, at least to the underlying hardware, we're going to get efficient systems eventually.
I'm on the left and the environmental AI angle is one of the most regressive and short sighted takes from these people. I put it in the same bucket as anti-immigration left-green policies. Our new political world order is putting greens on the side of conservative, regressive authoritarianism and I don't like it.
From that viewpoint it makes sense why young (optimistic, idealistic) people would want to vote republican.
Have you ever suffered from water scarcity? Because I've lived it. This is a facile argument. Data centers are standing on the backs of so many others while at the same time using their piles of cash to force them to do more with less. And their argument about why is because the technology is so life changing and amazing.
It's so sad that money and power blind people when there are viable technological alternatives available.
It's unclear what you really mean here, but "suffer" from "scarcity" could mean a few things.
Tech is not immune from abuses of power, but most water scarcity that causes suffering (people being literally thirsty or can't wash their clothes) is not a resource issue (amount of water available)- it's a political/infrastructure issue. Don't build a data center that takes away people's clean drinking water. Build water infrastructure where people live.
But of course that's not the problem described by the article, even though the writer would like to lightly imply that might be the case.
Edit to add: Water and water rights in the western USA is a political issue that politicians would have you believe is not up to them. It's a scam and an advertising campaign that has people buying low flush toilets when a lot of the water is wasted by big Ag companies.
My personal experience with suffering from water scarcity: Taking a loss on a crop of corn because the farm at the front of the ditch took more water than they were supposed to take. That same year, a ditch rider was threatened at gunpoint.
When your own personal livelihood depends on another business playing fair with water, these kinds of stories hit different.
It absolutely is a resource issue. It is also very much also a political issue.
If only there were some method for allocating scarce resources (such as water), by which everyone could indicate their perception of the relative value and some sort of equilibrium could be found.
Free markets fail when allocating resources like water. The problem is that when someone needs something to survive, the market stops working because the value is infinite to them. Another reason it doesn't work is because you can have people who need it but are too poor to pay what others will pay for it. If someone with more money comes in and buys all the water, they can then charge what they want for it or just dump it into the ocean if they feel like it. It also doesn't work when the product is common use, like a road or a park, because their is no profit in providing something everyone benefits from but will not directly pay to use. Property rights and markets are not the ultimate way to allocate all resources.
I think people get emotional when they talk about water, but it's easier when it's broken down into more specific categories-
Every person should have the right to clean easily accessible drinking water in their homes. We know how to do that, it just costs money but it should be a basic responsibility of every government.
Beyond that if you want to use water for whatever then you should have to compete for it in a market. Maybe the only other exception might be growing sustenance food. But those systems (as they are currently implemented across the world) quickly get political- but no one should starve because water is too expensive either.
The thing is that almost all of the time that's not what's at stake when talking about things like data centers- it's a strawman.
> when someone needs something to survive, the market stops working because the value is infinite to them
No, that's precisely the situation where markets work best: when resources are scarce. A single gallon of water isn't worth much to a data center or a farmer; you can't water a field with just one gallon. But it's potentially worth a lot to someone who's thirsty. Prices will go up, demand will drop, and supply will increase to meet demand, unless you kneecap that process by imposing artificial price controls - then you'll have shortages.
> you can have people who need it but are too poor to pay what others will pay for it
You're right, markets are not charities; they're only concerned with efficiency. Caring for people who can't take care of themselves isn't efficient, but it is the right thing to do. Even in this situation though you're better off with markets than without them, because it's way easier for people, organizations, and governments with excess resources to provide for the needy when they're operating in an efficient environment than in an inefficient one.
> If someone [...] buys all the water
That's called a monopoly and I agree that's not good, because if one person owns/controls everything it's no longer a free market; you're essentially back to central planning. Individuals and data centers should be able to buy their water in a competitive market, not one dominated by a single supplier.
> It also doesn't work when the product is common use, like a road or a park
Markets are actually great at allocating things like road space. NYC's congestion pricing is doing wonders for the efficiency of their road system right now, and tolls have been a thing basically forever. But I agree in principle it's hard for markets to allocate resources that there's no practical mechanism of charging for. Thankfully, water generally doesn't fall into that category.
> No, that's precisely the situation where markets work best: when resources are scarce. A single gallon of water isn't worth much to a data center or a farmer; you can't water a field with just one gallon. But it's potentially worth a lot to someone who's thirsty.
1 gallon might be worth nothing to a datacenter, but the datacenter is built and a company invested in it and it can't go unused. They will pay more to keep that datacenter open than the local who needs it to drink. It doesn't matter how much the local needs or wants the water, if they don't have the money, how are they going to out bid a global corporation?
> You're right, markets are not charities; they're only concerned with efficiency.
Markets aren't concerned with anything, they are a means to an end. And they have appropriate uses and inappropriate uses. Not everything needs to be 'efficient' sometimes things should be 'just' or 'humane' rather than 'efficient'
> Markets are actually great at allocating things like road space. NYC's congestion pricing is doing wonders for the efficiency of their road system right now
That isn't a market, that is a tax.
Look, markets are great, but I don't get this quasi-religious adherence to one mechanism amongst many as the be-all-end-all of solutions. It is good at some things, bad at others. Ayn Rand was just an author.
We should just create a World Government (like https://en.wikipedia.org/wiki/World_constitution) and be done with it, global focus on the most pressing issues like water, food and getting rid of nuclear weapons.
Who owns this publication and who paid these two authors to write it? What's their angle?
Very impressive the top comment on this nonsense hit piece is someone vaguely implying extrajudicial ‘retribution’ is the appropriate solution to a problem which doesn’t actually exist.
And we also vote for governments in the expectation they will regulate those companies.
To put it another way. "Voting with my wallet" is a very inefficient way to send signals that I want a behaviour change from the company. I'd rather they were told directly with the force of law behind that request.
Yes, so what change do you want, and are you going to act accordingly or just flail your arms meaninglessly?
I assume that you, personally, have a car, or will buy one in your lifetime. Someone produces that car, and the production uses about 10⁵l of water. A little more generally, depending on what kind of car. If you count the water use of a modern LLM in the same way, that purchase corresponds to approximately one LLM query per minute for the rest of your life.
If you don't care for cars, pick another product and do the math yourself. A pair of jeans maybe. It's really hard to do the math and find a product that doesn't use much more water than a year's worth of heavy AI use. So if it's water you care about, you have a lot of things you can affect much better.
If you'd rather blame an outgroup, bigco execs will do. Or Wall Street.
Farms or Tech? What is better?
Its obvious from an economic standpoint, so I looked up from a security standpoint what crops Spain produces and it seems to be non-essentials. So its purely for export purposes.
Tech will win if all is rational, but people have a soft heart for the status quo. Those poor farmers, what a sad story. I imagine they can probably rally the population against their own interests with the correct moral coating and maybe create a faux crisis of raising the prices of food by hoarding for a few months.
> seems to be non-essentials
I'd love to hear what non-essential food is less important that fricking AI data centers.
> its purely for export purposes.
Erm, yes? Spain is one of the main food producers of the EU, so that tracks. How is that an argument for anything?
Is this a form of the meme:
https://images.app.goo.gl/zYoRKHJsCrLgyf989
Or this one?
https://knowyourmeme.com/memes/we-should-improve-society-som...
Is there evidence of that? Is it badly written? Assuming everyone has already committed to the race to the bottom can only accelerate the feeling of inevitability around it.
> As the sector grows, consumption is expected to reach 90 million cubic meters by 2030, according to the water sector lobby Water Europe.
People who are trying to organize opposition to particular uses for water have a habit of citing the raw numbers without putting them in context, which works because there's so much water that the numbers are eye-wateringly large.
It's hard to find concrete stats on total water usage (as opposed to percentage changes), but one report I found that helps to put this number in context is this [0]:
> In addition, it takes 17,000 litres of water to produce a kilo of chocolate. According to statistics from 2019, Europe produced 3.7 million tonnes of chocolate, which equates to an eyewatering 63,625,200,000 litres of water.
Since a liter is 1/1000 of a cubic meter, we're looking at 63 million cubic meters for European chocolate alone, which places chocolate in the same ballpark as Europe's data centers.
Obviously data centers can and should work to conserve water (no misting in dry regions would help), but on the surface ~40% more water for data centers than for chocolate doesn't seem like that bad of a balance.
[0] https://www.euronews.com/green/2024/07/09/our-water-footprin...