Power the AI Boom—Without Making Families Pick Up the Tab

America’s sudden appetite for electricity is not a sign of decline. It reflects investment in artificial intelligence, data centers, advanced manufacturing, and a digital economy that the United States should intend to lead.

But opportunity does not repeal physics. Every new computing campus needs real generation, real wires, real fuel, and real reserves. If policymakers approve the load but fail to build the system behind it, households will eventually pay through higher bills, weaker reliability, or both.

The demand trend is already visible. The U.S. Energy Information Administration’s January outlook projected the strongest four-year growth in electricity use since 2000, with large computing facilities as the main driver. Its July forecast shows national generation climbing from 4,430 billion kilowatthours in 2025 to 4,636 billion in 2027. Over the same period, average residential electricity prices are forecast to rise from 17.3 to 18.7 cents per kilowatthour.

There is good news. The North American Electric Reliability Corporation reports that available bulk-power resources increased by more than 58.5 gigawatts ahead of summer 2026, outpacing the latest year-over-year increase in peak demand. Solar and battery additions supplied much of that gain, with some new natural-gas generation also contributing.

The longer-term picture is more demanding. NERC projects North American summer peak demand rising by more than 224 gigawatts over ten years, with data centers accounting for most of the increase. That figure carries uncertainty: projects can be delayed, canceled, or submitted in more than one queue. Texas planners have already reduced some requested data-center load assumptions sharply to reflect how facilities actually ramp. The lesson is not that the boom is imaginary. It is that planners must distinguish committed projects from speculative claims.

The Federal Energy Regulatory Commission’s June 18 action is therefore timely. The Commission ordered six regional grid operators to justify or reform the tariffs governing data centers and other large loads. The proceedings address faster studies, transparent transmission costs, co-location, flexible service, and coordinated study of new generation with nearby demand. Grid operators were also directed to explain how adequate generation will be available for new and existing customers.

That action begins a regulatory process; it does not mean a final nationwide tariff is already in place. The direction, however, is right.

The first principle should be simple: existing families and small businesses should not finance speculative infrastructure for some of the wealthiest corporations in the world. Large-load customers should demonstrate site control, meet development milestones, provide financial security, and pay the costs directly attributable to serving them. If a project is canceled after utilities begin construction, ordinary ratepayers should not be left holding the bill.

The second principle is bring your own new generation. A data center that arrives with dependable new supply—or agrees to curtail consumption when the grid is stressed—can connect more quickly and impose fewer costs on everyone else. Flexibility can help, but it must be measurable, enforceable, and visible to grid operators in real time.

The third principle is an American all-of-the-above buildout. Near-term growth will require more natural-gas generation and fuel infrastructure where systems need dispatchable power. Existing nuclear plants should be preserved where safe and economic, while licensing and construction barriers for new nuclear projects are reduced. Batteries, grid-enhancing technologies, reconductoring, and new transmission can make existing resources more useful. Viable coal capacity should not be retired on a political timetable before dependable replacement power is operating.

Forecasts show why supply matters. In an EIA stress scenario—not its baseline—faster data-center growth without additional generating capacity pushed modeled 2027 Texas wholesale prices 79 percent above baseline. That result is not destiny. It is a warning about approving demand faster than supply.

America should welcome the AI and manufacturing boom. But winning cannot mean subsidizing corporate campuses while families absorb the grid costs. The durable bargain is growth with accountability: build generation, modernize the network, require major new users to pay their fair share, and refuse to sacrifice reliability for a press release.

Power abundance is an America First advantage. It must be built—not assumed.