Legislative Issues
Artificial intelligence depends on physical infrastructure. Its economic benefits and local costs will depend on how policymakers plan for electricity, water, development, and community impacts.
What to Know
- Companies could invest nearly $7 trillion in global data-center infrastructure through 2030, with more than 40% of that spending in the United States.
- U.S. data centers used about 4.4% of the country’s electricity in 2023; that share could rise to 6.7%–12% by 2028.
- Data centers directly consumed about 46 million gallons of water per day in 2023, plus an estimated 576 million gallons indirectly through electricity generation.
- Large facilities can support up to 1,500 construction workers, but typically create 50 or more permanent on-site jobs.
- In 2025, Loudoun County, Virginia, ended most by-right data-center development and moved new projects into a public legislative-review process.
Artificial intelligence can feel like a purely digital technology, but it relies on a large physical system of servers, cooling equipment, substations, transmission lines, power plants, fiber connections, and industrial-scale buildings. The growth of that system has made data centers a major economic and policy issue.
The scale of this buildout has also created a national debate: how should communities capture the benefits of AI infrastructure while protecting households from higher utility costs, water stress, and poorly planned development? A recent Reason analysis argues that public discussion can overstate data centers’ national environmental footprint. That context matters, but national figures do not remove the need for local safeguards.
A $7 Trillion Infrastructure Boom
The expected investment explains why states and localities now compete for data-center projects. McKinsey estimates that companies could spend almost $7 trillion worldwide on data-center infrastructure through 2030. More than $4 trillion would go toward computing hardware, while the rest would support land, buildings, cooling systems, power generation, and transmission infrastructure.
Projected global data-center investment through 2030. McKinsey & Company.
More than 40% of that investment could occur in the United States. The spending can create demand for electricians, construction workers, engineers, equipment manufacturers, utilities, and local suppliers. It can also expand the physical capacity behind cloud services, scientific research, health-care systems, transportation networks, financial services, and AI tools.
Data centers are not, however, traditional mass-employment projects. McKinsey estimates that a typical large facility can employ up to 1,500 people during construction, including skilled tradespeople. Once operating, it may employ only 50 or more full-time workers on site.
That difference changes how policymakers should evaluate the public benefit. Rather than focusing only on permanent jobs, they should examine construction payroll, local supplier spending, tax revenue after incentives, workforce programs, and the cost of infrastructure needed to serve the project.
For example, McKinsey cites research showing that data-center construction, operations, and supporting businesses generated about $31 billion in supported economic output in Virginia in 2023. But local governments still need to test each proposed project on its own terms. Delays or cancellations can derail expected investment, while overly generous tax abatements can reduce the public return.
Water Use Requires Local Context
Water is one of the most visible concerns surrounding data-center development. Data centers use water directly through some cooling systems and indirectly through the electricity generation that powers their operations.
LBNL estimated that U.S. data centers directly consumed about 46 million gallons per day in 2023. It estimated a further 576 million gallons per day of indirect water consumption associated with electricity generation. Those estimates are significant, but their meaning depends on the comparison being made.
The U.S. Geological Survey reported that total U.S. water withdrawals were about 322 billion gallons per day in 2015. On that national scale, the data-center estimate represents a small share. However, this is not a perfect apples-to-apples comparison: USGS measures withdrawals, while the data-center estimate measures consumption.
U.S. water withdrawals by category in 2015. U.S. Geological Survey, public domain.
National context should not dismiss local concerns. A facility can have a small national footprint while still placing pressure on a drought-prone watershed or a municipal water system with limited capacity. Its impact also varies according to climate, cooling technology, workload, efficiency, and the local electricity mix.
Communities should therefore require developers to disclose projected water use, identify the water source, explain drought procedures, and report actual use once operations begin. Reclaimed water and more water-efficient cooling systems may be appropriate where local conditions support them.
Electricity Is the Central Infrastructure Challenge
Water concerns are local, but electricity demand creates an even broader infrastructure challenge. The Department of Energy reports that data-center electricity use rose from 58 terawatt-hours in 2014 to 176 terawatt-hours in 2023. DOE projects consumption could reach 325–580 terawatt-hours by 2028.
This growth can support investment in new generation, substations, storage, and transmission. Yet it also creates a direct household concern: if utilities build infrastructure mainly for a large new customer and the project does not materialize, existing customers could face part of the cost.
State regulators and utilities can reduce that risk by requiring upfront contributions, financial guarantees, minimum-demand commitments, and exit fees from large customers. Flexible-load agreements can also help some facilities reduce consumption during grid emergencies.
But local agreements solve only part of the problem. Even when a company and utility agree on project terms, regional interconnection and transmission processes can delay the new capacity required to serve the facility.
In June 2026, the Federal Energy Regulatory Commission directed six FERC-jurisdictional regional transmission organizations and independent system operators to justify or reform their rules for large electricity users: PJM, MISO, SPP, CAISO, ISO-NE, and NYISO.
FERC asked them to address five issues: faster study processes, transparent transmission costs, co-located generation, flexible large loads, and generation serving nearby major users. The agency’s action recognizes that data-center development is no longer only a local zoning question. It also depends on regional grid rules that determine when new power can connect and who pays.
Local Backlash Is a Planning Problem
Loudoun County shows both the opportunity and difficulty of rapid data-center growth. In 2000, its zoning administrator treated data centers similarly to office buildings, allowing by-right development in areas where offices were permitted. That policy helped Northern Virginia become the country’s largest data-center hub.
As facilities expanded, residents raised concerns about large buildings, backup-generator emissions, noise, transmission lines, water use, and development on previously undeveloped land. In 2025, the Loudoun County Board of Supervisors ended by-right development for most new data centers and required legislative review with public input.
Loudoun County’s data-center growth and policy timeline. Loudoun County.
This backlash extends beyond one county. Heatmap News reported that 25 U.S. data-center projects were canceled after local opposition in 2025, four times the number canceled in 2024. Heatmap based that finding on public records, project announcements, and press reports.
Neither unrestricted development nor indefinite moratoriums provide a stable answer. Communities can instead establish rules before individual projects arrive. Clear standards can address appropriate development zones, setbacks, noise, generator emissions, water reporting, landscaping, transmission routes, emergency operations, and decommissioning responsibilities.
A Practical Framework for Policymakers
Public officials do not have to choose between opposing data centers and approving every facility without conditions. The following principles connect infrastructure growth with consumer and community protection.
A Practical Framework for Policymakers
Build for a physical AI economy. AI’s cloud is physical. Policymakers, utilities, and developers must plan, power, and build the infrastructure that makes it possible.
Protect electricity customers. Large users should cover transparent, project-specific costs so households and small businesses do not subsidize infrastructure built mainly for a single facility.
Address water locally. Require clear disclosure of water demand, water sources, cooling technology, and drought procedures before approval.
Measure the total public return. Incentives should account for construction payroll, supplier activity, tax revenue, and infrastructure costs alongside long-term public obligations.
Set predictable local rules. Clear zoning and operating standards can give residents enforceable protections while giving developers a more reliable approval process.
Wrap Up
The data-center boom creates a significant economic opportunity, but the public case should not depend on treating every project as cost-free. Data centers can support investment, construction work, tax revenue, and technological capacity. Their benefits are strongest when developers make clear, measurable commitments to the communities hosting them.
The central policy challenge is aligning private development with public infrastructure. Faster approvals for power, transmission, and large-load connections should be paired with safeguards that prevent project-specific costs from shifting to existing utility customers.
A durable approach is both growth-oriented and accountable: build the infrastructure needed for AI while protecting household budgets, grid reliability, local water supplies, and community confidence.