The AI data center narrative is the perfect storm:
First most of the data center build out is happening in areas that have had little other opportunity so local resistance is muted. Abilene, Texas is referenced and is the kind of place my grandfather would lovingly say is where you go to learn to be a “dirt farmer”
Second, every environmentalist in the US is fighting 100 different battles with the most anti-regulation, pro-energy administration in decades (ever?) and has limited bandwidth.
And third, the AI narrative around national security, longevity, and super-intelligence-enabled abundance provides massive national coverage - the implications being that AI will solve any environmental and or human economic disasters that they enable.
Using GPUs as paperweights at night doesn't seem to be an efficient usage and it comes with its own costs. "Pull from the grid" is what they try not to do.
The Pacific Ocean is a bit of an exageration here.
There isn't a single battery this big today, but if batteries continue following the exponential growth curve they've been on then there probably will be in the next decade or so (if not sooner).
There is, and "some batteries" is an under count, however:
- The ratio between "some water" and "the pacific ocean" is a lot higher than 5
- On an exponential growth curve, a factor of 5 isn't all that much.
This probably isn't feasible for a data center being built today (although they could build solar to at least reduce their fossil fuel power generation needs during the day). But it probably will be for data centers being built in 5-10 years time.
Basically it's important to differentiate between "we can't quite do this yet" and "we'll likely never be able to do this". And powering a data center with rewnewables+batteries is definitely in the former category.
Do you have a source for this? I constantly point out that several cloudy days in a row often happens but am rebutted with graphs of the declining costs of solar modules and batteries
Nobody is suggesting to use solar as the sole source of electricity. That would obviously be insane and then leads to silly numbers like the ones you gave.
A MWh of solar is about 30 Euro/MWh if you levelize the capital costs.
The article discusses why LCOE is not a good estimate of the costs. Yes, combining different sources lowers the total cost (by damping intermittency). For Denmark it's offshore wind and solar in a 7 to 1 ratio (plus natural gas or biomethane power plants).
> Nobody is suggesting to use solar as the sole source of electricity.
The electricity price is not decoupled from the gas price because there's not enough energy storage, wind turbines, interconnections (that is the energy grid can't function without natural gas). The cost of building (and operating) all this goes into SLCOE.
No, it means that if you expect that sticking solar panels everywhere will solve all the problems for cheap (because solar panels are cheap), then you are up for a disappointment. I am of the opinion that handling out rose-tinted glasses will not go well.
Abilene is the confluence of three major resources. Massive wind farms that stretch from horizon to horizon, natural gas wells that are constantly having to flare off excess capacity, and a direct freeway connection to both Fort Worth and Dallas. In addition, land is dirt cheap.
I suspect their biggest problem will be water availability so they will need to use some other method for cooling.
I was thinking heat pumps hooked up to radiators and/or geo grids. Right now the cheapest method is evaporators which use a ton of water. It's expensive to get rid of a couple megawatts like that. Some data centers have their own microclimates that are 10-20C hotter on average.
China invested hugely into renewable energy. Their grid is strong. Ours is falling apart and our renewable energy contracts keep getting shredded. Gas powered data centers is beyond dystopian
That doom take has a few roots in truth, but it is mostly false. Our grid for the most part is very good and improving. A lot of the gloom is it is good for today, but here is why we are trying to expand it anyway - that is the gloom itself is helpful to get the needed changes made.
There is a lot of renewable energy in the US, and more is built all the time.
It's really not "doom" it's reality. Highly recommend googling the state of the us grid. It's under funded otherwise data centers would not be running on gas. Similarly one of the biggest anticipated bottlenecks for new data centers is power right now.
Although renewables are on the rise in use, huge projects have recently been cancelled in favor for gas. Meanwhile the us is going around blowing up and encouraging other gas producing nations to blow up gas infrastructure.
If you're worried abour grid health, how is adding distributed generation colocated with new loads dystopian?
Sure, it would be nice to be able to build more transmission lines and power stations wherever it makes sense for engineering to build them in order to build a strong grid. But that's hard to do with strong private land ownership and required environmental impact reporting.
Something something texas avoiding federal electric regulation.
It's unsustainable and bad for the local environments surrounding it. How many people need to die so people can chatgpt a social media post or summarize an article? These are real unanswered questions we are facing because the US has inadequate energy infrastructure.
You ever lose power and run a generator? The stink and sound is awful. Imagine powering more houses than there are in a suburban town with gas? Imagine doing that in hundreds of thousands of locations across the us. It's a dystopian thought.
I agree it is hard to stand up an effective grid to sustain technological innovations and products. It sounds like something we needed to be ready for like other countries are. Or maybe something that should happen first in order to be responsible and get ready for the future.
Last year, China installed 1.5GW of nuclear and 300GW of renewables. China's total coal use decreased as they decommissioned and under-utilized more coal than they installed.
This same pattern is occurring everywhere, regardless of local politics or local economic system. See Texas as another example.
It's because new renewables is superior at contemporary market prices. Markets have decided. Governments have decided. Everyone has decided.
It's so boring to relitigate this constantly on HN. It's like debating whether the sky is blue. Ridiculous that this comes up so often.
However you may try to put lipstick on a pig, the core of the matter is unchanged. China's coal use decreased, and they barely install any nuclear compared to renewables.
Nuclear is useful but it's hard to see it as a panacea. Renewable energy on the other hand is hard to beat. The economics of it keeps getting better, and previous estimates of the lifecycle of things like solar was grossly misrepresented.
Cancelling wind power contracts etc was a huge mistake.
"Environmentalists" is a large, diverse group and nuclear energy has been a controversial topic splitting the group for decades.
Many environmentalists are pro-nuclear, and viewed exclusively through an environmental lens, nuclear is likely the best energy source.
Other people share the "environmentalist" label because they care about clean air, unpolluted rivers, biodiversity, climate change, etc but they oppose nuclear on unrelated grounds (eg, as part of an anti nuclear weapon proliferation agenda) or out of fear of adverse events from damage to an energy facility.
The "pro-environment but anti-nuclear" subgroup held power within the Democrat party in the US through most of the cold war era. The "pro-environment, pro-nuclear" subgroup is now the largest group within the Democrat voting base, but some of the people and all of the regulations from the 1960s-1990s are still in power.
Yes, and that dates from the era when everyone thought that a "nuclear winter" (from an exchange of nuclear bombs between the US and USSR) was a more immediate risk than a long slow problem of climate change. That's what the French government carried out a terrorist attack against Greenpeace in New Zealand for: the Rainbow Warrior was protesting against French nuclear weapons testing.
Has everyone forgotten that the current crisis kicked off over the question of exactly how much uranium of what purity Iran is allowed to have? Do people really think that every country in the world should have multiple nuclear reactors?
yes, but that is not the point. The point is that countries pursued uranium reactors as a priority to produce plutonium for weapons. It did not / does not have to be this way, and the conflation of all nuclear reactors as either being at risk of meltdown or linked to nuclear weapons has resulted in the current situation. The green movement did us all a disservice by choosing not to clarify the difference, and here we are in 2026 still burning coal, which continues to produce more radioactive waste than all the nuclear reactors ever built - and released directly into the environment!
There are many claims that Thorium fuel cycle is nuclear proliferation safe, it is safer but not perfectly safe. Uranium-233 bread during the Thorium fuel cycle can used to construct nuclear weapons
"A declassified 1966 memo from the US nuclear program stated that uranium-233 has been shown to be highly satisfactory as a weapons material, though it was only superior to plutonium in rare circumstances."
Many nuclear energy opponents require absolute nuclear proliferation safety.
I was under the impression that Carl Sagan basically made up the concept of nuclear winter because of his political alignment.
And, yes, every country should have nuclear power and nuclear deterrence. Please contrast how the US negotiated with North Korea and Libya. Or what happened to Ukraine after they denuclearized. You might be impressed at how palatable peace becomes when leadership has to consider their personal safety and not just cynical economic and political calculus in their use of force.
Please. Show me a nuke that was under 10 billion over budget and less than 20 years late.
Seriously. Don't even try to insert nuclear power into the debate before you show it can be competently built, never mind safely run and waste safely disposed. But what about the French you say! The French are discovering they forgot to set aside money for decommissioning.
If you look at how much time and money the French poured into their nuclear plant construction project (and several other pre-decommissioning issues) things look rather less rosy.
Also, it was a very high priority for France - keep in mind Napoleon's quote about the relative importance of moral and material.
More critically: The logic of "The French did {extremely complex thing} fairly well in the late 1900's, therefore we can also" is very similar to the logic of "SpaceX designs and operates reusable rockets with hundreds of launches per year and no failures, therefore we can also". NO, sorry, you are just fantasizing about rocket science somehow being easy for you. SpaceX has lots of highly motivated would-be competitors - many of them far better financed than SpaceX was - but the hard fact is that zero of those can actually do what SpaceX is doing.
The costs were not small, but probably smaller than the costs of German Energiewende. Messmer plan needed not only money, but also capable heavy industry and consistent political support.
A lot of environmentalists are just degrowthers and NIMBYs. During the Biden administration multiple groups sued to block high voltage transmission lines for clean energy.
Even if modern renewables and batteries kill the need for future nuclear power plants, that doesn't excuse the consequences of decades of burning fossil fuels because environmental groups fought against nuclear power altogether.
We had better options back then, and we chose not to implement them while slowing down efforts to improve them: nuclear reactor designs could have been standardized to lower cost, even safer and more effective reactor designs could have been pursued years earlier, etc.
The costs--and opportunity costs--of inaction during that time were massive, and we're going to be paying them for generations. Renewables have a heavier lift ahead of them as a result, with less time to build out and upgrade the grid, transition to EVs, etc. The very least we can do is acknowledge the consequences.
Because it’s a dangerous technology where profit over security can have severe consequences.
Not to mention the new risks now that drone wars are a thing. All those Chemical plants are already valuable target, no need to additional nuclear ones.
>Because it’s a dangerous technology where profit over security can have severe consequences.
Yeah it's far better to have power plants kill a steady stream of people, but in a banal way that's hard to attribute, like coal power plants causing lung cancer.
No, but hydroelectric works pretty well. There's a reason they recycle aluminum, etc. (power-hungry industries—crypto-mining, ha ha) along the Columbia River, etc.
So build your data centers there. No reason to choose the least evil.
Hydro electric as a resource is probably mostly exploited - are there a lot of big hydro projects left to build? If there are they must be difficult or expensive or they would have already been built.
Even ignoring nuclear environmentalists are still the enemy here.
Green energy is diffuse. Fields of solar. Ridges dotted with turbines. Each unit needs a power cable running away from it, access roads, etc, etc. A lot of area has to be developed for a given payback (sellable power).
Environmentalists and have been instrumental in making it economically impossible to develop land cheap enough to make low value density projects like that pencil out.
The higher power cost states in the US would likely be dotted with all manner of solar infill if not for up front costs that these shortsighted and selfish people have imposed on any land alteration larger than approx SFH lot size. Farmer Johnson would love nothing more than to put up solar on that ~2ac hillside he owns but cannot farm economically. Neither he nor some 3rd party who would put up the panels will shell out half a mil for an EPA CG permit just to clear the vegetation because the panels will never pay that back in their lives.
The economics of compliance are why the only greenfield development that happens these days is value dense commercial stuff (shopping plaza, big box store, data center, office buildings etc) or dense SFH development.
The most rosy possible outlook is that we "just" wait for the hippie boomers who cooked this crap up to croak, shit can the clean water act and come up with some new way of regulating development that doesn't saddle these super low impact projects (it's hard to be lower impact than panels or turbines) with fixed-ish costs that are a non-starter except at huge scale.
But of course there are so many parties who are making rent off this broken system who won't go down without a fight.
It actually sits right about in the middle of all countries for percentage of GDP from industrial sectors.
It's pretty heavily fossil-fuel powered right now, but like most of the rest of the world Morocco is planning to capture most of its growth in energy demand with renewables. Because, getting back to the original discussion, it'd be idiotic to choose fossil fuels over renewables in 2026.
100% agree with you. Especially with the amount of sun they get, would be a great candidate for Solar.
Renewables also seem to be the best pathway for many nations to attain some level of energy independence (which is NOT the same as oil independence though).
Are you, like, just not very into reading things or remembering things or is it just one of those things where you're stuck in the 1980s and haven't updated your books since then.
And the sad part is that ai would be a perfect fit for intermittent energy sources: run inference 24/7 (on whatever you can muster, even if it's fossil it simply does not consume all that much compared to training) and over-build training to achieve whatever total throughout you need in time of energy abundance. How heavy would CO2 pricing have to be to make the market do this instead of the exact opposite?
What would be the economy of these data centers if they would run using power source with say, for example 25% capacity factor? What is the capital expenditure of a data center? What is the yearly operational expenditure of a data center?
I found that a 100 MW datacenter can cost roughly $3.35 billion, with a significant portion going to high-end GPUs like the B100 or H100. For electricity, 100 MW data center can incur annual power costs ranging from $41 million to over $131 million, depending on regional energy prices.
Compute has massive initial capital requirements, and a very short obsolescence window. It does not make sense to invest in compute and then not run it 24/7 at 100%.
Hardware sitting idle waiting for sunshine/wind is leaving money on the table. Especially if you can burn gas during that time and still be hugely profitable.
Would the world be inherently different if hardware cost was 4x of what it is per unit of training done? The world would be inherently different if we did not burn through fossil stockpiles at the speed we do, or not at all. It's market mechanisms failing to lead to a desirable outcome. Nothing new about that, plenty of problems of that kind have been solved before, but yes, this one is more difficult.
Very quick Googling suggests Wired’s estimate would be ~1.9% of US emissions.
AI data center investment is, at core, a bet on increasing the productivity of labor. That’s what businesses will pay for, and what will earn the big money.
If US labor productivity rises by more than 2%—and implicit in the size of this bet is a guess much higher—US carbon intensity goes down correspondingly, and these data centers end up as a win for the climate.
Problem is climate and ecosystem doesn’t give a fuck about some notional productivity or carbon intensity just the absolute number of pollution molecules going into the air
OK well now you have to look at changing the economy-wide energy mix or embracing de-growth. Switching data centers to 100% solar or nuclear or … solves Wired’s complaint but not this one.
The "climate" cares little about carbon intensity of labor or dollars. It cares about absolute tons of green house gases in the atmostphere and if you use more to produce more you still use more.
You should always assume these that benefit most from the technological forefront think they can outrun the output of climate change or any of the outputs.
It generally is called "effective altruism", eg, techno jesus will solve whatever problems creating techno jesus creates.
> If US labor productivity rises by more than 2%—and implicit in the size of this bet is a guess much higher—US carbon intensity goes down correspondingly, and these data centers end up as a win for the climate
I can promise you it won't happen in a million years. More productivity lends to more exploitation, because you can do more with the same unit of work, instead of getting the same result with less work.. Or at least we have decades of data proving that is what realistically happen.
So the only way to reduce emissions is either using carbon neutral sources (gas is... not?) or forbidding people from using energy in the first place (and let's be honest, that will not happen.)
That is a tough sell in the current environment. It's a regressive tax, so opposed on both ends of the political spectrum. People on the far right don't believe in climate change, and people on the far left don't believe in market efficiency. With 20% of the world's oil flow crimped in the Strait of Hormuz for who knows how long, higher energy prices is the last thing people want to contemplate.
In the longer run, a carbon tax is the best option. The fossil fuel price shock is a strong signal to produce energy through other means. There are major engineering initiatives around developing cheaper and safer nuclear energy. and it's cheaper now to deploy a solar farm than a coal plant.
A carbon tax would raise money to pay off national debts and encourage consumers and producers to figure out the most efficient way to accomplish their needs while minimizing their carbon footprint. It's a tough sell today, but this is they way to go for a better quality of life tomorrow.
>More productivity lends to more exploitation, because you can do more with the same unit of work, instead of getting the same result with less work..
But per-capita greenhouse emissions have been falling in much of the developed world? And you can't really claim with a straight face that productivity has been dropping from 2000 to today.
> But per-capita greenhouse emissions have been falling in much of the developed world?
Only by the deceptive accounting trick of not including the emissions associated with overseas production of the goods consumed by the "developed world".
If you include all the emissions that prop up the highest per capita consumption patterns on the planet then you see the highest per capita emissions attached to the highest consumers.
>Only by the deceptive accounting trick of not including the emissions associated with overseas production of the goods consumed by the "developed world".
That does increase US's emissions, but not enough to change the conclusion:
In history of the world there was almost never a reduction of greenhouse emissions. Only time that world did reduce greenhouse emissions was in times of economic crises and COVID-19.
Asia will do what US did in 1979, Jimmy Carter delivered this televised speech on July 15, 1979.
"Point three: To give us energy security, I am asking for the most massive peacetime commitment of funds and resources in our nation's history to develop America's own alternative sources of fuel -- from coal, from oil shale, from plant products for gasohol, from unconventional gas, from the sun."
Meanwhile anyone can run a capable LLM in their own living room, or even on their own mobile device, but nobody wants to talk about that. And no, 99.999% of slop does not need a math Olympics finalist model, sorry.
That has a body count. It's hard to know which people you're killing, but there's no doubt that at that point there's a number. Maybe... 10,000 or so? They'll die in heat waves and floods and tornadoes and food shortages in dry, poor countries. Obviously 10k is a tiny fraction of the full list of climate fatalities.
Most emissions are harder to reduce because industry isn't electrified (it runs directly on oil/coal not electricity). This is electricity usage. Electricity emissions are the easiest to avoid.
AI industry also promises to replace everyone, and they have access to large amounts of capital they could use for renewables.
That they punch down on everyone, including people outside of their country who don't benefit from the tax revenue they'll generate, with massive amounts of emissions, is a slap in the face.
It doesn't help that I think datacenters are the means of digital oppression rather than the means of digital empowerment.
Oppression is where you get people hooked on something but then lock it away and use your complete control over supply to extract rent. Gradually the rent will become so high that only the most abusive profit-seekers can afford to pay it.
Empowerment is owning technology so that its purpose is to work for you, rather than to extract rent from you. Tech that empowers individuals usually doesn't need giant chain-link fences around it.
That's not to mention talking people to suicide, which by the way is the far more likely way that it would kill your son or daughter, should you have or ever have one.
And for all that cost, are they bringing us real hope? The most ambitious people talking about this technology basically say that when it takes over all thought, there won't be much point in humans anymore.
We went from being evangelists of a message of hope to evangelists of a message of doom.
I'm not going to be one of those people. Fuck those people. I believe in the future. I will stand up and tell the next generation there is still hope, still compassion, still community and humanity and love out there. I fight for the users!!!
Have you ever been in India while they have those incredibly hot heatwaves in the summer? That kills thousands of people every year. This will only get worse the more greenhouse gasses we emit so slop machines can generate funny videos of the pope breakdancing.
> New gas projects linked to just 11 data center campuses around the US have the potential to create more greenhouse gases than the country of Morocco emitted in 2024
"Have the potential to", "Morocco". Presumably doesn't count the greenhouse gases emitted by Moroccans using overseas cloud services and AI.
The AI data center narrative is the perfect storm:
First most of the data center build out is happening in areas that have had little other opportunity so local resistance is muted. Abilene, Texas is referenced and is the kind of place my grandfather would lovingly say is where you go to learn to be a “dirt farmer”
Second, every environmentalist in the US is fighting 100 different battles with the most anti-regulation, pro-energy administration in decades (ever?) and has limited bandwidth.
And third, the AI narrative around national security, longevity, and super-intelligence-enabled abundance provides massive national coverage - the implications being that AI will solve any environmental and or human economic disasters that they enable.
> pro-energy administration
Very far from pro-energy when you give companies money to cancel energy projects that are not burning green house gases.
Abilene, Texas is also a great place to build a solar power plant with some batteries and reduce the gas bills for these datacenters.
EDIT: I am not suggesting that they don’t build gas turbines or go off grid. I’m saying they can save fuel by using solar when it’s there.
Everyone is talking about batteries but honestly you don’t need batteries.
- Data centers don’t sleep
- Data center load (for AI) could be shifted to follow the sun
- The energy requirements mean you aren’t likely to overbuild your solar farm
At night just stop running your GPUs and / or pull from the grid
<50% utilization of billions of dollars of capex. Brilliant idea.
Using GPUs as paperweights at night doesn't seem to be an efficient usage and it comes with its own costs. "Pull from the grid" is what they try not to do.
Seems crazy to build an expensive factory and then cheap out on the power source so it can only run when it is sunny out
There isn't a battery in the world big enough to provide enough power for data enter if this size.
Some batteries in this case is a bit like saying some water about the Pacific ocean.
A big AI data center uses about 1 GWh of power each night.
A large battery storage site is about 500MWh.
So this is totally doable and it’s also going to be economical as soon as the US has built enough LNG export capacity.
That's a big data enter from last decade. Today big data enters use 4GWh power in winter. So only 8 of them.
4GWh of power in what time frame?
Either way, in winter you’ll need the gas turbines, I didn’t claim otherwise.
The Pacific Ocean is a bit of an exageration here.
There isn't a single battery this big today, but if batteries continue following the exponential growth curve they've been on then there probably will be in the next decade or so (if not sooner).
There's a bit of a difference between 'some batteries' and at least five times the size of the largest battery ever build.
There is, and "some batteries" is an under count, however:
- The ratio between "some water" and "the pacific ocean" is a lot higher than 5
- On an exponential growth curve, a factor of 5 isn't all that much.
This probably isn't feasible for a data center being built today (although they could build solar to at least reduce their fossil fuel power generation needs during the day). But it probably will be for data centers being built in 5-10 years time.
Basically it's important to differentiate between "we can't quite do this yet" and "we'll likely never be able to do this". And powering a data center with rewnewables+batteries is definitely in the former category.
You will always use gas for cloudy days or maybe in winter. I didn’t claim solar should be the sole source of power.
[dead]
[flagged]
Some batteries? Battery-backed solar has the highest cost as a baseload power source due to high intermittency of solar power.
Do you have a source for this? I constantly point out that several cloudy days in a row often happens but am rebutted with graphs of the declining costs of solar modules and batteries
https://www.sciencedirect.com/science/article/pii/S036054422... Figure 3
When used as the sole source of electricity solar clocks at 565 EUR/MWh. Nuclear, for comparison, is at 141 EUR/MWh.
Nobody is suggesting to use solar as the sole source of electricity. That would obviously be insane and then leads to silly numbers like the ones you gave.
A MWh of solar is about 30 Euro/MWh if you levelize the capital costs.
The article discusses why LCOE is not a good estimate of the costs. Yes, combining different sources lowers the total cost (by damping intermittency). For Denmark it's offshore wind and solar in a 7 to 1 ratio (plus natural gas or biomethane power plants).
> Nobody is suggesting to use solar as the sole source of electricity.
You weren't clear on what you propose.
https://has-electricity-decoupled-yet.strommarktberatung.de
Germany has a lot of solar. It's suppressing prices below the electricity price you would get with pure gas power generation.
The electricity price is not decoupled from the gas price because there's not enough energy storage, wind turbines, interconnections (that is the energy grid can't function without natural gas). The cost of building (and operating) all this goes into SLCOE.
Well, I found the proposal clear and your response confusing.
There are reasons it might not work, not least of all political. But a link to a paper picking a bone with LCoE is talking past GP.
I wanted to remind that "just add solar" comes with the costs that are rarely mentioned.
I was very clear, even before the edit. I wrote "reduce the gas bills". Not "cut the gas line".
Konschubert, it's widely reported that German energy prices are some of the highest in the world.
Exempli gratia: https://www.cleanenergywire.org/news/germanys-household-powe...
And the proposal to fix this is more accounting games to transfer costs to different constituents.
Batteries are good to cover evening peaks or maybe a while night. Not multiple days of low solar, that’s not economical.
good point, I guess the market says we have to poison the earth
No, it means that if you expect that sticking solar panels everywhere will solve all the problems for cheap (because solar panels are cheap), then you are up for a disappointment. I am of the opinion that handling out rose-tinted glasses will not go well.
Abilene is the confluence of three major resources. Massive wind farms that stretch from horizon to horizon, natural gas wells that are constantly having to flare off excess capacity, and a direct freeway connection to both Fort Worth and Dallas. In addition, land is dirt cheap.
I suspect their biggest problem will be water availability so they will need to use some other method for cooling.
Maybe they can use the abundant oil for cooling as well :) that is how transformers are cooled
I was thinking heat pumps hooked up to radiators and/or geo grids. Right now the cheapest method is evaporators which use a ton of water. It's expensive to get rid of a couple megawatts like that. Some data centers have their own microclimates that are 10-20C hotter on average.
The Netherlands banned hyperscalers. But it's a rich country that can afford it.
At least a steel plant or refinery actually provides jobs for the locals. These American AI companies are just carpetbagger Yankies.
"First most of the data center build out is happening in areas that have had little other opportunity"
Speculation presented as fact...
Kind of self evident? If there were opportunity the land would have already been used for something more lucrative than empty.
fourth: It primarily benefits billionaires and corporations.
China invested hugely into renewable energy. Their grid is strong. Ours is falling apart and our renewable energy contracts keep getting shredded. Gas powered data centers is beyond dystopian
That doom take has a few roots in truth, but it is mostly false. Our grid for the most part is very good and improving. A lot of the gloom is it is good for today, but here is why we are trying to expand it anyway - that is the gloom itself is helpful to get the needed changes made.
There is a lot of renewable energy in the US, and more is built all the time.
It's really not "doom" it's reality. Highly recommend googling the state of the us grid. It's under funded otherwise data centers would not be running on gas. Similarly one of the biggest anticipated bottlenecks for new data centers is power right now.
https://www.fpri.org/article/2025/11/data-centers-at-risk-th...
https://www.nytimes.com/interactive/2026/04/27/opinion/elect...
Although renewables are on the rise in use, huge projects have recently been cancelled in favor for gas. Meanwhile the us is going around blowing up and encouraging other gas producing nations to blow up gas infrastructure.
https://thehill.com/policy/energy-environment/5851702-trump-...
https://www.insurancejournal.com/news/international/2026/04/...
https://www.aljazeera.com/news/2025/10/15/how-successful-is-...
It's only doom if we willfully do nothing to make it better or ignore it until it becomes a problem we can no longer fix.
If you're worried abour grid health, how is adding distributed generation colocated with new loads dystopian?
Sure, it would be nice to be able to build more transmission lines and power stations wherever it makes sense for engineering to build them in order to build a strong grid. But that's hard to do with strong private land ownership and required environmental impact reporting.
Something something texas avoiding federal electric regulation.
Gas = greenhouse gases. We needed to be reducing those decades ago, not increasing them.
It's unsustainable and bad for the local environments surrounding it. How many people need to die so people can chatgpt a social media post or summarize an article? These are real unanswered questions we are facing because the US has inadequate energy infrastructure.
https://sustainabilitydialogue.uchicago.edu/news/data-center...
You ever lose power and run a generator? The stink and sound is awful. Imagine powering more houses than there are in a suburban town with gas? Imagine doing that in hundreds of thousands of locations across the us. It's a dystopian thought.
I agree it is hard to stand up an effective grid to sustain technological innovations and products. It sounds like something we needed to be ready for like other countries are. Or maybe something that should happen first in order to be responsible and get ready for the future.
Don’t forget environmentalists have been fighting against nuclear energy for decades.
Update: Since I’m being downvoted I’m going to add a little sting:
Good luck with your solar power web servers and skipping flights.
Poor judgement really does have a price. It’s now inevitable that it’s going to get a bit warmer… everywhere. We’ll be lucky to hit Net Zero by 2100.
Last year, China installed 1.5GW of nuclear and 300GW of renewables. China's total coal use decreased as they decommissioned and under-utilized more coal than they installed.
This same pattern is occurring everywhere, regardless of local politics or local economic system. See Texas as another example.
It's because new renewables is superior at contemporary market prices. Markets have decided. Governments have decided. Everyone has decided.
It's so boring to relitigate this constantly on HN. It's like debating whether the sky is blue. Ridiculous that this comes up so often.
1 GW of renewables is not equivalent to 1 GW of nuclear or coal capacity,you need backup.
China is still building large amount of new coal power plants. In 2025, construction began on 83GW of new coal capacity – down from 98GW in 2024.
https://www.carbonbrief.org/rush-for-new-coal-in-china-hits-...
For comparision the total EU’s existing coal fleet is 109 GW.
https://ember-energy.org/latest-insights/coal-is-not-making-...
The growth Chinas coal consumption is slowing down, but still growth no reduction in 2026.
https://www.reuters.com/sustainability/boards-policy-regulat...
However you may try to put lipstick on a pig, the core of the matter is unchanged. China's coal use decreased, and they barely install any nuclear compared to renewables.
Nuclear is useful but it's hard to see it as a panacea. Renewable energy on the other hand is hard to beat. The economics of it keeps getting better, and previous estimates of the lifecycle of things like solar was grossly misrepresented.
Cancelling wind power contracts etc was a huge mistake.
"Environmentalists" is a large, diverse group and nuclear energy has been a controversial topic splitting the group for decades.
Many environmentalists are pro-nuclear, and viewed exclusively through an environmental lens, nuclear is likely the best energy source.
Other people share the "environmentalist" label because they care about clean air, unpolluted rivers, biodiversity, climate change, etc but they oppose nuclear on unrelated grounds (eg, as part of an anti nuclear weapon proliferation agenda) or out of fear of adverse events from damage to an energy facility.
The "pro-environment but anti-nuclear" subgroup held power within the Democrat party in the US through most of the cold war era. The "pro-environment, pro-nuclear" subgroup is now the largest group within the Democrat voting base, but some of the people and all of the regulations from the 1960s-1990s are still in power.
Good reply; factual. But in case you are not aware, “Democrat party” is a bit of a slur [1]
[1] https://en.wikipedia.org/wiki/Democrat_Party_(epithet)
"Environmentalists" have been fighting against many things for decades: eating meat, deforestation, whaling, air travel, mining, automobiles, fur.
Assuming it was really environmentalists who did it, why is nuclear energy the only thing they succeeded at stopping?
Yes, and that dates from the era when everyone thought that a "nuclear winter" (from an exchange of nuclear bombs between the US and USSR) was a more immediate risk than a long slow problem of climate change. That's what the French government carried out a terrorist attack against Greenpeace in New Zealand for: the Rainbow Warrior was protesting against French nuclear weapons testing.
Has everyone forgotten that the current crisis kicked off over the question of exactly how much uranium of what purity Iran is allowed to have? Do people really think that every country in the world should have multiple nuclear reactors?
Do people really think that uranium consuming reactors that produce plutonium are the only type of nuclear reactors?
Of the operating set of reactors today, what percentage are those vs., say, Thorium?
yes, but that is not the point. The point is that countries pursued uranium reactors as a priority to produce plutonium for weapons. It did not / does not have to be this way, and the conflation of all nuclear reactors as either being at risk of meltdown or linked to nuclear weapons has resulted in the current situation. The green movement did us all a disservice by choosing not to clarify the difference, and here we are in 2026 still burning coal, which continues to produce more radioactive waste than all the nuclear reactors ever built - and released directly into the environment!
There are many claims that Thorium fuel cycle is nuclear proliferation safe, it is safer but not perfectly safe. Uranium-233 bread during the Thorium fuel cycle can used to construct nuclear weapons
https://en.wikipedia.org/wiki/Thorium_fuel_cycle
https://en.wikipedia.org/wiki/Uranium-233#Weapon_material
"A declassified 1966 memo from the US nuclear program stated that uranium-233 has been shown to be highly satisfactory as a weapons material, though it was only superior to plutonium in rare circumstances."
Many nuclear energy opponents require absolute nuclear proliferation safety.
I was under the impression that Carl Sagan basically made up the concept of nuclear winter because of his political alignment.
And, yes, every country should have nuclear power and nuclear deterrence. Please contrast how the US negotiated with North Korea and Libya. Or what happened to Ukraine after they denuclearized. You might be impressed at how palatable peace becomes when leadership has to consider their personal safety and not just cynical economic and political calculus in their use of force.
Yes.
Please. Show me a nuke that was under 10 billion over budget and less than 20 years late.
Seriously. Don't even try to insert nuclear power into the debate before you show it can be competently built, never mind safely run and waste safely disposed. But what about the French you say! The French are discovering they forgot to set aside money for decommissioning.
This seems to be a western issue. China has been producing nuclear reactors quite cost effectively.
Radiant with their modular reactors seems to be doing quite well.
[dead]
If you look at how much time and money the French poured into their nuclear plant construction project (and several other pre-decommissioning issues) things look rather less rosy.
Also, it was a very high priority for France - keep in mind Napoleon's quote about the relative importance of moral and material.
More critically: The logic of "The French did {extremely complex thing} fairly well in the late 1900's, therefore we can also" is very similar to the logic of "SpaceX designs and operates reusable rockets with hundreds of launches per year and no failures, therefore we can also". NO, sorry, you are just fantasizing about rocket science somehow being easy for you. SpaceX has lots of highly motivated would-be competitors - many of them far better financed than SpaceX was - but the hard fact is that zero of those can actually do what SpaceX is doing.
Over 15 years France installed 56 nuclear reactors.
https://en.wikipedia.org/wiki/Nuclear_power_in_France#Messme...
The costs were not small, but probably smaller than the costs of German Energiewende. Messmer plan needed not only money, but also capable heavy industry and consistent political support.
https://theinnovationattorney.substack.com/p/the-messmer-pla...
https://carboncredits.com/nuclear-education-how-germany-lost...
Had France have such rich coal reserves as Germany, they would probably build coal power plants instead of nuclear power plants.
Germany 39022 millions short tons
France 176 millions short tons
https://en.wikipedia.org/wiki/List_of_countries_by_coal_rese...
Something tells me if they manage to get nuclear plants accepted, they will be built far, far away from the billionaires' doomsday bunkers.
A lot of environmentalists are just degrowthers and NIMBYs. During the Biden administration multiple groups sued to block high voltage transmission lines for clean energy.
As they should, now that we have renewables and batteries.
Even if modern renewables and batteries kill the need for future nuclear power plants, that doesn't excuse the consequences of decades of burning fossil fuels because environmental groups fought against nuclear power altogether.
We had better options back then, and we chose not to implement them while slowing down efforts to improve them: nuclear reactor designs could have been standardized to lower cost, even safer and more effective reactor designs could have been pursued years earlier, etc.
The costs--and opportunity costs--of inaction during that time were massive, and we're going to be paying them for generations. Renewables have a heavier lift ahead of them as a result, with less time to build out and upgrade the grid, transition to EVs, etc. The very least we can do is acknowledge the consequences.
Environmentalists are full of scammers.
Right wing idiots are against solar and wind. Left wing idiots are against nuclear… leaving us with no alternative other than gas and oil!
The common denominators are “idiots” and oil.
Because it’s a dangerous technology where profit over security can have severe consequences.
Not to mention the new risks now that drone wars are a thing. All those Chemical plants are already valuable target, no need to additional nuclear ones.
>Because it’s a dangerous technology where profit over security can have severe consequences.
Yeah it's far better to have power plants kill a steady stream of people, but in a banal way that's hard to attribute, like coal power plants causing lung cancer.
No, but hydroelectric works pretty well. There's a reason they recycle aluminum, etc. (power-hungry industries—crypto-mining, ha ha) along the Columbia River, etc.
So build your data centers there. No reason to choose the least evil.
Hydro electric as a resource is probably mostly exploited - are there a lot of big hydro projects left to build? If there are they must be difficult or expensive or they would have already been built.
In the Cascades they’re deconstructing dams as the benefits were oversold while the maintenance costs and environmental impacts were underestimated.
The Green movement in Australia was started over the successful blocking of a hydroelectric dam. Rivers are usually highly ecologically significant.
Which is why coal power plants are being shuttered everywhere in favor of gas and renewables (and I suppose nuclear, sometimes).
It's not nuclear or coal. There is wind, solar, etc.
Because the only options are coal or nuclear?
Wind and solar with storage is far better than nuclear.
Even ignoring nuclear environmentalists are still the enemy here.
Green energy is diffuse. Fields of solar. Ridges dotted with turbines. Each unit needs a power cable running away from it, access roads, etc, etc. A lot of area has to be developed for a given payback (sellable power).
Environmentalists and have been instrumental in making it economically impossible to develop land cheap enough to make low value density projects like that pencil out.
The higher power cost states in the US would likely be dotted with all manner of solar infill if not for up front costs that these shortsighted and selfish people have imposed on any land alteration larger than approx SFH lot size. Farmer Johnson would love nothing more than to put up solar on that ~2ac hillside he owns but cannot farm economically. Neither he nor some 3rd party who would put up the panels will shell out half a mil for an EPA CG permit just to clear the vegetation because the panels will never pay that back in their lives.
The economics of compliance are why the only greenfield development that happens these days is value dense commercial stuff (shopping plaza, big box store, data center, office buildings etc) or dense SFH development.
The most rosy possible outlook is that we "just" wait for the hippie boomers who cooked this crap up to croak, shit can the clean water act and come up with some new way of regulating development that doesn't saddle these super low impact projects (it's hard to be lower impact than panels or turbines) with fixed-ish costs that are a non-starter except at huge scale.
But of course there are so many parties who are making rent off this broken system who won't go down without a fight.
> the potential to create more greenhouse gases than the country of Morocco emitted in 2024.
Does that comparison mean anything to anybody?
For some context, it's a country with 40 million people.
But it’s hardly industrialized right?
So is that a fair comparison?
It actually sits right about in the middle of all countries for percentage of GDP from industrial sectors.
It's pretty heavily fossil-fuel powered right now, but like most of the rest of the world Morocco is planning to capture most of its growth in energy demand with renewables. Because, getting back to the original discussion, it'd be idiotic to choose fossil fuels over renewables in 2026.
100% agree with you. Especially with the amount of sun they get, would be a great candidate for Solar.
Renewables also seem to be the best pathway for many nations to attain some level of energy independence (which is NOT the same as oil independence though).
Are you, like, just not very into reading things or remembering things or is it just one of those things where you're stuck in the 1980s and haven't updated your books since then.
https://en.wikipedia.org/wiki/Economy_of_Morocco
Yeah, clearly I haven't done any meaningful reading regarding Morocco. I will read up more to fix the gaps in my knowledge. :-)
Another data point
"New AI data center in Utah will generate and consume more than twice the amount of power the entire state uses"
https://www.tomshardware.com/tech-industry/kevin-o-learys-9-...
I wouldn't be surprised if they use this as an excuse to open up public lands for extraction.
Yeah that’s quite the yardstick they’ve chosen
And the sad part is that ai would be a perfect fit for intermittent energy sources: run inference 24/7 (on whatever you can muster, even if it's fossil it simply does not consume all that much compared to training) and over-build training to achieve whatever total throughout you need in time of energy abundance. How heavy would CO2 pricing have to be to make the market do this instead of the exact opposite?
You can't price out these people, they have seemingly unlimited money.
Presumably chasing states and localities with generous tax breaks as well. That is no doubt what is really driving the choice of location.
With unlimited money, you'd be able to do unlimited over-build. At least that's what we thought until recently...
What would be the economy of these data centers if they would run using power source with say, for example 25% capacity factor? What is the capital expenditure of a data center? What is the yearly operational expenditure of a data center?
I found that a 100 MW datacenter can cost roughly $3.35 billion, with a significant portion going to high-end GPUs like the B100 or H100. For electricity, 100 MW data center can incur annual power costs ranging from $41 million to over $131 million, depending on regional energy prices.
Compute has massive initial capital requirements, and a very short obsolescence window. It does not make sense to invest in compute and then not run it 24/7 at 100%.
Hardware sitting idle waiting for sunshine/wind is leaving money on the table. Especially if you can burn gas during that time and still be hugely profitable.
Would the world be inherently different if hardware cost was 4x of what it is per unit of training done? The world would be inherently different if we did not burn through fossil stockpiles at the speed we do, or not at all. It's market mechanisms failing to lead to a desirable outcome. Nothing new about that, plenty of problems of that kind have been solved before, but yes, this one is more difficult.
New conspiracy theory: this is all secretly a big bitcoin mining buildout
Very quick Googling suggests Wired’s estimate would be ~1.9% of US emissions.
AI data center investment is, at core, a bet on increasing the productivity of labor. That’s what businesses will pay for, and what will earn the big money.
If US labor productivity rises by more than 2%—and implicit in the size of this bet is a guess much higher—US carbon intensity goes down correspondingly, and these data centers end up as a win for the climate.
Problem is climate and ecosystem doesn’t give a fuck about some notional productivity or carbon intensity just the absolute number of pollution molecules going into the air
[flagged]
You are supposing that with more productivity people will emit less carbon but there's no mechanism for that
More productivity means the employers just demand more from the workers
Yeah, this seems like a textbook case where one could apply Jevons Paradox.
OK well now you have to look at changing the economy-wide energy mix or embracing de-growth. Switching data centers to 100% solar or nuclear or … solves Wired’s complaint but not this one.
"with increases in productivity, we'll all work 2-4 hours a week and maintain the same output!"
No, you'll work 40 hours and just do 10 times more in that time. Same thing.
The "climate" cares little about carbon intensity of labor or dollars. It cares about absolute tons of green house gases in the atmostphere and if you use more to produce more you still use more.
You should always assume these that benefit most from the technological forefront think they can outrun the output of climate change or any of the outputs.
It generally is called "effective altruism", eg, techno jesus will solve whatever problems creating techno jesus creates.
> If US labor productivity rises by more than 2%—and implicit in the size of this bet is a guess much higher—US carbon intensity goes down correspondingly, and these data centers end up as a win for the climate
I can promise you it won't happen in a million years. More productivity lends to more exploitation, because you can do more with the same unit of work, instead of getting the same result with less work.. Or at least we have decades of data proving that is what realistically happen. So the only way to reduce emissions is either using carbon neutral sources (gas is... not?) or forbidding people from using energy in the first place (and let's be honest, that will not happen.)
Or the third solution: a carbon tax.
That is a tough sell in the current environment. It's a regressive tax, so opposed on both ends of the political spectrum. People on the far right don't believe in climate change, and people on the far left don't believe in market efficiency. With 20% of the world's oil flow crimped in the Strait of Hormuz for who knows how long, higher energy prices is the last thing people want to contemplate.
In the longer run, a carbon tax is the best option. The fossil fuel price shock is a strong signal to produce energy through other means. There are major engineering initiatives around developing cheaper and safer nuclear energy. and it's cheaper now to deploy a solar farm than a coal plant.
A carbon tax would raise money to pay off national debts and encourage consumers and producers to figure out the most efficient way to accomplish their needs while minimizing their carbon footprint. It's a tough sell today, but this is they way to go for a better quality of life tomorrow.
>More productivity lends to more exploitation, because you can do more with the same unit of work, instead of getting the same result with less work..
But per-capita greenhouse emissions have been falling in much of the developed world? And you can't really claim with a straight face that productivity has been dropping from 2000 to today.
https://ourworldindata.org/grapher/co-emissions-per-capita?c...
> But per-capita greenhouse emissions have been falling in much of the developed world?
Only by the deceptive accounting trick of not including the emissions associated with overseas production of the goods consumed by the "developed world".
If you include all the emissions that prop up the highest per capita consumption patterns on the planet then you see the highest per capita emissions attached to the highest consumers.
>Only by the deceptive accounting trick of not including the emissions associated with overseas production of the goods consumed by the "developed world".
That does increase US's emissions, but not enough to change the conclusion:
https://ourworldindata.org/consumption-based-co2
Emitting more greenhouse gas is only a win if your goal is to hasten and worsen the climate catastrophe.
Carbon intensity is not the relevant metric.
And there's no evidence or historic precedent backing your idea that it would go down anyway.
yo, "productivity of labor" isn't how they're selling this stuff. Its "replacement" of labor.
https://www.theguardian.com/world/2026/apr/01/iran-energy-cr...
It’s quite impressive how the world is unable to reduce greenhouse emissions.
In history of the world there was almost never a reduction of greenhouse emissions. Only time that world did reduce greenhouse emissions was in times of economic crises and COVID-19.
https://ourworldindata.org/co2-emissions
Asia will do what US did in 1979, Jimmy Carter delivered this televised speech on July 15, 1979.
"Point three: To give us energy security, I am asking for the most massive peacetime commitment of funds and resources in our nation's history to develop America's own alternative sources of fuel -- from coal, from oil shale, from plant products for gasohol, from unconventional gas, from the sun."
https://www.pbs.org/wgbh/americanexperience/features/carter-...
https://archive.ph/L1twq
Ah, I knew electricians still have a future.
Burning natural resources for bitcoin and chibi cartoon figures is the goal of this administration.
All while increasing natural gas prices through blockades and threats of secondary sanctions.
It will make oil billionaires (e.g., like the ones who founded the Daily Wire) very happy.
Our species is going to go extinct over samurai cat memes.
Meanwhile anyone can run a capable LLM in their own living room, or even on their own mobile device, but nobody wants to talk about that. And no, 99.999% of slop does not need a math Olympics finalist model, sorry.
[dead]
[flagged]
What?
Heavy fossil fuel use leads to health problems and given a large enough population deaths. Lung cancer, Asthma, heart issues etc
Guessing that is what GP means by body count
Emitting as much greenhouse gas as a country?
That has a body count. It's hard to know which people you're killing, but there's no doubt that at that point there's a number. Maybe... 10,000 or so? They'll die in heat waves and floods and tornadoes and food shortages in dry, poor countries. Obviously 10k is a tiny fraction of the full list of climate fatalities.
But isn't it kind of one-sided to single out data centers like that?
Most emissions are harder to reduce because industry isn't electrified (it runs directly on oil/coal not electricity). This is electricity usage. Electricity emissions are the easiest to avoid.
AI industry also promises to replace everyone, and they have access to large amounts of capital they could use for renewables.
That they punch down on everyone, including people outside of their country who don't benefit from the tax revenue they'll generate, with massive amounts of emissions, is a slap in the face.
Sure. But a race to the bottom isn't a vision for the future.
Tech used to offer society a vision for the future.
It doesn't help that I think datacenters are the means of digital oppression rather than the means of digital empowerment.
Oppression is where you get people hooked on something but then lock it away and use your complete control over supply to extract rent. Gradually the rent will become so high that only the most abusive profit-seekers can afford to pay it.
Empowerment is owning technology so that its purpose is to work for you, rather than to extract rent from you. Tech that empowers individuals usually doesn't need giant chain-link fences around it.
That's not to mention talking people to suicide, which by the way is the far more likely way that it would kill your son or daughter, should you have or ever have one.
And for all that cost, are they bringing us real hope? The most ambitious people talking about this technology basically say that when it takes over all thought, there won't be much point in humans anymore.
We went from being evangelists of a message of hope to evangelists of a message of doom.
I'm not going to be one of those people. Fuck those people. I believe in the future. I will stand up and tell the next generation there is still hope, still compassion, still community and humanity and love out there. I fight for the users!!!
Have you ever been in India while they have those incredibly hot heatwaves in the summer? That kills thousands of people every year. This will only get worse the more greenhouse gasses we emit so slop machines can generate funny videos of the pope breakdancing.
good
I hope the AI bubble bursts before our earth is too hot to survive on for millions of people.
Not to be a doomer but I'm confident we passed the point of no return decades ago.
Probably yeah, but we could at least try to not make it worse, though even that seems very hard for the tech grifters in silicon valley.
> New gas projects linked to just 11 data center campuses around the US have the potential to create more greenhouse gases than the country of Morocco emitted in 2024
"Have the potential to", "Morocco". Presumably doesn't count the greenhouse gases emitted by Moroccans using overseas cloud services and AI.
At least the example wasn't Vatican City.