'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-...
“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, which 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?
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 we were getting out of Qwen 3.5 297B when we started out with the project, which frees enough memory allows 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 hte 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.
In most communities these days you’d have better luck proposing to build a toxic waste processing facility than a data center.
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.
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.
Edit: When I say the middle of nowhere I mean say an hour’s drive from civilization, not the Arctic Circle.
Let’s put the data center in Palo alto or Atherton.
- 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.
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.
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.
The US does not use crude oil from the Gulf to generate electricity. The US could scale electricity production at least 100 times without needing to import anything from the Middle East.
> there will be more once rivers start drying up
Rivers will not be drying up because of data centers.
And yet, reserves are at historic lows. Markets are weird, huh?
> Rivers will not be drying up because of data centers.
That wasn't the claim made, rivers will be drying up because of climate change (they are already!), which is a threat to energy supply. Data centers will accelerate said climate change, but that's not even relevant to disprove your previous assertion of regulation being the only factor.
I'll also note you still haven't justified your core ideological assertion you tried smuggling in earlier in any way.
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."
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.
yes.
It just doesn't make sense! Surely they were manipulated to hate progress!