ZH version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
53% Positive
Analyzed from 4056 words in the discussion.
Trending Topics
#data#pollution#centers#power#more#air#center#minor#electricity#deaths
Discussion (105 Comments)Read Original on HackerNews
'The largest estimate for any single facility ever.'
'Minor source.'
I give you double-speak, gentlefolks of HN. Name it something - anything - and you frame it in those terms forever. It will take significant work to not refer to these as 'minor sources' when debating if they are a minor source.
From the researcher: 'We estimated between $53 million and $99 million in annual health damages from associated air pollution — among the largest health-damage estimates produced for a single facility to date.' [0]
Still enormous, but there is a significant difference between being 'THE X of all time, and being 'among the X of all time'.
Regarding 'minor source', in this case the expected additional premature deaths in the research range from 3.4 to 6.5 annually. [1] At the same time, air pollution leads to about 100,000 deaths a year in the US. While tragic at an individual level, adding 0.006% to the annual premature death count is fairly minor, just as reducing the deaths to 999,994 would be considered a pretty minor fix to the issue.
[0]: https://hsph.harvard.edu/news/analyzing-air-pollution-health...
[1]: https://www.pecva.org/wp-content/uploads/Health-Impacts-Vant...
[2]: https://sustainability.stanford.edu/news/identifying-sources...
I just love the irony of being absurdly callous and dismissive of human life while also getting the basic math wrong.
Nevermind that the 100,000 figure is across the nation of 300 million and includes things like automotive exhaust, while the other number is likely to be borne by that one specific community. Indeed, if you look at the study[0] your article cites -- which is looking at what we should be working on reducing -- industrial deaths are under 30,000, and electrical are around 10,000. So below half of your figure.
[0]: https://pubs.acs.org/estlcu/article/7/9/639/504911/Reducing-...
Reading these kinds of things make my blood boil, and I'm not someone who gets angry.
- Existing air pollution causes a ton of harm.
- Realistically, the optimal amount of air pollution is probably not zero.
- Logically, we should treat the cost/benefit of adding this "new" air pollution the same as the cost/benefit of removing existing air pollution.
- As an absolute number, the harm caused by this facility is very large, but as a percentage it is tiny.
People seem to be interpreting your conclusion as, "therefore it's completely fine to build this facility because only percentages matter, not absolute numbers." But I think your conclusion is more like, "our society has a complex relationship with air pollution and our existing choices reveal that we are willing to accept some air pollution sometimes, so you have to do a cost/benefit analysis, not just look at costs."
How many deaths would go from minor to major? How much impact to quality of life that isn't death would there be?
There is already very little to make these projects pay the externality costs; at this point paying enough remediation costs to reduce additional deaths should be the minimum.
That’s a good question. It seems in this case that 3 to 6 deaths a year qualifies as ‘minor’ for the government. I don’t know where the line should be but, based on the number of people that suffer and die from air pollution related causes in the US each year, I thought that 6 additional deaths probably meant it qualified as minor. I’d be just as interested as you to know what the answer is.
Edit to add: ‘minor’ in this case doesn’t mean people deserve to die or that we should be happy they are suffering, just that compared to some other even worse projects, this one is less important.
“It’s not that bad, just a little bad, so it’s okay”
We’re talking about air pollution and death rates, not financial risk.
A minor source should never lead to deaths or to put it otherwise No deaths = minor source
deaths = major source
Many deaths = majormost source
Now we just need define many but we already know its a major source of pollution that kills people
The air pollution does not matter, because global warming already exists and air is already polluted. And making all that worst also does not matter, because who cares about anything but what AI companies want.
> Based on Virginia Department of Environmental Quality (VA DEQ) air permit No. 74242, the facility is authorized to operate eight natural-gas simple-cycle turbines and more than fifty diesel generators, including emergency and black-start units
The article lumps data centers in with landfills, laundromats, and steel mills. But landfills, laundromats, and steel mills all involve physical waste products that can have chemical or biological effects on the surrounding environment. What waste products do data centers output?
The article talks about expanding a fossil fuel plant as the source of the pollution question in the Virginia example. This seems only tangentially related to data centers, which don't care what specific technologies are used to generate fungible electricity.
Are these disingenuous attempts to attribute pollution originating from power plants to one specific, cherry-picked application of electrical power (despite the increasing prevalence of clean generation technologies over polluting ones, and despite lots of other, unrelated drivers of increased demand for electrical power)? Or is there something more to these arguments that I'm missing?
Many local grids are currently unable to meet that demand, therefore various "solutions" so far include:
* buying up jet engines from COVID era bankrupted / downsized air fleets and running them continuously in banks along side data centres to spin generators.
You'll note these aren't the cleanest solution, are a bit noisy, and not renewable.
* building a data centre AND and onsite LNG gas extraction facility to directly on site gas supply and derived electrical power.
This one is expected to be the single largest CO2 source in mainland USofA (or some similar stat, it was recent news).
So these are fossil fuel polluting sources that are directly attached to specific data centres and not being used for non data centre purposes.
I don't think it's reasonable to regard concerns about emissions from banks of continuously running jet engines as "opposition to data centers" any more than concerns about, using the same jet engines to power charging stations in remote areas would, for example, amount to "opposition to EVs".
Energy production consequences aside, they produce tons of e-waste
We'd originally set this up to be able to locally run larger models, in the 300-400 billion parameter range, but the rate of improvement of open-weight models has been so fast that, coupled with extensive custom skill creation, we're now getting similar results out of Qwen3.8 27B to what we were getting out of Qwen 3.5 297B when we started out with the project, which frees enough memory to allow 20-25 users to have 256K context concurrently. Both the hardware, the software, and the models are improving at an accelerating rate.
Investing in data centers to support SaaS LLM providers today feels a bit like investing in mainframes and minicomputers in the late '70s, with a massive paradigm shift lurking right around the corner.
Actually, it's probably already closer to the early '80s, given that purpose-built local AI workstations are already available at price points lower than the inflation-adjusted initial price of the original IBM PC.
And even if you do inference at home you are not training the models. Moreover, most people certainly are not doing inference locally.
What? (Edit: I think you misread "in support of their primary concern, being AI" as "in support of AI"? People who complain about pollution are usually mostly concerned about AI, and they use arguments of pollution to strengthen their argument against AI related things.)
> And even if you do inference at home you are not training the models.
No, and training does use a large amount of energy on a large amount of hardware. But:
- That sort of workload doesn't really require many distributed data centers, only a few powerful ones. I believe training is also getting cheaper for the achieving higher levels of capabilities, but I don't think the efficiency advancement has been as dramatic as it has been for inference.
- I believe we are really hitting a wall of diminishing returns, especially at the high end of large models. There is lower demand for training, because models are good enough to have a reasonably long shelf life at this point. Year+ old models that were SOTA in their time are still useful today. The demand for training is going down.
> Moreover, most people certainly are not doing inference locally.
Maybe not, but I think most people who go out of their way to use LLMs because they find them useful for their work actually are. Though most inference is probably from people who do it accidentally (as a part of a search result or something) or students who don't have the resources to do it themselves. But basically every software engineer that I personally know that uses AI for programming is either running their inference on their own hardware, or is talking about building a rig to do it.
While I cautiously circle the AI programming concept, one thing close to my mind has definitely been whether the ultimately cloud-tied nature of it is what makes me much more stand offish to it.
I also use pi.dev as my agent harness, and ollama as my inference engine. Though I've been considering switching out the latter for something else, since the main advantage of ollama is the ease of switching models, which I don't really do often anymore.
In most communities these days you’d have better luck proposing to build a toxic waste processing facility than a data center.
https://ferc.gov/
https://ferc.gov/ https://nofrackedgasinmass.com/massachusetts-pipelines/
Of course building datacenters is absolutely not what eminent domain is for, but rules don't apply anymore, so they'll probably do it anyway until the supreme court stops them.
Somebody has to realize that gas turbines are backordered to the 2030s meanwhile the ROI on solar is getting impossible to ignore.
The only fair and workable solution is to make the data center investors pay for the grid upgrade before the data center is turned on. They can wait a decade for nuclear reactors, right?
E.P.A. Moves to Curb Public Input on Data Centers
https://www.nytimes.com/2026/08/25/climate/epa-data-centers-...
At the same time, the more the US government acts to support data centers, the greater the reaction will be against them. It is the classic situation where the more the government does, the less people will like it.
They will have to be robotized to a large degree. And the few humans necessary will have to be very well paid. The social cost is real.
Edit: When I say the middle of nowhere I mean say an hour’s drive from civilization, not the Arctic Circle.
Murmansk (population 295,374) Russia; Norilsk (178,018) in Russia; Tromsø (75,638) in Norway; Vorkuta (58,133) in Russia; Bodø (52,357) Norway; Harstad (24,703) in Norway; Kiruna, Sweden (22,841)
And of course, when the citizens selling THC beverages were concerned about the legislature making the product illegal.
Me thinks the Data Center people have been stingy with the checkbook.
Growth can be done carefully and well, or it can be cancer. The current data center boom is not being done carefully and well. How you do things matters a lot.
At least were not at this point yet:
"Mill and factory owners took to shooting protesters and eventually the Luddites were suppressed by legal and military force, which included execution and penal transportation of accused and convicted Luddites."
https://en.wikipedia.org/wiki/Luddites
Luddites wanted a rising tide to raise all ships.
Hackers used to be counter culture and want to push power down to individuals. If anything this techno fascist push to merge big tech with government while steamrolling humans is not hacker culture.
We should home brew data centers for open weight models and open labelled datasets.
The only threat to stable and affordable electricity supply is government regulations. The idea that we will from here on just have degrown and decline is insane. I'm not up for that. Which means we will always need more electricity, so time to get building more power plants. There is no bounds to how much electricity humans need, and the more we have the more prosperity we will have
As for water supply. Data centers barely make a blip on the water usage.
> The water, light and noise pollution?
The only one of these which actually has any validity is noise pollution, and the US has one of the lowest population densities of any country in the world, and data centres don't need to be close to population centres.
> The threat to jobs?
CEOs saying they are laying people off because of AI is not the same as anyone losing their job because of AI, but even if that was the case, it's about as smart as not using an electric drill to save jobs.
> The threat to truth and transparency?
AI does not only not threaten truth and transparency, it makes it easier to find the truth and to circumvent transperency restrictions.
> The threat to privacy?
It's not.
There are bounds to the Earth's capacity for our consumption. We won't be prosperous if ecosystems collapse.
That is straight up not true. Theres currently a threat due to the imprudently started war in the Middle East, there will be more once rivers start drying up and hydropower becomes unreliable or nuclear power plants can no longer be cooled.
> Which means we will always need more electricity, so time to get building more power plants. There is no bounds to how much electricity humans need, and the more we have the more prosperity we will have
This is an ideological dogma smuggled in as a self-evident truth. Please do make a point, not just an unsupported declaration.
Look, I think you misspelled "demand" there. It's an understandable error; those keys are so close together.
Data centers present a demand on resources. A demand that wasn't there previously. You hook a DC up to the grid, and it will demand more electricity from existing generation capability. The DC proposals, without provision for their own generation, present extra demand.
So yeah, unless the supply of those resources can be commensurately increased in some way, the DCs will present extra demand on what we already have, and that means someone, somewhere, will have less available.
- no NDAs signed with elected officials who then stonewall the public and hide behind them - paying the same prices for water and electricity that I pay - paying the same property taxes and income taxes that I pay - building in a fashion that doesn’t leave area residents stuck with bigger electric bills and water bills
All of these things can be done.
I have a tiny data centre. I get no sweetheart deals on the above tho bf s.
yes.
Let’s put the data center in Palo alto or Atherton.
But may I suggest we put it right next to Zuck's house? How about in SamA's back yard?
They are noisy buildings, designed to turn electricity into heat. And provide a small number of jobs.
It just doesn't make sense! Surely they were manipulated to hate progress!
The general population is uneducated enough to vote for Trump. Our population was easily infected with anti data center propaganda from China.
Next time you see anti data center propaganda, look at the account. Too often it appears to be an account from India or rural US. I imagine the former are being bought and the later is too uneducated. Data centers are one of The Wests