Good for them. No groundwater for data centers! Data centers should be run closed loop, if they can't, they shouldn't exist. AZ should also make them generate at least 70% of their power onsite.
I think the most cursed aspect in all of this is the fact that regions where evaporative coolers work best tend to have the scarcest water resources. Evaporative cooling is unreasonably effective when water is happy to evaporate. You can cut the power consumption of a datacenter by 70-80% in these regions if you are willing to sacrifice water.
I think in this case, solar energy is the best way to run the cooling systems, even if it requires an absurd amount of power to exercise compressors, etc. in order to improve heat concentration and rejection efficiency. As long as it's all green and theres a disaster recovery plan, who cares?
If you're doing large scale evap cooling in the Sonoran you're just removing water from an already stressed water table. I'd say put them somewhere cold and do district heating during the winter.
Does district heating actually work for heat from data centers? The exhaust temperature of computer equipment is only slightly above room temperature and the structures in need of heat would only be at slightly below room temperature when calling for heat. Heat distribution systems with tiny differentials like that tend to have poor efficiency, i.e. you're going to have consume a lot of energy on fans/pumps.
Meanwhile in a cold climate you can do cooling by just blowing outside air through a filter, so the alternative in those climates is that rather than running a compressor.
That kind of district heating already exists and it uses heat pumps to raise the temps. Basically you run a huge AC somewhere and dump the hot side not into air but into district heating.
I'm not a geologist but...does evaporation remove water from the water table? Where I live, my understanding is the answer is "no" because the vast majority of rain water is deposited in our groundwater. Things like agriculture are dangerous because the answer is "yes" - you're literally shopping the water away as an agricultural product.
Another dumb question: why are we building projects that need tons of power and water in the Sonoran desert instead of next to the Great Lakes
Depends on the aquifer. The assumption you're making is "natural precipitation recharges aquifers" and that assumption is not always true.
Sometimes it's true, but it takes 10,000 years so if you mean "recharges while our current civilization exists" then it's effectively false. This is the case for the Ogallala aquifer which supplies 27% of the irrigated land in the US.
Sometimes it's just completely false because the groundwater exists as a sealed bubble of water put there during the last ice age. That glass of water you just drank? Congratulations! It's 25,000 years old, and we know this through isotopic analysis. Your drinking water was mined and once it's gone it's gone forever.
So it's complicated. Generally speaking, deeper groundwater sources take longer to recharge if they recharge at all. I don't know the particular groundwater situation around Tucson.
> why are we building projects that need tons of power and water in the Sonoran desert instead of next to the Great Lakes
Evaporative cooling is effective and cheap where humidity is low. It doesn't work well where humidity is high.
It at least removes water from the local water table.
You pump water up from the local water table to run your evaporative cooler. The water evaporates. But the air was at 10-20% humidity. The water from the evaporative cooler will raise the humidity, a little bit, but not enough to make it rain. It may make it more likely to rain somewhere downwind a few miles, or a few hundred, but not here.
For your second dumb question: At least some of the Great Lakes have at times had issues with water level. (They want enough to allow ships to pass between the lakes.) The upside is, there the humidity is high enough that you're more likely to get the water back in the form of increased precipitation.
What should water be used for then? They do large-scale agriculture in the same area, which uses several orders of magnitude more water while generating substantially less value.
The water should be conserved to address long term water security in the face of a decades long drought and ongoing Tier 1 shortage. Tucson exists in an Active Management Area that heavily regulates expanded water usage for agriculture beyond existing grandfathered rights.
If a datacenter is a more valuable proposition (it may be?) then should have the ability to acquire and redirect existing resources being used for agriculture.
> Food that people and livestock eat is less valuable than data centers
This is objectively false, particularly when we consider that much of that food and water are exported. It’s also irrelevant in Tucson, which doesn’t have Central Valley syndrome.
You... do understand that it isn't exported to be incinerated, it's exported in return for money, and then people and livestock in other countries eat that food. The fact that there's a transaction in the middle of the process doesn't change the value proposition, just like the absurd overvaluation of all things AI doesn't change the value proposition there either.
We need food production as a matter of survival. That doesn’t mean all food production is inherently more valuable than everything else. We let most food spoil without being eaten because it’s more efficient to do that than treat every calorie as precious.
> just like the absurd overvaluation of all things AI doesn't change the value proposition there either
Literally does.
There are good arguments against a data centre in Tucson. “We could grow food with that water” isn’t one of them.
I also wonder why this isn't the go-to if you want to work in water-stressed places. Become a water distillation plant at the same time, have a win-win solution.
Visited NOVA recently and after Loudon let them pass under the radar(right by use) for as long as they have, they've recently to tried to put the kibosh on it, or at least slow it down.
What was once a rural county outside of Fairfax, Loudon has seen tremendous development thanks to data center expansion. Now Prince William County( one county over, westward) is trying to fight the same battle before it gets too big of a problem for them.
I’ve lived in Loudoun County for twenty-four years, so I’ve seen much of the development firsthand. The data center expansion, particularly in Ashburn and Sterling, has been amazing to watch.
However, I’m puzzled by this portion of your comment: ‘…has seen tremendous development thanks to data center expansion.’
The major residential growth in Ashburn, Broadlands, Sterling, Lansdowne, and surrounding areas significantly predates the data center boom. These communities were already experiencing rapid expansion and large influxes of new residents well before the data center industry took off here.
> The data center has been under discussion by local officials since at least 2023, according to emails between county administrators and real estate investment firm Diamond Ventures. The county has been under a non-disclosure agreement since at least June 2024.
That is insane. An NDA that the county was forced to sign? No wonder residents were outraged. I’m surprised this is legal.
> Update: Brendan Gallagher, Beale's SVP of development, said in a statement: “We are disappointed in the decision not to pursue this opportunity for Tucson. [...] We see it as a missed opportunity for the city, as this project potentially represents tens of millions of dollars in tax revenue, hundreds of millions of dollars in infrastructure to serve the community, and thousands of high-paying local and union jobs.”
and I'm "potentially" the next president of the U.S.
They would have to build a road to datacenter, datacenters themselves, all GPUs etc put together. That requires "thousands of high-paying local and union jobs" for short time. "hundreds of millions of dollars in infrastructure". Then a dozen of armed guards with dogs.
Good for them. No groundwater for data centers! Data centers should be run closed loop, if they can't, they shouldn't exist. AZ should also make them generate at least 70% of their power onsite.
I think the most cursed aspect in all of this is the fact that regions where evaporative coolers work best tend to have the scarcest water resources. Evaporative cooling is unreasonably effective when water is happy to evaporate. You can cut the power consumption of a datacenter by 70-80% in these regions if you are willing to sacrifice water.
I think in this case, solar energy is the best way to run the cooling systems, even if it requires an absurd amount of power to exercise compressors, etc. in order to improve heat concentration and rejection efficiency. As long as it's all green and theres a disaster recovery plan, who cares?
Arizona has the perfect conditions for solar energy (sunny 99% of the time), they really should have taken advantage of that.
If you're doing large scale evap cooling in the Sonoran you're just removing water from an already stressed water table. I'd say put them somewhere cold and do district heating during the winter.
Does district heating actually work for heat from data centers? The exhaust temperature of computer equipment is only slightly above room temperature and the structures in need of heat would only be at slightly below room temperature when calling for heat. Heat distribution systems with tiny differentials like that tend to have poor efficiency, i.e. you're going to have consume a lot of energy on fans/pumps.
Meanwhile in a cold climate you can do cooling by just blowing outside air through a filter, so the alternative in those climates is that rather than running a compressor.
That kind of district heating already exists and it uses heat pumps to raise the temps. Basically you run a huge AC somewhere and dump the hot side not into air but into district heating.
I'm not a geologist but...does evaporation remove water from the water table? Where I live, my understanding is the answer is "no" because the vast majority of rain water is deposited in our groundwater. Things like agriculture are dangerous because the answer is "yes" - you're literally shopping the water away as an agricultural product.
Another dumb question: why are we building projects that need tons of power and water in the Sonoran desert instead of next to the Great Lakes
Depends on the aquifer. The assumption you're making is "natural precipitation recharges aquifers" and that assumption is not always true.
Sometimes it's true, but it takes 10,000 years so if you mean "recharges while our current civilization exists" then it's effectively false. This is the case for the Ogallala aquifer which supplies 27% of the irrigated land in the US.
https://en.m.wikipedia.org/wiki/Ogallala_Aquifer
Sometimes it's just completely false because the groundwater exists as a sealed bubble of water put there during the last ice age. That glass of water you just drank? Congratulations! It's 25,000 years old, and we know this through isotopic analysis. Your drinking water was mined and once it's gone it's gone forever.
https://en.wikipedia.org/wiki/Fossil_water
Sometimes recharge can happen but only via snowmelt at high altitudes (and of course Tucson doesn't get much snow).
Sometimes it's completely true. The Edwards aquifer in Texas gets recharged every time it rains.
https://en.m.wikipedia.org/wiki/Edwards_Aquifer
So it's complicated. Generally speaking, deeper groundwater sources take longer to recharge if they recharge at all. I don't know the particular groundwater situation around Tucson.
> why are we building projects that need tons of power and water in the Sonoran desert instead of next to the Great Lakes
Evaporative cooling is effective and cheap where humidity is low. It doesn't work well where humidity is high.
It at least removes water from the local water table.
You pump water up from the local water table to run your evaporative cooler. The water evaporates. But the air was at 10-20% humidity. The water from the evaporative cooler will raise the humidity, a little bit, but not enough to make it rain. It may make it more likely to rain somewhere downwind a few miles, or a few hundred, but not here.
For your second dumb question: At least some of the Great Lakes have at times had issues with water level. (They want enough to allow ships to pass between the lakes.) The upside is, there the humidity is high enough that you're more likely to get the water back in the form of increased precipitation.
Isn't effective evaporative cooling a direct function of low water availability i the environment?
What should water be used for then? They do large-scale agriculture in the same area, which uses several orders of magnitude more water while generating substantially less value.
The water should be conserved to address long term water security in the face of a decades long drought and ongoing Tier 1 shortage. Tucson exists in an Active Management Area that heavily regulates expanded water usage for agriculture beyond existing grandfathered rights.
If a datacenter is a more valuable proposition (it may be?) then should have the ability to acquire and redirect existing resources being used for agriculture.
Not near Tucson they don't.
Food that people and livestock eat is less valuable than data centers used to chase the "AGI any day now" pipe dream? You're kidding right?
> Food that people and livestock eat is less valuable than data centers
This is objectively false, particularly when we consider that much of that food and water are exported. It’s also irrelevant in Tucson, which doesn’t have Central Valley syndrome.
You... do understand that it isn't exported to be incinerated, it's exported in return for money, and then people and livestock in other countries eat that food. The fact that there's a transaction in the middle of the process doesn't change the value proposition, just like the absurd overvaluation of all things AI doesn't change the value proposition there either.
> it's exported in return for money
Which is also paid for data centre access.
We need food production as a matter of survival. That doesn’t mean all food production is inherently more valuable than everything else. We let most food spoil without being eaten because it’s more efficient to do that than treat every calorie as precious.
> just like the absurd overvaluation of all things AI doesn't change the value proposition there either
Literally does.
There are good arguments against a data centre in Tucson. “We could grow food with that water” isn’t one of them.
Let the people eat and drink slop
* AI slop
It was going to use 1% of the available recycled water. That's about the equivalent usage of a medium size farm.
per year? Day? Month? Microsecond?
the available water as volume per time is already in the correct units.
Please apply this rule to Tucson's 40 golf courses.
Most Tucson residents undoubtedly agree. The golf courses are almost exclusively used by rich out-of-town visitors
But where will the poor elites go hang out then?
They can bugger off to Phoenix and get the same type of weather, landscape, etc.
They won't be as high of altitude though so their shots won't go as far. Too bad.
> No groundwater for data centers!
It's worst. They use tap water.
"...For the purposes of cooling, data centres mainly use potable water (water that is safe to drink or use for cooking, free from harmful contaminants). .." - https://www.twobirds.com/en/insights/2025/cooling-the-cloud-...
This particular data center was going to build a wastewater pipeline to cool their facility with currently-squandered wastewater.
I also wonder why this isn't the go-to if you want to work in water-stressed places. Become a water distillation plant at the same time, have a win-win solution.
Visited NOVA recently and after Loudon let them pass under the radar(right by use) for as long as they have, they've recently to tried to put the kibosh on it, or at least slow it down.
https://virginiabusiness.com/loudoun-county-supervisors-shak...
What was once a rural county outside of Fairfax, Loudon has seen tremendous development thanks to data center expansion. Now Prince William County( one county over, westward) is trying to fight the same battle before it gets too big of a problem for them.
I’ve lived in Loudoun County for twenty-four years, so I’ve seen much of the development firsthand. The data center expansion, particularly in Ashburn and Sterling, has been amazing to watch.
However, I’m puzzled by this portion of your comment: ‘…has seen tremendous development thanks to data center expansion.’
The major residential growth in Ashburn, Broadlands, Sterling, Lansdowne, and surrounding areas significantly predates the data center boom. These communities were already experiencing rapid expansion and large influxes of new residents well before the data center industry took off here.
Yeah, it was only farms in 1985. Now it's 95% residential sprawl and 5% data centers.
> The data center has been under discussion by local officials since at least 2023, according to emails between county administrators and real estate investment firm Diamond Ventures. The county has been under a non-disclosure agreement since at least June 2024.
That is insane. An NDA that the county was forced to sign? No wonder residents were outraged. I’m surprised this is legal.
> Update: Brendan Gallagher, Beale's SVP of development, said in a statement: “We are disappointed in the decision not to pursue this opportunity for Tucson. [...] We see it as a missed opportunity for the city, as this project potentially represents tens of millions of dollars in tax revenue, hundreds of millions of dollars in infrastructure to serve the community, and thousands of high-paying local and union jobs.”
and I'm "potentially" the next president of the U.S.
They would have to build a road to datacenter, datacenters themselves, all GPUs etc put together. That requires "thousands of high-paying local and union jobs" for short time. "hundreds of millions of dollars in infrastructure". Then a dozen of armed guards with dogs.
Datacenters do not employ thousands of jobs. They employ about 5.
Long term yes. Short term thousands.
and those definitely aren't union jobs.