These datacenters can be built in ways to limit this kind of noise pollution, but it appears local leaders do not think about things like this that can truly harm their constituents.
I don't know anything about this particular site, but I presume it's one of the new mega gpu sites.
I'm seeing many people in the comments with an early 2000's era concept of datacenters. The scale of these new sites is mind boggling. Take your idea of a typical datacenter building. Make it 4x bigger. Then put 4 of them together into a cluster. Then imagine 10 of those clusters at the site.
> Missouri campaign finance records show a political action committee — made up of labor unions that support data centers because of the jobs they create — spent almost $40,000 in the final weeks of the race on newspaper and digital ads and yard signs in support of the four council members booted from office.
Serious question, what jobs do datacenters create?
A small number of jobs for tradesmen (electricians, plumbers, etc.).
A small number of jobs for security guards.
Maybe a tiny number (one to three?) for individuals tasked with actual hardware swapping within the data center itself.
And all of the above assumes the data center owner does not "travel in" the requisite individuals on an "as needed" basis -- in which case the only jobs that may go to the locals is "security guard".
But all of the "sys-admin" management level work can be done remotely.
So the actual number of new jobs that arrive in the locality is likely on the order of 20-30 or fewer.
Yeah and that type of work bid usually goes to huge conglomerates. A local mom and pop electrician shop isn’t going to be building a datacenter, it’ll be something like Siemens.
A friend of mine is an independent electrician in the Columbus, OH area. Last summer he told me he was getting plenty of datacenter construction work, albeit it was in the form of subcontracted jobs from the larger firms who were awarded the contracts.
Even if the datacenter does hire some local labor for construction, that's still all temporary jobs. It's not an ongoing source of employment for locals.
I would imagine there aren't too many electricians in Yeehaw, Minnesota, trained and qualified to do gigawatt data center installs. So they'll freight in contractors to do that work, and maybe temporarily employ a few locals for a month or two for auxiliary stuff.
More generally, this is the universal playbook when someone wants to dump some megaproject on a community that doesn't want it: This will create X jobs and inject $Y into the local economy. Can you name one case where this actually happened? It's usually very few additional permanent jobs and, particularly for public-works stuff, millions or even billions in extra debt to pay off. But don't worry, this next thing we're working on once we get the local council to issue a permit to bulldoze your forest park, that will bring in jobs, we promise.
I work for an electrical contractor that does large data center projects and we almost always partner with a local contractor to provide labor from the local union(s).
sure, that only covers construction though. Once the thing is built they are going to travel in all the maintenance that needs to happen, and that local tradesman is not going to get to many new home construction jobs after it goes in. Who wants to live nearby a noise polluter like that?
How many of these are on-going jobs vs during construction and as-needed? I think you're right it'll be only security guard jobs. Even if they don't travel in workers, it's quick short-term tasks that maybe locals can perform, but that's not "creating jobs."
This argument has always been such a weird goalpost shift for me. Even at my full time job I am getting strung together by 3-12 month projects. Everyone works on projects. When this data center is done in a year, we'll (hopefully) need to build something else, keeping those people employed.
Like, of course it's creating a job. If you create a million 1-year jobs every year, that's a million jobs.
Because the "it'll create X jobs" implies it's ongoing. It's a disingenuous attempt to oversell the benefits because they know if they're transparent about it, suddenly it doesn't seem like such a great deal.
Actually if they were honest they'd say "It'll create x careers", which is a much better deal than just a job. Friend went from 17 an hour at his first DC to 100/hr last year with 70 people on his team.
> A small number of jobs for tradesmen (electricians, plumbers, etc.).
Its no car dealership but probably a reliable source of work-orders. Seems like a "gigascale" datacenter would be a large job for a tradesman to be a subcontractor within and afterward its scale means continuous upgrades/maintenance.
Is there any literature of ongoing economic impact of similar facilities?
In a town of 12K people I'd say it's incredibly unlikely. Most of if not all the labor to build it will be flown in, most of the labor to staff it will be moved in.
And once it's built it's not like a Walmart or something where you need enough staff to police the crowds...there are not crowds. There's some rack and stack needs, and some ongoing cabling needs generally,and some other stuff, but they are staffed as lightly as humanly possible.
I suppose w/ all the out of town labor to build it there will be more waitress and hotel cleaning jobs for a while...a town or over...where they can actually house the labor.
Oh, and they are getting an Olive Garden...which will probably employ more local labor.
Festus isn’t small because it’s in the middle of nowhere. It’s right there with Arnold, Barnhart, Crystal City, and other far south suburbs of St. Louis. The metro area can build it. It’s not like Boeing brings in remote labor from around the country every time they build an F-15.
It will raise $8-$18m/yr in property tax revenue for the county (depending on abatements), which will likely increase the local counties revenues by 30-50% and primarily go towards local schools, as well as an estimated 50-150 jobs.
If they require the datacenter to be a closed water system and pay for their own electricity, it's an extremely low environmental & industrial (all contained clean rooms, no air pollutants, risk to local water systems, etc.) once in a lifetime boon for the local municipality.
The council members (probably, again depending on abatements & water/energy policy) did represent their constituents well.
"If they require the datacenter to be a closed water system and pay for their own electricity..."
This assertion is doing a LOT of heavy lifting, and when it isn't true, it can cause huge externalities not just for the local community but possibly an entire region. It also does not address the noise problem.
Additionally, your jobs estimates are likely high and include short-term construction jobs which may not even go to locals anyway.
"Pay for their own electricity" is just not an actual thing that seems to be doable at this scale without those externalities. You're talking about sites that are, say, 800MW or 1GW or more... these use more power than tens of thousands of homes, and require entirely new power plants, interconnects, lines run, transformer costs, staff, etc. Typically, power companies amortize those costs and spread them over ratepayers, and when it's just part of the normal induced load of population growth, that's just how things go, and the system works.
In these cases, the rates that these DCs are paying for power are nowhere close to being able to fully absorb or offset the additional CapEx that the power companies are suddenly tasked with, even if they put up, say, the capital for IC, which is usually what's required. So the remaining new shortfalls get spread over the remaining ratepayers, ie, everyone else. If the demand wasn't induced by the build of these massive sites that, strictly speaking, aren't "necessary," then the rates wouldn't have to climb to accommodate them.
Frankly, there should be laws against power companies raising ratepayer rates to accommodate infrastructure investments driven solely by DC/fab load inducement.
Heating homes keeps humans alive who then go onto work, creating economic value.
A DC supplies, like, 12 jobs maybe. The 10,000 homes worth of power it uses supplies 20,000 jobs, maybe more if some older kids are working at the local Dairy Queen.
No, you don't understand. Sending bytes back and forth creates real economic value. We shouldn't heat homes and have supermarkets in this day and age, the margins are miniscule.
> Serious question, what jobs do datacenters create? Are there jobs for local residents?
If locals are qualified then yes.
The DC itself does not have many permanent staff (tech, facilities, security) but loads of work is contracted. I'd say that great majority of the work done in and around the DC campus is outsourced, and it creates work for plenty of people.
It doesn't create ongoing jobs. It creates short term work, and perhaps the occasional momentary task. The only permanent jobs will be physical security.
What do you think people working in construction actually do when the project is done? Just spend the rest of their life homeless? No, they go to the next project.
Are you talking about contractors just while the DC is under construction or after it’s built as well? Google wants to build one in my home town and I’m questioning what value it will bring to the community.
You're wrong. People are probably impressed by the dollar value number it takes to build a DC/campus and then expect that the number of hired people should be "proportionally" equally high. It doesn't work like that but DCs definitely create more than enough local jobs for qualified even after it's built
This. At least until we’re at a point where some guy in the Philippines operates a telepresence android this is definitely a net gain for the community.
With all due respect this is a vague as you can make a statement. What is more than enough local jobs? I drive by roughly 20 DCs on any given week between driving my kid to school and then to practice, the parking lots for the DCs are smaller than those of Chuck E Cheese and even at that size you never see more than 10-15 cars parked there. So not sure what more than enough is but it does not help much local economy
There's going to be continued support from local electricians, low voltage wiring vendors, various facilities service companies, HVAC, and now plumbers.. lots of plumbers. So many leaks. A site like this is going to have probably a few hundred full time people on site all the time in addition to the contracted folks.
A DC does not have a few hundred full time people on site 24/7, it's going to be less than a dozen. It takes much, much less people to run a data center than people think.
Electricians. Top skilled folks for the most part who can do industrial level conduit work and the type who can operate switching gear and control systems. There is enough of this ongoing work for these huge facilities to effectively employ a half dozen full time contractors or more. One of the facilities I work the most in has electrical contractors on-site every single day, with at least a few trucks in the parking lot. These are local union guys. Always something breaking, needing maintenance, or a new area of the facility being refreshed. The facility is over a decade old and the work has never slowed down.
Plumbers. Cooling these facilities takes vast amounts of plumbing work. And it's also typically some of the highest skilled plumbing needed outside of refinery and other manufacturing plant work. When you have 50 giant chillers running 24x7 at least one is undergoing some form of maintenance at any given time.
Probably overlapping with the above, but HVAC technicians. Again, the scale of these facilities means constant work being available as you are operating at miniature city sized installations.
Security guards of course. Not really material though. I've noticed more armed guards than before, with at least two on duty 24x7. As these places get more controversial, I imagine this sort of staffing will increase.
On-site (IT) technicians. For facilities these sizes, you will be staffing it 24x7 and have a large enough crew to get basic refresh projects done. Hard to really estimate this, but in the dozens of full time labor for these giant projects. Think folks who can pull cable, troubleshoot basic hardware, swap drives/bad RAM sticks, etc. For the larger refresh projects contractors typically get flown in during a surge so on-site staffing is relatively minimal, but very few facilities are operating "lights out".
Then you have facility management - highly skilled positions that know how to operate all the electric/mechanical and cooling equipment during emergencies. Every facility I've worked in is staffed by a crew of around half a dozen of these folks or so, with the top tier subject matter experts being flown in during critical emergencies. These are the guys generally coordinating all the contract labor above.
Probably a couple mid-tier network engineers and higher skilled sysadmin types as well depending on who is operating it. Everyone loves to pretend these are highly automated and copy/paste facilities hyperscalers are just perfect at executing - but there is a lot of "dirty" hands-on work to be done since that stuff is not nearly as perfect as advertised and often requires hands-on problem solving and on the spot hacks to get stuff going. As anywhere, how the sausage gets made is a lot uglier than the marketing.
Once you get out of the highly competent hyperscalers, the above numbers go way up. Enterprise datacenter operators are going to need far more on-site labor due to simply not being great at this sort of work. The stories I hear of some current builds are rather humorous in how many people it's taking to get stuff working - basically solving what should be automated via manual processes.
It's not a lot of jobs, but for these huge 100's of Megawatt facilities the low-end is probably in the 100+ range of FTE equivalent labor after construction is completed. Everyone but security and the basic "remote hands" type employees would be in the $100k+ salary range.
> It's not a lot of jobs, but for these huge 100's of Megawatt facilities the low-end is probably in the 100+ range of FTE equivalent labor after construction is completed. Everyone but security and the basic "remote hands" type employees would be in the $100k+ salary range.
For the size and resources they are using you can build a lot of other things here that will employ significantly North of 100 (this is too many but lets keep it for the sake of argument) employees with significantly North benefit to the population than F'ing data centers.
I mean yes, but that would require demand right? The demand is currently for building data centers, should you just wait around for better things to be in demand? It’s definitely a strategy, but it doesn’t seem like the obviously necessary strategy
If someone wants to build a DC out in the middle of nowhere in a non-populated areas of whatever state, be my guest, I won't complain. I was specifically talking about where I live ( https://www.datacentermap.com/usa/virginia/ ), few of the top-10 most affluent counties in USA are here but also the most DCs in the world are here. There is a huge demand here for non-Data Centers that would bring a lot more to the economy than buildings guarded and surrounded by fences that you have to drive next to every day
What are they giving up? You get property tax revenue, permit revenue, some construction trickle down, and you do get a few dozen jobs and in return you give up some some grandstanding/political posturing opportunities. Seems like an easy deal to me, but hey that's just me.
There will be at least 5 employees working as smart hands 24/7, so probably 3 shifts -- 15 people. Plus 1-2 security agents working 24/7, another 6 jobs. Plus a foreman with some maintenance crew for HVAC/electrical (not 24/7) so probably another 1-3 jobs.
That's a really sweet deal for a town with only 11k people and no other external investments on the horizon.
According to the U.S. Chamber of Commerce[1], 1688 while being built, 157 on-going jobs. I assume this is some 'average' datacenter; I didn't pursue methodology.
Almost none. Its entire value is in one-time construction regional purchasing and the ability to say the word “jobs” to the cameras. Occasionally they have the guts to charge market rates for resources or taxes but for the most part that’s heavily discounted. (See also e.g. The Dalles’ attempt to secretly sell much of its watershed to datacenters.)
As someone who lives in Northern Virginia, there are definitely ongoing jobs, but in this area they are mostly filled with H1B workers. The real money is in development
I hesitate to say it, but at least the datacenter companies haven't realized that federal railroad laws mean that the feds can preempt state and local governments with regards to railroads and yards ... though it may be hard to argue that a datacenter is a necessary part of a railroad.
It’d be a lot easier to argue that a railroad yard is a necessary part of a datacenter, and then eminent domain and pave it :)
But there’s some sensible planning in linking datacenters and railroads, honestly. Truck-shipping 44U fully-loaded cargo racks in standardized quarter-containers would a lot more sense in today’s AI-proliferation context. And I’d be up for seeing datacenters lose their natural lock-in resistance to customer migrations; “a competitor offered us a 5% discount plus freight refunds if we shift at least 5 cars of racks to them” is a lot easier when your datacenter has cargo crane capacity. There’s still a place for bespoke DCs but for the cog-in-the-cloud stuff that we have now, it’s not a bad idea!
(And, if you add a third rail for power-over-Ethernet, then you can start to have datacenter migrations that don’t cause an outage. Amtrak is already implementing the first stages of datacenter-grade connectivity for riders on their trains, though not amperes of the necessary degree yet.)
A railroad has to track a lot of data; what's in each car (declared by customer), what train it's attached to, where is it on maintenance schedules; similar for the rail and signal infrastructure, etc; in today's modern environment, they need multiple datacenters for high availability.
Something something route planning to reduce the number of coupling changes, etc, etc.
Edit: also, a lot of long distance fiber runs on railroad right of way, so datacenters at rail yards may be well placed for connectivity.
A railroad is the infrastructure for transporting commodities. In the modern digital economy, datacenters along with the whole internet infrastructure are the modern railroads, which need protection and deregulation for the sakes of safety, national security, economy etc etc. Maybe this argument works better if the others don't?
Maybe instead of performing mental gymnastics to expand the executive’s power well beyond beyond what Congress has legislated, we should just pass new laws
I am not sure this is a good idea. Having congress legislate just adds friction, removes chances of personal gain for the the executive branch, and has the potential of actually changing parts of it based on democratic debate. A good middle position would be to have the executive power do its thing and have congress legislate/approve long after the fact has been established.
I'm honestly surprised why local governments are so eager to make datacenter deals in the first place. I'm pro-progress, but a datacenter brings approximately nothing to the local economy. It doesn't employ any noteworthy number of people, it doesn't generate any real tax revenue, and it increases electricity costs for the region. So if the voters don't want it, that feels like their prerogative.
I don't know if it's the elected officials conflating data centers with the region becoming a bustling tech hub, rather than just a way for a Bay Area company to capitalize on cheap electricity... or if it's kickbacks.
Also a choice the local government can make! I don't know about this specific case but I suspect we'll see local governments get more sophisticated when negotiating with tech companies.
> and it increases electricity costs for the region
This doesn't need to be true. It would be both possible and reasonable to mandate subsidy by the datacenter as part of any deal so that costs don't go up for anyone else.
Yeah, that's what I've been thinking. If we charged twice as much money per kilowatt-hour for datacenter electricity compared to residential, it feels like the net revenue for electricity could be roughly the same to the power company, but then it wouldn't be nearly as annoying for the residents of the town having their prices spike way up.
Or, you know, the AI companies could actually supply their own power like I keep hearing tech bros mention is coming soon.
Possible and reasonable don't guarantee anything with big businesses. Around 2008, Atlanta had a major drought, and as the local government asked the citizens to conserve water, Coca Cola was bottling up the local water and sending it out on trucks. When the citizens complained, the government said it would cost too many jobs to stop the bottling.
You are engaging with a straw man that is literally the opposite of what I said. I said it would be possible and reasonable to mandate it, not intentionally look the other way, and not cross fingers and hope for beneficence.
You're still not engaging with what I said. Please see that "this government chose not to mandate" has zero relevance to whether a government mandate would be possible or reasonable.
I said "[datacenters] don't need to [increase electricity costs for others]. It would be possible to mandate...".
I said that because the person I was responding to said "a datacenter increases electricity costs for the region".
It CAN increase electricity costs for the region. It does not NEED to increase electricity costs for the region. And PREVENTION of increasing electricity costs for the region CAN be done by government mandate instead of hoping for profiteers to do less profiteering.
What this particular city council did with this datacenter is neither an inherent property of datacenters nor of city councils.
I wonder if the best way, and something that might be more likely to pass, is something like "progressive pricing".
Like the first N kilowatt hours are the regular price, and would cover the average case for most people (I don't know what the average amount of electricity used by a person is but the power companies absolutely know). Then the next M kilowatt hours are an increased price, and keep going as energy spikes up.
I think this could work just because this is how income tax works. Somehow that managed to get passed by congress and state legislatures.
> I am also saying that they will never choose to do it
If this article were posted when this campaign was just starting, this would be a top HN comment. Unfortunately, lazy nihilism runs deep in tech circles.
It seems like both of you have thoroughly missed the context of my subthread.
If your goal is to point out that people make choices, well, you're in the wrong thread branch and want to instead reply to a different part of the same message that I replied to. Because I never said or implied that they don't. Quite the opposite in fact.
Here is the context of my subthread, extracted, in two parts:
Part 1, the framing.
> "I don't know if [elected officials think a wrong thing about datacenters] or if it's kickbacks."
You see, the kickbacks option is already there. We all already understand that it could be kickbacks. Therefore bringing it up further would just be silly. I certainly have no reason to say that kickbacks aren't a possibility. The only part we need to address is OP believing that [thing is wrong].
Part 2, the [thing].
> "I'm pro-progress, but a datacenter brings approximately nothing to the local economy. It doesn't employ any noteworthy number of people, it doesn't generate any real tax revenue, and it increases electricity costs for the region."
That distills to:
> "a datacenter brings approximately nothing to the local economy"
That statement means either:
A (haven't): datacenters have in the past only ever brought nothing (and therefore they will in the future only ever bring nothing)
or
B (can't): datacenters cannot bring anything other than nothing (and therefore they will in the future only ever bring nothing)
And they're both wrong. A is wrong because past behavior does not imply future behavior. And B is wrong because in fact they can.
Technically, it creates construction revenue and jobs. If you’re a municipality with FOMO heading into a job-collapse recession and someone offers you jobs on a silver platter, you might get fired from the city council for refusing it. So it’s particularly interesting to see that citizens would rather refuse datacenters than gain from them. (I certainly agree.)
If it's allowed by regulation and zoning, they generally don't have a say in the first place. These stories are never about building another box-shaped building in an industrial zone. We're talking about rezoning, variances, or otherwise preferential treatment.
I think that they hear "$6 billion datacenter" and think that the town's economy is getting $6 billion in jobs rather than some foreign computer hardware company is getting $6B for computers that are housed in their town.
Data centers are high-dollar projects that sound great and officials are able to look like they're doing things to increase revenues, generate jobs, create infrastructure and put the community on the forefront of high-tech. Altogether, those are commendable increases, but the devil is in the details, in that all of those things do not occur at once and the costs associated with have a much longer payoff timeline.
Property tax and (in some cases) utility taxes are deeply attractive, especially in places with large industrial-zoned swaths of land nobody is really interested in.
This is really the only legitimate complaint that has any basis in reality.
But "region" is doing a lot of work here. This is typically a multi-state sized region. There are local congestion charges in some places, but overall it doesn't matter a whole lot to your electric bill if a large consumer goes in 200 miles away or across the road from you.
If it goes in across the road your local community gets the benefit of having about the least obnoxious industrial use of land possible. After construction there is very little truck traffic (e.g. much less wear and tear on local roads than a trucking terminal or manufacturing plant), and effectively is a giant office building in terms of impact on it's surroundings. In fact, until recently most of the datacenters were built in suburban office and light industrial parks and no one was the wiser.
There are legitimate complaints to be made about "datacenters" that also co-locate a natural gas or diesel power plant. But those complaints are towards building a power plant across the street, not a datacenter.
It's effectively as "free" of a tax base as you can get, assuming you don't negotiate stupid local tax abatements - which I suppose is a large caveat. Those should be simply outright illegal for everyone though, I don't see that as a datacenter specific thing. It also does effectively employ a few dozen to few hundred local tradesmen through the lifecycle of such a facility - since at these scales there is constant electric and plumbing work to be done. Usually the highest paid and highly skilled of such type of work. Many (most?) places are even using union labor for these bits.
The power problem exists broadly though. We spent a few generations not building out anything of material size and we are reaping what we have sewn. It was coming for us either way - datacenter AI bubble just brought it forward a some odd number of years. Just look at how hard it is to get a wind farm project off the ground due to NIMBY - both for the wind farm itself, and the 200 mile transmission line you might need to build to the closest major load centers. Effectively impossible.
That's only if you co-locate a power plant near it. With proper setbacks and decent design, there is very little to no noise pollution for the vast majority of these facilities.
Most folks near them do not even know they exist. Plus you typically put them in the middle of a field with berms around them, or in a light industrial park. Not across the street from homes.
Trucking traffic creates far more noise pollution. HVAC fans spinning at optimal speed simply are not a problem for the vast majority of facilities.
Generators running during a power outage? Sure. But those typically are relatively rare events. Testing each month for an hour is just not a material complaint to me.
I guess their point is that of all possible industrial usecases, data centers are the least obnoxious one. I live in one of the countries that actually manufactures things, unlike the US, and I find it hard to argue with that. Any noise pollution caused by data centers is far far less than most industrial setups. It's the same with every other resource, water, electricity, effect on local shared infrastructure like roads and commerce, etc,. Other industries are an order of magnitude worse.
Given that you _have_ to have some industrial setup unless you want to import everything (tokens, in this case), datacenters are far and away the best choice.
I'll add a qualifier to the above, modifying it to say that of all industrial setups generating atleast X dollars of economic value, datacenters are far and away the best in terms of impact on nbhd.
The jobs argument also falls apart, when you consider that it's essentially 100 jobs in return for just an office building worth of space. If you want a thousand job plant just build that as well next town over, it will take way way more space and other resources though. The reason that didnt happen even before this datacenter boom is because most manufacturing setups are fairly infeasible in rich countries like the US. I can't imagine the response to a textile plant or a steel plant if this is the response to datacenters.
I agree however, that if you colocate a gigantic power plant, then you get the worst of both worlds. Fewer jobs and the hindrance of a big power plant near residential areas. Grid expansion being slow in developed areas like most of the US is not surprising though.
But this is pretty much the best case scenario. Tolerating the power plant until the grid expands is the way to go I suppose.
> There are legitimate complaints to be made about "datacenters" that also co-locate a natural gas or diesel power plant. But those complaints are towards building a power plant across the street, not a datacenter.
Except of course there would be no complaints about the power plants if we did not need them in the first place to power the data centers.
> This is really the only legitimate complaint that has any basis in reality.
There are many, many others... You obviously do not live near ones, I live in Northern VA virtually surrounded by data centers and electricity costs are just part of the problem...
> gets the benefit of having about the least obnoxious industrial use of land possible
Or it could have been a lot less obnoxious residential use with parks and shit...
> Except of course there would be no complaints about the power plants if we did not need them in the first place to power the data centers.
And you only need stupid designs like tiny natural gas turbines on-site because NIMBY and lack of investment for a couple generations on the power infrastructure side. I find it difficult to be very sympathetic to our society on this issue, since I've been following it far before AI Datacenters became the thing to rage about. It was coming for us either way.
> There are many, many others... You obviously do not live near ones, I live in Northern VA virtually surrounded by data centers and electricity costs are just part of the problem...
I have lived near ones. Not datacenter alley scale, but nowhere in the world is at that level where you live. I had zero issues with them, and no one visiting even knew they existed. I've certainly seen horrible designs that should not have been permitted or built where they are, but a 500k sqft facility in the middle of 50 acres is just... not an issue to live near.
> Or it could have been a lot less obnoxious residential use with parks and shit...
Sure. Building a datacenter in the middle of a residential area is a bit silly. But we're not talking about that here. At some point you need industry to actually build things, and as industry goes this is about as light and least impactful to the local environment as it gets.
It’s a bit crazy that the elderly are blocking data center buildouts while they also expect to collect a big chunk of everyone’s payroll. Pick your lane.
Like, what did “the elderly” ever do for us? Aside from literally creating the civilization around us, and the infrastructure necessary to support it? Seems perfectly reasonable to expect reward for that, rather than pollution and increased energy costs.
Perhaps local gov that gets the gift of datacenter (aka hundreds of millions in taxes) should subsidize residents rooftop solar installations by a slight bit. Sounds like a fair deal to every party involved.
The coucilmembers probably got their bag already from whoever is building that datacenter, voting them out after the fact just means they don't have to clock in.
Generally this is good. Representatives should represent their populace and not monied interests. When they fail to do so they shoudl be removed. That's when democracy is operating correctly. But the article still contained the falsehood that data centers create jobs. This is just not accurate. Most data centers are acres of racks and HVAC with precious few humans to maintain them.
I wish when they write these storied they'd put the town's per capita income in brackets the way they do with politician's party affiliation or company's ticker. The "Fairfax of St. Louis" voting out half their legislature over a project means something very different than the "Newark of St. Louis" doing the same.
Because wealthy suburbs have said "not here, move your filthy industry somewhere else" since forever.
When the places that aren't swimming in jobs, the local government isn't swimming in property tax revenues and frankly probably can't even enforce the rules they're federally compelled to have without destroying everything says "take that somewhere else" it means something entirely different.
Take it from someone who lives in the St. Louis area, Festus is by no means a snooty suburb. It's a small town that has become a distant suburb of St. Louis due to sprawl spreading south into the adjacent county over the past 20 years.
For comparison, an upscale suburb of similar size (Town and Country, Missouri) has a median household income of $202,974, as compared to $59,041 for Festus. The average person you meet in Town and Country is likely to be a doctor, attorney, or executive. In Festus, the average person likely works in a factory, farm, or lead mine.
It will be interesting to see what we do with these enormous concrete boxes once we find a better way to do whatever we think all this 'ai' is going to do. 'Dead malls' that are being partially converted into pickleball courts and places for people to take their daily 'constitutional' indoors are going to seem quaint.
That is if the bubble doesn't pop because of other factors first that is...
> It will be interesting to see what we do with these enormous concrete boxes once we find a better way to do whatever we think all this 'ai' is going to do.
These are just prefab tip-up walls with metal roofs, bar joists, columns spaced 30-40 feet OC, and a 6” concrete slab.
It’s no different than an Amazon distribution center once all the electrical substations, switchgear, generators, and UPSes are gone.
... sure, again once it's all pointless it's still a gigantic concrete box. Tons of old hastily built automotive factories just rotting around detroit.
This is becoming a pretty clear wedge between red and blue. Why do you think Musk opened his diesel turbine driven data center in rural Mississippi? Big Tech is systematically targeting small municipalities across the US with promises of insane money to anyone willing to sell out their residents. Missouri being traditionally purple, it makes a lot of sense the flashpoint would be here.
It's not possible to identify operators who do not wish to be identified. The land is purchased by an LLC, developed by another LLC, and then sold to a third LLC for buildout. Once it's up and running it could be rented to a fourth LLC who has contracted with the actual client, or the client could just buy the involved LLCs (usually via another LLC).
The people are, their politicians are not. Overwhelmingly this is a problem of backroom deals with state and local Republicans subverting the electorate's will.
Is it still 'NIMBYism' if what they are building is something that none of the residents want at all? Like yeah no-one wants the new sewage plant in their backyard but they will still see the value in it.
Even as someone in tech I don't see the value in this scale of buildout. This technology is very new and no doubt in my mind that in 5 years we will be able to do everything these models do on a fraction of the resources.
Those are mostly more traditional data centers. 20 or 30 kilovolt racks with separate cages for different tenants, some meet-me rooms, and a bunch of telecom gear are the order of the day.
Internet traffic requires routing and switching. That’s all traditional DC equipment in terms of power and cooling. They don’t require 80 or 100 kilovolt racks like something stuffed full of AI accelerators.
Plugging this video about infrasound, which I only recently learned was a thing: https://www.youtube.com/watch?v=_bP80DEAbuo
These datacenters can be built in ways to limit this kind of noise pollution, but it appears local leaders do not think about things like this that can truly harm their constituents.
> local leaders do not think about things like this that can truly harm their constituents
Or they're paid to not think about it. I don't consider Occam's razor when politicians are involved
I don't know anything about this particular site, but I presume it's one of the new mega gpu sites.
I'm seeing many people in the comments with an early 2000's era concept of datacenters. The scale of these new sites is mind boggling. Take your idea of a typical datacenter building. Make it 4x bigger. Then put 4 of them together into a cluster. Then imagine 10 of those clusters at the site.
None of that is new whatsoever. AWS has been doing that for 15 years at this point. The cloud didn't start existing just because of AI.
Appreciate you making my point for me.
> Missouri campaign finance records show a political action committee — made up of labor unions that support data centers because of the jobs they create — spent almost $40,000 in the final weeks of the race on newspaper and digital ads and yard signs in support of the four council members booted from office.
Serious question, what jobs do datacenters create?
Are there jobs for local residents?
A small number of jobs for tradesmen (electricians, plumbers, etc.).
A small number of jobs for security guards.
Maybe a tiny number (one to three?) for individuals tasked with actual hardware swapping within the data center itself.
And all of the above assumes the data center owner does not "travel in" the requisite individuals on an "as needed" basis -- in which case the only jobs that may go to the locals is "security guard".
But all of the "sys-admin" management level work can be done remotely.
So the actual number of new jobs that arrive in the locality is likely on the order of 20-30 or fewer.
Yeah and that type of work bid usually goes to huge conglomerates. A local mom and pop electrician shop isn’t going to be building a datacenter, it’ll be something like Siemens.
A friend of mine is an independent electrician in the Columbus, OH area. Last summer he told me he was getting plenty of datacenter construction work, albeit it was in the form of subcontracted jobs from the larger firms who were awarded the contracts.
Even if the datacenter does hire some local labor for construction, that's still all temporary jobs. It's not an ongoing source of employment for locals.
They'll pay tons in taxes to energy company.
Local shops will absolutely be contracted to work on the project. A datacenter project like this can't find enough qualified electricians.
I would imagine there aren't too many electricians in Yeehaw, Minnesota, trained and qualified to do gigawatt data center installs. So they'll freight in contractors to do that work, and maybe temporarily employ a few locals for a month or two for auxiliary stuff.
More generally, this is the universal playbook when someone wants to dump some megaproject on a community that doesn't want it: This will create X jobs and inject $Y into the local economy. Can you name one case where this actually happened? It's usually very few additional permanent jobs and, particularly for public-works stuff, millions or even billions in extra debt to pay off. But don't worry, this next thing we're working on once we get the local council to issue a permit to bulldoze your forest park, that will bring in jobs, we promise.
I work for an electrical contractor that does large data center projects and we almost always partner with a local contractor to provide labor from the local union(s).
sure, that only covers construction though. Once the thing is built they are going to travel in all the maintenance that needs to happen, and that local tradesman is not going to get to many new home construction jobs after it goes in. Who wants to live nearby a noise polluter like that?
>Yeah and that type of work bid usually goes to huge conglomerates.
Which are exactly the kinds of entities that the trades unions and industry interest groups are most deeply in bed with.
How many of these are on-going jobs vs during construction and as-needed? I think you're right it'll be only security guard jobs. Even if they don't travel in workers, it's quick short-term tasks that maybe locals can perform, but that's not "creating jobs."
This argument has always been such a weird goalpost shift for me. Even at my full time job I am getting strung together by 3-12 month projects. Everyone works on projects. When this data center is done in a year, we'll (hopefully) need to build something else, keeping those people employed.
Like, of course it's creating a job. If you create a million 1-year jobs every year, that's a million jobs.
Because the "it'll create X jobs" implies it's ongoing. It's a disingenuous attempt to oversell the benefits because they know if they're transparent about it, suddenly it doesn't seem like such a great deal.
Actually if they were honest they'd say "It'll create x careers", which is a much better deal than just a job. Friend went from 17 an hour at his first DC to 100/hr last year with 70 people on his team.
> A small number of jobs for tradesmen (electricians, plumbers, etc.).
Its no car dealership but probably a reliable source of work-orders. Seems like a "gigascale" datacenter would be a large job for a tradesman to be a subcontractor within and afterward its scale means continuous upgrades/maintenance.
Is there any literature of ongoing economic impact of similar facilities?
In a town of 12K people I'd say it's incredibly unlikely. Most of if not all the labor to build it will be flown in, most of the labor to staff it will be moved in.
And once it's built it's not like a Walmart or something where you need enough staff to police the crowds...there are not crowds. There's some rack and stack needs, and some ongoing cabling needs generally,and some other stuff, but they are staffed as lightly as humanly possible.
I suppose w/ all the out of town labor to build it there will be more waitress and hotel cleaning jobs for a while...a town or over...where they can actually house the labor.
Oh, and they are getting an Olive Garden...which will probably employ more local labor.
Festus isn’t small because it’s in the middle of nowhere. It’s right there with Arnold, Barnhart, Crystal City, and other far south suburbs of St. Louis. The metro area can build it. It’s not like Boeing brings in remote labor from around the country every time they build an F-15.
> And once it's built it's not like a Walmart
Yet it creates infinitely more value than a supermarket.
It will raise $8-$18m/yr in property tax revenue for the county (depending on abatements), which will likely increase the local counties revenues by 30-50% and primarily go towards local schools, as well as an estimated 50-150 jobs.
If they require the datacenter to be a closed water system and pay for their own electricity, it's an extremely low environmental & industrial (all contained clean rooms, no air pollutants, risk to local water systems, etc.) once in a lifetime boon for the local municipality.
The council members (probably, again depending on abatements & water/energy policy) did represent their constituents well.
"If they require the datacenter to be a closed water system and pay for their own electricity..."
This assertion is doing a LOT of heavy lifting, and when it isn't true, it can cause huge externalities not just for the local community but possibly an entire region. It also does not address the noise problem.
Additionally, your jobs estimates are likely high and include short-term construction jobs which may not even go to locals anyway.
"Pay for their own electricity" is just not an actual thing that seems to be doable at this scale without those externalities. You're talking about sites that are, say, 800MW or 1GW or more... these use more power than tens of thousands of homes, and require entirely new power plants, interconnects, lines run, transformer costs, staff, etc. Typically, power companies amortize those costs and spread them over ratepayers, and when it's just part of the normal induced load of population growth, that's just how things go, and the system works.
In these cases, the rates that these DCs are paying for power are nowhere close to being able to fully absorb or offset the additional CapEx that the power companies are suddenly tasked with, even if they put up, say, the capital for IC, which is usually what's required. So the remaining new shortfalls get spread over the remaining ratepayers, ie, everyone else. If the demand wasn't induced by the build of these massive sites that, strictly speaking, aren't "necessary," then the rates wouldn't have to climb to accommodate them.
Frankly, there should be laws against power companies raising ratepayer rates to accommodate infrastructure investments driven solely by DC/fab load inducement.
It is way better than wasting power on house heating and watering your lawns. It actually creates economic value.
Heating homes keeps humans alive who then go onto work, creating economic value.
A DC supplies, like, 12 jobs maybe. The 10,000 homes worth of power it uses supplies 20,000 jobs, maybe more if some older kids are working at the local Dairy Queen.
This is deranged. It's like saying entire economy hinges on underwear manufacturing because everyone needs one pair.
No, you don't understand. Sending bytes back and forth creates real economic value. We shouldn't heat homes and have supermarkets in this day and age, the margins are miniscule.
> Serious question, what jobs do datacenters create? Are there jobs for local residents?
If locals are qualified then yes. The DC itself does not have many permanent staff (tech, facilities, security) but loads of work is contracted. I'd say that great majority of the work done in and around the DC campus is outsourced, and it creates work for plenty of people.
It doesn't create ongoing jobs. It creates short term work, and perhaps the occasional momentary task. The only permanent jobs will be physical security.
What do you think people working in construction actually do when the project is done? Just spend the rest of their life homeless? No, they go to the next project.
Are you talking about contractors just while the DC is under construction or after it’s built as well? Google wants to build one in my home town and I’m questioning what value it will bring to the community.
to build - yes. after it is built - no. so there is some temporary work but nothing permanent
You're wrong. People are probably impressed by the dollar value number it takes to build a DC/campus and then expect that the number of hired people should be "proportionally" equally high. It doesn't work like that but DCs definitely create more than enough local jobs for qualified even after it's built
This. At least until we’re at a point where some guy in the Philippines operates a telepresence android this is definitely a net gain for the community.
With all due respect this is a vague as you can make a statement. What is more than enough local jobs? I drive by roughly 20 DCs on any given week between driving my kid to school and then to practice, the parking lots for the DCs are smaller than those of Chuck E Cheese and even at that size you never see more than 10-15 cars parked there. So not sure what more than enough is but it does not help much local economy
I'll repeat a question asked somewhere else: what exactly are those local jobs after it's built? Can somebody care to list?
There's going to be continued support from local electricians, low voltage wiring vendors, various facilities service companies, HVAC, and now plumbers.. lots of plumbers. So many leaks. A site like this is going to have probably a few hundred full time people on site all the time in addition to the contracted folks.
A DC does not have a few hundred full time people on site 24/7, it's going to be less than a dozen. It takes much, much less people to run a data center than people think.
A dozen people couldn't handle the number of daily nvidia compute tray RMAs one of these datacenters produces.
oh there is shipping & handling or just handling? :)
Well they're liquid cooled, so there's some extra handling required. Then there's the sky high failure rate.
Electricians. Top skilled folks for the most part who can do industrial level conduit work and the type who can operate switching gear and control systems. There is enough of this ongoing work for these huge facilities to effectively employ a half dozen full time contractors or more. One of the facilities I work the most in has electrical contractors on-site every single day, with at least a few trucks in the parking lot. These are local union guys. Always something breaking, needing maintenance, or a new area of the facility being refreshed. The facility is over a decade old and the work has never slowed down.
Plumbers. Cooling these facilities takes vast amounts of plumbing work. And it's also typically some of the highest skilled plumbing needed outside of refinery and other manufacturing plant work. When you have 50 giant chillers running 24x7 at least one is undergoing some form of maintenance at any given time.
Probably overlapping with the above, but HVAC technicians. Again, the scale of these facilities means constant work being available as you are operating at miniature city sized installations.
Security guards of course. Not really material though. I've noticed more armed guards than before, with at least two on duty 24x7. As these places get more controversial, I imagine this sort of staffing will increase.
On-site (IT) technicians. For facilities these sizes, you will be staffing it 24x7 and have a large enough crew to get basic refresh projects done. Hard to really estimate this, but in the dozens of full time labor for these giant projects. Think folks who can pull cable, troubleshoot basic hardware, swap drives/bad RAM sticks, etc. For the larger refresh projects contractors typically get flown in during a surge so on-site staffing is relatively minimal, but very few facilities are operating "lights out".
Then you have facility management - highly skilled positions that know how to operate all the electric/mechanical and cooling equipment during emergencies. Every facility I've worked in is staffed by a crew of around half a dozen of these folks or so, with the top tier subject matter experts being flown in during critical emergencies. These are the guys generally coordinating all the contract labor above.
Probably a couple mid-tier network engineers and higher skilled sysadmin types as well depending on who is operating it. Everyone loves to pretend these are highly automated and copy/paste facilities hyperscalers are just perfect at executing - but there is a lot of "dirty" hands-on work to be done since that stuff is not nearly as perfect as advertised and often requires hands-on problem solving and on the spot hacks to get stuff going. As anywhere, how the sausage gets made is a lot uglier than the marketing.
Once you get out of the highly competent hyperscalers, the above numbers go way up. Enterprise datacenter operators are going to need far more on-site labor due to simply not being great at this sort of work. The stories I hear of some current builds are rather humorous in how many people it's taking to get stuff working - basically solving what should be automated via manual processes.
It's not a lot of jobs, but for these huge 100's of Megawatt facilities the low-end is probably in the 100+ range of FTE equivalent labor after construction is completed. Everyone but security and the basic "remote hands" type employees would be in the $100k+ salary range.
> It's not a lot of jobs, but for these huge 100's of Megawatt facilities the low-end is probably in the 100+ range of FTE equivalent labor after construction is completed. Everyone but security and the basic "remote hands" type employees would be in the $100k+ salary range.
For the size and resources they are using you can build a lot of other things here that will employ significantly North of 100 (this is too many but lets keep it for the sake of argument) employees with significantly North benefit to the population than F'ing data centers.
I mean yes, but that would require demand right? The demand is currently for building data centers, should you just wait around for better things to be in demand? It’s definitely a strategy, but it doesn’t seem like the obviously necessary strategy
If someone wants to build a DC out in the middle of nowhere in a non-populated areas of whatever state, be my guest, I won't complain. I was specifically talking about where I live ( https://www.datacentermap.com/usa/virginia/ ), few of the top-10 most affluent counties in USA are here but also the most DCs in the world are here. There is a huge demand here for non-Data Centers that would bring a lot more to the economy than buildings guarded and surrounded by fences that you have to drive next to every day
At least when i was at google, more than a decades ago at this point, hardware ops guys were locally sourced
When I was at (then) Facebook, this was mostly the case, but we also ran data centers with a hundred thousands of servers off a dozen local techs.
Facebook (and Google as well, IIRC) prided themselves on how few people they needed to run the datacenter.
Maybe I'm jaded but "we created 50 jobs" just doesn't hit that hard.
I mean it’s a warehouse sized building, unless you’re doing a call center boiler room in there how many more jobs you’d expect?
Me? Not many, but I also happened to work in adjacent roles that had me more informed than the average engineer.
People whose communities are affected? I think it'd be reasonable for them to expect more for it to be worth it.
What are they giving up? You get property tax revenue, permit revenue, some construction trickle down, and you do get a few dozen jobs and in return you give up some some grandstanding/political posturing opportunities. Seems like an easy deal to me, but hey that's just me.
There will be at least 5 employees working as smart hands 24/7, so probably 3 shifts -- 15 people. Plus 1-2 security agents working 24/7, another 6 jobs. Plus a foreman with some maintenance crew for HVAC/electrical (not 24/7) so probably another 1-3 jobs.
That's a really sweet deal for a town with only 11k people and no other external investments on the horizon.
According to the U.S. Chamber of Commerce[1], 1688 while being built, 157 on-going jobs. I assume this is some 'average' datacenter; I didn't pursue methodology.
[1] https://www.uschamber.com/assets/documents/ctec_datacenterrp...
Almost none. Its entire value is in one-time construction regional purchasing and the ability to say the word “jobs” to the cameras. Occasionally they have the guts to charge market rates for resources or taxes but for the most part that’s heavily discounted. (See also e.g. The Dalles’ attempt to secretly sell much of its watershed to datacenters.)
Many jobs during construction. A site like this is a substantial multi-year construction effort.
Long term permanent jobs.. not so much.
A few dozen jobs in a town of 12,000 is nothing to sneeze at.
As someone who lives in Northern Virginia, there are definitely ongoing jobs, but in this area they are mostly filled with H1B workers. The real money is in development
I hesitate to say it, but at least the datacenter companies haven't realized that federal railroad laws mean that the feds can preempt state and local governments with regards to railroads and yards ... though it may be hard to argue that a datacenter is a necessary part of a railroad.
It’d be a lot easier to argue that a railroad yard is a necessary part of a datacenter, and then eminent domain and pave it :)
But there’s some sensible planning in linking datacenters and railroads, honestly. Truck-shipping 44U fully-loaded cargo racks in standardized quarter-containers would a lot more sense in today’s AI-proliferation context. And I’d be up for seeing datacenters lose their natural lock-in resistance to customer migrations; “a competitor offered us a 5% discount plus freight refunds if we shift at least 5 cars of racks to them” is a lot easier when your datacenter has cargo crane capacity. There’s still a place for bespoke DCs but for the cog-in-the-cloud stuff that we have now, it’s not a bad idea!
(And, if you add a third rail for power-over-Ethernet, then you can start to have datacenter migrations that don’t cause an outage. Amtrak is already implementing the first stages of datacenter-grade connectivity for riders on their trains, though not amperes of the necessary degree yet.)
A railroad has to track a lot of data; what's in each car (declared by customer), what train it's attached to, where is it on maintenance schedules; similar for the rail and signal infrastructure, etc; in today's modern environment, they need multiple datacenters for high availability.
Something something route planning to reduce the number of coupling changes, etc, etc.
Edit: also, a lot of long distance fiber runs on railroad right of way, so datacenters at rail yards may be well placed for connectivity.
A railroad is the infrastructure for transporting commodities. In the modern digital economy, datacenters along with the whole internet infrastructure are the modern railroads, which need protection and deregulation for the sakes of safety, national security, economy etc etc. Maybe this argument works better if the others don't?
Maybe instead of performing mental gymnastics to expand the executive’s power well beyond beyond what Congress has legislated, we should just pass new laws
I am not sure this is a good idea. Having congress legislate just adds friction, removes chances of personal gain for the the executive branch, and has the potential of actually changing parts of it based on democratic debate. A good middle position would be to have the executive power do its thing and have congress legislate/approve long after the fact has been established.
Snowpiercer but with data centers? The breeze would help with cooling!
A LOT of fiber rides alongside railroad tracks. Easier to string and maintain.
I'm honestly surprised why local governments are so eager to make datacenter deals in the first place. I'm pro-progress, but a datacenter brings approximately nothing to the local economy. It doesn't employ any noteworthy number of people, it doesn't generate any real tax revenue, and it increases electricity costs for the region. So if the voters don't want it, that feels like their prerogative.
I don't know if it's the elected officials conflating data centers with the region becoming a bustling tech hub, rather than just a way for a Bay Area company to capitalize on cheap electricity... or if it's kickbacks.
> it doesn't generate any real tax revenue
This is a choice the local government can make. You can read Loudon County's (us-east-1 + everything else) explaining what it does with the data center revenue it gets https://www.loudoun.gov/6188/Data-Centers-in-Loudoun-County.
> it increases electricity costs for the region
Also a choice the local government can make! I don't know about this specific case but I suspect we'll see local governments get more sophisticated when negotiating with tech companies.
> and it increases electricity costs for the region
This doesn't need to be true. It would be both possible and reasonable to mandate subsidy by the datacenter as part of any deal so that costs don't go up for anyone else.
Yeah, that's what I've been thinking. If we charged twice as much money per kilowatt-hour for datacenter electricity compared to residential, it feels like the net revenue for electricity could be roughly the same to the power company, but then it wouldn't be nearly as annoying for the residents of the town having their prices spike way up.
Or, you know, the AI companies could actually supply their own power like I keep hearing tech bros mention is coming soon.
Possible and reasonable don't guarantee anything with big businesses. Around 2008, Atlanta had a major drought, and as the local government asked the citizens to conserve water, Coca Cola was bottling up the local water and sending it out on trucks. When the citizens complained, the government said it would cost too many jobs to stop the bottling.
You are engaging with a straw man that is literally the opposite of what I said. I said it would be possible and reasonable to mandate it, not intentionally look the other way, and not cross fingers and hope for beneficence.
It is the government that mandates things. Even in this article, it was the local council that sold them out.
> it was the local council that sold them out
You're still not engaging with what I said. Please see that "this government chose not to mandate" has zero relevance to whether a government mandate would be possible or reasonable.
I said "[datacenters] don't need to [increase electricity costs for others]. It would be possible to mandate...".
I said that because the person I was responding to said "a datacenter increases electricity costs for the region".
It CAN increase electricity costs for the region. It does not NEED to increase electricity costs for the region. And PREVENTION of increasing electricity costs for the region CAN be done by government mandate instead of hoping for profiteers to do less profiteering.
What this particular city council did with this datacenter is neither an inherent property of datacenters nor of city councils.
> Please see that "there was no government mandate" is not the same as "a government mandate isn't possible".
I agree with this, a government mandate is absolutely possible. But I am also saying that they will never choose to do it.
I wonder if the best way, and something that might be more likely to pass, is something like "progressive pricing".
Like the first N kilowatt hours are the regular price, and would cover the average case for most people (I don't know what the average amount of electricity used by a person is but the power companies absolutely know). Then the next M kilowatt hours are an increased price, and keep going as energy spikes up.
I think this could work just because this is how income tax works. Somehow that managed to get passed by congress and state legislatures.
> I am also saying that they will never choose to do it
If this article were posted when this campaign was just starting, this would be a top HN comment. Unfortunately, lazy nihilism runs deep in tech circles.
Well, maybe the next one will given that the one that didn't was just fired for it and now there's a lawsuit against the city and the developer.
You: "we should make this entity who's supposedly got the people's interest in mind extract concessions"
Them: "That entity seems to backstab the people every chance it gets"
You: "You're missing my point, the government could do it"
Perhaps you're missing the point. It's not that they can't. It's that they won't or they'll screw it up and defeat the point.
Thank you. At least someone understands what I was trying to day. You put that much better than I did.
It seems like both of you have thoroughly missed the context of my subthread.
If your goal is to point out that people make choices, well, you're in the wrong thread branch and want to instead reply to a different part of the same message that I replied to. Because I never said or implied that they don't. Quite the opposite in fact.
Here is the context of my subthread, extracted, in two parts:
Part 1, the framing.
> "I don't know if [elected officials think a wrong thing about datacenters] or if it's kickbacks."
You see, the kickbacks option is already there. We all already understand that it could be kickbacks. Therefore bringing it up further would just be silly. I certainly have no reason to say that kickbacks aren't a possibility. The only part we need to address is OP believing that [thing is wrong].
Part 2, the [thing].
> "I'm pro-progress, but a datacenter brings approximately nothing to the local economy. It doesn't employ any noteworthy number of people, it doesn't generate any real tax revenue, and it increases electricity costs for the region."
That distills to:
> "a datacenter brings approximately nothing to the local economy"
That statement means either:
A (haven't): datacenters have in the past only ever brought nothing (and therefore they will in the future only ever bring nothing)
or
B (can't): datacenters cannot bring anything other than nothing (and therefore they will in the future only ever bring nothing)
And they're both wrong. A is wrong because past behavior does not imply future behavior. And B is wrong because in fact they can.
Technically, it creates construction revenue and jobs. If you’re a municipality with FOMO heading into a job-collapse recession and someone offers you jobs on a silver platter, you might get fired from the city council for refusing it. So it’s particularly interesting to see that citizens would rather refuse datacenters than gain from them. (I certainly agree.)
Municipalities, at least in some states, can be sued for refusing development that meets existing regulation and zoning
If it's allowed by regulation and zoning, they generally don't have a say in the first place. These stories are never about building another box-shaped building in an industrial zone. We're talking about rezoning, variances, or otherwise preferential treatment.
A few steak dinners go a long way.
I think that they hear "$6 billion datacenter" and think that the town's economy is getting $6 billion in jobs rather than some foreign computer hardware company is getting $6B for computers that are housed in their town.
> I'm pro-progress
I think everyone is, by definition.
Data centers are high-dollar projects that sound great and officials are able to look like they're doing things to increase revenues, generate jobs, create infrastructure and put the community on the forefront of high-tech. Altogether, those are commendable increases, but the devil is in the details, in that all of those things do not occur at once and the costs associated with have a much longer payoff timeline.
Homeowner property tax would be 37% higher in Loudon County if not for all the datacenters. DCs are a great subsidy for the county coffers.
Property tax and (in some cases) utility taxes are deeply attractive, especially in places with large industrial-zoned swaths of land nobody is really interested in.
It's the second thing
> increases electricity costs for the region
This is really the only legitimate complaint that has any basis in reality.
But "region" is doing a lot of work here. This is typically a multi-state sized region. There are local congestion charges in some places, but overall it doesn't matter a whole lot to your electric bill if a large consumer goes in 200 miles away or across the road from you.
If it goes in across the road your local community gets the benefit of having about the least obnoxious industrial use of land possible. After construction there is very little truck traffic (e.g. much less wear and tear on local roads than a trucking terminal or manufacturing plant), and effectively is a giant office building in terms of impact on it's surroundings. In fact, until recently most of the datacenters were built in suburban office and light industrial parks and no one was the wiser.
There are legitimate complaints to be made about "datacenters" that also co-locate a natural gas or diesel power plant. But those complaints are towards building a power plant across the street, not a datacenter.
It's effectively as "free" of a tax base as you can get, assuming you don't negotiate stupid local tax abatements - which I suppose is a large caveat. Those should be simply outright illegal for everyone though, I don't see that as a datacenter specific thing. It also does effectively employ a few dozen to few hundred local tradesmen through the lifecycle of such a facility - since at these scales there is constant electric and plumbing work to be done. Usually the highest paid and highly skilled of such type of work. Many (most?) places are even using union labor for these bits.
The power problem exists broadly though. We spent a few generations not building out anything of material size and we are reaping what we have sewn. It was coming for us either way - datacenter AI bubble just brought it forward a some odd number of years. Just look at how hard it is to get a wind farm project off the ground due to NIMBY - both for the wind farm itself, and the 200 mile transmission line you might need to build to the closest major load centers. Effectively impossible.
Sure let’s completely ignore the noise pollution that makes living near one a constant hell
That's only if you co-locate a power plant near it. With proper setbacks and decent design, there is very little to no noise pollution for the vast majority of these facilities.
Most folks near them do not even know they exist. Plus you typically put them in the middle of a field with berms around them, or in a light industrial park. Not across the street from homes.
Trucking traffic creates far more noise pollution. HVAC fans spinning at optimal speed simply are not a problem for the vast majority of facilities.
Generators running during a power outage? Sure. But those typically are relatively rare events. Testing each month for an hour is just not a material complaint to me.
The cooling makes a lot of noise:
https://www.politico.com/news/2026/03/11/data-centers-ai-ele...
Add to that the health hazards that come from infrasound:
https://popwave.ai/benn-jordan/blog/data-centers-infrasound-...
People know they exist because they had to dig new wells because the water level sunk or the groundwater pollution reached high levels
https://cleantechnica.com/2025/12/02/massive-data-centers-ma...
Since managed aquifiers are rare, overall water consumption is an issue, regardless of cooling system:
https://harvardsciencereview.org/2026/02/28/re-architecting-...
As for the data enter owned power plants. Did you know that 1820 (global) gas turbines power the datacenters?
https://www.globalinforesearch.com/reports/3130730/data-cent...
I guess their point is that of all possible industrial usecases, data centers are the least obnoxious one. I live in one of the countries that actually manufactures things, unlike the US, and I find it hard to argue with that. Any noise pollution caused by data centers is far far less than most industrial setups. It's the same with every other resource, water, electricity, effect on local shared infrastructure like roads and commerce, etc,. Other industries are an order of magnitude worse.
Given that you _have_ to have some industrial setup unless you want to import everything (tokens, in this case), datacenters are far and away the best choice.
I'll add a qualifier to the above, modifying it to say that of all industrial setups generating atleast X dollars of economic value, datacenters are far and away the best in terms of impact on nbhd.
The jobs argument also falls apart, when you consider that it's essentially 100 jobs in return for just an office building worth of space. If you want a thousand job plant just build that as well next town over, it will take way way more space and other resources though. The reason that didnt happen even before this datacenter boom is because most manufacturing setups are fairly infeasible in rich countries like the US. I can't imagine the response to a textile plant or a steel plant if this is the response to datacenters.
I agree however, that if you colocate a gigantic power plant, then you get the worst of both worlds. Fewer jobs and the hindrance of a big power plant near residential areas. Grid expansion being slow in developed areas like most of the US is not surprising though.
But this is pretty much the best case scenario. Tolerating the power plant until the grid expands is the way to go I suppose.
> There are legitimate complaints to be made about "datacenters" that also co-locate a natural gas or diesel power plant. But those complaints are towards building a power plant across the street, not a datacenter.
Except of course there would be no complaints about the power plants if we did not need them in the first place to power the data centers.
> This is really the only legitimate complaint that has any basis in reality.
There are many, many others... You obviously do not live near ones, I live in Northern VA virtually surrounded by data centers and electricity costs are just part of the problem...
> gets the benefit of having about the least obnoxious industrial use of land possible
Or it could have been a lot less obnoxious residential use with parks and shit...
> Except of course there would be no complaints about the power plants if we did not need them in the first place to power the data centers.
And you only need stupid designs like tiny natural gas turbines on-site because NIMBY and lack of investment for a couple generations on the power infrastructure side. I find it difficult to be very sympathetic to our society on this issue, since I've been following it far before AI Datacenters became the thing to rage about. It was coming for us either way.
> There are many, many others... You obviously do not live near ones, I live in Northern VA virtually surrounded by data centers and electricity costs are just part of the problem...
I have lived near ones. Not datacenter alley scale, but nowhere in the world is at that level where you live. I had zero issues with them, and no one visiting even knew they existed. I've certainly seen horrible designs that should not have been permitted or built where they are, but a 500k sqft facility in the middle of 50 acres is just... not an issue to live near.
> Or it could have been a lot less obnoxious residential use with parks and shit...
Sure. Building a datacenter in the middle of a residential area is a bit silly. But we're not talking about that here. At some point you need industry to actually build things, and as industry goes this is about as light and least impactful to the local environment as it gets.
Do what you must, they've already won.
It’s a bit crazy that the elderly are blocking data center buildouts while they also expect to collect a big chunk of everyone’s payroll. Pick your lane.
Like, what did “the elderly” ever do for us? Aside from literally creating the civilization around us, and the infrastructure necessary to support it? Seems perfectly reasonable to expect reward for that, rather than pollution and increased energy costs.
None of that justifies the level of government checks they receive
> increased energy costs
Perhaps local gov that gets the gift of datacenter (aka hundreds of millions in taxes) should subsidize residents rooftop solar installations by a slight bit. Sounds like a fair deal to every party involved.
The coucilmembers probably got their bag already from whoever is building that datacenter, voting them out after the fact just means they don't have to clock in.
Generally this is good. Representatives should represent their populace and not monied interests. When they fail to do so they shoudl be removed. That's when democracy is operating correctly. But the article still contained the falsehood that data centers create jobs. This is just not accurate. Most data centers are acres of racks and HVAC with precious few humans to maintain them.
I wish when they write these storied they'd put the town's per capita income in brackets the way they do with politician's party affiliation or company's ticker. The "Fairfax of St. Louis" voting out half their legislature over a project means something very different than the "Newark of St. Louis" doing the same.
And why is that?
Because wealthy suburbs have said "not here, move your filthy industry somewhere else" since forever.
When the places that aren't swimming in jobs, the local government isn't swimming in property tax revenues and frankly probably can't even enforce the rules they're federally compelled to have without destroying everything says "take that somewhere else" it means something entirely different.
Take it from someone who lives in the St. Louis area, Festus is by no means a snooty suburb. It's a small town that has become a distant suburb of St. Louis due to sprawl spreading south into the adjacent county over the past 20 years.
For comparison, an upscale suburb of similar size (Town and Country, Missouri) has a median household income of $202,974, as compared to $59,041 for Festus. The average person you meet in Town and Country is likely to be a doctor, attorney, or executive. In Festus, the average person likely works in a factory, farm, or lead mine.
Thank you. That's exactly the kind of context I was looking for.
The politicians talked about distracting locals with a new Olive Garden. I’m guessing that was included to give us this answer.
It will be interesting to see what we do with these enormous concrete boxes once we find a better way to do whatever we think all this 'ai' is going to do. 'Dead malls' that are being partially converted into pickleball courts and places for people to take their daily 'constitutional' indoors are going to seem quaint.
That is if the bubble doesn't pop because of other factors first that is...
> It will be interesting to see what we do with these enormous concrete boxes once we find a better way to do whatever we think all this 'ai' is going to do.
These are just prefab tip-up walls with metal roofs, bar joists, columns spaced 30-40 feet OC, and a 6” concrete slab.
It’s no different than an Amazon distribution center once all the electrical substations, switchgear, generators, and UPSes are gone.
... sure, again once it's all pointless it's still a gigantic concrete box. Tons of old hastily built automotive factories just rotting around detroit.
This is becoming a pretty clear wedge between red and blue. Why do you think Musk opened his diesel turbine driven data center in rural Mississippi? Big Tech is systematically targeting small municipalities across the US with promises of insane money to anyone willing to sell out their residents. Missouri being traditionally purple, it makes a lot of sense the flashpoint would be here.
Red states are against the deals as well. Many people in Texas are fighting back but sometimes it’s too late because the deal was done in secret.
It's crazy that these are done in secret. From the article: "The operator of the data center hasn't been identified" -- that's shouldn't be allowed.
It's not possible to identify operators who do not wish to be identified. The land is purchased by an LLC, developed by another LLC, and then sold to a third LLC for buildout. Once it's up and running it could be rented to a fourth LLC who has contracted with the actual client, or the client could just buy the involved LLCs (usually via another LLC).
>Red states are against the deals as well.
The people are, their politicians are not. Overwhelmingly this is a problem of backroom deals with state and local Republicans subverting the electorate's will.
That data center is in the Memphis suburbs, 5 miles from the airport, in the third largest city in the state. Wouldn't call it rural
NIMBYism comes in all colors
Is it still 'NIMBYism' if what they are building is something that none of the residents want at all? Like yeah no-one wants the new sewage plant in their backyard but they will still see the value in it.
Even as someone in tech I don't see the value in this scale of buildout. This technology is very new and no doubt in my mind that in 5 years we will be able to do everything these models do on a fraction of the resources.
This is 100% not a political issue, red & blue are lining up against DCs. the DC capital of the world is Northern Virginia which is bluer than Bernie
Those are mostly more traditional data centers. 20 or 30 kilovolt racks with separate cages for different tenants, some meet-me rooms, and a bunch of telecom gear are the order of the day.
What exactly are you talking about?! :)
https://www.datacentermap.com/content/nova/
70% of global internet traffic flows through here
Internet traffic requires routing and switching. That’s all traditional DC equipment in terms of power and cooling. They don’t require 80 or 100 kilovolt racks like something stuffed full of AI accelerators.