I really don’t understand why data centers get such special treatment.
Somewhere gives Google etc a tax break for a data center. In exchange Google destroys a huge swath of land, Hoovers up electricity and water, and provides some temporary construction jobs.
In exchange the local residents get nothing… at least if it was a power plant or something they could get some cheaper electricity. Maybe they get slightly better internet connectivity? But from what I’ve read this doesn’t really seem to be the case.
> A firm announces a plan to build a new facility, but where? Local and state development officials compete to attract the firm with ever-more-generous tax breaks and subsidies. This scene plays out again and again – even though research shows that incentives do not substantially influence firm behavior, even in the face of media exposes about wasteful giveaways. Why? Governments hope to encourage jobs and business profits, and hubris leads officials to believe “this time will be different,” even if incentives have not worked before.
> But something more pervasive is also at work. My research with Stephen Ellis demonstrates the role of “business climate” in driving economic development professionals and government officials to engage in an incentives arms race. Officials feel they must offer incentives, because failing to compete to attract businesses will be interpreted as evidence that their locality is not business-friendly. States and localities will therefore continue to compete, to the point of giving away more than the value of the new firm or facility. Can American citizens find ways to prevent the negative effects of this no-win arms race?
It's interesting because the cities some people decry as not business friendly often get lots of job growth while charging companies a lot to expand there.
Cities like NYC, SF, Boston, Seattle usually offer zero incentives and even charge developers fees for new development.
If you live in Eastern Mass, like I do, there isn't much buildable land left unless you start using state forests and golf courses. The easiest to acquire land are old brownfield sites. But even there, it's more likely they will plaster over the original building to seal in the asbestos and lead paint, rather than build a new, denser structure.
Out at Interstate 495, there is a bunch of open space, but a lot of it is zoned conservation land. To take something out of conservation zoning, you need to pay all the taxes they would have incurred from the time it was zoned as conservation land to the current day.
Sometimes I think the only reasonable solution is to start razing in-city industrial property and building it over with dense, high-occupancy, 10+ story buildings. The old mill city I live in has done that with some success. Their solution to the parking problem was to build multiple city-owned parking garages and make parking costs explicit, rather than bundling them into the rent.
The main downside is that if you don't drive, you have to spend half an hour plus on a bus to get to a supermarket or other necessities. I guess the cost of a supermarket lease makes setting up a business there significantly riskier.
> Officials feel they must offer incentives, because failing to compete to attract businesses will be interpreted as evidence that their locality is not business-friendly.
I guess this is the crux but who really wants a data center? It’s a big flashy number but what does it really do for the community at the end of the day?
Because cities felt the devastating effects when industrial factories staffed with good union jobs went away, and yearn for their carbon copy replacement. These factories had ripple regional advantage effects beyond the factory workers. Armies of teamsters have to drive in and out of town to deliver inputs and outputs for industrial factories, and they all need to eat. Corporate and R&D types used to need to spend more time at the industrial factory. Put a factory in a region and a corporate office often follows. Put enough of them in the same region and you start to get an innovation hub as they all hang out and see each other at third spaces. Universities and innovation hubs mutually benefit and expand when distance matters.
So the industrial factory tax break model often did pay off. Data centers are selling the same story: give us tax breaks for big expensive capital investment and regional prosperity is yours. They often lie about even the direct number of jobs. But the implied regional advantage is definitely dead when it is all cloud and zoom, rather than widgets and happy hours.
I don’t think all of them. I believe data center are a net negative for the residents in the area. I live in South Texas and the county here really gave it their all to get Space X down here. It has become a big employer in this area. There’s a few locals I know that work there as HVAC or welding. The higher skilled jobs move in people from outside here, but they contribute to the local economy by shopping, entertainment, eating, and taxes.
Yeah, something like that happened nearby. They set up this bike repair store near my place and started repairing bikes. No one asked me if I even wanted this store nearby. I've been living here for like a decade and they gave me nothing. They offered to repair my bike for a fee, but I can already do that at home for free. We should ban businesses if they won't pay everyone within the vicinity the fee they demand.
I think there was some kind of corruption as to how they get to just repair bikes without giving me anything. I've been a local for way longer than these people.
EDIT: And yes, they do get government money. SF has a city program to encourage local businesses or something so they get grants. Besides all businesses are eligible for SBA loans and no one asked me if they should be.
A bike shop or other small retail business brings jobs and benefits to the local community. A data center brings pollution and practically no new permenant jobs or benefits specifically to that community, only to the cloud.
But don't worry about it, this is something an $850/hr consultant is paid to not understand.
What makes datacenters more corruptible than a real estate developer building a mall or office building? Unlike Microsoft or Google, your local mom&pop developer doesn't have an audit committee or a policy on ethics/corruption, so it's probably easier for them to bribe local politicians as well.
>even housing produces far more economic activity than a shed full of servers
Objectively not, given they're able to outbid all everyone else for the same land. You might not like a massive datacenter being built to serve ads or generate AI slop, but the fact that investors are willing to put money into it, and no one else can outbid them means the market expects datacenter will generate more future "economic activity" than any other possible use for the land. Whether a $1 generated by adtech is worth more than $0.5 (or whatever) generated by a car factory is a separate discussion, but arguing over how much "economic activity" is the wrong way to approach this.
Datacenters have very little side-effects. The whole environmental "costs" are overblown. They consume little space and water when compared to agriculture and you don't need to have them in the city center. They can exist in remote and desolate areas. As far as electricity, they pay their bills themselves most of the time.
Compare that to most other industries where you have pollution, noise, truck traffic, low paying and neck-breaking jobs, need import/export ports, etc. The footprint of datacenters is rather low.
Most countries have a sh%t ton of space and some have lots of water (ie: Chile). There is little reason not to have them.
As I understand, data centers from hyperscalers (Microsoft, Google, Amazon, Apple, Meta, Oracle, IBM, Alibaba, etc.) don't create many long-term jobs. Mostly they create short term jobs in the construction industry. Data centers from non-hyperscalers, like Equinix, probably create more long-term jobs because companies rent space, then need (near constant) changes to their hardware that requires "remote hands". I think hyperscalers mostly install monolithic hardware that requires almost no regular changes, except for break-fix.
Having worked in a hyper scaler on a system to orchestrate hw breakfix, it requires a lot of hands.
We replaced thousands of hard drives alone per day...
And racks are constantly coming and going. Hyperscaler DCs are busy places.
This is good first hand information. If you had a guess, how much did these people make per hour? Someone who (1) locates a physical server (amoungst thousands) and (2) replaces a hard drive has roughly the same skillset as someone stocking supermarket shelves in 2025. I guess then make max 20 USD per hour.
>but not compared to similar businesses of their capital value
So? it's not like if hyperscalers weren't building datacenters, the billions that would otherwise be spent on GPUs would be spent on 10 car factories or whatever. The only reason the billions was being invested in the first place was because there's a craze for AI datacenters
When I was last working in Chile on a SaaS product for use in Chile, we deployed everything to US-EAST. We still had local CDN caching, but it is always nice to be closer to your hardware. I can't tell from the article what types of datacenters these are, but if these are new Azure and Google Cloud availability zones, this will be a great for latency.
It looks like a maximum security prison so the locals cannot revolt against "AI".
Of course no jobs are created. The data centers are there for cheaper energy, laxer environmental regulations and for the ability to process U.S. citizens' data and build files on them where it would be illegal to do that on U.S. soil.
So these companies receive subsidies like in the U.S. How does the government shutdown affect these subsidies in the U.S.? Are SNAP benefits for corporations being halted as well or does it only affect poor people?
The same thing happened here in NZ with AWS. The job creation stories is almost always bullshit.
Down here in NZ the govt subsidizes an aluminium smelter which uses 13% of the national electricity just to provide 3,000 jobs. In the media headlines 3,000 jobs seems quite measurable and impactful compared to many customers overpaying for electricity by an mysterious undetermined amount. It's obvious which way the politicians are going to lean.
> Down here in NZ the govt subsidizes an aluminium smelter which uses 13% of the national electricity just to provide 3,000 jobs
I'm not in favour of giving the smelter owners a sweet deal but I believe there is some nuance which is lost in your comment.
When you say subsidize I assume you're talking about the price the smelter pays for electricty (I'm not aware of any direct subsidy).
Until about 2022 the transmission lines out of Manapouri heading north could only handle a part of what Manapouri could produce, other lines headed towards Tiwai Point to feed the smelter with the balance of Manapouri's output. This meant that negotiating electricity prices with the smelter owners was tricky because it was perfectly clear that there was nowhere else to take the electricity. In the past five years more capacity has been added to allow electricity from Manapouri to reach the National Grid and so, I presume, this significantly dilutes downward price pressure from the smelter.
I'm not sure how these matters are handled in NZ, but in the EU they would just raise prices so private consumers have to save electricity while the industries are subsidized.
I really don’t understand why data centers get such special treatment.
Somewhere gives Google etc a tax break for a data center. In exchange Google destroys a huge swath of land, Hoovers up electricity and water, and provides some temporary construction jobs.
In exchange the local residents get nothing… at least if it was a power plant or something they could get some cheaper electricity. Maybe they get slightly better internet connectivity? But from what I’ve read this doesn’t really seem to be the case.
https://scholars.org/contribution/how-competition-attract-bu...
> A firm announces a plan to build a new facility, but where? Local and state development officials compete to attract the firm with ever-more-generous tax breaks and subsidies. This scene plays out again and again – even though research shows that incentives do not substantially influence firm behavior, even in the face of media exposes about wasteful giveaways. Why? Governments hope to encourage jobs and business profits, and hubris leads officials to believe “this time will be different,” even if incentives have not worked before.
> But something more pervasive is also at work. My research with Stephen Ellis demonstrates the role of “business climate” in driving economic development professionals and government officials to engage in an incentives arms race. Officials feel they must offer incentives, because failing to compete to attract businesses will be interpreted as evidence that their locality is not business-friendly. States and localities will therefore continue to compete, to the point of giving away more than the value of the new firm or facility. Can American citizens find ways to prevent the negative effects of this no-win arms race?
It's interesting because the cities some people decry as not business friendly often get lots of job growth while charging companies a lot to expand there.
Cities like NYC, SF, Boston, Seattle usually offer zero incentives and even charge developers fees for new development.
>Cities like NYC, SF, Boston, Seattle usually offer zero incentives and even charge developers fees for new development.
That's not a good thing. Those cites also have the lowest rates of new residential construction (in %) and housing shortages.
If you live in Eastern Mass, like I do, there isn't much buildable land left unless you start using state forests and golf courses. The easiest to acquire land are old brownfield sites. But even there, it's more likely they will plaster over the original building to seal in the asbestos and lead paint, rather than build a new, denser structure.
Out at Interstate 495, there is a bunch of open space, but a lot of it is zoned conservation land. To take something out of conservation zoning, you need to pay all the taxes they would have incurred from the time it was zoned as conservation land to the current day.
Sometimes I think the only reasonable solution is to start razing in-city industrial property and building it over with dense, high-occupancy, 10+ story buildings. The old mill city I live in has done that with some success. Their solution to the parking problem was to build multiple city-owned parking garages and make parking costs explicit, rather than bundling them into the rent.
The main downside is that if you don't drive, you have to spend half an hour plus on a bus to get to a supermarket or other necessities. I guess the cost of a supermarket lease makes setting up a business there significantly riskier.
> Officials feel they must offer incentives, because failing to compete to attract businesses will be interpreted as evidence that their locality is not business-friendly.
I guess this is the crux but who really wants a data center? It’s a big flashy number but what does it really do for the community at the end of the day?
Tax revenue.
https://www.google.com/search?q=nova+tax+revenue+data+center...
> A data center costs the county $0.04 per $1 of tax revenue received, whereas normal businesses cost about $0.25 per $1 of revenue
Because cities felt the devastating effects when industrial factories staffed with good union jobs went away, and yearn for their carbon copy replacement. These factories had ripple regional advantage effects beyond the factory workers. Armies of teamsters have to drive in and out of town to deliver inputs and outputs for industrial factories, and they all need to eat. Corporate and R&D types used to need to spend more time at the industrial factory. Put a factory in a region and a corporate office often follows. Put enough of them in the same region and you start to get an innovation hub as they all hang out and see each other at third spaces. Universities and innovation hubs mutually benefit and expand when distance matters.
So the industrial factory tax break model often did pay off. Data centers are selling the same story: give us tax breaks for big expensive capital investment and regional prosperity is yours. They often lie about even the direct number of jobs. But the implied regional advantage is definitely dead when it is all cloud and zoom, rather than widgets and happy hours.
> In exchange the local residents get nothing
Isn't that the case for a lot of these corporate welfare programs?
I don’t think all of them. I believe data center are a net negative for the residents in the area. I live in South Texas and the county here really gave it their all to get Space X down here. It has become a big employer in this area. There’s a few locals I know that work there as HVAC or welding. The higher skilled jobs move in people from outside here, but they contribute to the local economy by shopping, entertainment, eating, and taxes.
I think so yeah. They generally seem to be pretty disingenuous. Which is a shame because these corps have more than enough resources.
Yeah, something like that happened nearby. They set up this bike repair store near my place and started repairing bikes. No one asked me if I even wanted this store nearby. I've been living here for like a decade and they gave me nothing. They offered to repair my bike for a fee, but I can already do that at home for free. We should ban businesses if they won't pay everyone within the vicinity the fee they demand.
I think there was some kind of corruption as to how they get to just repair bikes without giving me anything. I've been a local for way longer than these people.
EDIT: And yes, they do get government money. SF has a city program to encourage local businesses or something so they get grants. Besides all businesses are eligible for SBA loans and no one asked me if they should be.
Okay. Did the bike shop get a tax break to be set up? If not then your ironic analogy has failed.
O wow I didn’t realize bike repair shops were 1M+ sq ft!
You could fit every bike in your city in there lol.
A bike shop or other small retail business brings jobs and benefits to the local community. A data center brings pollution and practically no new permenant jobs or benefits specifically to that community, only to the cloud.
But don't worry about it, this is something an $850/hr consultant is paid to not understand.
The obvious conclusion to me is that corruption is involved
Agreed. Difficult to prove but it's the obvious explanation.
What makes datacenters more corruptible than a real estate developer building a mall or office building? Unlike Microsoft or Google, your local mom&pop developer doesn't have an audit committee or a policy on ethics/corruption, so it's probably easier for them to bribe local politicians as well.
they're the worst use of land possible
even housing produces far more economic activity than a shed full of servers
>even housing produces far more economic activity than a shed full of servers
Objectively not, given they're able to outbid all everyone else for the same land. You might not like a massive datacenter being built to serve ads or generate AI slop, but the fact that investors are willing to put money into it, and no one else can outbid them means the market expects datacenter will generate more future "economic activity" than any other possible use for the land. Whether a $1 generated by adtech is worth more than $0.5 (or whatever) generated by a car factory is a separate discussion, but arguing over how much "economic activity" is the wrong way to approach this.
> Objectively not, given they're able to outbid all everyone else for the same land.
they're often subsidised by hopeful governments, and every tax they would pay is offset by "licensing"
it's not a level playing field
Datacenters have very little side-effects. The whole environmental "costs" are overblown. They consume little space and water when compared to agriculture and you don't need to have them in the city center. They can exist in remote and desolate areas. As far as electricity, they pay their bills themselves most of the time.
Compare that to most other industries where you have pollution, noise, truck traffic, low paying and neck-breaking jobs, need import/export ports, etc. The footprint of datacenters is rather low.
Most countries have a sh%t ton of space and some have lots of water (ie: Chile). There is little reason not to have them.
As I understand, data centers from hyperscalers (Microsoft, Google, Amazon, Apple, Meta, Oracle, IBM, Alibaba, etc.) don't create many long-term jobs. Mostly they create short term jobs in the construction industry. Data centers from non-hyperscalers, like Equinix, probably create more long-term jobs because companies rent space, then need (near constant) changes to their hardware that requires "remote hands". I think hyperscalers mostly install monolithic hardware that requires almost no regular changes, except for break-fix.
Having worked in a hyper scaler on a system to orchestrate hw breakfix, it requires a lot of hands. We replaced thousands of hard drives alone per day... And racks are constantly coming and going. Hyperscaler DCs are busy places.
This is good first hand information. If you had a guess, how much did these people make per hour? Someone who (1) locates a physical server (amoungst thousands) and (2) replaces a hard drive has roughly the same skillset as someone stocking supermarket shelves in 2025. I guess then make max 20 USD per hour.
Busy places but not compared to similar businesses of their capital value I imagine.
>but not compared to similar businesses of their capital value
So? it's not like if hyperscalers weren't building datacenters, the billions that would otherwise be spent on GPUs would be spent on 10 car factories or whatever. The only reason the billions was being invested in the first place was because there's a craze for AI datacenters
When I was last working in Chile on a SaaS product for use in Chile, we deployed everything to US-EAST. We still had local CDN caching, but it is always nice to be closer to your hardware. I can't tell from the article what types of datacenters these are, but if these are new Azure and Google Cloud availability zones, this will be a great for latency.
Edit: It looks like all the major cloud providers have Chile AZs except Amazon, which has one planned: https://aws.amazon.com/blogs/aws/coming-soon-aws-south-ameri...
It looks like a maximum security prison so the locals cannot revolt against "AI".
Of course no jobs are created. The data centers are there for cheaper energy, laxer environmental regulations and for the ability to process U.S. citizens' data and build files on them where it would be illegal to do that on U.S. soil.
So these companies receive subsidies like in the U.S. How does the government shutdown affect these subsidies in the U.S.? Are SNAP benefits for corporations being halted as well or does it only affect poor people?
a fence and some rent-a-cops won't stop anything but a trivial revolt
The same thing happened here in NZ with AWS. The job creation stories is almost always bullshit. Down here in NZ the govt subsidizes an aluminium smelter which uses 13% of the national electricity just to provide 3,000 jobs. In the media headlines 3,000 jobs seems quite measurable and impactful compared to many customers overpaying for electricity by an mysterious undetermined amount. It's obvious which way the politicians are going to lean.
> Down here in NZ the govt subsidizes an aluminium smelter which uses 13% of the national electricity just to provide 3,000 jobs
I'm not in favour of giving the smelter owners a sweet deal but I believe there is some nuance which is lost in your comment.
When you say subsidize I assume you're talking about the price the smelter pays for electricty (I'm not aware of any direct subsidy).
Until about 2022 the transmission lines out of Manapouri heading north could only handle a part of what Manapouri could produce, other lines headed towards Tiwai Point to feed the smelter with the balance of Manapouri's output. This meant that negotiating electricity prices with the smelter owners was tricky because it was perfectly clear that there was nowhere else to take the electricity. In the past five years more capacity has been added to allow electricity from Manapouri to reach the National Grid and so, I presume, this significantly dilutes downward price pressure from the smelter.
There's also the job creation for the extra energy generation...
I'm not sure how these matters are handled in NZ, but in the EU they would just raise prices so private consumers have to save electricity while the industries are subsidized.
And that might understandable if the company contributed tax or national security to the state which subsidizes it. But it's usually not like that.