I found a lot of value in this article. Out of frustration with people who are alarmist over how much water a datacenter "consumes" compared to households, I've probably erred too often towards:
'People sometimes invoke the idea that water moves through a cycle and never really gets destroyed, in order to suggest that we don’t need to be concerned at all about water use. But while water may not get destroyed, it can get “used up” in the sense that it becomes infeasible or uneconomic to access it.'
Side note, this personal anecdote from the author caught me off guard: "my monthly water bill is roughly 5% of the cost of my monthly electricity bill". I'm in the American southwest (but not arid desert like parts of Arizona/Nevada/Utah), and my monthly water cost averages out annually to ~60% of the cost of electricity. Makes me wonder if my water prices are high, if my electricity prices are low, if my water usage is high or my electricity usage is low.
Biggest alarmist is movement against Nestle using water for bottled water in California. They don’t even use as much as an average golf course.
How much water is wasted on golf courses in these arid regions? Or growing water intensive crops like alfalfa that isn’t even directly used to feed people.
Yep, 1.6 trillion gallons of water from the Colorado river goes into irrigation for alfalfa[1]. Google's total water consumption across all data centers in 2023 was 6.4 billion gallons[2].
People are sounding the alarm about water usage in AI data centers while ignoring the real unsustainable industries like animal agriculture.
Talking about wasteful. There 16,000 golf courses that use 312,000 gallons a day[1]. Thats 1.82 trillion gallons annually. Only 28 million people play golf course on a course. Google's MAU is 90%+ of US population, beef or milk consumptions i would guess that 90% of population consumes it at least once a month. We're focusing on things that everyone uses but the things that less than 10% of the populations partake in. Why do we have golf courses in arid regions that have severe water shortages? Before places like LA county spends $8 billion on a toilet to tap system[2], maybe shut down the golf courses first.
It’s a great example of using large numbers without context to scare people.
Say “6.4 billion gallons” in isolation and people will be horrified. Put it in context relative to something like alfalfa farming and it doesn’t even appear on the same scale.
I've remembered the fact that a million seconds is ~11 days and a billion seconds is ~32 years since I was a kid. Still feels pretty ridiculous as an adult, no-one who didn't know it has even guessed close (and some who try to work it out were way off).
I just had to google what a trillion is in years, and the answer made me realise I don't instinctively understand the relationship between a billion and a trillion either!
Each "-illion" is 1000x bigger than the previous one.
If you have some cubes 1 cm on a side (about the size of a sugar cube), you can make a bigger cube out of them with 10 little cubes along each edge. Now you have 1 big cube made from 1000 smaller cubes.
Your bigger cube is now 10cm x 10cm x 10 cm. Easy enough to pick up in one hand.
Now do it again. Make a bigger cube with 10 of those cubes along each edge. Now you have a cube 1 meter on each side. Too big to pick up by hand but it would still fit in the back of a pickup truck.
This 1-meter cube contains 1 million sugar cubes.
Do it again: With 1000 of the previous cubes, make a cube 10 meters on a side. This cube is the size of a 3-story house, and it contains 1 billion sugar cubes.
Now do it once more: With the house-sized cubes, make a 10x10x10. Now the cube is about the size of a football stadium. It contains 1 trillion sugar cubes.
Take 4 of these stadiums, call each sugar cube $1, and you have the market cap of Nvidia.
[Note: This is US usage. In older UK English, some of the "-illion" words mean different things than they do here.]
You pay for fuel for your car => Saudi monarchy gets it share because they supply it => while they completely waste 20% for “supercars” and vanity, they still have enough money to do whatever they want including => they grow alfalfa next to you to feed their local cattle
You are overlooking location. The ideal place to grow crops is a place with great soil, good weather, a long growing season, and abundant water, but there aren't a lot of those. Of those four things, water is the only one that can be reasonably transported.
Data centers have fewer constraints. It should be possible to place more or all of them in places where water is abundant.
My comment was just focused on total water use. I agree that location does matter, and that data centers should be placed where water is abundant.
It still doesn't change my concern about how unsustainable growing alfalfa is. Trillions of gallons to grow an inefficient animal feed crop while we're told by the evening news to take shorter showers (8 minute shower is ~16 gallons of water) and let our lawns die.
You are overlooking location. The ideal place to place a datacenter is a place with cheap land, cheap electricity, good backbone connectivity, and close to users, but there aren't a lot of those.
Water is not evenly distributed.aData centers are not bieng located where there is excess water, they are bieng located in areas where they have access to the critical infrastucture they need,and the use of domestic potable water supplys to cool there operations is done as it reduces there land
and infrastructure requirements, is quick, and they care nothing about the costs of electricity and water, while they drive up costs for the people who live in the surounding areas.
People NEED water, data does not.
People NEED agriculture, they do not NEED data.
conflating the water uses of things to people is false.
Vastly cheaper to just have an efficient water market. But the current system makes farmer either use their water allocations for agriculture or not have that water at all.
I remember doing the calculations on the Nestle plant that caused a big storm a few years ago. The plant sat on several acres of land, which if converted into an alfalfa farm, would have consumed the same amount of water. The surrounding area was littered with alfalfa farms so it wasn't an unfair comparison. Meanwhile that bottling plant employs dozens of people, far more then a farm would have.
Do you think that leading mothers away from natural breast feeding into buying nestle formula is good for babies?
Terminator crops that won't propagate from seed also have the same SaaS feel, making two natural solutions for human sustinanace up for rent, directing science research away from negative conclusions about both these scenarios, creating dependency on both these industries sounds like a win for money but not human wellbeing. Sign me up for the tribe that says no to that!
Great examples of the sort of nonsense I was talking about.
Formula is fine. How do I know that? If it weren't, if there were any evidence it was bad, the FDA would not allow it to be sold in the US. But there's a wide variety of such formulas here.
Perhaps you think formula is fine for we smart first worlders, but poor dumb third world mothers have to have their agency removed and us determine what they can choose? Strong pass on that; this is no longer the age of European colonialism.
Terminator crops... oh boy. This one is hilariously ridiculous. Their sin is that they don't allow the farmer to retain seeds and plant them again. You know what else has that property? All hybrid seeds! These have been around for nearly a century. No hybrids breed true if you try to replant them; the uniform combinations of genes in the first generation hybrid becomes randomized in the second generation.
You aren't complaining about terminator seeds, you are complaining about modern agriculture.
What I'm complaining about is everything being for rent. And how we get there in mamy cases is by turning a blind eye to knock on effects of these technologies. There's a reciprocal immune response between mother and baby through nursing. I haven't heard of a formula that has this feature. That link I posted claimed that approximately 212,000 babies died a year in areas without clean water when supplied with formula. That's a bleak outcome, is that just the cost of progress for you?
You're again engaging in the patronizing approach of removing any agency from those involved in these countries. Some mothers there would certainly benefit from access to formula, just as they do in your country. You remove this because you don't consider them able to make choices for themselves. It's the same horrid mindset of the colonialists, seeing the third worlders as primitives "just down from the trees".
I agree about the agency, except the game is rigged, once you stop breast feeding it's hard to start again. So a mother is sold on a way to feed a baby that can free her up to work and anyone can toss a bottle to the kid, but then water quality becomes problematic, and the baby is the most vulnerable. You can't fall back to breast feeding. And the formula is no doubt funded by third world debt arrangements. Nestle didn't have the foresight to think about water quality, and it resulted in a tragic outcome. Even though they also sell water, they could have been the hero and made twice as much money if they were in the loop (not sure if they were in the water business back then).
You can waste the energy you spent cleaning it and pumping it around. But between nuclear and solar we ought to have an overabundance of that.
In a market economy, if it becomes "economically infeasible" to purify used water, the price goes up slightly, and suddenly it makes a lot more sense to treat dirty water, or even seawater.
You see the same type of argument against oil or mineral use; the idea that we'll run out. But people who argue we'll run out almost always look at confirmed reserves that are economical to extract right now. When prices rise, this sends a signal to prospectors and miners to go look for more, and it also makes far more reserves economical.
For example, Alberta's oil sands were never counted as oil reserves in bygone decades, because mining it made no sense at the time. But the economy grew per capita and overall, prices rose, and suddenly Canada is an oil-rich nation.
A similar dynamic applies to water and everything else.
Of course there are finite amounts of oil and uranium and so on, but the amounts just on this one planet are absolutely mind-boggling. The Earth has a radius of 6400km, and our deepest mines are 3-4km. We may expect richer mineral deposits (not oil) as we go further down.
Keep following this price logic and at a certain point it'll make sense to mine the far side of the moon, the asteroid belt, and so on ad infinitum.
You are not technically wrong, but you are economically wrong.
The water cycle _could_ require spending grid energy to filter/pump water into an economically usable state. Instead if water was better managed, we would not need to build additional grid capacity for water management.
Your argument basically boils down to "If energy was unlimited, we could be wasteful!", which, again, is technically true, but ignores the economic reality.
From what I can glean from Google, the sun moves 1500 cubic kilometers of water from the ocean into the air every day, around 500,000 cubic kilometers a year (ie, a stupendous amount).
Apparently around 10% of that makes it up the various mountains and comes back down as rivers - that's 50,000 cubic kilometers.
And for scale, human "consumption" is 5000 cubic kilometers.
I agree we should be careful and intelligent about how we use water and where we get it from, but I fail to be alarmed.
Every degree of global warming raises the amount of water the air can hold by 7%. That's what's going on in California recently. We only need to put our finger on the scale to really fuck things up. We don't have to stand on it.
Also heat island effect. We don't have to move the needle in Yosemite to make downtown LA into a death trap.
What's your tidy "Me worry?" explanation for aquifer depletion?
Water used for nuclear reactor cooling can only be returned to the environment if its temperature is within 0.5 deg F of the local source temperature. I live near a facility that is on the river with several man made cooling lakes. During the winter, there is constant fog and ice by the roads. So much so, that the road to the facility itself has covered bridge crossing one of the lakes.
During drought, the capacity of the plant is reduced due to lack of cooling capacity.
And remember, the reactor is used to generate high pressure steam which produces electricity, hot water and low grade steam. Even with high efficiency gas turbines and heat integration, there is a significant amount of steam that needs to be condensed before it can be feed back into the reactor.
If it's the same molecule but downhill and mixed in with some other ones, it's just x number of joules and y number of dollars' worth of infrastructure away from being among its own kind and uphill from your tap again.
We get blasted with an uncountable number of these joules from above (the sun) and below (nuclear). Our economy is generating an exponentially increasing number of dollars.
I understand wanting to be careful with resources, but not to the point where frugality becomes a goal in and of itself.
That's like saying fossil fuels don't actually pollute or emit greenhouse cases, because we're just X joules away from sequestering it back from the atmosphere.
Desalination, and pumping water over thousands of miles is extremely expensive. Sure, you're not wrong, but the values of X and Y are uneconomical.
I don't think they're uneconomical. Fresh, clean water is astonishingly cheap; of course people are using it to grow almonds and alfalfa in the desert.
Just charge people what the water is worth and they'll stop, or water companies will be able to afford much more treatment capacity.
You have a point about sequestering CO2 molecules, but:
a) I'm sure this will get cheaper over time, just like every other technology
b) we should be using solar and nuclear for everything
People grow almonds in the desert specifically because they have access to artificially cheap water. In the U.S. lots of land comes with water rights: e.g. if a river or creek passes through your land you can use x% of the water to irregate your crops. Some of these water rights date back to the 1800s and they're locked in.
The water rights can be clawed back a couple ways: if they're unused for X years, or in times of drought.
There's an exception for droughts though: farmers with trees (that would die if unwatered) still get priority, while people that grow crops that replenish each season (like wheat) don't.
So this leads to perverse incentives where these farmers need to find a way to use ALL of their water, every year, or they'll lose access to their absurd water rights from the 1800s, and they need to use it on trees so it doesn't get clawed back during a multi year drought.
So, they end up planting the most water-hungry trees they can grow on their land (almonds), then they get to sell them to the world at artificially low prices because the water that was used to grow them is almost free.
Because you can find it in "concentrated" form (think entropy), all in an aquifer or a river, and these are everywhere. But these dry up because of our usage and the climate, and when they do you still have the same amount of water on the planet, it's just not as easily accessible. It's super spread out, it's too far away, it needs a lot of expensive processing to make usable, or all of the above.
What's cheaper and easier for you, to condense a cup of water from the air or to just turn on the faucet?
> we should be using solar and nuclear for everything
Why solar? Energy is not lost/consumed in the universe, so why not collect it from anywhere else. Energy is astonishingly cheap, that's why we use so much of it. If you know what I mean...
While I do agree the hysteria around water use is unfounded, it's just patently false to say that fresh water cannot be wasted, pointing out that the molecule is just in a harder to access state is pedantry.
But the reason I hammer on about this point is precisely due to the hysteria. In the popular imagination, we spray x million gallons of water onto a golf course, and it just evaporates, never to be seen again. It is the alarmism that alarms me.
True. However, substitution of one good for another, or bringing online another source, won’t be instantaneous and thus otherwise needless human suffering will occur. The raw numbers don’t capture that.
Here in Michigan, my water price is also about 5% of my electric bill. Which is also small, we barely used the AC this summer.
Water billing here is (frustratingly) not progressive: the first thousand gallons costs the same as the tenth or hundredth thousand gallons. It's cheap, we're surrounded by fresh water on the surface and you can stick a well down through 80-100 feet of glacial sand and gravel and get drinkable water basically anywhere.
I was surprised to learn that 70% of my township's municipal water is used by only 15% of the households: basically, those that irrigate their lawns daily.
>I'm in the American southwest (but not arid desert like parts of Arizona/Nevada/Utah)
Doesn’t matter whether you are in the desert or not, only matters if you are in a shared watershed with them. There is huge agricultural demand for water and water rights in those areas which translates to high prices for the areas where they can source water (like your presumably more-watered location)
I think the water-usage stuff regarding data centers is really lacking context in online discourse – and yet, I still believe that freshwater usage really needs to be more of a concern for people, generally. I'm not 'anti-AI' but, I cringe a bit every time someone dismissively says "water cycle" to dismiss concerns around freshwater because, some aquifers are not going to recharge in a meaningful timeframe. That water isn't 'destroyed' – but if a town is tight on water already, it's not necessarily coming back, practically speaking.
I would like to know how much water is taken by a datacenter vs. the same size space of apartments. I can see why it could be considered a bad choice for communities long term if a datacenter takes more.
I’ve driven through The Dalles. It’s a very small town. A search shows a population of 15,000 and declining annually.
It’s also right on a big river. The article you linked said that Google was spending nearly $30 million to improve the city’s water infrastructure so there are no problems.
Talking about this in terms of percentages of a small town’s water supply while ignoring the fact that the city is literally on a giant river and Google is paying for the water infrastructure is misleading.
That's because it's a large industry and nobody lives there. This pattern appears all over the place. The paper mills in the pacific northwest consume large multiples of the water used by their little towns.
That's not the point, the question was whether an apartment building would use the same amount of water and clearly an apartment would consume substantially less water.
No, the question was whether "the same size space of apartments" (i.e. apartment buildings occupying the same land area as the datacenter) would use more or less water than the datacenter.
Under reasonable assumptions, the apartments would use more water.
- Google's datacenter complex in the Dalles covers ~190 acres.
- Typical density for apartment buildings is 50 units/acre, meaning you'd have 9,500 units on 190 acres.
- Average household size in the US is 2.5, so the 9,500 units would have a population of 23,750.
- According to the original article, per capita domestic water usage in the U.S. is 82 gallons per day, meaning a total water consumption of 710M gal/yr for the apartments. And this doesn't count the substantial indirect water usage you'd need to support this population.
- The Google datacenter uses 355M gal/yr (per the Oregonian article).
- 710M > 355M
Now, it would be somewhat ridiculous to replace the entire Google datacenter with apartment buildings in a rural town with declining population, but that was the original question...
If you replace the area of that data center with apartments, as the question suggested, it would add half again to the local population, which could indeed use 30% of the city water.
I'm not understanding the logic. You want to add more population to the city? That doesn't seem fair but I'll concede I may not understand the point you're trying to make.
Assuming that the population is the same in the city and you just move residents into an apartment complex. I don't understand how you would get the same water consumption, am I missing something? Evaporative cooling is extremely water heavy and these facilities also have the normal HVAC you'd expect. Everything just seems to point to more water usage not less.
Some quick napkin math using averages (data center designs vary). One of Google's larger and thirstier data centers, in Oklahoma, is said to use 833 million gallons per year (that's about 2500 acre-feet, in useful terms). It occupies about 250 acres, most of which looks to be parking lots but whatever. The number of households that can be supported on 1 acre-foot per year ranges from 2 to 6 depending (Las Vegas on one end, San Francisco on the other).
You said apartments specifically and this urban form usually starts at 50 dwellings per acre, minimum, which would lead me to say the apartments use more water. The break-even point in this equation is 2-5 households per acre.
With no AC and gas hot water, my monthly water bill is ~150% of my electric (that water cost is not including the wastewater that is billed on the water metering).
My water usage is pretty average and my electric usage is apparently hilariously low.
I track my water usage and electricity usage every month. I'm confused why the cost ratio is off by an order of magnitude from the author. The base monthly charge of my water bill ignoring any usage is more then 10% of my largest electricity bill (so maybe that's the answer right there).
'People sometimes invoke the idea that water moves through a cycle and never really gets destroyed, in order to suggest that we don’t need to be concerned at all about water use. But while water may not get destroyed, it can get “used up” in the sense that it becomes infeasible or uneconomic to access it.'
Side note, this personal anecdote from the author caught me off guard: "my monthly water bill is roughly 5% of the cost of my monthly electricity bill". I'm in the American southwest (but not arid desert like parts of Arizona/Nevada/Utah), and my monthly water cost averages out annually to ~60% of the cost of electricity. Makes me wonder if my water prices are high, if my electricity prices are low, if my water usage is high or my electricity usage is low.