Excellent article. I would also like to point out what a very small percentage of total water use is for households. When people think of water conservation, they typically think of low-flow showers and low-water-use dishwashers. Household appliances actually do not use that much water as a percentage, and by far the biggest use within households is home lawns and gardens. So if you want to encourage water conservation, focus on agriculture and the other large water users mentioned in this article.
The percentage of total water use that comes from household appliances is low almost everywhere in the US. The data in this article bears this out, although Potter does not mention it.
The transportation cost, coastlines, mountain ranges, etc, do not change the percentages by very much.
I think that Robert may be pointing toward the larger scope of the water use issue, specifically in regions like Colorado Springs and Denver in Colorado. A large portion of the municipal water is pumped up and over the continental divide (since majority of the snowmelt occurs on the Western side of the divide, whereas large developments occur mostly on the eastern side). In this sense, household water use percentages may still be relatively low in comparison to other uses. However, due to the laws of prior appropriation in Western states, once that water is pumped up and over, it continues to flow downstream and is no longer available for use to upstream users (very often for agriculture like in the Lower Arkansas Valley). “Buy and Dry” is a common phrase used to describe practices like this.
Ok, maybe it doesn’t happen in the USA, but I live in Melbourne, Australia, which is a coastal city surrounded by hills. Australia’s largest river system, the Murray-Darling Basin, is on the other side of the hills, but Melbourne’s water supply does not come from the Murray-Darling catchment (partly for economic reasons but also political ones).
Of the water supplied by Melbourne water (which covers the entire metro area) 66% is for residential use.
What you say applies to some of the self-supplied water described above. Some home owners associations supply their own water and typically would have 100% residential use. But new homes are built, infrastructure needs to be upgraded, etc, so you can come up short.
“Residential use” does not only include drinking water, showers, toilets and other household appliance (what I was talking about in my original comment). It also includes large amounts for landscaping, gardens, and lawns plus perhaps golf courses. So growing plants is a significant part of residential usage, and few people realize it.
Even your example of Melbourne does not violate the general rule. Just look at the data.
According to ChatGPT:
“ From the Australian Bureau of Statistics data pertaining to Victoria:
• Agriculture used 66% of water consumption.
• Households accounted for 8%…
Urban water in Melbourne: household (~59%)
From Sustainable Gardening Australia, it’s reported that just over one-third (≈ 33–35%) of household water is used in gardens.”
So the actual number is 40%, not 60%.
And the water comes from a Murray-Darling Basin whose water use is dominated by agriculture.
More from ChatGPT:
“ Agricultural / Individual Users from Murray-Darling Basin
• This is by far the dominant sector in water use—accounting for several thousand gigalitres.
• In 2022–23, individual users (mostly irrigation) used approx 5,379 GL.
• Historically, agriculture has claimed 95% of diverted water from the Basin
Urban water usage (for cities, towns, and municipalities) is relatively small in comparison.
I think you didn’t read what they said. Agricultural water use is from the Murray-Darling system, on the other side of the mountains. Urban water use is from local water supplies. These are separate sources, so conservation on one is irrelevant to conservation on the other. The claim being made is that the water sources for the city of Melbourne are primarily used residentially.
(I’m a little skeptical that there’s *no* significant agriculture, or golf courses, or lawns, on the same water source as the city. But I do believe that most agricultural use in the state as a whole is on a completely separate water system.)
OK, I guess I misread your comment on the water source. I missed the “not.” I thought you were saying that the river *is* the water source.
Sorry for that.
But I stand by my original claim. My original claim was NOT about “residential use.” That was your claim.
As I stated before:
“Residential use” (the stat you use) does not just include drinking water, showers, toilets, dishwashers, and other household appliances (what I was talking about in my original comment). It also includes larger amounts for landscaping, gardens, and lawns plus perhaps golf courses and other uses. So growing plants is a significant part of water usage even in residential areas, and few people realize it.
So using the From Sustainable Gardening Australia estimate that over one-third (≈ 33–35%) of household water is used in gardens, the actual number for Melbourne is less than 40%, not 60% that you stated. Melbourne water usage for drinking water, showers, toilets and other household appliances might actually be well under 40% of total water use.
I question the accuracy of the water consumption of semiconductor manufacturing. Micron is breaking ground outside Syracuse on what is planned to be 4 chip fabs, thanks to $20B in various public subsidies if all 4 fabs are built.
The Onondaga County Water Authority will build a half billion dollar, 25 mile 4' diameter line from Lake Ontario to the Micron facility. It has capacity of 50 million gallons daily that Micron may fully consume (see OCWA launches biggest upgrade ever to double water supply for Micron-spurred growth. Who will pay? https://archive.ph/BKiax). It nearly doubles area water consumption.
In addition to water, 8 million gallons are so toxic it cannot be returned to the watershed. Some will be trucked out of state to injection wells. Waste water contains proprietary trade secret chemicals. 40 million gallons of waste will be treated at a $1.5B plant and returned to the watershed via the Oswego River, close to the county's intakes.
History doesn't repeat, but it seems to rhyme. Recalling Onondaga Lake was the most polluted Lake in the US and is not swimmable of fish able decades after Allied Chemical closed and remediation. Lake Ontario fisheries have many human consumption restrictions from industrial pollution decades ago.
The article is about water, but Micron’s electricity consumption will be over 500MW. One can only speculate the impact on residential and commercial rates.
Clear data is hard to find in the semiconductor industry, but keep in mind that most facilities recycle the majority of their water and often supplement it with air scrubbers to make their own h2o.
That said, the micron facility will no doubt have an impact on the local community.
"66 million gallons per day is about 6% of the water used by US golf courses, and it's about 3% of the water used to grow cotton in 2023."
This is a confusing sentence. It makes it sound like 66Mgal, the daily consumption for data centers, is 3% of the *yearly* water used to grow cotton. Granted, the next paragraph specifies that it's 2.5Bgal per day for cotton, and anyone could check the numbers themself. But I saw someone on Twitter get tripped up by a screenshot of the sentence I quoted.
This is likely the big money paragraph of the piece, you need it to be airtight. I would suggest rephrasing the sentence, maybe just by removing "in 2023".
DCs use evaporative cooling because it is extremely efficient (COP in the thousands). It's very different from your PC which just uses water as a heat transport medium.
Excellent article Brian. It would be very interesting to see a follow up that explores how the economics of desalination have changed over time. In particular now that we have intermittent sources of renewables, and desalination is not particularly time sensitive. Is there a path to economically feasible large scale irrigation of the West? Or does that remain a pipe dream?
great article but i think you missed grey/gray water, the water that is used in agriculture and horticulture and then gets back into aquifers, rivers, lakes and oceans to cause much harm, and health impacts for drinking water. in most cases the blue and green water components of food is small compared to the grey water https://www.sciencedirect.com/science/article/pii/S001393511930218X
Is there any reason why data centres or any other new large users cannot be required to build additional processing and storage facilities. Could they not re-cool the water and recycle it?
I would like to read a similar post on Indian water usage, or atleast peruse the data for India's case. Would you know where I might be able to find it?
Another fantastic effort ! Learned a lot ..... didn't know that the vast majority of the water evaporates. It reminds me (through analogy) of the time I learned the vast majority of the weight we gain/lose as humans is mechanically through the breathing process (the body's evapotranspiration).... makes so much sense once you think about it.
Excellent article. I would also like to point out what a very small percentage of total water use is for households. When people think of water conservation, they typically think of low-flow showers and low-water-use dishwashers. Household appliances actually do not use that much water as a percentage, and by far the biggest use within households is home lawns and gardens. So if you want to encourage water conservation, focus on agriculture and the other large water users mentioned in this article.
That’s an oversimplification.
Transporting water long distances is very expensive, particularly when nature puts a mountain range in the road.
So if you’ve got a city on the coast next to a mountain range, much of the cheaply available water might well end up being used for domestic purposes.
So while it may not be that significant when taking a nationwide view, it can be very important locally.
I do not understand your logic at all…
The percentage of total water use that comes from household appliances is low almost everywhere in the US. The data in this article bears this out, although Potter does not mention it.
The transportation cost, coastlines, mountain ranges, etc, do not change the percentages by very much.
I think that Robert may be pointing toward the larger scope of the water use issue, specifically in regions like Colorado Springs and Denver in Colorado. A large portion of the municipal water is pumped up and over the continental divide (since majority of the snowmelt occurs on the Western side of the divide, whereas large developments occur mostly on the eastern side). In this sense, household water use percentages may still be relatively low in comparison to other uses. However, due to the laws of prior appropriation in Western states, once that water is pumped up and over, it continues to flow downstream and is no longer available for use to upstream users (very often for agriculture like in the Lower Arkansas Valley). “Buy and Dry” is a common phrase used to describe practices like this.
Ok, maybe it doesn’t happen in the USA, but I live in Melbourne, Australia, which is a coastal city surrounded by hills. Australia’s largest river system, the Murray-Darling Basin, is on the other side of the hills, but Melbourne’s water supply does not come from the Murray-Darling catchment (partly for economic reasons but also political ones).
Of the water supplied by Melbourne water (which covers the entire metro area) 66% is for residential use.
What you say applies to some of the self-supplied water described above. Some home owners associations supply their own water and typically would have 100% residential use. But new homes are built, infrastructure needs to be upgraded, etc, so you can come up short.
“Residential use” does not only include drinking water, showers, toilets and other household appliance (what I was talking about in my original comment). It also includes large amounts for landscaping, gardens, and lawns plus perhaps golf courses. So growing plants is a significant part of residential usage, and few people realize it.
Even your example of Melbourne does not violate the general rule. Just look at the data.
According to ChatGPT:
“ From the Australian Bureau of Statistics data pertaining to Victoria:
• Agriculture used 66% of water consumption.
• Households accounted for 8%…
Urban water in Melbourne: household (~59%)
From Sustainable Gardening Australia, it’s reported that just over one-third (≈ 33–35%) of household water is used in gardens.”
So the actual number is 40%, not 60%.
And the water comes from a Murray-Darling Basin whose water use is dominated by agriculture.
More from ChatGPT:
“ Agricultural / Individual Users from Murray-Darling Basin
• This is by far the dominant sector in water use—accounting for several thousand gigalitres.
• In 2022–23, individual users (mostly irrigation) used approx 5,379 GL.
• Historically, agriculture has claimed 95% of diverted water from the Basin
Urban water usage (for cities, towns, and municipalities) is relatively small in comparison.
• In 2022–23, urban systems used around 361 GL.
I think you didn’t read what they said. Agricultural water use is from the Murray-Darling system, on the other side of the mountains. Urban water use is from local water supplies. These are separate sources, so conservation on one is irrelevant to conservation on the other. The claim being made is that the water sources for the city of Melbourne are primarily used residentially.
(I’m a little skeptical that there’s *no* significant agriculture, or golf courses, or lawns, on the same water source as the city. But I do believe that most agricultural use in the state as a whole is on a completely separate water system.)
OK, I guess I misread your comment on the water source. I missed the “not.” I thought you were saying that the river *is* the water source.
Sorry for that.
But I stand by my original claim. My original claim was NOT about “residential use.” That was your claim.
As I stated before:
“Residential use” (the stat you use) does not just include drinking water, showers, toilets, dishwashers, and other household appliances (what I was talking about in my original comment). It also includes larger amounts for landscaping, gardens, and lawns plus perhaps golf courses and other uses. So growing plants is a significant part of water usage even in residential areas, and few people realize it.
So using the From Sustainable Gardening Australia estimate that over one-third (≈ 33–35%) of household water is used in gardens, the actual number for Melbourne is less than 40%, not 60% that you stated. Melbourne water usage for drinking water, showers, toilets and other household appliances might actually be well under 40% of total water use.
So even in Melbourne, my claim still holds.
Excellent read.
I question the accuracy of the water consumption of semiconductor manufacturing. Micron is breaking ground outside Syracuse on what is planned to be 4 chip fabs, thanks to $20B in various public subsidies if all 4 fabs are built.
The Onondaga County Water Authority will build a half billion dollar, 25 mile 4' diameter line from Lake Ontario to the Micron facility. It has capacity of 50 million gallons daily that Micron may fully consume (see OCWA launches biggest upgrade ever to double water supply for Micron-spurred growth. Who will pay? https://archive.ph/BKiax). It nearly doubles area water consumption.
In addition to water, 8 million gallons are so toxic it cannot be returned to the watershed. Some will be trucked out of state to injection wells. Waste water contains proprietary trade secret chemicals. 40 million gallons of waste will be treated at a $1.5B plant and returned to the watershed via the Oswego River, close to the county's intakes.
History doesn't repeat, but it seems to rhyme. Recalling Onondaga Lake was the most polluted Lake in the US and is not swimmable of fish able decades after Allied Chemical closed and remediation. Lake Ontario fisheries have many human consumption restrictions from industrial pollution decades ago.
The article is about water, but Micron’s electricity consumption will be over 500MW. One can only speculate the impact on residential and commercial rates.
Semiconductor use data comes from this paper, which pegs it at around 110-115 billion liters (~30 billion gallons) a year: https://www.sciencedirect.com/science/article/pii/S2212371719300150
I agree that if one Micron cluster could use up to 50 million gallons a day then this would have to be a serious underestimate.
Clear data is hard to find in the semiconductor industry, but keep in mind that most facilities recycle the majority of their water and often supplement it with air scrubbers to make their own h2o.
That said, the micron facility will no doubt have an impact on the local community.
"66 million gallons per day is about 6% of the water used by US golf courses, and it's about 3% of the water used to grow cotton in 2023."
This is a confusing sentence. It makes it sound like 66Mgal, the daily consumption for data centers, is 3% of the *yearly* water used to grow cotton. Granted, the next paragraph specifies that it's 2.5Bgal per day for cotton, and anyone could check the numbers themself. But I saw someone on Twitter get tripped up by a screenshot of the sentence I quoted.
This is likely the big money paragraph of the piece, you need it to be airtight. I would suggest rephrasing the sentence, maybe just by removing "in 2023".
Or rephrase it as “3% of the daily use for cotton in 2023”.
Can I ask the dumb question: I've seen water-cooled PCs. They don't _consume_ any liquid. Why are datacenters any different?
DCs use evaporative cooling because it is extremely efficient (COP in the thousands). It's very different from your PC which just uses water as a heat transport medium.
Excellent article Brian. It would be very interesting to see a follow up that explores how the economics of desalination have changed over time. In particular now that we have intermittent sources of renewables, and desalination is not particularly time sensitive. Is there a path to economically feasible large scale irrigation of the West? Or does that remain a pipe dream?
great article but i think you missed grey/gray water, the water that is used in agriculture and horticulture and then gets back into aquifers, rivers, lakes and oceans to cause much harm, and health impacts for drinking water. in most cases the blue and green water components of food is small compared to the grey water https://www.sciencedirect.com/science/article/pii/S001393511930218X
The notion that "AI" is a legitimate use of water is laughable.
More or less laughable than the idea that golf or alfalfa or electricity are?
Excellent article!
Wondering if there is an integration possibility with Hannah's article today on AI carbon footprint with AI water usage..?
Is there any reason why data centres or any other new large users cannot be required to build additional processing and storage facilities. Could they not re-cool the water and recycle it?
I would like to read a similar post on Indian water usage, or atleast peruse the data for India's case. Would you know where I might be able to find it?
Great article btw!
Great post! Is there an explanation beyond pricing for the more-than-double household usage per day in the US vs UK/FR/Germany?
Another fantastic effort ! Learned a lot ..... didn't know that the vast majority of the water evaporates. It reminds me (through analogy) of the time I learned the vast majority of the weight we gain/lose as humans is mechanically through the breathing process (the body's evapotranspiration).... makes so much sense once you think about it.
I did not see wheat in the irrigation table.