ES version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
39% Positive
Analyzed from 3927 words in the discussion.
Trending Topics
#data#energy#centers#water#power#center#more#don#prices#electricity
Discussion (118 Comments)Read Original on HackerNews
Why is that?
Where I live my energy bill is a fixed cost. It's a super expensive fixed cost, since I'm living in Germany, but it is fixed none the less. It increases year by year because all costs only go up, but it remains completely unaffected by unrelated infrastructure being built around where I live.
I've been to a data center before due to work and my impression of the place was really good. It's kinda invisible from the outside, everything is extremely clean. There's no noise on the outside and the excess heat is used to get people cheap warm water or something like that.
I'm quite confused why people suddenly hate on them. Feels like I'm missing a whole lot of context.
In the United States at least there's been plenty of polling on why people oppose data centers. Consistently, it's water usage and water contamination fears, which is exactly what you see most often in social media and discussed online. I think a lot of high-minded rationalizations are just trying to dodge the embarrassing fact that people simply think data centers are draining municipal and natural water sources.
Monthly private spending on data center construction:
2015: $168 million
2024: $1.59 billion
10x increase over 9 years (26% annual growth rate)
2026: $4.36 billion
3x increase over 2 years (73% annual growth rate)
Let me reiterate: MONTHLY spending.
Compound this with presumably we've gotten better and not worse at building data centers (if nothing else, just building larger data centers would create economies of scale), the construction dollar to power consumption ratio has probably moved in the wrong direction for your argument.
The 2015 - 2024 growth curve would conceivably encourage more consistent usage of baseload and, where insufficient, gradual construction of additional supply. That would have the expected effect as seen in the paper.
The latter growth (which is obviously accelerating), especially after an extended period of the former growth, would not necessarily do this. We in fact know this to be true because the rate-limiting step on data center construction is lack of energy.
So obviously there is not ample excess baseload just sitting around. Obviously it's non-trivial to add additional capacity at the speed data centers need it.
Here's a recent source: https://pacinst.org/publication/how-much-water-do-data-cente...
Farms use an order of magnitude more than either. You don’t see any demands whatsoever on agriculture to reduce water usage.
> How can higher demand lead to lower prices? … a bump in demand can allow for higher utilization of the existing generation … Provided the bump does not violate capacity constraints
I.e. data centers reduce electricity prices, as long as you don’t build new generation to support them. Whereas data centers, nationally are projected to increase energy use between 325 to 580 TWh annually by 2028 [1]. There is no way you are going to meet that without building new generation assets to meet that demand.
[1] https://www.energy.gov/articles/doe-releases-new-report-eval...
"We estimate that data centers caused average retail electricity rates to fall modestly in the United States from 2015 to 2024 using an instrumental variables approach."
But the data center backlash really started 2 years ago. It's 2026 now. Getting new energy generation built and approved takes time.
"We caution that future supply constraints could reverse the effect."
There it is.
it's my power bill, dude. and everyone else's around me. your paper stops measuring that data right at the time when everyone's bills started increasing sharply.
Is it any wonder people hear that and go "hmm, no thanks" when the prospect of a massive whirring data centre next to their house is raised?
The SV LLM crowd know a lot about LLM psychology but seemingly very little about human psychology
They're detached from the great unwashed masses. It's not that they don't know anything about human psychology. They don't care about those people or their concerns, historically. And that's what is coming out in this interfacing with the broader public.
Here in Switzerland new AI datacenters are also causing some uproar but a lot less because regulations are rigid enough in most places that you can only build them if you do so cleanly. You can't just take all the water, electricity, make noise and polute.
I live next to a waste valorisation plant and they have very over sized water cooling systems because they need to keep the noise below regulation. They also for example have under pressure in the trash receiving bunker to prevent smells from escaping. All regulations that cost money but are the benefit of all.
About time.
Yeah, Elon's a big fan of this tactic.
Then these big buyers go ahead and pay 50% of what everyone else is paying.
Also, I live in Northern Virginia where we have tons of data centers, and energy prices have not spiked.
That is now being consumed and now Loudon County will also start engaging in electricity price wars with data centers. New regulations to segregate markets is supposed to kick in on Jan 1 2027, in fact.
But also, it seems to have increased in price quite a bit lately.
https://fred.stlouisfed.org/series/APUS35A72610
(Of course I might simply be projecting.)
I can almost guarantee there's a good bit of that also involved in addition to all the other concerns and issues folks also rightly have.
Three Mile Island
"2027 – Nuclear Regulatory Commission to decide whether to issue operating license to Constellation Energy, with the decision expected in May. If granted, Constellation expects to bring the reactor back online later in the year."
https://www.witf.org/2026/07/28/three-mile-island-reactor-re...
Michigan’s Palisades nuclear power plant
https://www.canarymedia.com/articles/nuclear/americas-first-...
Duane Arnold Energy Center in Iowa
https://www.reuters.com/business/energy/nextera-energy-partn...
https://en.wikipedia.org/wiki/Duane_Arnold_Energy_Center#Res...
Though we really should get better at building nuclear power plants. I'm think more of small ones to power ships though, not land based ones. Ships are one of the few places where oil/coal are used that there are not realistic renewable alternatives.
https://x.com/GrantChalmers/status/1422472474487263233
These new AI data centers are now built with their own power generation capacity. They're too electricity hungry to be supplied by the local grid, requiring megawatts to gigawatts of generation. If they're grid-connected at all (for training / inference racks + cooling) it is secondary supplementation.
There is a concerted effort by foreign governments to slow down American progress in fielding AI capability. Attempts to generate civil unrest and propaganda against $new_thing are included in this.
> These new AI data centers are now built with their own power generation capacity
And now you've got all the noise and pollution from that to deal with.
The outside agitators are in the news. The U.S. just happens to be a little more vulnerable to "straight to the people" propaganda campaigns.
S'okay. Facts and reality don't matter. Everything's peachy-keen, folks. Just wander on by and downvote reality and it'll magically go away... :sigh:
Also, adding a few data centers to a large grid shouldn’t bump up the energy prices, it’s only when you add many across the area. This is difficult to pin down because if one side of a region builds many data centers without their own power generation, the whole region could see eventual price increases. Fortunately most (all?) states regulate utility price increases, but still you don’t want to max out the price bump year after year.
It’s hard to discuss because different facilities have different configurations, and the local impact depends on that plus local characteristics.
Given the externalities of large facilities like this, impacted neighbors should definitely get a vote. Maybe we need a standard to define the impact radius based on the facility characteristics. This would add transparency and make it easier for people to decide without FUD and distortion. Unfortunately I don’t have much confidence in anyone coming up with such a standard without altering it to benefit their side, now that it’s become a political football.
What about concerns about air pollution from unregulated diesel generators? Or using large amounts of water in drought prone areas?
99% is an interesting figure because that's roughly how much my energy bill has increased despite no significant change in my usage due to data center demand on the grid
Regardless hopefully this restarts nuclear power generation
Not if they get a sweetheart tax deal. Why do you think folks are so pissy? It’s like a sports stadium next to your house in that the principals get tax breaks, eminent domain, and other considerations, but it makes noise every day(not just weekends) and you don’t get to go to the game.
The problem, from the cases I have looked at, seems to be that towns want to do dumb shit with this money rather than just cut residents a tax break.
tbh I'm more worried about the resource curse from the sudden change in budget
Unless local governments are getting bribed en masse, literally no municipality would want a datacenter if there wasn't a net benefit. I checked in my area and the datacenters are receiving tax breaks, but they also have to pay into the local school districts and fund the energy infrastructure buildout. Just because tax breaks are part of the deal doesn't mean there isn't a net benefit for the community.
There are times when Silicon Valley looks out of touch and there are times when Silicon Valley just looks like idiots with hubris. This is more the later.
They know very little about the job market impact of AI, and are mostly retired so they don't care anyway.
Can we put them on Mars, and maybe send Elon to tend them?
Lots of these commentaries miss the following: - noise and uncomfortable vibrations making nearby habitual living challenging - rises costs of power and water - more traffic and disruption to traveling patterns
The first is a big deal. These companies have a history of refusing to address noise and disruptions to neighbors personal health. They simply don’t care because they only care about the money. Its greed at an immense level. Until you’re ready to accept you would be comfortable with a data center as your neighbor you have 0 skin in making decisions for others who do.
And the rich bourgeoisie people who are so misguided about this: mark Cuban, Kevin O’Leary famous shark tank billionaires, who claim “you’re experiencing Chinese propaganda about data centers” and probably even anyone on these forums who makes a strong claim without experience, would never ever accept living next to a data center. They are sideline commentators and should take responsibility for the harms to local people.
But the sketchiest part of the bubble is when you directly bolt a no name data center company to Nvidia's cash pile, and you start turning cash into ever-lengthening receivables collection time, that's why people are skeptical of data centers: even if they can't tell a Google data center from a pivoting crypto bros data center, they smell the bullshit.
People also know they are being lied to, Oracle claims their New Mexico data center will run on Bloom fuel cells. I doubt Bloom is ready to scale to data center size without proof of concept in different applications at large if not directly comparable scale.
1. Lawmakers literally do not care what you think (study from Princeton) [1];
2. Private utilities are more expensive than public utilities [2]; and
3. Nobody goes to war over a solar panel [3].
All of this is relevant here.
The privatization of utilities has shown to be a massive mistake, in part based on debunked economics (ie the tragedy of the commons). Demand is very inelastic so they have to be highly regulated. The electricity price is set by lawmakers so how do they make profits? By building more infrastructure because that cost can be passed on to consumers.
So if a hyperscalar needs 6GW of power, you need to build massive transmission lines and power plants. That's great for private utility profits. And that's what we're all paying for. The DCs are usually getting subsidized electricity as well as not paying for the extra infrastructure.
So many things would be better if we'd had mass adoption of renewable power but that's bad for the wealthy. You install a solar panel and it just generates electricity every year. It can be built incrementally and quickly. It also tends to utilize land that is of little value otherwise, such as in the Southwest US. But fossil fuels make people wealthy. An oil or natural gas well declines in production over time. That means you have to keep building them. That's more profit.
This whole saga is revealing to a lot of people just how little lawmakers care about the voters. They'll ignore protests and vote anyway. They'll do it in secret. If they get voted out, who cares? They'll magically find their way into jobs for these companies.
[1]: https://act.represent.us/sign/problempoll-fba
[2]: https://energyandpolicy.org/utility-profit-report/
[3]: https://www.theenergymix.com/no-one-goes-to-war-over-a-solar...
For example, imagine you're worried about what goes on inside those DCs, are worried about the lack of accountability from Big Tech and would like to see sensible consumer-protection regulations. Can you express that anywhere? Sure, you can leave random comments somewhere, but that doesn't matter. OTOH, you can actually join a local NIMBY group and slow things down. It's not what you want, but it's better than nothing. Same goes for people worried about clean energy and other issues.
I haven't seen people really care and mobilize since maybe BLM?
The only lever the government seems to have is tax breaks, and the only lever the people seem to have is NIMBYism. What a joke.
I HATE that this is somehow becoming a culture war issue.
These companies have a lot of money behind them - if they're not willing to put in the effort to convince local residents that the data center will be good for them, I feel like that's on them. They could easily do some handouts and fix this. Offer to pay extra taxes to the state or whatever, fund a new power plant, pay everyone who lives nearby the DC some thousands of dollars, whatever.
But have they disproportionately amplified them? I follow some energy experts, and the consensus for AI-focused mega-datacenters seems to be "no". Some interesting reports:
• Artificial Intelligence: Generative AI's Environmental and Human Effects - https://www.gao.gov/products/gao-25-107172
• FERC (Federal Energy Regulatory Commission): "…approximately half of this [$5.1 billion] cost will be borne by customers…" - https://www.ferc.gov/news-events/news/commissioner-clements-...
• Federal Reserve Bank of Dallas: Processing Power: The Effect of Data Centers on Wholesale Electricity Markets - https://www.dallasfed.org/~/media/documents/research/papers/...