President Trump has a characteristically understated message for communities resisting AI data centers: Accept them or risk becoming “backwards and poor.” He says the facilities will bring jobs, prosperity, and lower taxes, while opposition will please China and kill America’s technological “Golden Goose.” Subtle it’s not.

Still, beneath the presidential megaphone is a serious question. Are data centers economic engines that short-sighted communities reject at their peril, or resource-hungry industrial projects whose benefits are often oversold? The answer matters because these facilities aren’t ordinary office parks. They can represent billions of dollars in investment, consume electricity on the scale of a small city, and employ surprisingly few people once the construction crews pack up.

Trump’s claim therefore deserves more than either automatic applause or outrage. It needs the increasingly rare Washington treatment known as arithmetic.

The Case for Trump’s Claim

Servers, Shovels, and a Whole Lot of Concrete

The strongest argument for Trump’s position begins with the enormous amount of money required to build an AI data center. Hyperscale campuses require land, concrete, electrical equipment, cooling systems, fiber connections, substations, backup power, security, and highly specialized construction. That creates substantial demand for electricians, engineers, equipment operators, pipefitters, truck drivers, contractors, and suppliers.

Recent research summarized by Georgia Tech found that, during the first three years after a data center opened, employment in the host county rose by roughly 0.9 percent, wages by 1.1 percent, business establishments by 1 percent, and household income by 0.7 percent. The estimated effects grew over time. Those aren’t civilization-altering numbers, but neither are they imaginary. For a county struggling to attract private investment, even modest gains can matter.

Construction is only the first layer. Operating facilities purchase electricity, maintenance, security, landscaping, equipment, and professional services. Employees and contractors spend money in nearby businesses. A data center can also justify upgrades to power, broadband, roads, and other infrastructure that may support additional development later.

The argument isn’t that every server rack personally strolls downtown and buys lunch. It’s that a very large capital investment can generate business throughout a local supply chain.

The Tax Base Has Entered the Chat

Data centers can also generate property, sales, use, and business-tax revenue. Because the buildings and equipment are so expensive, even a facility with a relatively small permanent workforce can add considerable taxable value. If the revenue exceeds the public cost of serving the site, local governments may gain money for schools, emergency services, roads, tax relief, or debt reduction.

This is the most plausible basis for Trump’s promise of “far lower taxes.” A community doesn’t necessarily need thousands of permanent employees to benefit fiscally if a highly valuable industrial property contributes heavily to the tax base while requiring comparatively few public services.

The administration’s Ratepayer Protection Pledge is intended to strengthen that case. Participating companies commit to build, bring, or buy the new power their facilities require; pay for associated generation and grid infrastructure; and negotiate separate utility rates that they owe whether or not they use all the subscribed electricity. The pledge also calls for local hiring, workforce development, and investment in community resilience.

If those promises become enforceable contracts, a host community could receive new revenue and infrastructure without forcing households to subsidize the project.

The AI Race Doesn’t Run on Positive Thinking

Trump’s also making a national competitiveness argument. Artificial intelligence requires physical infrastructure. The United States can’t lead in advanced computing, cybersecurity, scientific research, medicine, logistics, manufacturing, and defense by hoping that somebody stores all those processors in a tasteful shed overseas.

Data centers are the industrial backbone of the digital economy. Communities that host them may become part of a broader technology cluster, attracting suppliers, engineers, startups, training programs, university partnerships, and additional investment. Brookings notes that communities can negotiate for research partnerships, workforce pipelines, computing resources for universities, and support for local entrepreneurs rather than settling for an isolated server warehouse.

There’s also a collective-action problem. Any single town may prefer that the facility go somewhere else while still enjoying AI services, cloud computing, streaming, electronic banking, digital health records, and reliable online commerce. But if every community says “not here,” the country can’t build the infrastructure on which those services—and American competition with China—depend.

Opportunity Rarely Waits in the Parking Lot

Companies have options. A community that rejects a proposed campus might not receive a revised offer. The investment might simply move to another county or state.

That possibility matters most in places with a declining industrial base, weak population growth, or limited access to large private investments. Passing on a project can mean losing construction work, tax revenue, infrastructure upgrades, and a foothold in a growing industry. Once a neighboring region develops the power capacity, skilled labor, fiber network, and supplier base, later investment may cluster there as well.

In that narrower sense, Trump has a point: reflexively rejecting all data-center development could leave some communities economically behind. A town can’t spend decades telling every new industry to build elsewhere and then act surprised when “elsewhere” collects the investment.

The Case Against Trump’s Claim

The Permanent-Job Cupboard Can Look Awfully Bare

The first objection to Trump’s claim is that a billion-dollar building isn’t the same thing as a billion-dollar local economy. Data centers are extraordinarily capital-intensive but comparatively light on permanent labor.

The Georgia Tech research found that metropolitan counties captured most of the measurable employment and business growth, while nonmetropolitan counties saw few broad gains. Rural facilities often employ fewer than 100 permanent workers and import specialized services from outside the county. Another analysis of Texas data centers found no statistically discernible net employment growth overall, although it did identify construction activity associated with new facilities. In plain English, a project can create a great deal of visible work without permanently increasing the region’s total number of jobs.

Construction workers eventually finish construction. The ribbon gets cut, the politicians locate the scissors, and a massive facility may then operate with a workforce smaller than a busy supermarket.

A Tax Windfall After the Tax Giveaway

Tax revenue is also less impressive when governments give much of it away to win the project. States and localities routinely compete through sales-tax exemptions, property-tax abatements, subsidized land, infrastructure spending, and other incentives. The larger the subsidy, the more difficult it becomes to determine whether the community gained a valuable taxpayer or merely purchased an expensive press release.

The fiscal calculation must include roads, water and sewer systems, emergency services, substations, transmission lines, financing costs, and the risk that a facility closes or uses less capacity before public investments are repaid. The World Resources Institute reports that data centers can generate real tax revenue, but incentives can sharply reduce the net gain, and communities may remain responsible for infrastructure built to serve a facility that later downsizes or relocates.

Trump’s claim assumes the project produces lower taxes. A poorly negotiated project can do the opposite: privatize the profits, socialize the infrastructure, and send residents a commemorative utility bill.

The Golden Goose Has a Meter Running

Electricity is the largest and most immediate concern. According to Lawrence Berkeley National Laboratory, data centers consumed about 4.4 percent of U.S. electricity in 2023 and could consume between 6.7 and 12 percent by 2028. A later update estimated that they could account for 11.8 percent by 2030, with a plausible range from 9.5 to 15.3 percent. Those projections reflect the rapid growth of AI equipment and cooling demand.

New generation and transmission can eventually increase supply, but they take time and money. If a facility connects before adequate capacity is available—or if regulators allow utilities to spread upgrade costs across everyone’s bills—households and small businesses can pay more. The Georgia Tech study found an electricity-price increase of about 5 percent in areas where the effect could be measured cleanly.

Trump’s Ratepayer Protection Pledge responds directly to that concern, but a national pledge isn’t self-enforcing. Electricity rates are established through utilities, contracts, state regulators, cooperatives, and local authorities. The actual protection depends upon binding tariffs, cost-allocation rules, long-term payment commitments, and enforcement when a project is delayed, downsized, or canceled.

Critics answer that the phrase “they will pay” is less protective than a signed agreement specifying exactly who pays, for what, for how long, and under which penalty.

Water, Noise, Land, and Other Things That Don’t Fit in a Campaign Post

Communities oppose data centers for reasons beyond electricity prices. Cooling systems can consume substantial water. The World Resources Institute reports that a midsized facility may use up to 300,000 gallons per day, while a large facility may use as much as 5 million gallons, which is roughly the demand of a small town. The local effect depends upon the cooling technology, climate, water source, and whether the operator uses drinking water, groundwater, or reclaimed wastewater.

Large campuses can also consume hundreds of acres, displace farmland, alter nearby property uses, and create persistent noise from cooling equipment, generators, and electrical infrastructure. On-site gas generation and diesel backup systems can add air pollution. These costs won’t be identical everywhere, but that’s precisely why communities ask questions before approving a project.

Opponents therefore argue that wanting dependable water, affordable electricity, reasonable setbacks, and a good night’s sleep isn’t proof that residents hate technology. Sometimes people object because they’ve read the proposal, which is an almost suspiciously old-fashioned approach to civic life.

“No” Can Be an Economic Decision Too

Trump’s formulation leaves no room for opportunity cost. Land, water, power capacity, tax incentives, and infrastructure dollars used for a data center can’t simultaneously support housing, manufacturing, agriculture, small businesses, or another major employer. A community may rationally conclude that a different use would create more jobs, consume fewer resources, or better fit its long-term plans.

Local economies also differ. A metropolitan county with a strong supplier network and technical workforce may capture meaningful spillover benefits. A rural county may receive tax revenue but few lasting jobs. A water-rich region with excess generating capacity faces a different calculation from a drought-prone town with an already strained grid.

Critics conclude that the available evidence doesn’t support the blanket claim that communities declining data centers will become poor. Some may reject an excellent opportunity; others may avoid a bad deal. Public opposition is now broad and bipartisan: an August 2026 Annenberg survey found 61 percent opposed local data-center construction, up from 49 percent in the spring. That doesn’t prove the public is correct about every project, but opponents say it does make blanket dismissals of local resistance difficult to defend.

Welcome the Servers, Keep the Invoice

From my perspective, Trump’s right about the large national question but wrong about the sweeping local claim.

America needs data centers. Artificial intelligence will influence economic power, national defense, medicine, science, manufacturing, and communications whether we find the trend exciting, unsettling, or both. Surrendering the underlying infrastructure to China would be foolish. Private investment, new technology, abundant energy, and American innovation deserve encouragement.

But no, a community doesn’t become “backwards and poor” merely because it rejects one data-center proposal. That line is classic Trump: memorable, combative, and inflated until the nuance requires supplemental oxygen. The economic evidence is mixed. Benefits are real but uneven, permanent employment is often modest, and the costs can land on neighbors who never agreed to become silent partners in a trillion-dollar industry.

AI infrastructure can’t be economically successful if its business model depends on quietly shifting power, water, and construction costs onto families and small businesses.

Conservatism should defend private enterprise, but it shouldn’t confuse free enterprise with corporate welfare. It should value national strength without bulldozing local authority. It should respect property rights, contractual responsibility, and the right of citizens to question a project that could reshape their community for decades.

Christian stewardship adds another obligation: count the full cost and treat our neighbors as people, not obstacles between a developer and a permit. Jesus asked, “which of you, intending to build a tower, sitteth not down first, and counteth the cost” (Luke 14:28). That principle applies rather neatly to buildings full of towers of servers.

A responsible data-center agreement should therefore require:

• Binding utility rates and long-term payment guarantees that prevent costs from being shifted to households and small businesses.
• New generation and grid capacity sufficient to serve the project without weakening reliability or consuming capacity already needed by the public.
• Public water-use estimates, sustainable sourcing, drought protections, and reclaimed-water requirements where practical.
• An independent net-fiscal analysis that subtracts tax breaks, infrastructure expenses, financing costs, and public services from projected revenue.
• Performance-based incentives, local hiring and apprenticeship commitments, measurable community benefits, and clawbacks when promises aren’t met.
• Enforceable standards for noise, emissions, setbacks, traffic, farmland protection, site security, and eventual decommissioning.
• Transparent negotiations, public hearings, and meaningful local consent rather than secret agreements followed by a ceremonial vote.

If a company accepts those conditions, a data center may be an excellent investment. If it demands subsidies, cheap water, priority power, weakened zoning, and immunity from consequences, the community should show it the door.

Trump’s “Golden Goose” can lay valuable eggs. But before celebrating, local officials should make sure the goose belongs to the community, pays for its feed, and isn’t nesting on the electric meter.

America should build the infrastructure needed to lead the AI era. It should also insist that the corporations profiting from that infrastructure bear its true costs. That’s not backwardness. It’s stewardship, market discipline, and common sense, which are the three things most likely to keep a community prosperous after the groundbreaking photographers have gone home.