Table of contents
ARTIFICIAL INTELLIGENCE & SOCIETY
PART 8
We all still have to eat.
Local officials and company boards still decide what gets built, what gets shut down, whose neighborhood gets the giant computer building, and whose kids grow up next to the noise and the power lines. A small set of owners still gets the leftover money when machines make more than people need to sell their hours just to survive.
That last part is the one that matters most here.
In What If Most of Us Don’t Have to Work Anymore?, I argued that a future with far less required work could expand human freedom, but only if society separates survival from a job and shares the gains from automation. This piece doesn’t reopen the question of whether most jobs disappear. It takes that idea seriously and asks a simpler follow-up.
If AI and robots do more of the work, who gets the money that used to go into paychecks?
The leftover money goes to whoever owns the tools
There’s a simple pattern here that people like to dress up in complicated language. When a company finds a way to produce more with fewer people, the savings don’t show up first as something everybody shares. They show up as profit for whoever owns the tools that made the savings possible.
That is not a conspiracy theory. It is how ownership usually works. The leftover money follows the owner before it follows the worker.
Picture a warehouse that needs fewer people because software and robots handle picking, packing, routing, and inventory. The savings hit the company books before they hit anybody’s kitchen table. Or picture a computer program that can draft contracts, sort medical paperwork, write code, or handle customer support. The company that owns that program and the computers under it captures more of the value those tasks used to pay in wages. In both cases the story starts as convenience and ends as a claim on the leftover money.
The International Monetary Fund has already warned that AI can widen the gap between rich and poor partly because money from owning the tools becomes more important, and that money tends to pile up among people who already own those tools.[1][2] Its budget work also argues that social protection and taxes can broaden the gains, while warning that clumsy taxes on automation create their own problems.[3]
None of that settles the end of the story. It does describe the path we’re on, and the open question is what we do with that path while there is still room to choose.
People can still invent new jobs. Companies can still hire. Wages can still rise in some places. History is full of technologies that killed tasks and created others. The ownership question doesn’t require pretending that process has already ended. It requires noticing what happens if the money-making tools keep ending up in fewer hands while the wage door gets narrower for more people.
A thin monthly check can keep a person alive as a customer. It doesn’t automatically give that person a share of the systems producing the extra wealth. I wouldn’t confuse those two outcomes.
What people actually own when they say they “own AI”
When people say “AI,” they often mean a chatbot on a phone screen.
What actually makes the money is wider than that. The leftover money follows the tools under the chat window, not the chat window itself.
The big AI programs and the software around them are not free-floating ideas. They are trained, hosted, updated, licensed, and wrapped inside companies. Industry already produces the large majority of notable AI systems, and the most capable ones are often the least transparent about how much computer power and data they used.[4]
Training and running advanced AI takes enormous computer power. Epoch AI estimates that five giant cloud companies (Amazon, Google, Meta, Microsoft, and Oracle) held about 71 percent of the world’s AI computer power by late 2025, up from 63 percent in early 2024.[5] Many leading AI labs still depend on those same firms for research and for answering user requests. Stanford’s 2026 AI Index reports that global AI computer power has grown roughly 3.3 times per year since 2022, with Nvidia chips accounting for more than 60 percent of the total.[4] That pile-up sits under almost every cutting-edge demo people argue about in public, whether they notice the foundation or not.
That doesn’t mean one company owns the future. It does mean the physical foundation of AI ability is already highly concentrated in a few hands.
AI programs don’t run on wishful thinking. They run on electricity, cooling, substations, power lines, and sites that can host that equipment. The International Energy Agency estimates that electricity use by computer warehouses grew about 17 percent in 2025, that AI-focused warehouses grew even faster, and that electricity use by those buildings could roughly double from 485 TWh in 2025 to about 950 TWh by 2030.[6] Spending by a handful of large technology companies has already moved into the hundreds of billions of dollars a year.[6]
If energy, land, water, and power-grid access become bottlenecks, whoever controls those bottlenecks shapes who can build, who can compete, and who gets priced out. There are neighborhoods already learning that lesson when a substation expansion or cooling plant shows up next door.
AI in the physical world becomes a different kind of economic power when it can move through warehouses, factories, farms, construction sites, and delivery routes. The International Federation of Robotics reported about 542,000 industrial robots installed worldwide in 2024, more than double the annual installations of a decade earlier, with Asia accounting for most new deployments.[7] Those machines are owned assets. So are the systems that connect them, the maintenance contracts, and the buildings built around them.
Ownership often arrives through stock, board seats, licenses, cloud contracts, and exclusive partnerships rather than through a single person holding a key to a robot closet. Concentration in chips and chip factories, plus strategic stakes and cloud deals between giant cloud companies and labs, already shape who can compete at the top.[8]
Put those pieces together and the picture gets clearer.
The danger is not only that jobs shrink. The danger is that a small set of owners can claim almost all the leftover money while everyone else lives on monthly checks calibrated just high enough to keep demand alive.
That future is not finished. Calling it inevitable rule by a few rich owners would oversell the evidence. Treating the pile-up as imaginary would undersell how much of the money-making tools already sit in a few hands.
A check is not the same thing as ownership
A monthly survival check, wage insurance, and free or cheap public services all matter in a world where wages stop being the main bridge to survival. Part 7 already opened that door.[9]
But a check is not a claim on the machines making the money. Why treat a monthly payment as the whole answer when the machines (and the energy under them) are where the extra wealth actually sits?
Think about the difference in ordinary language. If you receive money because somebody else owns the robots and decides you should remain a customer, you are dependent on that political and corporate bargain continuing. If you own a share of the money-making asset, or if the public owns a lasting stake that pays people a yearly check, your claim is attached to the extra wealth itself.
Those are not identical forms of security. What does freedom mean if you eat only by permission of whoever owns the fleet?
Alaska’s Permanent Fund is not an AI policy. It is still one of the clearest living examples of a public claiming a share of resource wealth and paying residents a yearly check. The fund was built so oil wealth would benefit Alaskans beyond the extraction boom, and residents have received annual payments for decades.[10] Researchers and policy advocates now argue for similar permanent funds, shared-data payments, or public stakes tied to AI tools and shared resources.[11][12] The comparison is imperfect, and it is still useful because it shows a society treating a money-making resource as something ordinary people can claim a share of.
I don’t think any one of those proposals is settled doctrine. Some versions sound like national investment funds. Some sound like broader employee ownership. Some sound like public options for computer power or energy. Some sound like cooperatives or city ownership of local equipment. Some keep markets for scarce goods while guaranteeing broader access to essentials.
The point is not to crown a favorite acronym. The point is that preparation can’t stop at consolation payments after ownership has already piled up in a few hands. If machines do more of the work, the real question becomes who owns the machines, the energy that powers them, the land they sit on, the data they trained on, and the companies that control the tools underneath.
When survival becomes permission
Part 7 asked whether a post-work future could expand freedom. This piece asks what happens if we answer that question with monthly payments alone.
If most required work fades and society replaces wages with checks calibrated just high enough to keep people alive as customers, the bargain shifts. You do not only lose income. You depend on whoever decides whether the payment continues, what it buys, and what you have to accept to keep it.
That dependency can run through government programs or through private companies that own the machines and negotiate the political deal. The IMF’s budget work already treats social protection and taxes as tools to broaden AI gains, which is useful. It also shows how much of the fight will land in public budgets rather than in private paychecks.[3] Neither version is freedom in the sense Part 7 described. Both can leave residents hosting the power load while a small ownership class captures the leftover money.
The technology does not force that outcome. It opens a choice. We can drift toward a country, and maybe a world, of dependents. Or we can push for broader ownership and real voice while the tools are still being built.
Your neighborhood is already on the map
Computer warehouses and power upgrades are not abstract future problems. They are local politics now.
Good Jobs First documents that many data center projects receive large property tax breaks, often with limited public disclosure of who got what.[13] Research from the Federation of American Scientists found that in states reporting subsidy costs, data center programs can cost hundreds of millions of dollars per year, while fully built sites employ relatively few permanent workers per dollar of public subsidy.[15] Columbia Law School’s climate program describes Lancaster, Pennsylvania, negotiating a binding community benefits agreement after officials and residents weighed noise, energy, diesel backup generation, and local costs against the project’s gains.[14]
Residents should have a say when companies seek tax breaks to build in their towns. The host community should benefit from the technology advance, not only absorb the noise, water use, and grid strain. Community benefit agreements can require local hiring, clean energy commitments, payments to schools or sustainability funds, and relief when utility bills rise.[13][14][16] Lancaster’s agreement includes $20 million in direct community payments, local hiring plans, and clean energy benchmarks tied to enforcement money held by the city.[14][16] Those deals are imperfect. They are still a concrete way to claim a share before ownership hardens.
A few companies owning most of it is a path, not a finished verdict
I want to be careful here, because future essays love to sound finished. Who benefits from pretending the ownership question is already closed?
We don’t know how open-source AI, national industrial policy, rules that limit how big one company can get, energy constraints, chip supply shocks, or new competitors will reshape who owns those tools. We don’t know whether robot fleets stay concentrated in a few hands or become cheaper and more widely held. We don’t know whether societies choose public stakes early or wait until the political fight is uglier.
What we can see now is the direction of travel.
Ownership of AI computer power is already skewed toward a small number of giant cloud companies.[5] Notable AI development is dominated by industry.[4] Electricity demand from computer warehouses is rising fast enough to become a local politics problem, not only a tech story.[6] Industrial robot installations have scaled into the hundreds of thousands per year.[7] Budget offices are already debating how to tax and share AI gains because money from owning the tools can outrun money from wages.[1][2][3]
If those trends continue without ownership rules, the likely default is familiar: private leftover claim first, public repair later. This means thinner monthly checks, more dependency, and less bargaining power for ordinary people when the repair finally arrives.
That is why I keep saying preparation.
Preparation is not predicting the exact year when a job category vanishes. Preparation is deciding, while there is still room to decide, whether the tools of an automated economy will be broadly held, publicly staked, cooperatively governed, heavily taxed and redistributed, or left almost entirely to whoever got there first with money and computer power.
What shared ownership can look like
I’m not going to pretend I have the blueprint. I don’t.
I do think the menu is more interesting than “a monthly check versus nothing.” For instance, ownership rules can attach people to money from owning things, not only to money from a paycheck.
Employee ownership, citizen shares, and funds that hold stock in AI and robotics companies can attach ordinary people to ownership income rather than only to wage income. Joseph Blasi and others have argued for state and national permanent funds that invest in AI tools and pay out that ownership income over time.[11] For example, a public fund that holds a stake in AI computer buildings and energy can pay residents a yearly check even when wages no longer carry most households.
If AI depends on publicly created knowledge, the data we leave online, airwaves, water, land, and room on the power grid, societies can charge for those shared uses and recycle the proceeds into citizen payments or public services. Alaska’s oil model is the nearest large American precedent, not a perfect copy-paste.[10][12]
Computer power, energy, and internet access can include public or nonprofit options so cutting-edge access is not only a private club. That doesn’t require the government owning every lab. It does require refusing the assumption that the only legitimate owners are the current giant cloud companies.
Local robot fleets, care robots, delivery systems, and energy projects can be owned by the communities that host them. Automation doesn’t have to arrive only as a distant corporate lease. Imagine a town owning part of the delivery fleet that moves through it, rather than only hosting the warehouse noise.
Even when ownership stays private, taxing extraordinary AI and automation profits can fund services and a guaranteed floor of income. The IMF’s own work treats that as a live budget problem rather than a slogan.[3]
Every option has tradeoffs. Public ownership can become bureaucratic capture. Private ownership can become rule by a few. Cooperatives can fail to scale. Taxes can slow useful investment if they are clumsy. Yearly checks can become thin if the underlying claim is weak.
So the question is not which pamphlet wins. The question is whether we design ownership before leftover money hardens into a political fact that only a few people can revise. That is the distinction I care about most.
Who decides is part of who owns
Ownership is not only a money story. It is a power story.
The company that owns the AI program can decide acceptable use, pricing, where it is available, and which customers matter. The company that owns the cloud can decide whose startup gets computer power when capacity is scarce. The owners of land and energy can decide which communities absorb the cost of the build-out. The owners of robot fleets can decide which plants stay open and which regions keep production. Who decides which towns host the power load, and who only gets the bill?
If most people only meet that system as users and check recipients, they don’t merely have less money. They have less say. There is no clean separation between who owns the tools and who gets a voice once those tools sit under everyday life.
That is why who gets a voice belongs next to who gets the money. Competition rules, public stakes, rules that force companies to show what they own, local rules about where computer warehouses go, worker voice, and citizen payments are different tools aimed at the same problem: a money-making system that can feed people while excluding them from control.[1][3][13]
When a town approves tax breaks for a new computer warehouse without a public benefits package, it is voting on dependency in advance. The company gets a claim on leftover money and grid access. Residents get the hosting costs and, later, a political fight over whether anyone else should share in the upside.[13][15]
I don’t want a future where machines work and a small ownership class writes the rules for everyone else’s free time. I also don’t want a future where fear of that outcome makes us refuse useful automation and invent busywork just to preserve the old wage ritual.
There is a third path, and it starts earlier than most people want to admit. Perhaps that is the hardest part: acting before the harm is finished enough to feel undeniable.
It starts with asking, while those tools are still being built, who eats, who decides, and who owns the leftover money.
What we can do
If Part 7 was about whether a post-work future can be free, Part 8 is about whether freedom can survive when ownership piles up in a few hands. Cash helps. Services help. Meaning and community still matter for the hours that used to be filled by required work. None of that replaces a claim on the tools that make the money.
You do not need an economics degree to participate in this design fight.
Before the vote
When a computer warehouse, substation expansion, or power upgrade is proposed near you, request the zoning packet and attend the public hearing. Ask whether tax incentives are publicly disclosed and who negotiated them. FAS research notes that many states do not disclose which companies receive data center subsidies, and companies sometimes hide behind non-disclosure agreements on energy and water details.[15]
Before approvals finalize, push for a community benefits agreement. Good Jobs First and Columbia Law School’s climate program describe agreements that can require local hiring, noise and water protections, clean power commitments, and direct payments to schools, broadband, or ratepayer relief when utility costs rise.[13][14]
Contact city council or county commissioners and ask for full public disclosure of every tax break, payment in lieu of taxes, and infrastructure subsidy tied to the project.[13][15]
File open-records requests for emails, draft agreements, and meeting materials while the deal is still moving. Guadalupe County, Texas, residents sued in 2025 alleging insufficient public notice, private negotiations, and conflicts of interest around roughly $150 million in proposed county tax abatements for two large AI computer warehouse projects.[17]
Support state or federal proposals for public stakes, worker ownership, or permanent funds that pay residents from shared resources, not only from monthly survival checks.[11][12]
Track who owns the big cloud and energy bottlenecks in your region. Epoch AI’s compute concentration data and IEA electricity analysis are not only national stories. They show which towns host the build-out and which towns only get the bill.[5][6]
When the room is full but leadership still says yes
This is the part people skip in polite policy essays. You organize. You pack the hearing. Officials still approve the tax break, the zoning change, or the development agreement anyway.
That does not mean you lost forever. It means the fight moves to different levers, and the clock gets shorter.
Document what happened. Record the vote, who moved it, what notice was given, and whether required hearings actually happened in public. Lawsuits over data center approvals often target open-meeting violations, weak environmental review, and tax abatements approved without the process state law requires.[17][18][19] You are building a record, not venting on social media.
Appeal inside the local process on deadline. Many zoning and land-use decisions can be appealed to a planning commission, board of adjustment, or state agency within a fixed number of days. Miss the window and a bad approval can become much harder to unwind. Check your county clerk’s office or a local land-use attorney for the appeal form and the calendar.
Ask a court to pause the project. Opponents have sought injunctions and temporary restraining orders to stop construction or site work while a case moves. In Pine Island, Minnesota, a judge paused a data center project after finding that allowing it to proceed before environmental review finished could cause irreparable harm.[18] Courts do not always side with residents. Taylor County, Texas, dismissed a rural data center challenge after a judge ruled the plaintiffs lacked standing because their properties would not be directly affected; residents appealed.[17] Standing and timing matter. So does hiring counsel who knows local land-use law.
Sue over process, not vibes. Bloomberg Law reported a growing pattern of communities suing local officials over approval steps: inadequate notice, closed sessions used for decisions that should have been public, zoning changes slipped through without required review.[19] Forsyth County, North Carolina, walked back a zoning change that added computer warehouses as a permitted use after residents and the Southern Environmental Law Center sued over notice and zoning procedure.[19] Process violations are concrete claims. “We don’t want it” is not always enough on its own.
Push ballot measures and moratoriums where your state allows them. Several states and local governments have paused new data center tax breaks or project consideration after backlash.[19] Rules vary sharply by state. Some ballot paths get blocked in court. Others have slowed or stopped projects. New York imposed a statewide pause on new hyperscale computer warehouse approvals in 2025.[19]
Use recall and regular elections. Cascade Locks, Oregon, voters recalled two port commissioners who backed a proposed $100 million computer warehouse and defeated two other supporters in the next election; the port and developer then dropped the project.[20] Recall is not available everywhere. Where it exists, the petition rules, signature thresholds, and timing are strict. Even when recall fails, the next regular election still removes mayors, county commissioners, port directors, and council members who treated public opposition as background noise.
Escalate to state regulators when the fight is really about the grid. If the dispute is power lines, substations, or who pays for grid upgrades, state public utility commissions and energy offices may have separate review paths from the city council vote. Texas Governor Greg Abbott ordered regulators in 2025 to audit new projects seeking grid access amid statewide backlash.[19] Grid access can be a bottleneck even when local zoning already passed.
Keep negotiating for a community benefits agreement after approval. A bad vote does not always kill a benefits package. Lancaster, Pennsylvania, secured a binding agreement with direct payments and clean energy benchmarks even as officials weighed local costs.[14][16] Late leverage is weaker than early leverage. It is still leverage.
None of this is a guarantee. Developers sue back when towns pause projects.[19] Courts sometimes dismiss challenges.[17] Officials sometimes treat opposition as a communications problem rather than a legitimacy problem. The point is that “they approved it anyway” is not the end of the menu.
The longer ownership fight
Vote and organize on hosting decisions before non-disclosure agreements lock residents out of the conversation.[15]
Refuse the false choice between banning AI and accepting rule by a few with a thin monthly check. Pilot yearly checks, public stakes, and broader ownership while the politics are still negotiable.
Data Center Watch tracked hundreds of opposition cases nationwide and billions of dollars in blocked or delayed projects amid local pushback, including recalls, lawsuits, and renegotiated deals.[21] The pattern is not theoretical. Communities are already testing which levers actually move officials who approved tax breaks residents never wanted.
The work is quieter than a demo video. It is also the work that decides whether abundance becomes shared freedom or thin dependence.
I don’t know which mix a decent society will choose. I’m pretty sure the mix matters more than the slogan. In the end, abundance without shared ownership is just a prettier form of dependence.
Machines can create abundance without creating freedom. Freedom requires that ordinary people share in what the machines produce, and in the decisions about how that production is governed.
Who gets rich when AI does the work should not be a trivia question about stock prices. It should be a design question about what kind of society we want.
Frequently Asked Questions
Is this arguing that AI has already ended most jobs?
No. This article supports the longer-term ownership question opened in Part 7. Current evidence still shows substantial AI exposure and transformation rather than a settled end of work. The ownership problem matters because leftover money can pile up in a few hands even while many jobs remain.[1][2][9]
Why isn’t a monthly survival check enough on its own?
A guaranteed floor of income can protect survival when wages weaken. It doesn’t automatically create a stake in AI programs, computers, energy, land, or robot fleets. Without ownership or lasting public claims on the leftover money, monthly checks can leave people dependent on whoever controls the money-making tools.
Who owns AI computer power today?
Ownership is concentrated. Epoch AI estimates that five large cloud companies held about 71 percent of all AI computer power worldwide by late 2025. Many leading AI labs still rely on those providers for training and for answering user requests.[5]
What does energy have to do with AI ownership?
Advanced AI depends on electricity, cooling, land, and room on the power grid. IEA analysis shows rapidly rising electricity demand from computer warehouses and very large spending by a small set of technology companies. Energy limits and local building fights shape who can build and who absorbs local costs.[6]
Are AI permanent funds or shared-data payments a real policy idea?
They are active proposals, not settled law. Alaska’s Permanent Fund shows that a public can claim resource wealth and pay residents yearly checks. Researchers have proposed similar funds or fees on shared resources tied to AI tools and the data we leave online. Designs differ, and none is automatic.[10][11][12]
Does a few companies owning most of it mean rule by a few is inevitable?
No. Concentration is a present path with an uncertain ending. Free open AI programs, public options, competition rules, energy limits, broader stock ownership, and political choices can all change the path. The argument here is to design ownership on purpose rather than assume markets will share the leftover money on their own.
Can residents do anything when a data center seeks tax breaks nearby?
Yes. Public hearings, disclosure demands, and community benefit agreements are live tools. Lancaster’s agreement shows a city securing direct payments, local hiring plans, and clean energy benchmarks before build-out hardens.[14][16] Good Jobs First argues residents should push for benefits when a project cannot be stopped, and should resist tax breaks without public return when they can.[13]
What if officials approve a data center or tax break despite public opposition?
The fight often moves to appeals, lawsuits, ballot measures, recalls, and the next election. Minnesota courts paused at least one project after finding environmental review was insufficient.[18] Texas residents have sued over alleged open-meeting and tax-code violations around abatements.[17] Cascade Locks, Oregon, voters recalled port commissioners and the developer later dropped its project.[20] Outcomes vary by state law, standing rules, and timing. None of it is automatic.[17][19][21]
References
[1] International Monetary Fund. “AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity.” January 14, 2024. https://www.imf.org/en/blogs/articles/2024/01/14/ai-will-transform-the-global-economy-lets-make-sure-it-benefits-humanity
[2] International Monetary Fund. “Gen-AI: Artificial Intelligence and the Future of Work.” January 14, 2024. https://www.imf.org/en/publications/staff-discussion-notes/issues/2024/01/14/gen-ai-artificial-intelligence-and-the-future-of-work-542379
[3] International Monetary Fund. “Broadening the Gains from Generative AI: The Role of Fiscal Policies.” June 11, 2024. https://www.imf.org/en/publications/staff-discussion-notes/issues/2024/06/11/broadening-the-gains-from-generative-ai-the-role-of-fiscal-policies-549639
[4] Stanford Institute for Human-Centered Artificial Intelligence. “The 2026 AI Index Report - Research and Development.” 2026. https://hai.stanford.edu/ai-index/2026-ai-index-report
[5] Epoch AI. “Five hyperscalers now own over two-thirds of global AI compute.” Data Insight. https://epoch.ai/data-insights/hyperscalers-control-most-compute
[6] International Energy Agency. “Key Questions on Energy and AI” (executive summary and related coverage). April 2026. https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary
[7] International Federation of Robotics. “World Robotics 2025 report - INDUSTRIAL ROBOTS.” September 25, 2025. https://ifr.org/ifr-press-releases/news/global-robot-demand-in-factories-doubles-over-10-years
[8] Vipra, Jai, and Anton Korinek. “On Labs and Fabs: Mapping How Alliances, Acquisitions, and Antitrust are Shaping the Frontier AI Industry.” arXiv:2406.01722. June 2024. https://arxiv.org/abs/2406.01722
[9] Andrew Drasen. “What If Most of Us Don’t Have to Work Anymore?” A Vision of Hope Insights. August 18, 2026. https://www.avisionofhopebook.com/insights/artificial-intelligence-society/what-if-most-of-us-dont-have-to-work-anymore
[10] Alaska Permanent Fund Corporation. “About the Alaska Permanent Fund.” https://apfc.org/
[11] ImpactAlpha. “Joseph Blasi: Give workers a stake in AI’s upside through state and federal ‘permanent funds’.” https://impactalpha.com/joseph-blasi-give-workers-a-stake-in-ais-upside-through-state-and-federal-permanent-funds-qa/
[12] Bloomberg Tax. “Alaska Is a Model for Taxing AI as an Extracted Natural Resource.” https://news.bloombergtax.com/tax-insights-and-commentary/alaska-is-a-model-for-taxing-ai-as-an-extracted-natural-resource
[13] Good Jobs First. “Community Benefit Agreements with Data Centers Can Help Mitigate Harms.” https://goodjobsfirst.org/community-benefit-agreements-with-data-centers-can-help-mitigate-harms/
[14] Columbia Law School Sabin Center for Climate Change Law. “Community Benefits Agreements and Data Center Development.” Climate Law Blog, May 28, 2026. https://blogs.law.columbia.edu/climatechange/2026/05/28/community-benefits-agreements-and-data-center-development/
[15] Federation of American Scientists. “Fair AI-Fueled Data Center Development for Communities.” https://fas.org/publication/community-benefit-agreements-data-center-development/
[16] City of Lancaster, Pennsylvania. “Community Benefits Agreement Summary” (Lancaster AI Hub). https://www.cityoflancasterpa.gov/wp-content/uploads/2025/11/Community-Benefits-Agreement-Summary-1-1.pdf
[17] Law.com / Texas Lawyer. “Lawsuit Challenges Central Texas County’s Tax Breaks for AI Data Centers.” July 22, 2026. https://www.law.com/texaslawyer/2026/07/22/lawsuit-challenges-central-texas-countys-tax-breaks-for-ai-data-centers/
[18] MinnPost. “In pausing data centers, MN courts did what local officials wouldn’t.” July 2026. https://www.minnpost.com/greater-minnesota/2026/07/in-pausing-data-centers-minnesota-courts-did-what-local-officials-wouldnt/
[19] Bloomberg Law. “Data Center Foes’ Newest Tack: Sue Locals Over Approval Process.” 2025–2026. https://news.bloomberglaw.com/litigation/data-center-foes-newest-tack-sue-locals-over-approval-process-100
[20] The Oregonian / OregonLive. “Cascade Locks rejects plan for $100 million data center in the Columbia River Gorge.” July 20, 2023. https://www.oregonlive.com/silicon-forest/2023/07/cascade-locks-rejects-plan-for-100-million-data-center-in-the-columbia-river-gorge.html
[21] Data Center Watch. “$64 billion of data center projects have been blocked or delayed amid local opposition.” https://www.datacenterwatch.org/report