Key Takeaways

  • Data centers can bring substantial capital investment, but their long-term fiscal value depends on taxes, operating jobs, and infrastructure obligations.
  • Utility contracts and rate structures will determine whether developers or other customers absorb the cost of new generation and grid upgrades.
  • Water availability, construction timelines, and uncertain future demand are becoming central factors in site approvals.

The data center boom is quickly becoming an infrastructure financing test. Communities are being asked to approve large campuses promoted as sources of investment, construction work, and tax revenue. At the same time, utilities may need to build substations, transmission lines, generating capacity, and water systems to support them.

That trade-off is growing harder to ignore. The International Energy Agency reported that global data center electricity demand increased 17% in 2025. It expects data centers to account for about half of U.S. electricity-demand growth through 2030.

Artificial intelligence is adding urgency. Electricity use by AI-focused facilities rose 50% in 2025, according to the IEA, while total global data center consumption could increase from 485 TWh in 2025 to 950 TWh in 2030. Microsoft, Google, and Meta are among the major users and operators of the hyperscale infrastructure behind that expansion.

A billion-dollar construction announcement does not automatically translate into a billion-dollar public benefit. Much of the spending can go toward servers, electrical equipment, cooling systems, and specialized construction. A campus may support a sizable temporary workforce during development but employ fewer people once it is operating.

Tax treatment matters, too. Incentives intended to attract projects can reduce revenue from equipment or property that would otherwise expand the local tax base. Officials therefore need to compare the expected recurring revenue with the cost of roads, utility connections, emergency services, water capacity, and other public obligations.

The electricity requirement is particularly significant. U.S. data centers consumed approximately 176 TWh in 2023, equivalent to about 4.4% of national electricity use, according to data from Lawrence Berkeley National Laboratory. Its updated assessment projects that the sector could consume as much as 12% of U.S. electricity by 2028.

Who pays to prepare the grid for that load?

If a utility spreads expansion costs across its broader customer base, households and existing businesses can end up helping finance infrastructure built primarily for a new industrial user. Regulators can reduce that exposure through dedicated tariffs, minimum billing commitments, upfront contributions, collateral requirements, and contract provisions addressing projects that are delayed, downsized, or canceled.

Demand forecasting creates another wrinkle. Utilities often plan assets over decades, while computing hardware and AI economics can shift much faster. A transmission project built around an anticipated campus could become underused if the developer changes its plans. Conversely, waiting for demand to materialize can leave a region without enough capacity to connect projects on schedule.

Water introduces a similar allocation question. A 100-megawatt U.S. data center may require about 530,000 gallons per day across the electricity supply chain, including roughly 190,000 gallons consumed onsite. Actual use varies considerably with cooling design, climate, and the local generation mix. In water-stressed areas, average annual consumption may also hide the importance of demand during hot, dry periods.

Public scrutiny is already increasing. Fortune has reported on mounting opposition as data centers account for a larger share of new U.S. power demand. That resistance can affect permitting schedules, utility proceedings, and the political durability of incentive packages.

For developers, transparent commitments may become a competitive advantage. That can include disclosing expected peak electricity demand, setting water-use targets, funding dedicated infrastructure, and accepting financial responsibility if projected load does not arrive. Efficiency metrics such as Power Usage Effectiveness can help compare facility designs, although operational reporting provides a clearer picture than design estimates alone.

For local governments and utilities, the central issue is not whether data centers create economic activity. They do. The more useful question is whether contracts, tariffs, and tax agreements assign costs to the parties creating them while protecting existing customers from speculative infrastructure risk. Communities that settle those details before construction are more likely to capture the upside without quietly transferring the bill elsewhere.