Great writeup! I've done some work on Codecarbon [1] in the past, it's a pretty standard tool for measuring ML/AI workload emissions, it would be awesome to have your findings contributed to that project as well.
We can pretend that international trade and energy price arbitrage doesn't exist to make us feel nice and happy; but the reality is a Watt spent in Sweden increases the marginal cost of energy by the same amount as a Watt spent in Texas. These things are tied together by foreign exchange, imported energy intensive goods, petroleum products and directly imported power. It's time we stop being naive about energy use.
One of the reasons to build them near populated areas is that it makes it easier to repurpose waste heat to, for example, heat houses in colder climates.
These are future plans and they lack specificity. I don't think they will come to fruition. The problem is that waste water from data centers isn't hot enough.
First link: "With the ambition to phase out its coal-fired plant by 2025, Denmark’s third largest district heating company, Fjernvarme Fyn, has established Denmark’s biggest heat pump plant. The plant will recover 215,000 MWh of energy from Meta’s hyperscale Tietgenbyen data centre and redistribute the surplus heat to the district heating network of the city of Odense."
As it clearly says on the page, this project was completed in 2019. The commercial use began in early 2020.
Even in the very sentence you quoted they are writing about plans to retire a coal plant by 2025 in future tense. This should be a very obvious hint that the text was written some time ago - unless, of course, the reader is not aware of what year we currently are living in.
In 2024, Fjernvarme Fyn reported ([1], page 19) they were recovering enough heat from that very site to heat almost 7000 homes in Odense and that the number can still rise significantly as the data center expands.
Probably so they have no problems getting people who work there and provide technical services. I mean that seriously without any sarcasm or tongue-in-cheek, it seems the most likely explanation to me.
Because that is where the existing roads, power and internet connections are. Might not be that significant investment to build those in context of scale. But still it simplifies many steps in building.
They need to be near a high bandwidth fiber connection, but with relatively cheap land. So the outer edges of major cities are prime locations, with Ashburn VA being the prime example. There is no reason to put them further out because the added costs don't make sense and the data centers actually have minimal impact on the surrounding population outside of hyperventilating conspiracy theorists, Ashburn VA also being the prime example.
This is why people in left wing areas need to oppose datacenters bans. Regulate them. Require green energy investments that more than offset their load. Use some of that flowing AI money for the betterment of society.
I think "people in left wing areas" is weird phrasing. There is nothing inherent about an area that makes it left or right wing. "People who live under governments who are responsive to collective issues such as climate and energy prices" would make more sense.
The reasoning effort knob is becoming a really important variable regarding how much impact AI models will have on energy consumption. Medium and max can differ by as much as two orders of magnitude in their token consumption. We also have to consider the amplification of compute demand that occurs by way of the tool calls themselves. Tokens are mostly a constant cost thing, but how they are interpreted can mean the difference between a 5 minute cpu bound operation and a 50uS cached database lookup. At some point, the CPU is going to become the bottleneck again.
And eating a single burger results in the equivalent Co2 emissions of 17,000 requests in Sweden and 2,000 requests in Texas.
Reducing/stopping meat consumption and replacing it with healthy plant-based tofu, seitan, tempeh etc has a far bigger impact (and as a bonus, living beings don't have to suffer)
AFIACT, if I make a single request, the part that uses the most power is my monitors while they display the result and I read/use it.
Which makes me suspect that if I do something complex involving 2,000 requests, the power used is less than I'd use to achieve the same goal using my monitors and other office equipment. That depends on the goal of the complex action, of course.
It might not help the cause, but ultimately it's an uncomfortable truth. It's easy to dismiss a fact as 'moralising'.
It's simply a fact that animals suffer when they are factory farmed and then killed. It's a fact that they are living sentient beings. You can be totally okay with that, but the reality is that most people are only okay with it because it doesn't happen right in front of them.
I used to eat meat, btw. And I also used to attack people for 'moralising'. I now realise this was my brain trying to defend itself; to try to make sense of why I would pet a dog but eat a pig (everyone else is doing it, this person is just different and wrong, go away bad vegan person!)
I also had this sudden realization a year ago about the cruelty of animal usage in society and went vegan and my only regret is that I didn't do it sooner.
One other angle is there are the beginnings of consideration of cruelty to AI's. That's great that people are starting to consider it, to try to be compassionate out of the abundance of caution that it might matter.
But to then just straight up pay to have tens to hundreds of obviously sentient animal killed for you per year (and yes egg laying hens and dairy cows are killed in their youth, when they don't want to die) seems like a blind spot a lot of people have. I had it myself until recently, so I don't cast any judgment, but I also wish it was easier to make people just see the truth on this
Good for you for accepting an uncomfortable truth and making a change. I'm not sure where you are based, in some places it's easier to eat out as a vegan than in others. That said, the more people the better, and the easier it gets for everyone.
I do live in a big city, and I've found the eating part to be a non issue after like a one month learning curve. I now miss absolutely nothing, and am healthier and spend less on food. The hard part is dealing with social situations, but it's not that bad.
That's actually kind of a scary statistic... I probably eat a burgur at most every month (beef maybe once a week at most) but probably make more than 2000 requests a day.
Your comment makes me really contextualize my AI usage and want to reduce it for ecological reasons.
0.5 Pwh/yr - worldwide data center energy
4-5 PWh/yr - meat total farm to table
166 PWh/yr - worldwide total energy production
the true bottleneck and cost we have is energy, carbon is a polite proxy
if you can meaningfully cut down the energetic cost, that's very valuable either way
maybe tomorrow you'll want to spin up a GPU on the cloud to briefly run a model, pick a data center:
DC #1: EU Efficient, but has ~100ms additional latency
DC #2: TX Runs on jet engine turbines, neighbors complain of noise and air pollution
env impact is something worth considering regardless of industry
Tell you what, I'll keep eating beef, walking to work, and not using AI. Closer to how our ancestors lived for hundreds of thousands of years, so it can't be too off the mark.
I trap and kill hare, and make cheese from fresh milk from my friend's dairy farm. But no, I do not kill the cows that I eat.
What do immigrants have to do with this? I shared my opinion that living like our ancestors did is the way I'd like to continue my life. If you have a counter opinion, can you offer it?
> And eating a single burger results in the equivalent Co2 emissions of 17,000 requests in Sweden and 2,000 requests in Texas.
And China is building 6-8 new mega-scale lignite coal power plants *per month.* Each one of those power plants emits more CO2 than 8.3 million people consuming medium amounts of animal products *for life.*
For the record I'm not denigrating your point. I'm taking it a step further. We're not focusing on the major contributors of CO2 emissions. It's not data centers and it's not meat consumption.
> China emits half the emissions per capita of the US.
China emits more CO2 than any other country by a wide margin. It's not less damaging to the environment just because they have more people. Even if we pretended that that were the case, [China emits more CO2 per capita than 32 of 38 OECD countries.](https://ourworldindata.org/grapher/co-emissions-per-capita?c...) They are an egregious and unrepentant polluter, in every way imaginable, and in many ways you cannot even imagine.
Any discussion of reducing CO2 emissions which doesn't include a large section on China is not taking the subject seriously. But to be clear, we should also discuss emissions by the other large polluters like the U.S. and Canada.
Worry not, AI people! Greenhouse gas exemptions will be made for the future of work, maths and art, the salvation of the humankind, the genie that is out of the bottle, etc, if only at the expense of taxing your and your dog's fart.
The comments section reminds me of The Dead Great-Grandfather Test
[Science fiction writer Bruce] Sterling was using the term specifically to describe the hairshirt green lifestyle that is driven by eco-anxieties. For hairshirt-green types, life is all about saving water, recycling, composting, reducing eco-footprints and various other behaviors marked by a kind of fearful, non-generative retreat from living. Permanent existential hibernation.
Sterling's rule of thumb for spotting acting-dead behaviors is a great one: if it's something your dead great-grandfather can do better than you, it's a case of acting dead. Your dead great-grandfather uses no water or plastic, and is actually recycling himself as we speak, not just his possessions. Try and top that.
But acting dead goes beyond hairshirt-green behaviors. While spartan frugality is a virtue, when it becomes the entire purpose of your life, there's a problem. For a portion of the dying American middle class, frugality has turned into a life purpose.
This is a fallacy. People 100 years like your great grandfather ago solved problems they had 100 years ago. People today solve problems they have today.
I think it’s along the lines of “premature optimization is the root of all evil”. These little micro optimizations take your attention and energy away from places that might be less visible and less enjoyable but more consequential.
Whoever responds "we need to stay within our CO2 budgets, otherwise we'll be in real trouble in 20 years" with "being a hippie and organizing your life around recycling" is really building up some sort of weird straw man and not really usefully engaging in solving the problem.
Is getting in touch with nature, appreciating the value in things and protecting complex ecosystems really 'Acting Dead'?
I would say 'Acting Dead' is actually sitting at home ordering on Amazon/Temu, watching brainless TikTok/Instagram/Netflix/YouTube videos, and talking to a transformer model all day.
Thanks for the ragebait. Reminding myself that the flaw here is obvious, there’s no need to respond, arbitrary „tests” aren’t binding to anyone, protecting the future is perfectly defensible as a moral stance even at cost of present convenience.
too many things can happen in the ai provider's backend which won't be observable from the outside. for starters how will you even know what hardware and inference backend is being used by RandomAiProvider3284.com? you can't. i looked at your berget-ai/co2-calculator repo and you're expected to provide this yourself. it also uses "measuredResponseTimeSeconds" as a metric even though it's impossible to know how much of that time was spent rate limiting the user during hours of high activity
> for starters how will you even know what hardware and inference backend is being used by RandomAiProvider3284.com? you can't.
From the article:
A typical request to the open model Kimi K3 emits about 0.4 grams of CO₂ in Sweden, 1.6 grams in Ireland, 2.5 grams in Germany and 3.2 grams in Texas, on identical hardware — 8.5 times more in Texas than in Sweden.
-> So the comparison is based on the assumption that identical hardware is used.
Also from the article:
Emissions estimates are based on the current average emissions intensity of each regional electricity grid, which has been the norm in greenhouse gas accounting.
So it's just an incredibly stupid way to compare the power grids in Texas vs Sweden. It's not even accurate because the marginal watt during the day looks very different than the overall grid average.
Well yes, it’s not like GPUs belch CO2 from their bellies.
The point is that they use a lot of electricity, and this is an active concern to many Americans who see electricity prices go up and polluting energy sources being used more as data centers roll into their state.
I agree, but AI seems like one of the things that could easily follow the sun/best source of green (and therefore cheap) energy. I (a European, unfortunately born in the UK) wouldn't mind a bit of latency on my Claude Code sessions if I knew it was so that everything was being run from solar in north Africa.
For 2025 the grid carbon intensity in Texas was 363gCO2eq/kWh, while it was 22gCO2eq/kWh in Sweden, so 16,5x more in Texas. Right now it's 427 in Texas vs. 50 in Finland ie. 8,54x more, like in the article (yes I looked the wrong country at first).
The reason is half of Texas electricity production is still fossil fuels (gas & coal) while mostly hydro, nuclear and wind in Sweden. So it's not really about ai, the models or the hardware. Fossil fuels make CO2, that's it.
Fair comparison. Texas just needs to flood the valleys of its famous mountain chain and collect the melted snow with dams to match Sweden's 40% hydro electricity production.
75 comments
[ 10.9 ms ] story [ 206 ms ] thread[1] https://github.com/mlco2/codecarbon
Power connections.
https://danskfjernvarme.dk/viden-vaerktoejer/udgivelser/taen...
https://danmarksdagblad.dk/article/datacenter-paa-15-million...
https://pulse.microsoft.com/da-dk/transform-da-dk/government...
I don't think anywhere would want to be entirely reliant on a single datacentre, at least not at the scale of district heating use in Denmark.
(I gave some pages in Danish, you can search "Denmark datacentre district heat" yourself if you like.)
AtNorth's are under construction: https://www.atnorth.com/news/selected-group-will-utilize-sur... and https://www.atnorth.com/news/denmark-data-center-to-supply-h...
Microsoft's is operating: https://danskfjernvarme.dk/aktuelt/nyheder/2026/dataskyen-se... (I can't find an English source.)
"Will" means "in the future".
But you can do your own searching, you seen determined that this doesn't exist.
Even in the very sentence you quoted they are writing about plans to retire a coal plant by 2025 in future tense. This should be a very obvious hint that the text was written some time ago - unless, of course, the reader is not aware of what year we currently are living in.
In 2024, Fjernvarme Fyn reported ([1], page 19) they were recovering enough heat from that very site to heat almost 7000 homes in Odense and that the number can still rise significantly as the data center expands.
[1] https://stateofgreen.com/en/wp-content/uploads/2024/12/SoG_W...
Reducing/stopping meat consumption and replacing it with healthy plant-based tofu, seitan, tempeh etc has a far bigger impact (and as a bonus, living beings don't have to suffer)
That will put some sweat on some brows :)
Which makes me suspect that if I do something complex involving 2,000 requests, the power used is less than I'd use to achieve the same goal using my monitors and other office equipment. That depends on the goal of the complex action, of course.
The "only the biggest thing matters" is a faulty logic. If you truly care about the environment then we need to improve things across the board.
And moralizing about living beings doesn't help your cause here.
It's simply a fact that animals suffer when they are factory farmed and then killed. It's a fact that they are living sentient beings. You can be totally okay with that, but the reality is that most people are only okay with it because it doesn't happen right in front of them.
I used to eat meat, btw. And I also used to attack people for 'moralising'. I now realise this was my brain trying to defend itself; to try to make sense of why I would pet a dog but eat a pig (everyone else is doing it, this person is just different and wrong, go away bad vegan person!)
One other angle is there are the beginnings of consideration of cruelty to AI's. That's great that people are starting to consider it, to try to be compassionate out of the abundance of caution that it might matter.
But to then just straight up pay to have tens to hundreds of obviously sentient animal killed for you per year (and yes egg laying hens and dairy cows are killed in their youth, when they don't want to die) seems like a blind spot a lot of people have. I had it myself until recently, so I don't cast any judgment, but I also wish it was easier to make people just see the truth on this
Your comment makes me really contextualize my AI usage and want to reduce it for ecological reasons.
maybe tomorrow you'll want to spin up a GPU on the cloud to briefly run a model, pick a data center:
env impact is something worth considering regardless of industryWhat do immigrants have to do with this? I shared my opinion that living like our ancestors did is the way I'd like to continue my life. If you have a counter opinion, can you offer it?
Just go vegan for a week or something and you’ve offset your AI water use for a few lifetimes.
And China is building 6-8 new mega-scale lignite coal power plants *per month.* Each one of those power plants emits more CO2 than 8.3 million people consuming medium amounts of animal products *for life.*
For the record I'm not denigrating your point. I'm taking it a step further. We're not focusing on the major contributors of CO2 emissions. It's not data centers and it's not meat consumption.
China emits more CO2 than any other country by a wide margin. It's not less damaging to the environment just because they have more people. Even if we pretended that that were the case, [China emits more CO2 per capita than 32 of 38 OECD countries.](https://ourworldindata.org/grapher/co-emissions-per-capita?c...) They are an egregious and unrepentant polluter, in every way imaginable, and in many ways you cannot even imagine.
Any discussion of reducing CO2 emissions which doesn't include a large section on China is not taking the subject seriously. But to be clear, we should also discuss emissions by the other large polluters like the U.S. and Canada.
[Science fiction writer Bruce] Sterling was using the term specifically to describe the hairshirt green lifestyle that is driven by eco-anxieties. For hairshirt-green types, life is all about saving water, recycling, composting, reducing eco-footprints and various other behaviors marked by a kind of fearful, non-generative retreat from living. Permanent existential hibernation.
Sterling's rule of thumb for spotting acting-dead behaviors is a great one: if it's something your dead great-grandfather can do better than you, it's a case of acting dead. Your dead great-grandfather uses no water or plastic, and is actually recycling himself as we speak, not just his possessions. Try and top that.
But acting dead goes beyond hairshirt-green behaviors. While spartan frugality is a virtue, when it becomes the entire purpose of your life, there's a problem. For a portion of the dying American middle class, frugality has turned into a life purpose.
https://ribbonfarm.com/2011/12/08/acting-dead-trading-up-and...
I would say 'Acting Dead' is actually sitting at home ordering on Amazon/Temu, watching brainless TikTok/Instagram/Netflix/YouTube videos, and talking to a transformer model all day.
From the article:
A typical request to the open model Kimi K3 emits about 0.4 grams of CO₂ in Sweden, 1.6 grams in Ireland, 2.5 grams in Germany and 3.2 grams in Texas, on identical hardware — 8.5 times more in Texas than in Sweden.
-> So the comparison is based on the assumption that identical hardware is used.
Also from the article:
Emissions estimates are based on the current average emissions intensity of each regional electricity grid, which has been the norm in greenhouse gas accounting.
The point is that they use a lot of electricity, and this is an active concern to many Americans who see electricity prices go up and polluting energy sources being used more as data centers roll into their state.
The reason is half of Texas electricity production is still fossil fuels (gas & coal) while mostly hydro, nuclear and wind in Sweden. So it's not really about ai, the models or the hardware. Fossil fuels make CO2, that's it.
Like 2/3rds or 4/5ths of it on a fossil heavy grid.