The secret deals, NDAs, soaring power bills, and vanishing water behind the data centers training AI, and how communities are fighting back and winning.
Did some research and abused Claude on fact checking, arguments, etc. Maybe this will add some interesting information to this discussion:
Data centers don't need to keep growing as large and centralized as current plans assume. Distributed edge sites reduce grid strain and connect faster; smaller models handle most routine queries at a fraction of the energy; and renewable-powered, low-water facilities (seawater cooling, near-zero-water designs) are already running commercially, not hypothetical. Together these can meet real AI demand with far less pressure on power and water supplies than the hyperscale build-out path assumes.
That said, planners are right that some centralized, high-density capacity remains genuinely necessary for training and tightly-coupled workloads, and small models can't yet replace frontier ones for complex tasks. So this isn't an argument to halt data centers — it's an argument that growth must be bounded and sited responsibly, with distributed/efficient options used wherever they can substitute for new mega-campuses.
**The priority that should govern this trade-off:** homeowners' access to affordable, reliable electricity and water — and the public health consequences of losing either — must outrank data center revenue or speed-to-market. Concretely, that means:
- Data centers should not draw down shared water supplies or strain residential grids in water-stressed or already-constrained regions, regardless of projected financial return.
- New capacity should be required to bring its own power (on-site renewables, waste-heat recovery, closed-loop cooling) rather than compete with households for existing grid and water allocations.
- Permitting and rate structures should protect residential utility costs and supply reliability first; data center growth should absorb the cost of new generation/transmission it requires, not pass it to ratepayers.
- Health-relevant local impacts (grid instability, water table effects, noise, air quality from backup generation) should be weighed above build-out speed or corporate cost savings in siting decisions.
In short: the technology to grow AI compute sustainably already exists. Whether it gets used depends on whether policy treats homeowner power/water security and public health as the binding constraint — rather than treating them as costs to be offset once financial commitments are already made.
Thank you. It seems the daily outrage is by design- to keep us distracted and exhausted while they remake our world into one uninhabitable for ordinary people.
Strong piece, Mitch. The NDAs and secret approvals are real. But I'd add one layer: even when communities win the transparency fight, the core problem remains. The reason data centers need so much water, land, power, and grid capacity is architectural — hyperscale infrastructure designed for web apps in 2010 is being forced to run AI workloads it was never engineered for.
Legal and community advocacy is necessary. But the structural answer is going to the source of power and building AI-native infrastructure from the ground up — so these facilities don't need to show up in communities at all. We've been tracking where this market is heading from an operator perspective. Free report: aiofthecoast.dcxps.com
I had to examine the website of DCC, and their statement.
The (DCC) Data Center Coalition as a non-profit org., has written that it advocates, (translated: they lobby in secret/NDAs, Non-Disclosure Agreements) for greater access to "clean" energy, and "tax policies" to support the growth & success of these DATA Centers.
Let's recall what took place with this admin, the EPA & Biden's clean energy plan.
Also, I may add, at the expense of health, well being and survival of you and your communities, town and cities.
Farm lands and forestries are being acquired by DCC members - Mark Zuckerberg's Meta, Jeff Bezo's Amazon, Alphabet's Google, Microsoft, Elon Musk's xAIs, large campaign
donors to Donald Trump's askings; however, these massive Data Centres with Gas Turbines are releasing deadly chemicals & toxin particles in the air we breathe, in our waterways & soils.
These Data Warehouses' operate 24hrs x 365 days, an overutilization of our water & electricity, not only deplete the needs of humanity, we are forced to pay increased rates, masked as grid-upgrades etc
This tech may offer some solutions down the line...
Atoco is a California startup built on metal-organic frameworks, a class of porous crystalline materials invented by Omar Yaghi, the UC Berkeley chemist who shared the 2025 Nobel Prize in Chemistry with Susumu Kitagawa and Richard Robson for developing them. These frameworks, called MOFs, are engineered at the atomic level so that a few grams hold an internal surface area the size of a sports arena, and they are tuned to trap only water molecules out of the surrounding air, even in extremely dry regions. Atoco packs this material into a unit the size of a 20-foot shipping container, where the MOFs silently pull H2O from the air and store it in their microscopic cavities before releasing it as water. Because the material grabs only H2O, the water comes out near distilled quality, free of PFAS, microplastics, and other contaminants. The company has tested the technology in Death Valley, one of the driest places on Earth, and confirmed it can still produce clean water there. Yaghi serves as Atoco's Chief Science Officer.
The output depends heavily on conditions and on how the unit is powered. On-grid units that run on electricity cycle more often and produce roughly 2,000 to 4,000 liters, about 1,057 gallons at the top end, of high-purity water per day, while off-grid units that run on ambient or waste heat produce up to about 1,000 liters a day, so the headline figure of 4,000 liters is a best case rather than a guarantee. As of spring 2026 the technology is preproduction, with a commercial prototype demonstrated for Bloomberg in an Orange County parking lot in April 2026 and volume production slated for later in the year, pending field tests. Atoco has not released pricing, though it estimates the water will cost a few cents per liter, which is more expensive than desalinated water, with the selling points being purity, decentralization, and resilience rather than low cost. Interest in the technology has risen partly because the Iran war exposed the vulnerability of large centralized desalination plants that became military targets, and Atoco is marketing the units to data centers, semiconductor plants, hospitals, farms, and any water-stressed region, though it would take roughly a dozen units to supply even a single water-efficient data center, and a competitor called AirJoule, a joint venture with GE Vernova, is pursuing a similar MOF-based harvester producing about 2,000 liters a day on a comparable timeline.
Yes. I'm already familiar with the early work involved with this update. They even make a small plastic box contraption that yields an average of a litre and half per day for home use. The purity of the water is a great selling point for off grid uses. It's simple physics that works well in the desert. There is some evidence that it was a practical idea used in ancient Egypt using the same condensation principles.
I own three shares of GE Verona. A very innovative focused technology company spun off from the old General Electric Co.
This is one of the types of technology that I was referring to in my "demands". Another is using heat based capture and electricity regeneration systems developed for Formula One racing. Also there are kinetic based electricity regeneration modules that absorb vibrations and g force loads from movement in race cars that can be modified for large structures to take up energy releases from machinery and give back current.
...and there is nothing wrong with covering every square inch of the exterior of a data center with solar cells AND if they were built vertically the height could accommodate some wind generators on top, too.
I also think that maybe using AI to ask itself how to build a super energy efficient data center could go a very long way to solving some of the problems with developing energy savings in their design ing through the completion process.
Awhile back I did ask how many offshore wind generators it would take to power the numbers of data centers expected to come on line by 2035 based on their needed TerraWatt load. 37,000. There are currently less than 100 off the East Coast. Seems s bit high...
If data centers have to be built the following notes should be considered before any breaking of ground takes place...
A. They must be built on brownfields. Brownfields are already spoiled from previous industrial usage and can be cleaned up during construction by the rich bastards who demand data centers for the increase of their wealth.
No breaking of any fresh ground any where. Especially in areas that are deemed arable land suitable for crops or have some amount of historical references to battlefield, neglected cemetery, designated as historic property, etc. Also near or are on protected wetlands, upland wetlands or forested lands currently suitable for recreational use. That alone stops eminent domain takings by developers of family farms or set asides previously designated for such purposes.
B. Data center construction must be VERTICAL to decrease the footprint. No more sprawl over large acreages. Going vertical has a lot of better outcomes for heat loading, and decreased water usage. It's an engineering challenge and not a giveaway to the damn development crowd. The footprint size is the reason. Sprawling is not to be allowed. Go up in stories or go home. Too, the minor cost of elevators is absorbed through better engineering. Expectations should be 40-80 story in range and going deeper into the ground wouldn't be out of the ordinary. This alone can force data centers to be closer to urban areas where tall structures are the norm. Fire protection costs go down as rural fire department resources would not be driven up for specialized apparatus.
C. Water usage must come from gray water sources or sea water. No more using up treated potable water necessary for human consumption. One more damn engineering challenge that must be met, not given away at the expense of municipal resources. No proffer amounts can alleviate the long term problems to local infrastructure costs over time. Recouping those additional costs over the expected life cycle of a data center is pure folly. Localities believing the developers projected costs to their taxpayer base, especially with the giving of incentives, are likely in the back pockets of the tech bros. Citizens must demand full accountability BEFORE and after by transparency of each entire project through out the process. Water is the most valuable resource on the planet.
D. Power consumption must be from renewable energy. No usage of dozens or hundreds of large displacement diesel engines. No usage of large scale gas turbine engines for powering them, either. A data center usage of power using more hydrocarbon fuel than a fleet of aircraft flying to Europe and back every day? That is purely stupid... So, build up to scale these demands for every single one of these data centers by all renewable energy... Do it without the extreme air pollution created by back up power generation on site.
Donald Trump will just have to give up his unfounded dreaded fear of "windmills" and realize that offshore wind generators needed to create power for his data center bros will need to be fast tracked and not ever delayed by mere whim. It's just an engineering challenge easily met by the vast riches of his under taxed data center bromancers.
E. Local and state support for data centers must include heavy oversight and complete inputs on public health and safety through strict compliance
to building and fire codes and established zoning requirements. Demand nothing less!
As the data center industry evolves under the above conditions the engineering challenges will be met and over time can be seen as a working partnership that may prove worthy. People have control if they exercise the better options to keep with health and safety in this endeavor.
Thanks for this very informative article. We tend to hear about these economic, environmental and social impacts as one offs. Thank you for showing the cumulative impact by addressing them all in a single article.
The larger question is the fact that governments have done so little to curb the size, reach and power of these massive corporations. Most of the companies constructing these facilities have market cap valuations exceeding the GDP of many countries in the world. Over a century ago, the US government took action to break up Bell Telephone and US Steel. Given the financial and political power of these companies, would any similar attempt today have any chance of success?
Huge issues, big potential, large downsides. It's well above my pay grade and I don't have the answers. I just know we need to get all this figured out sooner rather than later.
The key ingredient in breaking up these malignant companies is "political will." And this year we have an opportunity to increase it! I read recently that about 70% of Americans don't want a data center anywhere near them (which is why they're getting dumped in poor rural areas).
I think far too many electeds are just thinking "tax base." We don't tax the rich, and we have to sacrifice the future for "tax base."
I don't think any establishment Democrat is going to "get" this, as they rely on big donors (or conversely: the absence of Big Money for an opponent).
One question I have is : Why, actually, do we need all these data centers? Is it anything more than the big tech companies trying to scrape more data to take more money and freedoms from us?
A comment about the power plants. Gas turbines, of course, burn fossil fuels, which contribute to climate change. But not all gas turbines are alike. Combined-cycle plants with modern pollution control are pretty clean, with zero particulates, no sulfur oxides, and pretty low nitrogen oxides. On the other hand, single-cycle peaker plants emit far more pollution of all kinds.
Wanna guess what the difference is?
$$$$$$$.
Combined-cycle plants are far more costly to buy, build, and operate.
Single-cycle rapid startup machines are cheap and quick.
I'll leave you to guess which species of gas turbine is preferred by the data center builders.
Also, of course, any fossil fuel generator is bad for climate.
MOFs have been around for a long time and are pretty remarkable. The big question is how fast water can move in or out of the material — the mass transfer rate. I will have to look at what this startup is doing.
I think I know the reasons I may write about this in the future. It's a concept that was around long before these data centers and focuses more on community development and privacy during conversations than on pulling the wool over the consumer's eyes. But recently I think that's exactly what's been happening.
As I noticed in several projection graphings on other web sites, too. It's like they really don't want anyone to know until they break ground.
Pennsylvania had clusters with data centers seemingly on top of one another in concentrations. Lazy map making? Or different data center developers fighting over scarce land?
I think it has more to do with avoiding public scrutiny and competition than anything else. Of course maximizing profits over people will be the end play for most large corporations.
Did some research and abused Claude on fact checking, arguments, etc. Maybe this will add some interesting information to this discussion:
Data centers don't need to keep growing as large and centralized as current plans assume. Distributed edge sites reduce grid strain and connect faster; smaller models handle most routine queries at a fraction of the energy; and renewable-powered, low-water facilities (seawater cooling, near-zero-water designs) are already running commercially, not hypothetical. Together these can meet real AI demand with far less pressure on power and water supplies than the hyperscale build-out path assumes.
That said, planners are right that some centralized, high-density capacity remains genuinely necessary for training and tightly-coupled workloads, and small models can't yet replace frontier ones for complex tasks. So this isn't an argument to halt data centers — it's an argument that growth must be bounded and sited responsibly, with distributed/efficient options used wherever they can substitute for new mega-campuses.
**The priority that should govern this trade-off:** homeowners' access to affordable, reliable electricity and water — and the public health consequences of losing either — must outrank data center revenue or speed-to-market. Concretely, that means:
- Data centers should not draw down shared water supplies or strain residential grids in water-stressed or already-constrained regions, regardless of projected financial return.
- New capacity should be required to bring its own power (on-site renewables, waste-heat recovery, closed-loop cooling) rather than compete with households for existing grid and water allocations.
- Permitting and rate structures should protect residential utility costs and supply reliability first; data center growth should absorb the cost of new generation/transmission it requires, not pass it to ratepayers.
- Health-relevant local impacts (grid instability, water table effects, noise, air quality from backup generation) should be weighed above build-out speed or corporate cost savings in siting decisions.
In short: the technology to grow AI compute sustainably already exists. Whether it gets used depends on whether policy treats homeowner power/water security and public health as the binding constraint — rather than treating them as costs to be offset once financial commitments are already made.
Thank you. It seems the daily outrage is by design- to keep us distracted and exhausted while they remake our world into one uninhabitable for ordinary people.
Strong piece, Mitch. The NDAs and secret approvals are real. But I'd add one layer: even when communities win the transparency fight, the core problem remains. The reason data centers need so much water, land, power, and grid capacity is architectural — hyperscale infrastructure designed for web apps in 2010 is being forced to run AI workloads it was never engineered for.
Legal and community advocacy is necessary. But the structural answer is going to the source of power and building AI-native infrastructure from the ground up — so these facilities don't need to show up in communities at all. We've been tracking where this market is heading from an operator perspective. Free report: aiofthecoast.dcxps.com
A comprehensive review with a compelling call-to-action for average Joes like me.
¡Bravò, Mitch, bravò!
Nicely done good and faithful savant!
Techno Mordor
I had to examine the website of DCC, and their statement.
The (DCC) Data Center Coalition as a non-profit org., has written that it advocates, (translated: they lobby in secret/NDAs, Non-Disclosure Agreements) for greater access to "clean" energy, and "tax policies" to support the growth & success of these DATA Centers.
Let's recall what took place with this admin, the EPA & Biden's clean energy plan.
Also, I may add, at the expense of health, well being and survival of you and your communities, town and cities.
Farm lands and forestries are being acquired by DCC members - Mark Zuckerberg's Meta, Jeff Bezo's Amazon, Alphabet's Google, Microsoft, Elon Musk's xAIs, large campaign
donors to Donald Trump's askings; however, these massive Data Centres with Gas Turbines are releasing deadly chemicals & toxin particles in the air we breathe, in our waterways & soils.
These Data Warehouses' operate 24hrs x 365 days, an overutilization of our water & electricity, not only deplete the needs of humanity, we are forced to pay increased rates, masked as grid-upgrades etc
Excellent analysis. One issue… the graphic at the top is not correct (arrows point to the wrong places)
Thanks, Jeff. We updated the map.
A basic question for anyone who has expertise in this area: What are these data centers needed for?
Data about what?
To process that data faster and faster, for what?
I am skeptical of "technology" as it's roughly 50% helpful and 50% harmful.
This tech may offer some solutions down the line...
Atoco is a California startup built on metal-organic frameworks, a class of porous crystalline materials invented by Omar Yaghi, the UC Berkeley chemist who shared the 2025 Nobel Prize in Chemistry with Susumu Kitagawa and Richard Robson for developing them. These frameworks, called MOFs, are engineered at the atomic level so that a few grams hold an internal surface area the size of a sports arena, and they are tuned to trap only water molecules out of the surrounding air, even in extremely dry regions. Atoco packs this material into a unit the size of a 20-foot shipping container, where the MOFs silently pull H2O from the air and store it in their microscopic cavities before releasing it as water. Because the material grabs only H2O, the water comes out near distilled quality, free of PFAS, microplastics, and other contaminants. The company has tested the technology in Death Valley, one of the driest places on Earth, and confirmed it can still produce clean water there. Yaghi serves as Atoco's Chief Science Officer.
The output depends heavily on conditions and on how the unit is powered. On-grid units that run on electricity cycle more often and produce roughly 2,000 to 4,000 liters, about 1,057 gallons at the top end, of high-purity water per day, while off-grid units that run on ambient or waste heat produce up to about 1,000 liters a day, so the headline figure of 4,000 liters is a best case rather than a guarantee. As of spring 2026 the technology is preproduction, with a commercial prototype demonstrated for Bloomberg in an Orange County parking lot in April 2026 and volume production slated for later in the year, pending field tests. Atoco has not released pricing, though it estimates the water will cost a few cents per liter, which is more expensive than desalinated water, with the selling points being purity, decentralization, and resilience rather than low cost. Interest in the technology has risen partly because the Iran war exposed the vulnerability of large centralized desalination plants that became military targets, and Atoco is marketing the units to data centers, semiconductor plants, hospitals, farms, and any water-stressed region, though it would take roughly a dozen units to supply even a single water-efficient data center, and a competitor called AirJoule, a joint venture with GE Vernova, is pursuing a similar MOF-based harvester producing about 2,000 liters a day on a comparable timeline.
Atoco's official website is: https://atoco.com
Its press release on the atmospheric water harvesting prototype is at https://atoco.com/press/atoco-unveils-atmospheric-water-harvesting-on-grid-prototype/.
The most detailed independent reporting on the project I've come across is the Bloomberg feature from May 12, 2026, titled "Startup Atoco Unveils Device That Pulls Drinking Water From Air," at https://www.bloomberg.com/news/features/2026-05-12/startup-atoco-unveils-device-that-pulls-drinking-water-from-air (Bloomberg may be behind a paywall. The Mining.com writeup covers much of the same reporting if the Bloomberg link is blocked.)
Yes. I'm already familiar with the early work involved with this update. They even make a small plastic box contraption that yields an average of a litre and half per day for home use. The purity of the water is a great selling point for off grid uses. It's simple physics that works well in the desert. There is some evidence that it was a practical idea used in ancient Egypt using the same condensation principles.
I own three shares of GE Verona. A very innovative focused technology company spun off from the old General Electric Co.
This is one of the types of technology that I was referring to in my "demands". Another is using heat based capture and electricity regeneration systems developed for Formula One racing. Also there are kinetic based electricity regeneration modules that absorb vibrations and g force loads from movement in race cars that can be modified for large structures to take up energy releases from machinery and give back current.
...and there is nothing wrong with covering every square inch of the exterior of a data center with solar cells AND if they were built vertically the height could accommodate some wind generators on top, too.
I also think that maybe using AI to ask itself how to build a super energy efficient data center could go a very long way to solving some of the problems with developing energy savings in their design ing through the completion process.
Awhile back I did ask how many offshore wind generators it would take to power the numbers of data centers expected to come on line by 2035 based on their needed TerraWatt load. 37,000. There are currently less than 100 off the East Coast. Seems s bit high...
Denver, CO....a one year moratorium on data centers was passed recently but one got through, shame on city council for allowing this.
Were NDAs used?
If data centers have to be built the following notes should be considered before any breaking of ground takes place...
A. They must be built on brownfields. Brownfields are already spoiled from previous industrial usage and can be cleaned up during construction by the rich bastards who demand data centers for the increase of their wealth.
No breaking of any fresh ground any where. Especially in areas that are deemed arable land suitable for crops or have some amount of historical references to battlefield, neglected cemetery, designated as historic property, etc. Also near or are on protected wetlands, upland wetlands or forested lands currently suitable for recreational use. That alone stops eminent domain takings by developers of family farms or set asides previously designated for such purposes.
B. Data center construction must be VERTICAL to decrease the footprint. No more sprawl over large acreages. Going vertical has a lot of better outcomes for heat loading, and decreased water usage. It's an engineering challenge and not a giveaway to the damn development crowd. The footprint size is the reason. Sprawling is not to be allowed. Go up in stories or go home. Too, the minor cost of elevators is absorbed through better engineering. Expectations should be 40-80 story in range and going deeper into the ground wouldn't be out of the ordinary. This alone can force data centers to be closer to urban areas where tall structures are the norm. Fire protection costs go down as rural fire department resources would not be driven up for specialized apparatus.
C. Water usage must come from gray water sources or sea water. No more using up treated potable water necessary for human consumption. One more damn engineering challenge that must be met, not given away at the expense of municipal resources. No proffer amounts can alleviate the long term problems to local infrastructure costs over time. Recouping those additional costs over the expected life cycle of a data center is pure folly. Localities believing the developers projected costs to their taxpayer base, especially with the giving of incentives, are likely in the back pockets of the tech bros. Citizens must demand full accountability BEFORE and after by transparency of each entire project through out the process. Water is the most valuable resource on the planet.
D. Power consumption must be from renewable energy. No usage of dozens or hundreds of large displacement diesel engines. No usage of large scale gas turbine engines for powering them, either. A data center usage of power using more hydrocarbon fuel than a fleet of aircraft flying to Europe and back every day? That is purely stupid... So, build up to scale these demands for every single one of these data centers by all renewable energy... Do it without the extreme air pollution created by back up power generation on site.
Donald Trump will just have to give up his unfounded dreaded fear of "windmills" and realize that offshore wind generators needed to create power for his data center bros will need to be fast tracked and not ever delayed by mere whim. It's just an engineering challenge easily met by the vast riches of his under taxed data center bromancers.
E. Local and state support for data centers must include heavy oversight and complete inputs on public health and safety through strict compliance
to building and fire codes and established zoning requirements. Demand nothing less!
As the data center industry evolves under the above conditions the engineering challenges will be met and over time can be seen as a working partnership that may prove worthy. People have control if they exercise the better options to keep with health and safety in this endeavor.
also, see this? https://mitchthelawyer.substack.com/p/theyre-building-data-centers-across/comment/267755932
Thanks for this very informative article. We tend to hear about these economic, environmental and social impacts as one offs. Thank you for showing the cumulative impact by addressing them all in a single article.
The larger question is the fact that governments have done so little to curb the size, reach and power of these massive corporations. Most of the companies constructing these facilities have market cap valuations exceeding the GDP of many countries in the world. Over a century ago, the US government took action to break up Bell Telephone and US Steel. Given the financial and political power of these companies, would any similar attempt today have any chance of success?
Huge issues, big potential, large downsides. It's well above my pay grade and I don't have the answers. I just know we need to get all this figured out sooner rather than later.
The key ingredient in breaking up these malignant companies is "political will." And this year we have an opportunity to increase it! I read recently that about 70% of Americans don't want a data center anywhere near them (which is why they're getting dumped in poor rural areas).
I think far too many electeds are just thinking "tax base." We don't tax the rich, and we have to sacrifice the future for "tax base."
I don't think any establishment Democrat is going to "get" this, as they rely on big donors (or conversely: the absence of Big Money for an opponent).
One question I have is : Why, actually, do we need all these data centers? Is it anything more than the big tech companies trying to scrape more data to take more money and freedoms from us?
Thank you for your thorough coverage and your call to action.
The interview with Jim Acosta and Aaron Brockovich really caught my attention.
Your skillfully crafted writing is getting seen by everyone in my orbit!
💜
A comment about the power plants. Gas turbines, of course, burn fossil fuels, which contribute to climate change. But not all gas turbines are alike. Combined-cycle plants with modern pollution control are pretty clean, with zero particulates, no sulfur oxides, and pretty low nitrogen oxides. On the other hand, single-cycle peaker plants emit far more pollution of all kinds.
Wanna guess what the difference is?
$$$$$$$.
Combined-cycle plants are far more costly to buy, build, and operate.
Single-cycle rapid startup machines are cheap and quick.
I'll leave you to guess which species of gas turbine is preferred by the data center builders.
Also, of course, any fossil fuel generator is bad for climate.
So many issues, right? Did you see this possible partial solution? https://mitchthelawyer.substack.com/p/theyre-building-data-centers-across/comment/267755932
MOFs have been around for a long time and are pretty remarkable. The big question is how fast water can move in or out of the material — the mass transfer rate. I will have to look at what this startup is doing.
One of the first things that hits me: why are NDAs even allowed? For anything? It feels like a crime against humanity to me.
I think I know the reasons I may write about this in the future. It's a concept that was around long before these data centers and focuses more on community development and privacy during conversations than on pulling the wool over the consumer's eyes. But recently I think that's exactly what's been happening.
Many of the locational dot placements on the map are waaaaay off.
Noted and thank you. We updated the graphic.
Thanks for following up, Mitch.
As I noticed in several projection graphings on other web sites, too. It's like they really don't want anyone to know until they break ground.
Pennsylvania had clusters with data centers seemingly on top of one another in concentrations. Lazy map making? Or different data center developers fighting over scarce land?
I think it has more to do with avoiding public scrutiny and competition than anything else. Of course maximizing profits over people will be the end play for most large corporations.