Data centers should do everything they can to reduce water usage.
That said, this is masterful scapegoating. The ag lobby must be gleeful if they're not directly responsible for this narrative.
In 2003 my middle school in Central Pennsylvania had this exact same problem. All our water fountains had to have signs posted stating that the water was not safe to drink. Many of my classmates had the purple-tinged skintone that is characteristic symptom of consuming the polluted water.
The issue stems from high input, fossil-fuel based farming, and most of society simply looks the other way because no one has figured out a cheaper way to produce enough food. Data centers are just a red herring.
The short answer is nitrate poisoning. It usually happens in infants before it hits adults. The reduction in water volume in the aquifer is reducing the dilution of existing nitrate contamination from agriculture.
From the article:
> Morrow County, Oregon, has recorded nitrate readings as high as 73 parts per million (ppm) in household wells—more than ten times the state’s legal ceiling of 7ppm—following reports that local data centres are intensifying aquifer contamination.
From the CDC:
> The first reported case of fatal acquired methemoglobinemia in an infant due to ingestion of nitrate contaminated well water in the United States occurred in 1945 [Comly 1945]. This condition is also termed “Blue Baby Syndrome”.
Cyanosis is the medical term is but I should clarify its not like their skin was purple all the time. What happens is that if you drink a lot of the polluted well water then when your veins narrow (like from cold) it gets a more purplish hue. This happens naturally (like when your lips are cold after swimming) but its just more noticeable if you are drinking polluted well water from farming runoff.
A thing kids would do at my school is stick their hand out the window of the bus in the fall/winter and then compare whose hand is "more purple". Dumb kid stuff. I still remember a single person I knew whose cheeks would be super purple every time we came inside from playing in the winter. Almost like the character in the willy wonka movies.
So like its well known the color is simply more noticeable when you are being constantly exposed to nitrates. The person I am thinking of -- he was almost certainly drinking bad well water at home, perhaps for his whole life. I can't imagine that was only from the fountains at school. As far as I know its not a huge deal but like the article is saying... it quite obviously has knock on health effects.
This article is misleading. The data centers role here was to evaporate a tiny bit of already polluted water and increase the concentration of the pollutant.
The increase is estimated to be around .1% with reasonable assumptions.
The primary driver of pollution was agriculture but the data centers can be attributed to .1% at most. Is it a big deal? Not in my opinion.
Putting more water on the fields is the issue. It's not clear what portion of the water Amazon is contributing or how the concentration of that water compares to other users.
Presumably untreated irrigation water is the primary use in the county, but it's pretty hard to find numbers on it.
I didn't say radiation exposure, I said radioactive particulates in my body. 0.1% increase in radiation exposure is a high altitude plane flight. Nothing to be concerned about as you say. 0.1% increase in radioactive particulates in my body is a Fukishima sand grain in a clam. Or a leaky medical iodine cell. Or a speck of Chernobyl ash stirred up by passing soldiers and inhaled. That's a much bigger concern.
The risk from nitrates is cumulative. .1% may not seem much, but it can be well be the trigger for enormous problems. My best guess is that the limit was already very close at its max of 10 mg/L (10 ppm) and the data center tipped the scale.
Maybe they're (ab)using the findings of the data centers to get better water treatment, maybe they could have done more effort themselves?
But of course money and laws are far more important than healthy citizens, so why do anything about it when you don't have to?
Companies kinda tend to forget that if there is nobody left to consume stuff, you're also out of business. Governments kinda tend to forget that if there is nobody left to pay the taxes, they're also out of business.
Lets assume that .1% is cumulative, but so is the effect from agriculture which is 99.9%. Agriculture would affect the person 1000x than the data centre (if you accept the .1% figure). This sounds ridiculous so it is necessary to put it in concrete terms.
Of course. The issue could be that the previous parties combined "successfully" could uphold the legal limit, but now that suddenly the data centers also participate it's above the limit.
Who's to blame? Tricky legal question. My personal opinion is everybody involved should take their percentage of the blame, but i doubt they're willing to pay or act upon it.
Datacenters didn't put the pollutants in the water. Datacenters used water, but as their process does not use the pollutants the concentration of pollutants in water increased very slightly.
Background radiation is 10msv per day, so 3650 in a year. A cross country flight (6hrs according to google flights) is 40msv, so 1% of yearly background exposure. However I did mention a "short" flight, so if you scale it down for a 2 hr flight, you get 0.3%, which is admittedly higher than 0.1%, but not too far off.
Typical global yearly background radiation dose according to Canadian nuclear safety commission is around 2.5 mSv per year [0].
Ignoring the major issue that the location dependent variance is orders of magnitude more significant than 0.1% we can reason that 0.1% is an extra 2.5 uSv over an entire year.
Thats like a 20 minute flight... or half a dental xray, or if you planned to eat like 20 bananas this year you better cut them out.
Now the fact that there are all these other obvious ways to cut down your radiation before you even begin to think about datacenters, isn't unique for this radiation dose thing. It's true for the water consumption as well. Your Chatbots are not the cause of your water usage its your diet and material goods.
Ok I don’t know a lot about water purification, but playing the devil’s advocate: isn’t it the responsibility of the water purifiers to get rid of nitrates and ensure the drinking water stays within legal limits?
Or is it somehow a very difficult / impossible process to do this?
Those are residential systems. At the municipal level it is more often done with bacteria in large tanks.
It's not unlike brewing beer in that you need to adapt your recipe constantly to account for variations in inputs in order to get a consistent product out the other end.
I was about to say, I happen to have aquascaping / aquariums as a hobby and getting the right bacteria in your water and keeping those in balance is a crucial part of keeping nitrates down.
People have wells at their homes, their water isn't being purified by someone else. Centralized supplies in the area are going to treat and monitor their water to ensure that it is below the applicable limits.
It's reasonable to treat the aquifer home wells draw from as a public good and prevent contamination of it. Whether Amazon has particular responsibility in this case is a different question than that.
It is but if a party mixes in another stream that normally speaking would not have found its way into the water purification system that system may no longer work. Amazon could of course just use closed loop systems like they are required by law to use elsewhere but chose not to.
Amazon chose not to because it’s almost twice as efficient to use open loop cooling towers vs a closed loop. COP of 4 for an air-cooled chiller but add a cooling tower and your COP goes up to ~7.
The comments here make you think about whether this is an issue or not. But, regardless, I wish that Amazon execs were literally shackled to their data centers. If they move to avoid taxation, like when Bezos moved to Florida, at least they would have to drag those data centers with them. If there truly are pollution issues, at least they can share that with their new neighbors. Bezos certainly seems to care about fitness, so if anyone can drag a data center that is shackled to his neck, it would be Bezos.
Accountability and responsibility is necessary for any society to function. When corporations get all the profit while they externalize all the cost the result is a dystopia.
The rich should be forced to suffer the consequences of their actions, otherwise they have no incentive to respect the health or even life of the rest of citizens.
Headline is very misleading. The article is saying the data center might be making an existing problem slightly worse, but it’s not the source of the underlying problem.
Evaporative cooling (like a "swamp cooler" for residential homes) is how most data centers in the US are cooled. The water is primarily consumed by evaporation. When you continually evaporate water from a system, eventually the remaining water in the system gets concentrated in salts and other minerals and is dumped and replaced with fresh water.
Much of the day/season, evaporative cooling is not needed and data centers can pull in outside air. Ultimately you state the main reason in your comment: using outside air + evaporative cooling is cheaper and consumes less power than any other approach.
In a lot of cases, even if the server chips themselves are liquid cooled (for example, in an NVIDIA GB200 rack), then liquid is then air cooled through a cooling distribution unit (basically a giant radiator.
While air cooling doesn't scale, air also is not a great heat carrier when you cram that much power to a small space.
Today's supercomputers (AI or not) can't cool themselves off with air. Too much heat in a too confined space. Direct Liquid Cooling is a must.
However, you can use closed-loop liquid cooling (like Europe), but open-loop is cheaper since it skips the "pump the heat out from water to atmosphere" part and "who cares about the water anyway, there's monies to be made".
Putting money above the environment always makes me angry though. It's like burning the walls of your house to stay warm.
Yet look at the prices on e.g. Hetzner and how profitable US cloud is. They can easily afford to do closed loop.
Whenever there’s a real environmental problem the answer is usually “there’s a right way to do it that lacks these issues but it’s slightly more expensive.”
Adding in an evaporative cooling tower almost doubles the efficiency so it costs almost half as much to cool a data center using cooling towers vs air cooled chillers alone.
This article states that wastewater from the center contains nitrates at 56 ppm, but the local wells are testing above 70 ppm. It also targets the vector of contamination as being from percolation to groundwater from water treatment plants, which I'm guessing only a fraction of wastewater has the ability to leak through.
This feels incredibly disingenuous, or at the least, incredibly poor journalism.
The original piece by rolling stone describes Amazon's role:
- Experts say Amazon’s arrival supercharged this process. The data centers suck up tens of millions of gallons of water from the aquifer each year to cool their computer equipment, which then gets funneled to the Port’s wastewater system. All of the data center water gets mixed into the dirty lagoon wastewater, which only increases how much water the Port must then discard over the fields. As Greg Pettit, who served at the DEQ for 38 years and led the development of Oregon’s Groundwater Quality, explains, “the more water you put on, the faster you’re going to drive the nitrogen through the soil and down into the aquifer.” -
So even if the datacenter expelled pure H2O it would still be a problem by increasing the hydraulic head and speeding up the spread of the agricultural runoff polluted water that already was in groundwater?
Or rather, a hypothesis disguised as an explanation. It ultimately said these nitrates come from farms, and this new datacenter is exacerbating it. The root cause is still the farms, though.
An explanation is just a theory that accounts for the facts. A hypothesis is a theory that accounts for the facts while also providing falsifiability and testability
.
They didn't provide any accounting from facts. Did farms usage of nitrates increase? We don't know. Are there more or less farms? All we know is that they assert the datacenter now exists and since then nitrates have increased. It doesn't account for all the variables nor provide a likely explanation for how this could be true; it is only a possibility of many.
Data centers should do everything they can to reduce water usage.
That said, this is masterful scapegoating. The ag lobby must be gleeful if they're not directly responsible for this narrative.
In 2003 my middle school in Central Pennsylvania had this exact same problem. All our water fountains had to have signs posted stating that the water was not safe to drink. Many of my classmates had the purple-tinged skintone that is characteristic symptom of consuming the polluted water.
The issue stems from high input, fossil-fuel based farming, and most of society simply looks the other way because no one has figured out a cheaper way to produce enough food. Data centers are just a red herring.
What type of contamination causes purple-tinted skin? Googling isn’t turning up much (but then…maybe I should try Kagi instead these days)
The short answer is nitrate poisoning. It usually happens in infants before it hits adults. The reduction in water volume in the aquifer is reducing the dilution of existing nitrate contamination from agriculture.
From the article:
> Morrow County, Oregon, has recorded nitrate readings as high as 73 parts per million (ppm) in household wells—more than ten times the state’s legal ceiling of 7ppm—following reports that local data centres are intensifying aquifer contamination.
From the CDC:
> The first reported case of fatal acquired methemoglobinemia in an infant due to ingestion of nitrate contaminated well water in the United States occurred in 1945 [Comly 1945]. This condition is also termed “Blue Baby Syndrome”.
https://archive.cdc.gov/www_atsdr_cdc_gov/csem/nitrate-nitri...
Cyanosis is the medical term is but I should clarify its not like their skin was purple all the time. What happens is that if you drink a lot of the polluted well water then when your veins narrow (like from cold) it gets a more purplish hue. This happens naturally (like when your lips are cold after swimming) but its just more noticeable if you are drinking polluted well water from farming runoff.
A thing kids would do at my school is stick their hand out the window of the bus in the fall/winter and then compare whose hand is "more purple". Dumb kid stuff. I still remember a single person I knew whose cheeks would be super purple every time we came inside from playing in the winter. Almost like the character in the willy wonka movies.
So like its well known the color is simply more noticeable when you are being constantly exposed to nitrates. The person I am thinking of -- he was almost certainly drinking bad well water at home, perhaps for his whole life. I can't imagine that was only from the fountains at school. As far as I know its not a huge deal but like the article is saying... it quite obviously has knock on health effects.
This article is misleading. The data centers role here was to evaporate a tiny bit of already polluted water and increase the concentration of the pollutant.
The increase is estimated to be around .1% with reasonable assumptions.
The primary driver of pollution was agriculture but the data centers can be attributed to .1% at most. Is it a big deal? Not in my opinion.
Putting more water on the fields is the issue. It's not clear what portion of the water Amazon is contributing or how the concentration of that water compares to other users.
Presumably untreated irrigation water is the primary use in the county, but it's pretty hard to find numbers on it.
> .1% at most. Is it a big deal? Not in my opinion.
Depends. .1% increase in the number of calories going into my body? Not a big deal.
.1% increase in the number of radioactive particulates going into my body? Big deal.
>.1% increase in the number of radioactive particulates going into my body? Big deal.
Why would that be a big deal? It is a .1% increase.
Yearly average exposition is about 600 mrem, don't take risks with those aditional 0.6 mrems!
I didn't say radiation exposure, I said radioactive particulates in my body. 0.1% increase in radiation exposure is a high altitude plane flight. Nothing to be concerned about as you say. 0.1% increase in radioactive particulates in my body is a Fukishima sand grain in a clam. Or a leaky medical iodine cell. Or a speck of Chernobyl ash stirred up by passing soldiers and inhaled. That's a much bigger concern.
Can you frame it into more concrete terms like how much more likely it is to cause harm? And put agriculture by side.
The risk from nitrates is cumulative. .1% may not seem much, but it can be well be the trigger for enormous problems. My best guess is that the limit was already very close at its max of 10 mg/L (10 ppm) and the data center tipped the scale.
Maybe they're (ab)using the findings of the data centers to get better water treatment, maybe they could have done more effort themselves?
But of course money and laws are far more important than healthy citizens, so why do anything about it when you don't have to?
Companies kinda tend to forget that if there is nobody left to consume stuff, you're also out of business. Governments kinda tend to forget that if there is nobody left to pay the taxes, they're also out of business.
Lets assume that .1% is cumulative, but so is the effect from agriculture which is 99.9%. Agriculture would affect the person 1000x than the data centre (if you accept the .1% figure). This sounds ridiculous so it is necessary to put it in concrete terms.
Of course. The issue could be that the previous parties combined "successfully" could uphold the legal limit, but now that suddenly the data centers also participate it's above the limit.
Who's to blame? Tricky legal question. My personal opinion is everybody involved should take their percentage of the blame, but i doubt they're willing to pay or act upon it.
> Who's to blame? Tricky legal question
The polluters.
Datacenters didn't put the pollutants in the water. Datacenters used water, but as their process does not use the pollutants the concentration of pollutants in water increased very slightly.
>.1% increase in the number of radioactive particulates going into my body? Big deal.
So... a short flight? There's background radiation everywhere, and less radiation protection higher up.
The increase during the flight is nowhere close to .1%, many many orders of magnitude more.
According to this chart, it's approximately correct?
https://xkcd.com/radiation/
Background radiation is 10msv per day, so 3650 in a year. A cross country flight (6hrs according to google flights) is 40msv, so 1% of yearly background exposure. However I did mention a "short" flight, so if you scale it down for a 2 hr flight, you get 0.3%, which is admittedly higher than 0.1%, but not too far off.
If we're looking at yearly exposure, as opposed to realtime exposure then sure.
Typical global yearly background radiation dose according to Canadian nuclear safety commission is around 2.5 mSv per year [0].
Ignoring the major issue that the location dependent variance is orders of magnitude more significant than 0.1% we can reason that 0.1% is an extra 2.5 uSv over an entire year.
Thats like a 20 minute flight... or half a dental xray, or if you planned to eat like 20 bananas this year you better cut them out.
Now the fact that there are all these other obvious ways to cut down your radiation before you even begin to think about datacenters, isn't unique for this radiation dose thing. It's true for the water consumption as well. Your Chatbots are not the cause of your water usage its your diet and material goods.
[0] https://www.cnsc-ccsn.gc.ca/eng/resources/fact-sheets/natura...
Not what I said: https://news.ycombinator.com/item?id=46116920
Ok I don’t know a lot about water purification, but playing the devil’s advocate: isn’t it the responsibility of the water purifiers to get rid of nitrates and ensure the drinking water stays within legal limits?
Or is it somehow a very difficult / impossible process to do this?
Apparently, none of the methods used in normal water purification remove nitrates:
https://www.freshwatersystems.com/blogs/blog/how-to-remove-n...
Those are residential systems. At the municipal level it is more often done with bacteria in large tanks.
It's not unlike brewing beer in that you need to adapt your recipe constantly to account for variations in inputs in order to get a consistent product out the other end.
https://www.epa.gov/sites/default/files/2019-08/documents/de...
I was about to say, I happen to have aquascaping / aquariums as a hobby and getting the right bacteria in your water and keeping those in balance is a crucial part of keeping nitrates down.
Plants also use nitrates as a nutrient.
People have wells at their homes, their water isn't being purified by someone else. Centralized supplies in the area are going to treat and monitor their water to ensure that it is below the applicable limits.
It's reasonable to treat the aquifer home wells draw from as a public good and prevent contamination of it. Whether Amazon has particular responsibility in this case is a different question than that.
It is but if a party mixes in another stream that normally speaking would not have found its way into the water purification system that system may no longer work. Amazon could of course just use closed loop systems like they are required by law to use elsewhere but chose not to.
Amazon chose not to because it’s almost twice as efficient to use open loop cooling towers vs a closed loop. COP of 4 for an air-cooled chiller but add a cooling tower and your COP goes up to ~7.
Not sure, but we know that it is definitely not AWS’ job to pollute it
The comments here make you think about whether this is an issue or not. But, regardless, I wish that Amazon execs were literally shackled to their data centers. If they move to avoid taxation, like when Bezos moved to Florida, at least they would have to drag those data centers with them. If there truly are pollution issues, at least they can share that with their new neighbors. Bezos certainly seems to care about fitness, so if anyone can drag a data center that is shackled to his neck, it would be Bezos.
Accountability and responsibility is necessary for any society to function. When corporations get all the profit while they externalize all the cost the result is a dystopia.
The rich should be forced to suffer the consequences of their actions, otherwise they have no incentive to respect the health or even life of the rest of citizens.
Headline is very misleading. The article is saying the data center might be making an existing problem slightly worse, but it’s not the source of the underlying problem.
I’m a dummy: why do data centers have to consume water again?
I have both computers and air conditioning and neither consume water.
I’m assuming of course that evaporation cooling is cheaper and consumes less energy than closed cycle cooling with a forced air heat sink.
Evaporative cooling (like a "swamp cooler" for residential homes) is how most data centers in the US are cooled. The water is primarily consumed by evaporation. When you continually evaporate water from a system, eventually the remaining water in the system gets concentrated in salts and other minerals and is dumped and replaced with fresh water.
Much of the day/season, evaporative cooling is not needed and data centers can pull in outside air. Ultimately you state the main reason in your comment: using outside air + evaporative cooling is cheaper and consumes less power than any other approach.
In a lot of cases, even if the server chips themselves are liquid cooled (for example, in an NVIDIA GB200 rack), then liquid is then air cooled through a cooling distribution unit (basically a giant radiator.
While air cooling doesn't scale, air also is not a great heat carrier when you cram that much power to a small space.
Today's supercomputers (AI or not) can't cool themselves off with air. Too much heat in a too confined space. Direct Liquid Cooling is a must.
However, you can use closed-loop liquid cooling (like Europe), but open-loop is cheaper since it skips the "pump the heat out from water to atmosphere" part and "who cares about the water anyway, there's monies to be made".
Putting money above the environment always makes me angry though. It's like burning the walls of your house to stay warm.
That’s what I thought.
Yet look at the prices on e.g. Hetzner and how profitable US cloud is. They can easily afford to do closed loop.
Whenever there’s a real environmental problem the answer is usually “there’s a right way to do it that lacks these issues but it’s slightly more expensive.”
Air cooling doesn’t scale. Try having thousands of computers all in your room.
Well sure but why does it need to be open cycle and actually consume water? That’s the problem part, not internal water cooling.
It absolutely does not have to be open cycle, but that saves some money and externalizes the problem in the form of heating up the ground water.
Adding in an evaporative cooling tower almost doubles the efficiency so it costs almost half as much to cool a data center using cooling towers vs air cooled chillers alone.
Source: https://www.rollingstone.com/culture/culture-features/data-c...
This article states that wastewater from the center contains nitrates at 56 ppm, but the local wells are testing above 70 ppm. It also targets the vector of contamination as being from percolation to groundwater from water treatment plants, which I'm guessing only a fraction of wastewater has the ability to leak through.
This feels incredibly disingenuous, or at the least, incredibly poor journalism.
The original piece by rolling stone describes Amazon's role:
- Experts say Amazon’s arrival supercharged this process. The data centers suck up tens of millions of gallons of water from the aquifer each year to cool their computer equipment, which then gets funneled to the Port’s wastewater system. All of the data center water gets mixed into the dirty lagoon wastewater, which only increases how much water the Port must then discard over the fields. As Greg Pettit, who served at the DEQ for 38 years and led the development of Oregon’s Groundwater Quality, explains, “the more water you put on, the faster you’re going to drive the nitrogen through the soil and down into the aquifer.” -
So even if the datacenter expelled pure H2O it would still be a problem by increasing the hydraulic head and speeding up the spread of the agricultural runoff polluted water that already was in groundwater?
https://www.rollingstone.com/culture/culture-features/data-c...
[dead]
[dead]
[flagged]
Read the article, it explains it clearly.
does it though? It provides a hypothesis, not an explanation.
It actually provides an explanation and not a hypothesis.
Or rather, a hypothesis disguised as an explanation. It ultimately said these nitrates come from farms, and this new datacenter is exacerbating it. The root cause is still the farms, though.
An explanation is just a theory that accounts for the facts. A hypothesis is a theory that accounts for the facts while also providing falsifiability and testability .
They provide an explanation, not a hypothesis.
They didn't provide any accounting from facts. Did farms usage of nitrates increase? We don't know. Are there more or less farms? All we know is that they assert the datacenter now exists and since then nitrates have increased. It doesn't account for all the variables nor provide a likely explanation for how this could be true; it is only a possibility of many.
Did you read the article?