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.
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.
> 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.
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).
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.
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.
> 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.
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.
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.
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.
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.
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.
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'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.
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.
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.
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.)
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.
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.
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
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.
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.
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.
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.
> 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.
Local shops will absolutely be contracted to work on the project. A datacenter project like this can't find enough qualified electricians.
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).
>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.
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.
> 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.
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.
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.
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.
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.
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.
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.
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'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.
Municipalities, at least in some states, can be sued for refusing development that meets existing regulation and zoning
> I'm pro-progress
I think everyone is, by definition.
> 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.
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.
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.
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.
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.)
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.
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.
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.
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
Snowpiercer but with data centers? The breeze would help with cooling!
Do what you must, they've already won.
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.
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.
NIMBYism comes in all colors
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