AI Data Centers Can Warm Surrounding Areas By Up To 9.1C 71
An anonymous reader quotes a report from New Scientist: Andrea Marinoni at the University of Cambridge, UK, and his colleagues saw that the amount of energy needed to run a data centre had been steadily increasing of late and was likely to "explode" in the coming years, so wanted to quantify the impact. The researchers took satellite measurements of land surface temperatures over the past 20 years and cross-referenced them against the geographical coordinates of more than 8400 AI data centers. Recognizing that surface temperature could be affected by other factors, the researchers chose to focus their investigation on data centers located away from densely populated areas.
They discovered that land surface temperatures increased by an average of 2C (3.6F) in the months after an AI data center started operations. In the most extreme cases, the increase in temperature was 9.1C (16.4F). The effect wasn't limited to the immediate surroundings of the data centers: the team found increased temperatures up to 10 kilometers away. Seven kilometers away, there was only a 30 percent reduction in the intensity. "The results we had were quite surprising," says Marinoni. "This could become a huge problem."
Using population data, the researchers estimate that more than 340 million people live within 10 kilometers of data centers, so live in a place that is warmer than it would be if the data centre hadn't been built there. Marinoni says that areas including the Bajio region in Mexico and the Aragon province in Spain saw a 2C (3.6F) temperature increase in the 20 years between 2004 and 2024 that couldn't otherwise be explained. University of Bristol researcher Chris Preist said the findings may be more complicated than they look. "It would be worth doing follow-up research to understand to what extent it's the heat generated from computation versus the heat generated from the building itself," he says. For example, the building being heated by sunlight may be part of the effect.
The findings of the study, which has not yet been peer-reviewed, can be found on arXiv.
They discovered that land surface temperatures increased by an average of 2C (3.6F) in the months after an AI data center started operations. In the most extreme cases, the increase in temperature was 9.1C (16.4F). The effect wasn't limited to the immediate surroundings of the data centers: the team found increased temperatures up to 10 kilometers away. Seven kilometers away, there was only a 30 percent reduction in the intensity. "The results we had were quite surprising," says Marinoni. "This could become a huge problem."
Using population data, the researchers estimate that more than 340 million people live within 10 kilometers of data centers, so live in a place that is warmer than it would be if the data centre hadn't been built there. Marinoni says that areas including the Bajio region in Mexico and the Aragon province in Spain saw a 2C (3.6F) temperature increase in the 20 years between 2004 and 2024 that couldn't otherwise be explained. University of Bristol researcher Chris Preist said the findings may be more complicated than they look. "It would be worth doing follow-up research to understand to what extent it's the heat generated from computation versus the heat generated from the building itself," he says. For example, the building being heated by sunlight may be part of the effect.
The findings of the study, which has not yet been peer-reviewed, can be found on arXiv.
Farm pasture versus concrete buildings? (Score:4, Insightful)
Are they comparing farm pasture temperature readings versus temperature readings of concrete buildings and paved parking lots?
I dunno understand. (Score:2, Funny)
Can't they just open a window or something?
Re:I dunno understand. (Score:5, Funny)
Most data centers don't run windows. They run systemd.
Re: (Score:2)
I'm an old man. I yell at systemd.
7 KM away (Score:5, Insightful)
> the team found increased temperatures up to 10 kilometers away. Seven kilometers away, there was only a 30 percent reduction in the intensity. "
This entire research should be about comparing data center produced heat on-site and nearby spill over to other high energy use/high heat production industries such as steel mills.
Re:7 KM away (Score:5, Interesting)
The comparison is always a cost/benefit analysis. Steel mills actually deliver a useful product, so the costs of their existence is justified. An "AI" datacenter produces hallucinations, its existence is completely unjustifiable.
Re: (Score:2)
Re: (Score:2)
Far from it, just not when they're coming from a datacenter.
Re: (Score:3)
An AI data center can replace a legion of human workers. So the heat emissions can be offset if those humans cease to exist.
Re: (Score:2)
Re: (Score:2)
A human operates at anywhere from 75-150W of energy. At 150W, a MW of energy is 6666 people. A rack in a datacenter right now consumes around 5-15kW, so a megawatt is around 100 racks.
AI datacenters are attempting to scale that up to 100kW per rack.
And none of that includes cooling and ancillary - this is just pure compute power consumption.
Re: (Score:2)
Username checks out.
Re: (Score:3)
Depends on the scale a steel plant may use a lot. The npp that was my first job was sending a full GW of power to a steel manufacturer back in the day. There's a lot more going on in a steel factory than just heating a kettle - there are many kettles, most produce something that has to be formed and annealed, which means heating it up and maintaining it so, plus there are the extra operations after it is done, etc.
But that notwithstanding, you're absolutely right, keeping stuff cool at scale does take crazy
cold storage (Score:2)
Do you have any sense of the temperature differential at different distances from the cold storage building?
Did it attract lots of birds during the winder?
Re: (Score:1)
> the team found increased temperatures up to 10 kilometers away. Seven kilometers away, there was only a 30 percent reduction in the intensity. "
This entire research should be about comparing data center produced heat on-site and nearby spill over to other high energy use/high heat production industries such as steel mills.
Why? So Vampire-AI can suck taxpayers dry abusing strawman WhatAbout excuses?
Steel mills deliver steel. A necessary resource for First World survival and expansion. AI delivers delusion and cats playing guitars. A resource only deemed necessary to the children entertained by it.
We should be focused on studying the effects of Before-AI to After-AI, including the considerable impact of humans becoming permanently unemployable. Not clamoring for strawman arguments to feed the distraction machine.
Re: (Score:2)
What "considerable impact?" Famous economist Adolph Hitler discovered that if you just shoot them, peoples' recurring maintenance cost fall to nearly zero.
This also (mostly) worked for the American south prior to 1860, which is where "Hitler and team" got many of their ideas from.
Though "Hitler and team" did come up with more efficient disposal methods for equipment that no longer worked.
AI is bound to be even more efficient [wikipedia.org] if anything that is not keeping it running and likely expanding is consuming resources it wants or needs.
Re:Farm pasture versus concrete buildings? (Score:5, Interesting)
From page 5:
"These results are dramatically impressive, especially considering that the typical LST increase caused by the quintessential example of compound of anthropogenic activities - the urban heat island effect - has been estimated in the 4 and 6 [degrees] C interval. This apparent step function emphasize the clear effect of AI hyperscalers on their surrounding areas, so much that it can match the impact of "islands" of higher temperatures: therefore, we call this the data heat island effect."
Re: (Score:1)
That doesn't at all answer the question that was asked.
Re:Farm pasture versus concrete buildings? (Score:5, Insightful)
Since concrete buildings and paved parking lots are part of the urban heat island effect, yes it does.
The question could be phrased more generally: how much of the *data* heat island effect is because it's a data center and not another type of building.
The answer, apparently: a lot.
Re: (Score:3)
I understood the question to be whether the study controlled for other changes in land use in the surrounding area. For example, northern Virginia has built a ton of new data centers close together over the last decade -- in many cases, replacing pastures. Attributing the results of the whole set to individual data centers would be a methodological error.
Re:Farm pasture versus concrete buildings? (Score:5, Informative)
From the summary:
"Recognizing that surface temperature could be affected by other factors, the researchers chose to focus their investigation on data centers located away from densely populated areas."
You'll have to look at the study data to see if that completely addresses your concern, but unsurprisingly the professional researchers have put some thought into what controls a study like this might need.
Re: (Score:2)
"away from densely populated areas" is both subjective and not the same as "away from other development".
This kind of article should wait until the study has been peer-reviewed, not flood the zone with unreviewed slop.
Re: Farm pasture versus concrete buildings? (Score:3)
Yes, it's a pre-print, but calling it "slop" is hardly warranted.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
One might then reasonably ask, "why are we wasting time on unreviewed findings?"
Re: (Score:2)
Are they comparing farm pasture temperature readings versus temperature readings of concrete buildings and paved parking lots?
The development around previously-agrarian weather sites into urban heat islands is already a big contributor to the rise in the temperature record attributed to anthropogenic CO2 emissions causing climate change. Now they'll just have a way to double the claimed harm -- the CO2 emissions from the power plants fueling the data centers, and the waste heat from the centers themselves sited on ex-farmland.
Re: (Score:2)
Are they comparing farm pasture temperature readings versus temperature readings of concrete buildings and paved parking lots?
No.
They're comparing "measurements over the last 20 years" to define Before-AI, to the end result we have now, regardless of what comprises the After-AI.
No point in even questioning if there is a measured difference between the natural jungle and the concrete jungles we create while destroying jungles. We've known that answer for decades before AI.
Re: (Score:2)
RTFS
They acknowledged that very issue.
We are (Score:2)
We're cooked, literally
Re: (Score:3)
This is why (Score:1)
Re: (Score:2, Informative)
Nah, he wants to distract from his legal issues. Naughty uncle touchy.
Re: (Score:1)
Dat some super-serial comeback, grampa. Well-reasoned, well-articulated and full of valid arguments.
Re: In a Word (Score:3)
The only person claiming this is about AGW is you. Poor reading comprehension, or intentional disingenuousness? Folks can come to their own conclusions.
Re: (Score:2)
It is possible that this paper wasn't intended to pass peer-review but was instead sent to arXiv to get media attention. Two basic proofreading failures in one sentence?
Re:In a Word (Score:5, Insightful)
If global warming is indeed a hoax, why can't any real 'murican researchers disprove it? You'd be hailed as a hero to the MAGA camp and might be allowed to kiss dear leaders ring. Everyone keeps coming to the same conclusion except for a few fringe nutjobs who start going off about angels and the bible.
Re: (Score:2)
global warming alarmists
So, please explain to me why, in detail, why Trump wants Greenland. Go deep into it: you know, what's happening at the Arctic that's prompting geopolitical maneuvers and the threat of Russia and China and whatnot in the region. Add an explanation for why that thing is happening. Is it a mystery? Does it have causes? Why is it happening now and not, let's say, 200 years ago? Are those causes, maybe, supernatural? No? Then come back and compare that explanation with what you just said. We'll wait.
Re: (Score:3)
Simple math would tell you dissipating millions of watts of energy results in heat.
Build greenhouses around them! (Score:4, Interesting)
Combined with the CO2 rich exhaust from the nearby power station and the warmed cooling water, the hotter greenhouses will enable excellent plant growth and productivity. If we need to put massive electric heaters out in the open, make the best of it!
Re: (Score:1)
Are you trolling? Many plants cannot tolerate warm weather.
Re: (Score:1)
Are you trolling? Many plants cannot tolerate warm weather.
Are you trolling? Don't you know that many of the places where these datacenters are being set up already have greenhouses for the cooler months? GP is suggesting that when life gives you lemons, you grow lemons during over winter.
Re: (Score:2)
Hey we might get some of that tariff free domestic coffee out of this deal.
Re: (Score:2)
Hey we might get some of that tariff free domestic coffee out of this deal.
Or US-grown bananas, as suggested by Lutnick June last year, given how banana producers were ripping-off America by not producing their bananas locally, thus the heavy tariffs against them. Once America is growing its own tropical fruits, it will stop being a Banana Republic in name only.
Re: (Score:2)
Growth and bulk productivity yes, for certain species of plant. Nutritional value no. Mineral uptake is impacted after a certain CO2 concentration which is why you need to get the soil right for growing in CO2 enriched greenhouses.
So not so good for feeding large human populations. Expect to pay large amounts of tax to deal with the impact of increased migration due to hunger.
Re: (Score:2)
Bullshit Metrics Per Usual (Score:4, Insightful)
"more than 340 million people live within 10 kilometers of data centers"
NOT ALL DATA CENTERS ARE FUCKING AI DATA CENTERS.
I'm getting sick n tired of all the bullshit metrics.
By the definition of "data center" that these people use in one stance (AI) to talk about heat, then they shift to a different definition to cover "people live X distance away) is for an entirely different class. Things like carrier hotels and co-locations. Ya'know, the shit that just delivers your fucking internet to your house and cross-connects to other ISPs / Carriers. THESE ARE DATA CENTERS! But they're not "gigawatt" data centers. Often they're not even "megawatt" data centers.
Re: (Score:2)
"more than 340 million people live within 10 kilometers of data centers"
NOT ALL DATA CENTERS ARE FUCKING AI DATA CENTERS.
Living proof right there that using AI to think for you makes you stupid. The article was about all data centers, not just AI.
9.1C (Score:1)
Thanks for using metric. I don't know why they're talking about temperature and heating when the units are clearly referring to charge. Anyway, 9.1 Coulombs isn't that much charge over an area that large, and they should look into using better wire insulation. Or conductors with a higher work function.
Alaska (Score:2)
The answer is obvious: build the AI data centers near the Arctic Circle. Granted, there are territorial disputes there, but AI can be sure to break some ice.
Re: (Score:1)
Wouldn't it be better to build them and lift them into orbit?
Re: (Score:2)
Not really. The vacuum of space is a great insulator.
Re: (Score:2)
Re: (Score:2)
You beat me to it. I was going to suggest building them on the northern coasts of Canada, Alaska, Norway, Greenland and Iceland. Have fiber-connected submarine networks running to the nearest major cities, be it Utqiagvik, Halifax, Ivato, Oulu.... If Russia ends the Ukraine war and comes off sanctions, they could even be lined all over Russia's Arctic coast, from Murmansk to the Bering Straits
Then two things: they won't be depriving a lot of people of power, since all those places are sparsely populat
Warming (Score:2)
Just move the AI data centres to Northern Europe then. Tropical AIland.
A small price to pay... (Score:3)
...for text-to-pr0n.
Animal brains are much more efficient (Score:2)
I keep thinking about human brains which have many orders of magnitude higher processing capability than an entire data center but only needs about 50 Watts of power.
I think there must be a better way.
Just USE it (Score:3)
Put greenhouses all around and use the excess heat and warm water.
Or:
Viticulture -- wine grapes are extremely sensitive to microclimate.
A few degrees warmer could allow varieties that wouldn't normally thrive in that region.
This is actually already happening naturally in places like England and Belgium due to climate change generally.
Other heat-loving crops, lavender, certain stone fruits, Mediterranean herbs, even olives in marginal climates.
Extended growing seasons, even for normal crops, warmer nights mean longer frost-free periods.
The irony would be delicious -- AI data centers accidentally creating French-style wine regions in Northern Europe.
There's probably a business case somewhere for a vineyard deliberately sited next to a large data center, using the waste heat both for the microclimate and potentially for heating greenhouses directly.