The amount of people here in the comments happily suggesting to let Google use the clean water for their AI datacenters and return dirty water to use in crops is a bit worrying
Correct me if I'm wrong, but isn't water for cooling a closed loop? The water is used to cool, presumably it becomes water vapor and is re-condensed when cooled and used again.
Either way, prices should determine what an effective use of resources should be. It signals the scarcity, allows it to flow to the most productive resources, encourages new production and sources, and provides revenue.
> prices should determine what an effective use of resources should be
I have $1,000,000,000 and an insatiable appetite for both material and domination. My 9 neighbors, stupid naive fucks that they are, only have $100,000 in total and do not have imaginations sufficient to even begin to want all materials and power in the world.
So of course, when the sole owner of water comes along and offers to sell it, I buy it all for $100,001. I can really never have enough water, especially as I need to power wash my driveway everyday. (I absolutely cannot stand the sight of grime.)
Anyways I guess my point is, I’m glad we all understand that price determines efficiency. Once my 9 neighbors die of dehydration, I’ll be able to gather more materials and power with less obstruction and competition. Hooray!
Guess what people usually use to cool water vapor...
It does make sense that datacenters would be cooled just like your water-cooled PC but that's probably not very sustainable given the fact that they don't do so.
I'm not sure but I'm guessing gray water (or treated waste water) is not suitable for cooling purposes? Particle charge in small pipes and scaling may be a problem. Also, collecting gray water or channeling treated waste water - depending on the location that might be a problem.
Not that I'm in favor of using drinking water for cooling slop factories, but I guess the reason we don't see waste water being used for cooling is cost (unless governments start mandating that...)
I believe (happy to be corrected!) it's the same reason juice has little to no fibre: particles in the liquid could potentially clog the data centre cooling systems. But Google should just include the filtering cost as part of their operational expenses
Stupid question: datacenters need water for cooling right? But they don't boil that water, ie it comes out of the datacenter just a little warmer? If that is the case does it matter to the city? The warmer water can still be used for agriculture or any other common usage.
There are multiple ways. Closed loops, well not big deal you fill up and there is slight evaporation losses, but you could ship that in in tanker truck maybe once every few years.
Next is open loop cooling using secondary loop. Take a river, lake or sea. Pump some water from it, pass through heat exchanger and pump back out. Manageable for most of the year. Worse version is pump ground water and return it to these. Depletes the ground water...
And finally evaporative cooling. Which is boiling, but not at boiling point. Water goes to sky. No immediate return to local ground water or downstream the river... In this case you actually do in sense use up the water. Kinda like burning fossil fuels returns co2 to atmosphere. It will later turn to biomass, but that is a separate cycle.
The CO2 cycle is problematic because of timelines. We are releasing millions of years of CO2 accumulation.
Rain is more of a location problem. The evaporated water returns as rain quickly, but maybe somewhere else, such as over ocean. And the aquifer compresses and loses water retention ability.
It's not a stupid question but: technically, after passing through Google's facility that is now gray water, and you can't use that for agriculture or any other 'common usage' without a whole raft of work and you can't just dump it into the aquifer either.
But if it just went through some heat exchangers, it's not like if it was dirty? As far as I know, nuclear power plants return the water they consume to the rivers they extracted it from.
> it comes out of the datacenter just a little warmer
Exact values matter. Some power plants had been found dumping +10 C water into lakes/rivers, while they had permit only for +5, and it totally destroyed local ecosystem. And most efficient (in terms of money) is evaporation cooling, where at least part of water is "lost".
A lot of it gets converted to water vapor in the evaporative coolers, so it doesn't flow out -- it becomes humidity or clouds. The coolers do also produce waste water, but with all the minerals left behind after evaporation it's not suitable for drinking.
More people should scrutinize the methodology behind these AI data center water usage reports.
One widely cited Berkeley Lab figure includes the water evaporated from reservoirs behind hydroelectric dams.
Excluding that factor cuts their water usage estimate by more than half.
On AI & water, looks like all US data center usage (not just AI) ranges from 628M gallons a day (counting evaporation from dam reservoirs used for hydro-power) to 200-275M with power but not dam evaporation, to 50M for cooling alone. [0]
Even this article mentions how ridiculous this article is. Google's gigantic data center uses a third of the water in the Dalles, a miniscule community of only 17000 people. So they are using the water of 5000 homes, i.e. nothing.
I know google fiber kinda flumped, but if they are already doing their own power generation for data centers they might decide to sell that power to the public too. What is really scary is that I foresee a day where these big tech companies will see it is more profitable to serve utilities to people than web services. Then, after they have a monopoly in most areas, they will enshitify it too.
I don't think that will happen. Being utility is hard and margins are not great unless you get some government money like credit. And even those might go away with change in regime.
There just isn't enough margin or "free money" for someone like Google.
This is true, to supply software you can build it once and replicate endlessly, to supply email you need to run servers. That's commoditised and the team just sees a slider controlling number of servers.
But to provide power or internet you need to dig up the roads and lay a wire to every house. It's a totally different kind of business to which a tech person is completely unaccustomed. It would be more likely for a plumber or electrician to do such a thing. It's true a tech company could buy wholesale fiber access and provide internet on top of that, like they provide email on top of wholesale servers, but that's only one part of the business.
Tech companies are struggling to even build datacenters right now because of underestimating the work involved. They're really not used to things that don't scale by themselves.
Why on earth do they want water from the national forest when the massive Columbia River is right there!? Is it too expensive to treat the river water? /s
This accusation ignores the fact that the formal species description (Etnier, 1976) was thoroughly peer-reviewed before publication and has withstood decades of scrutiny—-even from biologists employed by TVA. Etnier himself acknowledged that genetic affinities between snail darters and stargazing darters would be instrumental in fully understanding their degree of relatedness and divergence (Etnier, 1976:487).
Regardless of whether snail darters are genetically distinct from stargazing darters based on current scientific data, the Little Tennessee River population in the 1970s would have nevertheless represented a population separated from the core population of stargazing darters by over 700 river miles of habitat highly modified by nearly 100 years of impoundment projects. As such, the Little Tennessee River population may still very well have qualified for protection under ESA as a sub-species or as a distinct population segment, as the ESA is applied today.
It was transparently an attempt to block the dam. Thank god Congress of that time wasn’t that convinced by the guy running dam protests suddenly finding an endangered species. The Wikipedia editor is not fully informed. These days this is the canonical example of scientific malpractice in this pseudo environmental approach.
Yes, corporations are not saints and stories about nestle? contracts that allow to sell them bottled glacier water boil my blood equally.
At the same time I cannot not to notice that only environmentalists, activists and watchdogs are quoted in the article. One (bark) got even quotes from 2 different people.
Not a single scientific entity, like university etc was presented.
From a linguistics perspective "environmentalists worry" is a phrase designed to trigger a certain response. It sets up an us versus them scenario with "us" being anti environmentalists. "Concerns over" would include the reader. More interesting for the journalism for some.
At this moment I just assume by default that those “watchdogs”, “environmentalists”, “nonprofits” are mix of nimby-ists and/or thinly veiled attempts of extracting money
(it’s a nice things you got here. It would be a shame if some rare species of a frog would be found here. A small donation for the great cause/good, of course, would help us to work on ensuring that nobody gets in harms way).
Nah. I pretty much said my option about types of organizations who were mentioned in the article, and the herd immediately assumed that I’m pro-corpo or whatever (enough to see in answers to my comments). Logic isn’t strong suit for many people, I accept that.
There are multiple documentations where people have problems with their water level due to data centres [0]. It's also no longer news that data centres are planned/run in arid/desert areas [1]. Scepticism to Google's intentions before the damage is done should be allowed.
This comment made me curious is such a thing actually happens.
As it turns out "greenmailing" is a thing, but not from environmental groups. Here's what claude found for me:
<ai>
The concern isn't baseless—there are documented cases of parties using environmental law as leverage, particularly California's CEQA. But empirical studies show only ~13% of such lawsuits actually come from environmental groups; the majority come from labor unions, business competitors, and NIMBYs hijacking environmental review for unrelated purposes. In this specific case, WaterWatch has a 40-year track record on Oregon water issues and the concerns about fish habitat are supported by the Confederated Tribes of Warm Springs—so the 'thinly veiled shakedown' framing doesn't really fit
</ai>
I hope doing that research didn't spend too much water!
Spoken like someone who hasn't interacted with the real world in quite some time. There are plenty of third world countries you can have a look at to see how it's going without those pesky rules.
Where did I say that rules do not apply/shouldn’t apply? I specifically stated my opinion about many types of activists. I’ll repeat my other comment - article quotes only activists/enviromentalists/watchdogs, without any mention of their qualifications in the subject matter. Executive director at Bark, conservation director for environmental group (wtf does it even mean?) are not qualifications
The amount of people here in the comments happily suggesting to let Google use the clean water for their AI datacenters and return dirty water to use in crops is a bit worrying
Yes, clearly it should be the other way round; people should be drinking clean water and Google should use their waste water for cooling.
Correct me if I'm wrong, but isn't water for cooling a closed loop? The water is used to cool, presumably it becomes water vapor and is re-condensed when cooled and used again.
Either way, prices should determine what an effective use of resources should be. It signals the scarcity, allows it to flow to the most productive resources, encourages new production and sources, and provides revenue.
> prices should determine what an effective use of resources should be
I have $1,000,000,000 and an insatiable appetite for both material and domination. My 9 neighbors, stupid naive fucks that they are, only have $100,000 in total and do not have imaginations sufficient to even begin to want all materials and power in the world.
So of course, when the sole owner of water comes along and offers to sell it, I buy it all for $100,001. I can really never have enough water, especially as I need to power wash my driveway everyday. (I absolutely cannot stand the sight of grime.)
Anyways I guess my point is, I’m glad we all understand that price determines efficiency. Once my 9 neighbors die of dehydration, I’ll be able to gather more materials and power with less obstruction and competition. Hooray!
Guess what people usually use to cool water vapor...
It does make sense that datacenters would be cooled just like your water-cooled PC but that's probably not very sustainable given the fact that they don't do so.
Ownership matters and some owners may have preferred uses independent of market price.
I might not want to sell the spare room in my house to creepy stranger, and I shouldn't have to outbid them if I already own it.
Who owns the water?
why bother with using a radiator to cool the water back down when it's cheaper to dump it/evaporate it when it's hot.
>Either way, prices should determine what an effective use of resources should be.
And with the AI frenzy, generating slop is considered way more important than people having access to water.
I'm not sure but I'm guessing gray water (or treated waste water) is not suitable for cooling purposes? Particle charge in small pipes and scaling may be a problem. Also, collecting gray water or channeling treated waste water - depending on the location that might be a problem.
Not that I'm in favor of using drinking water for cooling slop factories, but I guess the reason we don't see waste water being used for cooling is cost (unless governments start mandating that...)
I believe (happy to be corrected!) it's the same reason juice has little to no fibre: particles in the liquid could potentially clog the data centre cooling systems. But Google should just include the filtering cost as part of their operational expenses
Stupid question: datacenters need water for cooling right? But they don't boil that water, ie it comes out of the datacenter just a little warmer? If that is the case does it matter to the city? The warmer water can still be used for agriculture or any other common usage.
There are multiple ways. Closed loops, well not big deal you fill up and there is slight evaporation losses, but you could ship that in in tanker truck maybe once every few years.
Next is open loop cooling using secondary loop. Take a river, lake or sea. Pump some water from it, pass through heat exchanger and pump back out. Manageable for most of the year. Worse version is pump ground water and return it to these. Depletes the ground water...
And finally evaporative cooling. Which is boiling, but not at boiling point. Water goes to sky. No immediate return to local ground water or downstream the river... In this case you actually do in sense use up the water. Kinda like burning fossil fuels returns co2 to atmosphere. It will later turn to biomass, but that is a separate cycle.
The CO2 cycle is problematic because of timelines. We are releasing millions of years of CO2 accumulation.
Rain is more of a location problem. The evaporated water returns as rain quickly, but maybe somewhere else, such as over ocean. And the aquifer compresses and loses water retention ability.
It's not a stupid question but: technically, after passing through Google's facility that is now gray water, and you can't use that for agriculture or any other 'common usage' without a whole raft of work and you can't just dump it into the aquifer either.
But if it just went through some heat exchangers, it's not like if it was dirty? As far as I know, nuclear power plants return the water they consume to the rivers they extracted it from.
11 replies →
Don't they reuse the water by cooling it outside the data center? Most power plants do that.
4 replies →
If they can return it to the river how can't it flow to agriculture?
8 replies →
> it comes out of the datacenter just a little warmer
Exact values matter. Some power plants had been found dumping +10 C water into lakes/rivers, while they had permit only for +5, and it totally destroyed local ecosystem. And most efficient (in terms of money) is evaporation cooling, where at least part of water is "lost".
A lot of it gets converted to water vapor in the evaporative coolers, so it doesn't flow out -- it becomes humidity or clouds. The coolers do also produce waste water, but with all the minerals left behind after evaporation it's not suitable for drinking.
More people should scrutinize the methodology behind these AI data center water usage reports.
One widely cited Berkeley Lab figure includes the water evaporated from reservoirs behind hydroelectric dams.
Excluding that factor cuts their water usage estimate by more than half.
On AI & water, looks like all US data center usage (not just AI) ranges from 628M gallons a day (counting evaporation from dam reservoirs used for hydro-power) to 200-275M with power but not dam evaporation, to 50M for cooling alone. [0]
So not nothing, but also a lot less than golf.
[0] https://www.construction-physics.com/p/i-was-wrong-about-dat...
Even this article mentions how ridiculous this article is. Google's gigantic data center uses a third of the water in the Dalles, a miniscule community of only 17000 people. So they are using the water of 5000 homes, i.e. nothing.
It's possible to be both anti-golf, and anti-water-waste-in-tech.
[flagged]
2 replies →
How far we have fallen from the "Do no evil" marketing.
I know google fiber kinda flumped, but if they are already doing their own power generation for data centers they might decide to sell that power to the public too. What is really scary is that I foresee a day where these big tech companies will see it is more profitable to serve utilities to people than web services. Then, after they have a monopoly in most areas, they will enshitify it too.
I don't think that will happen. Being utility is hard and margins are not great unless you get some government money like credit. And even those might go away with change in regime.
There just isn't enough margin or "free money" for someone like Google.
This is true, to supply software you can build it once and replicate endlessly, to supply email you need to run servers. That's commoditised and the team just sees a slider controlling number of servers.
But to provide power or internet you need to dig up the roads and lay a wire to every house. It's a totally different kind of business to which a tech person is completely unaccustomed. It would be more likely for a plumber or electrician to do such a thing. It's true a tech company could buy wholesale fiber access and provide internet on top of that, like they provide email on top of wholesale servers, but that's only one part of the business.
Tech companies are struggling to even build datacenters right now because of underestimating the work involved. They're really not used to things that don't scale by themselves.
Why on earth do they want water from the national forest when the massive Columbia River is right there!? Is it too expensive to treat the river water? /s
[flagged]
[flagged]
Who are "these guys"? Are they the same people mentioned in the article? Why would you assume that everyone caring about the environment is the same?
If you are referring to this: https://en.wikipedia.org/wiki/Snail_darter_controversy
It says:
This accusation ignores the fact that the formal species description (Etnier, 1976) was thoroughly peer-reviewed before publication and has withstood decades of scrutiny—-even from biologists employed by TVA. Etnier himself acknowledged that genetic affinities between snail darters and stargazing darters would be instrumental in fully understanding their degree of relatedness and divergence (Etnier, 1976:487).
Regardless of whether snail darters are genetically distinct from stargazing darters based on current scientific data, the Little Tennessee River population in the 1970s would have nevertheless represented a population separated from the core population of stargazing darters by over 700 river miles of habitat highly modified by nearly 100 years of impoundment projects. As such, the Little Tennessee River population may still very well have qualified for protection under ESA as a sub-species or as a distinct population segment, as the ESA is applied today.
It was transparently an attempt to block the dam. Thank god Congress of that time wasn’t that convinced by the guy running dam protests suddenly finding an endangered species. The Wikipedia editor is not fully informed. These days this is the canonical example of scientific malpractice in this pseudo environmental approach.
I hope you also take a similar view of “corporate needs”. Because their abuses of people and the environment are many.
Yes, corporations are not saints and stories about nestle? contracts that allow to sell them bottled glacier water boil my blood equally.
At the same time I cannot not to notice that only environmentalists, activists and watchdogs are quoted in the article. One (bark) got even quotes from 2 different people.
Not a single scientific entity, like university etc was presented.
1 reply →
From a linguistics perspective "environmentalists worry" is a phrase designed to trigger a certain response. It sets up an us versus them scenario with "us" being anti environmentalists. "Concerns over" would include the reader. More interesting for the journalism for some.
At this point, Google could be a drop-in replacement for the corporate villain in any 1980s/1990s action movie.
Cue the White Knights defending the poor misunderstood megacorp from their keyboards.
80s and 90s films had better villains.
Has google even had a preserved fetal pig delivered to anyone?
8 replies →
At this moment I just assume by default that those “watchdogs”, “environmentalists”, “nonprofits” are mix of nimby-ists and/or thinly veiled attempts of extracting money
(it’s a nice things you got here. It would be a shame if some rare species of a frog would be found here. A small donation for the great cause/good, of course, would help us to work on ensuring that nobody gets in harms way).
>I just assume by default
Gitmo couldn't get me to admit to this degree of intellectual cowardice
Blessed are those who have not seen and yet have believed.
2 replies →
It's so refreshing to see someone fully disclose their ignorance in a comment, rather than pretend they're arguing in good faith.
Nah. I pretty much said my option about types of organizations who were mentioned in the article, and the herd immediately assumed that I’m pro-corpo or whatever (enough to see in answers to my comments). Logic isn’t strong suit for many people, I accept that.
2 replies →
There are multiple documentations where people have problems with their water level due to data centres [0]. It's also no longer news that data centres are planned/run in arid/desert areas [1]. Scepticism to Google's intentions before the damage is done should be allowed.
[0] https://youtu.be/RfzwLtWkVjg?si=OTjSgW6kBDR3-0So
[1] https://youtu.be/Mn5ttoXxAe8?si=EvN2jnIkA9yRAvJb
This comment made me curious is such a thing actually happens.
As it turns out "greenmailing" is a thing, but not from environmental groups. Here's what claude found for me:
<ai> The concern isn't baseless—there are documented cases of parties using environmental law as leverage, particularly California's CEQA. But empirical studies show only ~13% of such lawsuits actually come from environmental groups; the majority come from labor unions, business competitors, and NIMBYs hijacking environmental review for unrelated purposes. In this specific case, WaterWatch has a 40-year track record on Oregon water issues and the concerns about fish habitat are supported by the Confederated Tribes of Warm Springs—so the 'thinly veiled shakedown' framing doesn't really fit </ai>
I hope doing that research didn't spend too much water!
Do you still work for Google?
No. Any other questions?
Spoken like someone who hasn't interacted with the real world in quite some time. There are plenty of third world countries you can have a look at to see how it's going without those pesky rules.
Projecting much?
Where did I say that rules do not apply/shouldn’t apply? I specifically stated my opinion about many types of activists. I’ll repeat my other comment - article quotes only activists/enviromentalists/watchdogs, without any mention of their qualifications in the subject matter. Executive director at Bark, conservation director for environmental group (wtf does it even mean?) are not qualifications
1 reply →