Here's the link:

https://www.theguardian.com/us-news/2026/may/13/utah-approves-datacenter-backlash

A plan to create one of the world’s largest datacenters, a gargantuan project spanning an area more than twice the size of Manhattan, has provoked a furious public backlash in Utah amid concerns over its vast energy use and impact upon the state’s stressed water supplies.

The Stratos artificial intelligence datacenter footprint will cover more than 40,000 acres (62 sq miles) over three sites in Box Elder county in north-western Utah. The facility will require about 9GW of power, which is more than the entire state of Utah currently consumes, and suck up a significant amount of water in an area that has been hit by severe drought in recent years....(end clip)

This is a project with so much environmental impact, I have to wonder if there aren't ways to better engineer the sourcing of power and coolant. And if a baking desert is really a suitable site location for something so critically dependent on cooling and water. The present plan seems almost like the engineering equivalent of trolling. Also, why not at least take advantage of the solar efficiency of a desert? But no, they're going to put in gas turbines. It's this kind of idiocy which makes the US currently look so...idiotic in the eyes of other developed countries.

1 hour ago, TheVat said:

This is a project with so much environmental impact, I have to wonder if there aren't ways to better engineer the sourcing of power and coolant. And if a baking desert is really a suitable site location for something so critically dependent on cooling and water. The present plan seems almost like the engineering equivalent of trolling. Also, why not at least take advantage of the solar efficiency of a desert? But no, they're going to put in gas turbines. It's this kind of idiocy which makes the US currently look so...idiotic in the eyes of other developed countries.

They are not only popping up in the US. Canada and the EU is working hard on the idea of AI sovereignty (which would be the charitable way to to describe it- it is to me not clear what kind of agreements are in place to actually enforce it properly). Most that I am aware of will rely on LNG due to perceived flexibility and faster implementation. Some have proposed closed-loop cooling systems, but I have not seen details how exactly that works. The one in Utah will have a much more massive impact. However, given the rate they are popping, it is hard to overlook that the mid-sized ones will likely also have considerable issues.

That being said, a positive element is that some of the data centers will connect their large capacity to the grid, and the promise here is that they can easily scale use up and down and either consume overcapacity or feed into the grid when needed, thus providing more stability.

If they can figure out some way of cooling besides water or not requiring it, the project could be worthwhile IMO but is not the only environmental issue. Providing power is a relatively trivial problem compared to the water usage issue. More solar panels, windmills, turbines or whatever can be built for power but getting sustainable sources of water in the desert has always been a difficult proposition.

1 hour ago, CharonY said:

They are not only popping up in the US. Canada and the EU is working hard on the idea of AI sovereignty (which would be the charitable way to to describe it- it is to me not clear what kind of agreements are in place to actually enforce it properly). Most that I am aware of will rely on LNG due to perceived flexibility and faster implementation. Some have proposed closed-loop cooling systems, but I have not seen details how exactly that works. The one in Utah will have a much more massive impact. However, given the rate they are popping, it is hard to overlook that the mid-sized ones will likely also have considerable issues.

That being said, a positive element is that some of the data centers will connect their large capacity to the grid, and the promise here is that they can easily scale use up and down and either consume overcapacity or feed into the grid when needed, thus providing more stability.

Why does water for cooling have to be a problem? Surely any system can run with closed loop cooling, provided you build adequate radiators? And in Utah why can’t they build their own solar farm to power the thing? I cam’t see why they should be allowed to be parasites on the infrastructure built for the citizens.

Edited May 13May 13 by exchemist

So the argument against solar was from what I understand the ability to scale production up and down as needed. Solar was considered to be too tricky. But I have not seen really a quantitative argument. My guess is that there are more considerations and up-front costs that they would rather avoid, but this is only my gut feeling. I am sure there are some studies/report on cost/benefit somewehre.

The closed loop argument is also a bit tricky, and I only skimmed some of the issues, so am not really knowledgeable on that front, either. But from what I understand, closed-loop system cost more energy to run, so depending on how the power gets in, the ecological costs are moved further upstream. Otoh there are ideas of also using the heat e.g., for greenhouses and other buildings. I am pretty sure that folks have or are currently doing heavy calculations on the overall burden (carbon emission, water usage etc.). Ultimately, someone has to pay the price.

Data centers and AI are big business and big profits for Billionaires and investors.
They can afford to pay the high prices for power, and in so doing,, they drive up energy costs, or cause draughts, for the idiots who buy into this technology to have it think for them.
We are our own worst enemies, and there are plenty of unscrupulous people willing to take advantage of, and get insanely rich off of, us stupid people.

With companies it comes down to relative costs. Open loop is cheap (pipe from the river, dump hot water back in the river, the hell with riparian life), and there's some evaporative loss. Closed loop is expensive, you need glycol and inhibitors in the waters, cooling towers or other means of dumping the heat, and any use of dumped heat is even more expensive. But water use is minimal. I'll guess that closed energy loops are even more expensive (like fairy tales at this stage), where you reclaim waste heat to power turbines which feed joules back to the chips (With whatever mechanical losses that would involve). There's also thermogalvanic batteries being developed, which seem like a potential solution down the road. I still think companies need to think longterm and use solar in the desert, reaping the longterm benefit of not paying to pipe in NG and thus generating with fewer moving parts

Edited May 14May 14 by TheVat

4 hours ago, MigL said:

Data centers and AI are big business and big profits for Billionaires and investors.
They can afford to pay the high prices for power, and in so doing,, they drive up energy costs, or cause draughts, for the idiots who buy into this technology to have it think for them.
We are our own worst enemies, and there are plenty of unscrupulous people willing to take advantage of, and get insanely rich off of, us stupid people.

There also seems to be a gold rush situation where communities are trying to draw in data centers by providing tax benefits and other things. It does seem a bit short-sighted to me, as I suspect it is not clear how much money they are going to ultimately bring to those communities (beyond the investment in building the infrastructure).

9 hours ago, CharonY said:

There also seems to be a gold rush situation where communities are trying to draw in data centers by providing tax benefits and other things. It does seem a bit short-sighted to me, as I suspect it is not clear how much money they are going to ultimately bring to those communities (beyond the investment in building the infrastructure).

Very little from cases I've seen. Huge buildings with three employees on-site.

I have SO many reservations about AI. I've always lamented our tendency to use our brains to make us stronger and more capable of aggression instead of just focusing on being smarter, which would seem to be our strong suit.

14 hours ago, TheVat said:

With companies it comes down to relative costs. Open loop is cheap (pipe from the river, dump hot water back in the river, the hell with riparian life), and there's some evaporative loss. Closed loop is expensive, you need glycol and inhibitors in the waters, cooling towers or other means of dumping the heat, and any use of dumped heat is even more expensive. But water use is minimal. I'll guess that closed energy loops are even more expensive (like fairy tales at this stage), where you reclaim waste heat to power turbines which feed joules back to the chips (With whatever mechanical losses that would involve). There's also thermogalvanic batteries being developed, which seem like a potential solution down the road. I still think companies need to think longterm and use solar in the desert, reaping the longterm benefit of not paying to pipe in NG and thus generating with fewer moving parts

Seeing as the AI companies make billions - and forecast they will make billions more from AI investments - I should think they should be required to make their facilities environmentally neutral and not a net drain on public utilities. If that makes them look harder at their return on investment calculation, that’s fine too.

On 5/13/2026 at 6:07 PM, exchemist said:

Why does water for cooling have to be a problem? Surely any system can run with closed loop cooling, provided you build adequate radiators?

From what I have read, somewhere around 20C or less is where operation of a data center begins being affected negatively so you have to have a sink that's cooler than 20C. Unfortunately, it is often well above that temperature in the Utah desert, especially in the summer even at 1500+ meters like much of Box Elder County. Other than a water source, it seems to me one would have to use artificial means (A/C) to get adequate cooling. this would substantially add to the center's already large amount of energy use. The problem isn't intractable but all of the current solutions push construction costs to as much as several times that of just being able to use all of the available water.

50 minutes ago, npts2020 said:

From what I have read, somewhere around 20C or less is where operation of a data center begins being affected negatively so you have to have a sink that's cooler than 20C. Unfortunately, it is often well above that temperature in the Utah desert, especially in the summer even at 1500+ meters like much of Box Elder County. Other than a water source, it seems to me one would have to use artificial means (A/C) to get adequate cooling. this would substantially add to the center's already large amount of energy use. The problem isn't intractable but all of the current solutions push construction costs to as much as several times that of just being able to use all of the available water.

Then Utah is a damned silly place to build the thing.

11 hours ago, exchemist said:

Then Utah is a damned silly place to build the thing.

Pretty much.

On 5/14/2026 at 1:54 AM, TheVat said:

Open loop is cheap (pipe from the river, dump hot water back in the river, the hell with riparian life), and there's some evaporative loss.

Not that cheap even where environmental protection laws are lax. Lots of civil works on the intakes, buffer storage, and outfalls, and the water treatment costs (filtration, biological control, corrosion control etc.) are much higher than closed loop systems. From what I could glean from the Guardian article, (and referenced impact study), no one is seriously proposing this.

On 5/14/2026 at 1:54 AM, TheVat said:

Closed loop is expensive, you need glycol and inhibitors in the waters, cooling towers or other means of dumping the heat, and any use of dumped heat is even more expensive. But water use is minimal.

I don''t disagree entirely. But the situation is more complex than this, particularly if @npts2020 's information regarding the temperature dump shortfall (ambient wet bulb > target temperature of 20^oC) is correct. This would pretty much enforce some degree of refrigeration cycle in the mix.

I've attached a manufacturer's semi-technical brochure to give some indication of the current state of play in datacentre cooling system architecture. (In principle, not so different to other modern sectors). The trend seems to be moving away from water based systems where feasible.

Araner_Whitepaper_DataCenter.pdf

It may turn out that the solution is computers that operate at higher temps but that is no trivial task at present and could turn out to be impossible to change to the degree required to significantly reduce the amount of cooling needed for operation.

8 hours ago, npts2020 said:

It may turn out that the solution is computers that operate at higher temps but that is no trivial task at present

Is it? My gear never sees a room temperature below 25⁰C, even at night. The only unit that complains now and then is my old phone, but that's when the body is hand hot. Are you really sure that individual components are prone to failure at below say 45^oC?

Your 20⁰C isn't just a recommended AC setting for a server room is it? These numbers make a huge difference to what is feasible in large scale installations.

On 5/15/2026 at 7:33 AM, exchemist said:

Then Utah is a damned silly place to build the thing.

Actually, thinking about it, I can see why 'high and dry' might appeal. But that's based on an experience of condensation and trace sulphides being the main environmental causes of failures in electronic equipment.

16 hours ago, sethoflagos said:

This would pretty much enforce some degree of refrigeration cycle in the mix.

I've attached a manufacturer's semi-technical brochure to give some indication of the current state of play

Thanks. And yes, I meant "or other means" as including refrigeration. In places like W VA where many datacenters have popped up, that's led to some serious noise issues with adjacent properties. As I recall, the chillers make quite a racket 24/7. The industry has been called into court on noise complaints - likely that's part of the shift towards very remote areas.

13 hours ago, sethoflagos said:

Is it? My gear never sees a room temperature below 25⁰C, even at night. The only unit that complains now and then is my old phone, but that's when the body is hand hot. Are you really sure that individual components are prone to failure at below say 45^oC?

Your 20⁰C isn't just a recommended AC setting for a server room is it? These numbers make a huge difference to what is feasible in large scale installations.

Actually, thinking about it, I can see why 'high and dry' might appeal. But that's based on an experience of condensation and trace sulphides being the main environmental causes of failures in electronic equipment.

Ambient temperature is not the temperature at which an electronic device operates, it is always much warmer. Also, the temps at which components fail is usually significantly higher (mostly in the 150C+ range) than where their operation is being affected by things like slowing computing speeds and higher energy usage. Odds are, the difference isn't significant enough for you to notice on your devices but a data center would probably notice even a percent or two change. 20C might be the recommended AC setting but it is also where most electronic devices begin to have their operation affected so engineers have designed workarounds that allow fairly efficient operation to significantly higher temps.

46 minutes ago, npts2020 said:

Odds are, the difference isn't significant enough for you to notice on your devices but a data center would probably notice even a percent or two change. 20C might be the recommended AC setting but it is also where most electronic devices begin to have their operation affected so engineers have designed workarounds that allow fairly efficient operation to significantly higher temps.

I get it that there is an inverse relation between temperature and electron speed through a logic gate, and that processor clock speeds are selected according (in part) to this constraint.

Admittedly, I've not done any machine code programming since 1MHz was a fair clock rate, but Google's AI is telling me that current generation processors don't start clock speed temperature throttling until chip temperature reaches 90-100^oC suggesting that ambient air cooling should be viable in any climate providing heat transfer efficiency is appropriately dealt with at cabinet, rack, circuit board, and chip level, as an integrated part of the design.

Of course, if these aspects are not being adequately addressed by the manufacturers, then datacentres would be obliged to crudely dial down the temperature of their data hall ventilation and rely on diffusion to do the business.

That's the compromise available, and I guess reality must lie somewhere between these extremes. Somewhere no doubt there will be a bunch of accountants making decisions on topics they are ill-equipped to comprehend.

As one of the most staunchly Republican US states, how much would an 'anything but renewables' energy choice be a necessary condition for government support for such a massive project in Utah, despite of - even because of - so much energy in Utah already coming from solar and wind ? How much is the massive energy demand of AI seen as a spanner in the spokes of renewable energy growth - an opportunity to shift away from that course - by people who don't really know much or care about AI?

My cynicism - full disclosure here - runs very deep.

10 hours ago, sethoflagos said:

Somewhere no doubt there will be a bunch of accountants making decisions on topics they are ill-equipped to comprehend.

And CEO's who fully understand only one or two parts of the business/programming/engineering equation required.

ASHRAE has technical standards for cooling. This gives a general outline but the meat of it is behind a paywall.