Data Centers: Benefits, Costs, and the Special Case of AI

Paul Kruger, Sept 2026

Introduction: The Invisible Infrastructure Becomes Visible

Data centers have become one of the most important—and increasingly controversial—pieces of modern infrastructure. Every email, video stream, online purchase, cloud storage access, and AI query depends on computers housed in data centers somewhere in the world.

My intent is neither to advocate nor argue against these facilities. It is to inform the reader so they can form an educated opinion divorced from political and financial rhetoric.

For decades, data centers remained largely invisible to the public. That has changed rapidly. The explosive growth of artificial intelligence is creating demand for enormous new facilities packed with increasingly powerful computers. Communities are now being asked to approve massive data-center projects requiring huge amounts of electricity, substantial infrastructure investment, and—depending on their cooling technology and location—significant quantities of water.

The question is not simply whether data centers are good or bad. The more useful question is: What benefits does a particular data center bring to a community, and are those benefits sufficient to justify its costs?

Why Data Centers Matter

Modern economies depend on always-available digital infrastructure. Data centers make it possible for businesses to store and process data remotely, for hospitals to access digital records, for researchers to analyze large data sets, and for consumers to use online services at global scale.

They can also improve computing efficiency compared with thousands of smaller, poorly utilized server rooms. Large operators can consolidate workloads, invest in advanced cooling and power-management systems, and run equipment closer to full capacity. Where they are supplied with low-carbon electricity and designed efficiently, centralized facilities can reduce the energy intensity of digital services.

General Advantages

• Digital reliability and resilience. Data centers support redundant networks, backup power, cybersecurity controls, disaster recovery, and secure storage for critical services. This can reduce downtime for businesses, public agencies, healthcare providers, and other institutions.

• Cloud access for smaller organizations. Instead of purchasing and maintaining their own servers, small businesses, schools, nonprofits, and startups can rent computing and storage as needed. This lowers upfront technology costs and makes sophisticated software more accessible.

• Economic activity during construction. Large projects can support substantial, often well-paid construction work for electricians, technicians, heavy-equipment operators, and other trades. Data centers also purchase local services, materials, and maintenance support.

• Tax-base potential. A facility's buildings and equipment can produce government revenue, especially where local tax policy does not exempt major portions of the investment.

• Innovation infrastructure. Data centers enable research, high-performance computing, software development, telemedicine, digital media, industrial automation, and other data-intensive activities. They are a foundational utility for a digitally connected economy.

General Drawbacks

• Very high electricity demand. Data centers operate continuously and require power not only for servers but also for cooling, networking, lighting, and backup systems. The International Energy Agency estimates that data centers used about 1.5% of global electricity in 2024, or roughly 415 terawatt-hours, with consumption projected to reach about 950 terawatt-hours by 2030.

• Grid strain and potential rate impacts. New facilities can require new substations, transmission lines, generation capacity, and distribution-system upgrades. If utilities recover those costs broadly rather than directly from the data-center customer, households and small businesses may face higher bills.

• Water use. Many facilities use water-based cooling, while electricity generation can consume additional water off-site. Water impacts are especially contentious in drought-prone areas or places with limited water-treatment capacity. Large facilities can use millions of gallons a day in some operating conditions.

• Emissions and air pollution. A data center running on a fossil-heavy grid indirectly increases greenhouse-gas emissions. On-site diesel backup generators can also contribute local air pollution during testing and outages.

• Land-use and community impacts. Large campuses can change a community's character through land clearing, traffic, transmission corridors, industrial-scale buildings, continual equipment noise, and generator testing.

• Limited permanent employment. Construction employment can be significant, but it is temporary. Operating a highly automated data center generally requires a much smaller ongoing workforce than a manufacturing plant or large office campus of comparable capital cost.

• Weak public returns when incentives are excessive. Tax breaks, discounted electricity, subsidized infrastructure, and expedited permitting can reduce the public value of a project.

What Changes With AI Data Centers

AI data centers are not merely conventional data centers with a different label. Training and serving advanced AI models often relies on dense clusters of graphics processing units and other accelerators. These chips consume substantial power and produce much more concentrated heat than many traditional enterprise computing workloads.

As a result, AI-oriented facilities may need higher-capacity electrical connections, more aggressive cooling, denser server racks, larger backup systems, and faster network interconnections. Some use liquid cooling rather than—or in addition to—conventional air cooling. These design changes can improve performance and sometimes efficiency, but they also make the facilities more resource-intensive in absolute terms.

A conventional data center may use electricity comparable to tens of thousands of homes, while newer AI-focused hyperscale facilities can be far larger. A single modern AI data center may use as much power as 100,000 homes. The International Energy Agency reports that electricity consumption by AI-focused data centers increased by approximately 50% during 2025, substantially faster than overall data-center electricity consumption, and projects that electricity use by AI-focused facilities will triple between 2025 and 2030.

Benefits of AI Data Centers

• Faster scientific and medical research. AI computing can support protein modeling, drug discovery, climate simulation, materials research, medical-image analysis, and other computationally intensive work.

• Productivity and new services. AI infrastructure may help companies automate repetitive tasks, improve customer service, analyze complex data, strengthen fraud detection, translate languages, and build more capable software tools.

• Domestic capacity and technological competitiveness. Nations and regions may view advanced computing infrastructure as strategically important. Local AI capacity can reduce reliance on foreign computing resources and support domestic research, startups, and technology sectors.

• Potential for grid modernization. Large, creditworthy data-center buyers can justify investments in new generation, transmission, storage, and grid modernization. This is a public benefit only if the projects add genuinely cleaner and reliable capacity.

• Opportunity to accelerate clean power procurement. Major technology firms can use long-term power-purchase agreements to finance renewable generation. However, claims of "100% renewable" operation should be evaluated carefully: annual renewable-energy certificates are not necessarily the same as having clean electricity available at the same location and hour the facility is operating.

The Electricity Problem

The largest issue surrounding modern data centers is electricity. These global percentages can be misleading, however. Data centers are geographically concentrated. A facility may represent only a tiny fraction of worldwide electricity consumption while becoming a major customer of a particular local utility.

U.S. data centers consumed roughly 180 terawatt-hours in 2024 and accounted for approximately half of the increase in U.S. electricity consumption in 2025. That creates an important question for communities: Who pays for the additional electricity infrastructure?

A new facility may require transmission lines, substations, generation capacity, and other infrastructure. If those costs are spread across all utility customers, existing residents and businesses could effectively subsidize a private development. A reasonable policy might require the new customer to bear an appropriate share of the incremental infrastructure costs it causes.

Water: An Important but Complicated Issue

Cooling is another major consideration. Computers produce heat, and that heat has to be removed. Some data centers use air cooling, while others use evaporative or liquid-based cooling systems. Water consumption can therefore vary dramatically from one facility to another.

Some newer systems use closed-loop cooling, which can substantially reduce ongoing water consumption. But that does not eliminate the local water question. A facility in a water-rich region may have a very different impact from an identical facility in an area experiencing drought or competing demands on its water supply.

Consequently, communities should ask not simply "How much water does the data center use?" but rather "How much water will this facility consume here, under normal and drought conditions, and who else competes for that water?"

Land, Noise, and the Industrial Footprint

Data centers can also have significant physical impacts. Large facilities can occupy hundreds of acres when the buildings, electrical equipment, substations, cooling systems, security areas, roads and future expansion are considered.

Backup generators and other mechanical equipment can produce noise. Large electrical infrastructure can alter the visual character of an area. There can also be opportunity costs: land devoted to a data center cannot simultaneously be used for housing, agriculture, commercial development, conservation or other purposes.

The Emerging Regulatory Landscape: Limits, Bans, and Moratoriums

As communities grapple with data center impacts, regulatory pressure on the industry has intensified dramatically. What was once a quiet infrastructure question has become a significant political issue across the country, with actions spanning from local restrictions to statewide moratoriums and federal advocacy.

The Scale of Regulatory Action

In 2025, more than 200 bills were introduced across all 50 states aimed at regulating data centers, with over 40 bills enacted into law. That pace has accelerated in 2026. By mid-2026, over 500 local jurisdictions—municipalities and counties—had enacted construction moratoriums or restrictions on new data center development. More than 150 local restrictions on data centers have been successfully implemented nationwide.

These restrictions are not confined to one region or one political ideology. Bipartisan action has emerged in Republican-led and Democratic-led states, rural and urban areas, and communities nationwide. A March 2026 Gallup poll found that seven in ten Americans would oppose the nearby construction of data centers for artificial intelligence—higher than the 53% of respondents who said they would oppose living near a nuclear power plant.

State and Local Moratoriums

Maine is poised to become the first state to implement a comprehensive data center construction moratorium, pausing new projects until November 2027. Several other states are moving toward statewide restrictions.

In September 2026, governors in three states announced concrete action: New Hampshire's governor confirmed she would seek a multi-year data center moratorium; Connecticut's governor declared the state "will not host massive AI data centers"; and Oregon's governor ordered a pause on data center development on state-owned land through July 2027.

New York Governor Kathy Hochul signed an executive order pausing approval for all data centers consuming 50 megawatts of energy or more. The state legislature passed the Responsible Data Center Development Act, which would place a one-year moratorium on permits for "large data centers," defined as facilities with peak demand of 20 megawatts or more.

At the local level, the acceleration has been even more dramatic. As of July 2026, the number of local data center moratoriums had jumped from 300 in late June to over 500 by August. Denver introduced a temporary pause until May 2027. Nashville implemented a permit moratorium and regulations banning facilities over 500,000 square feet. Other communities have enacted permanent bans.

Regulatory Focus Areas

State and local regulation of data centers is rapidly expanding across multiple policy domains:

• Energy procurement and utility rates. Higher or specialized utility rates for large-load data centers ensure those customers pay the full cost to serve them, minimizing cost shifting to existing customers. More than 65 special tariffs have been proposed or approved across more than 30 states.

• Water consumption. Mandatory water-use reporting and cost-recovery requirements. Minnesota established a separate water permitting requirement for data centers in 2025, with other states following suit.

• Renewable energy requirements. State legislation increasingly requires or incentivizes facilities to source power from new clean generation rather than existing grid supplies.

• Zoning and siting restrictions. Communities are restricting or banning data center construction in certain areas, limiting facility size, or requiring comprehensive environmental impact assessments.

• Employment and labor standards. Some states are imposing requirements for job creation, wages, and local hiring tied to data center projects.

• Transparency and disclosure. The Data Center Transparency Act and similar measures prohibit nondisclosure agreements regarding electricity and water usage, requiring developers to publish operational data.

Federal Tensions and the Ratepayer Protection Pledge

Federal policy has moved in two directions. In July 2025, an executive order aimed to rapidly build out data center infrastructure by easing federal regulatory burdens for facilities requiring more than 100 MW of new electricity load.

In response to state and local pressure, however, major data center developers and technology companies signed the Ratepayer Protection Pledge in March 2026, committing to cover the full cost of new electric generation resources needed to meet their energy demands. This pledge represents an acknowledgment of a core concern driving regulation: the risk that data center projects would shift infrastructure costs to existing ratepayers.

At the federal legislative level, Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez introduced a bill to pause data center construction nationwide, signaling that data center regulation has become a significant political issue.

Why Communities Are Acting

Community opposition cuts across party lines and geography, driven by several common concerns:

• Electricity rates. Residents worry that their utility bills will rise if utilities pass infrastructure costs to all customers.

• Water stress. In water-constrained regions, large water-consuming facilities raise anxiety about competition with agriculture, drinking water supplies, and ecosystem needs.

• Environmental and quality-of-life impacts. Noise, land-use changes, industrial development character, and the pace of change motivate local pushback.

• Limited local benefits. Communities question whether promised job creation and tax revenue justify the public costs and risks, especially when companies receive tax breaks and subsidies.

• Speed of deployment. The rapid scale of hyperscale AI data center development outpaces community planning and regulatory adaptation, leaving communities feeling unable to prepare or negotiate favorable terms.

A Better Framework for Community Decision-Making

Communities do not need to treat data centers as either automatically beneficial or automatically harmful. A responsible approval process should require clear, enforceable conditions.

Key Requirements for Communities

• Make resource use public. Developers should disclose projected and actual electricity demand, peak load, water withdrawals, water source, water-consumption method, emissions, generator use, and anticipated infrastructure needs.

• Protect existing utility customers. Regulators can require large facilities to pay for grid upgrades they cause, commit to minimum usage levels, and use tariffs that prevent costs from shifting to households and small businesses.

• Prioritize clean, additional power. Developers should pursue new low-carbon generation and storage that genuinely adds supply, rather than relying only on paper claims or existing clean power already serving the grid.

• Use water responsibly. Projects should favor cooling designs appropriate to the local climate, reuse water where safe and feasible, avoid potable water when alternatives exist, and face stricter limits in water-stressed regions.

• Limit local pollution and noise. Permits can set enforceable noise standards, restrict generator testing hours, require emissions controls, and require monitoring that is available to the public.

• Evaluate the net economic benefit. Governments should compare expected tax revenue with foregone taxes, utility upgrades, land-use costs, public-service demands, and the relatively limited number of permanent jobs.

• Include communities early. Residents should have access to clear impact studies before zoning and incentive decisions are finalized, rather than after infrastructure commitments make a project difficult to change.

• Plan for accountability. Agreements should include reporting requirements, performance standards, clawbacks if promised investments do not materialize, and mechanisms for revisiting permits if actual water or power use exceeds projections.

Bottom Line

Data centers provide essential digital infrastructure and can support innovation, reliable online services, construction activity, and public revenue. AI data centers may also enable important advances in science, health, productivity, and computing capability.

Yet their benefits are not automatic, and their costs can be substantial. AI-focused facilities make the tradeoffs sharper because they can consume exceptionally large amounts of electricity, cooling capacity, water, land, and grid infrastructure while creating relatively few permanent jobs compared with their size and public incentives.

The regulatory wave sweeping across communities reflects a legitimate demand for transparency, accountability, and balance. Communities should engage seriously with data center proposals but need not accept them on terms favorable primarily to developers. The most productive debate focuses not on whether data centers are good or bad, but on what conditions, safeguards, and community benefits are necessary for a particular project to proceed.

For policymakers and community leaders, the challenge is developing clear frameworks that permit beneficial projects while protecting existing residents and infrastructure. A balanced approach—one that insists on transparency, requires developers to bear appropriate infrastructure costs, protects existing ratepayers, and evaluates net community benefit—serves communities far better than blanket approval or blanket rejection.

 

Note: this author is a heavy user of AI technology and I have found it to be of great value not just for my code writing and web site creation, but as a research tool and excellent editor of my writing. Were it not for AI, I would find it impossible to produce the valumn and quality of work that I believe I provide. I add this footnote not to defend any negative uses of AI but to illustrate the fact that in and of itself, AI is neither good nor bad. It is a tool to be used and the user determines the outcome, not the AI itself.

It's not refuteable. We have moved into an age of instant date, social media, games, messaging apps and so on. None of these are possible without the data center. The issue becomes where, how many, their impact on our economy both national and personal (utility costs, environment etc). The people who benifit from their existance, not just the tech giants who profit, desrve to have a say.

//