If an AI data center is proposed near your town, the first question is not whether artificial intelligence is good or bad. It is who pays for the infrastructure that makes the project possible.
That question is newly relevant because the data-center boom is moving from a national tech story into local planning rooms. Fox News highlighted Dallas on August 10 after Cushman & Wakefield ranked it the world's No. 1 primary data-center market, and Meta published a same-day letter promising a new community compact around its AI buildout.
The short version: a data center can bring construction, taxes, and some permanent jobs, but the deal is only as good as the power, water, ratepayer, land-use, and public-benefit terms written into it.
Do this first
Ask whether ordinary ratepayers are protected. AI data centers need large, reliable loads of electricity. Cushman & Wakefield said global capacity under construction approached 31.7 gigawatts in 2025, while planned capacity across the Americas rose from 46.1 gigawatts in 2024 to 191.3 gigawatts by the end of 2025. Those numbers matter locally because new substations, transmission lines, backup generation, and interconnection work can become utility costs unless regulators or contracts assign them to the project.
Ask what happens before power is available. The same Cushman report said average power-delivery timelines for new large-load requests are about 4.4 years globally and roughly five years across the Americas and EMEA. A town may approve land use long before the grid work is finished, so residents should ask what construction can begin, what milestones trigger public commitments, and what happens if the company delays or changes plans.
Separate one-time construction work from permanent jobs. Big projects can support construction employment, but a finished hyperscale data center may not employ as many local workers as residents expect. Ask for separate numbers for construction jobs, full-time operations jobs, local hiring commitments, wages, apprenticeships, and contractor requirements.
Put water, noise, emergency services, and road wear into the deal. Cooling systems, backup generators, truck traffic, security needs, and emergency-response planning are not abstract concerns. They affect water systems, nearby neighborhoods, fire departments, and local roads. A useful agreement should say what the operator will monitor, what it will pay for, and what remedies apply if limits are missed.
Do not count a community fund twice. Meta said on August 10 that it is launching a Future Is For Everyone Fund and a community compact for places where it works, including investment in schools, public services, energy-price protection, jobs, and environmental commitments. That may be meaningful, but residents should ask whether a fund is new money on top of taxes and infrastructure obligations, or whether it replaces benefits the community expected anyway.
Check these details
Start with the rate case or utility agreement. The key issue is not the headline size of the project; it is whether the data-center customer pays for dedicated grid upgrades, backup equipment, and connection costs or whether those costs are spread across residential and small-business bills.
Then read the tax-incentive package. A project can be praised for growing the local tax base while also receiving abatements, exemptions, or infrastructure reimbursements. Ask for a plain-language estimate of how much the local government gives up, how much it receives, and when the tradeoff changes.
Next, look for enforceable milestones. A promise to invest, hire, or donate is weaker than a contract that ties benefits to dates, permits, operating capacity, or tax receipts. The strongest local agreements explain what happens if the company sells the site, misses a deadline, scales down the project, or changes the power plan.
Finally, ask who gets a public update after approval. Communities should not have to wait for rumors or a future election to learn whether power use, water use, noise, emergency calls, traffic, or promised benefits are matching the original case for approval.
Common mistakes
The first mistake is treating national AI competition as a reason to skip local math. The United States may want more AI infrastructure, but a town still has to protect its utility customers, land, roads, and emergency services.
The second mistake is confusing a brand-name company with a simple deal. Meta, Google, Microsoft, Amazon, Oracle, and other major players may have the balance sheets to make strong commitments. That does not mean every local agreement automatically assigns costs clearly.
The third mistake is focusing only on opposition. Brookings noted in July that data-center fights have become a visible proxy for broader anxiety about AI, jobs, and political power. But the practical local question is narrower: what terms would make the project accountable enough for the people who live nearby?
When to slow down
Slow down if the utility filing is not public, if officials cannot explain who pays for dedicated infrastructure, if water use is described only in averages, if tax incentives are presented without a payback timeline, or if community benefits are described as voluntary goodwill rather than written obligations.
Also slow down if a proposal asks for fast zoning approval while leaving the power source, rate structure, emergency plan, or environmental monitoring for later. In a market where power access is the defining constraint, those are not finishing details. They are the deal.
Bottom line
An AI data center can be a serious economic-development project. It can also become a quiet transfer of risk from a technology company to residents if the local agreement is vague. Before a town says yes, it should know who pays the power bill, who pays for upgrades, who monitors impacts, and who enforces the promises after the ribbon-cutting.
