Washington talks about artificial intelligence as if it were mainly a matter of chips, software and national security. But AI also depends on something far more basic: electricity.
Data centers require transmission lines, substations, turbines, solar panels, steel, copper and concrete. The AI economy will not run on slogans. It will run on physical infrastructure, and the central question is whether that infrastructure strengthens American industry or deepens our dependence on foreign supply chains.
The scale is already visible. Loudoun County, Virginia has become one of the world’s most concentrated data-center markets. Utilities are now facing power requests that would have sounded extraordinary only a decade ago, with hyperscale campuses measured in hundreds of megawatts and some interconnection requests reaching 1,000 megawatts. The electric grid was not built for that kind of shock.
The concern is not theoretical. Families in communities like mine are watching productive farmland disappear under solar panels while utility bills keep rising. Those concerns are legitimate. But the argument over solar is being framed too narrowly. The real question is not whether more power generation is coming. It is who builds it, who profits from it and who pays for it.
In the long run, nuclear power is the right answer. It provides the steady baseload that data centers require, and any serious country should be building more of it. But long run and near term are not the same thing. Nuclear takes years to permit, finance and construct. It is the right strategic answer, but it will not solve the immediate power gap now facing utilities, governors and grid operators.
That is where solar comes in.
That does not mean treating solar as a cure-all or pretending it can replace baseload power. It means recognizing that the near-term power gap is real, and that temporary infrastructure choices still have permanent industrial consequences.
Solar can be deployed faster than new nuclear plants and major transmission expansions. It is not the permanent answer to everything. It is the bridge. And the real argument is whether that bridge will be built by American workers using American-made components, or whether foreign producers will once again capture the industrial upside while Americans absorb the cost.
We do not have to speculate about whether domestic capacity exists. American companies are already proving that it does. Suniva restarted solar-cell production in Georgia and announced a second plant in South Carolina. Nextpower says it has opened or expanded more than 25 U.S. factories with manufacturing partners since 2021, building out a domestic supply base for solar-tracking systems. The capacity is here. The workers are here. What is still missing is the political will to treat that as a national advantage worth defending.
That is the choice in front of us.
We can meet the power demands of the AI economy with American-made energy equipment, generating American wages and strengthening American supply chains. Or we can repeat one of the most familiar failures of the last generation. We saw it in steel. We saw it in semiconductors. We saw it in pharmaceuticals. Demand rises here. Production moves elsewhere. American workers are told to be satisfied with the leftovers.
That is not an industrial strategy. It is surrender.
There is one more point policymakers cannot ignore: large data-center customers should pay for this infrastructure, not ordinary ratepayers.
Working families are already stretched. Manufacturers are already carrying heavy energy costs. Small businesses do not need another increase pushed onto them through the monthly utility bill. Hyperscale data centers are private enterprises making extraordinary profits. If utilities must build new capacity to serve them, those costs should be borne through direct agreements with those customers, not loaded onto families, retirees and productive businesses that did not create the demand.
Congress, the Federal Energy Regulatory Commission, state utility commissions and governors should make three principles clear: new large-load power demand should be paid for by the customers creating it; new energy infrastructure should favor domestic manufacturing; and emergency buildout should not become another channel for import dependence. Pennsylvania regulators have already moved in that direction, emphasizing that customers driving new infrastructure needs should be responsible for the associated costs.
The farmland concern is real. Ratepayer concerns are real. Neither is an argument for doing nothing. They are arguments for doing this correctly.
The AI economy is being built right now. The infrastructure behind it should strengthen this country, not hollow it out. Build what is needed. Build it quickly. Build it here. And make the companies driving this new demand carry the cost instead of shifting it onto everybody else.
John A. Burtka III is Director of Advancement and Outreach for the Coalition for a Prosperous America and a former automotive manufacturing executive.
This article originally appeared on The Detroit News: AI’s electricity demand shouldn’t fall on ratepayers | Opinion
Reporting by John Burtka III / The Detroit News
USA TODAY Network via Reuters Connect
