The artificial intelligence boom is producing extraordinary demand for electricity, and American households should not be required to subsidize it.
The White House is preparing to bring technology companies, data center developers and utilities together around a Ratepayer Protection Pledge. Amazon, Google, Meta, Microsoft, OpenAI, Oracle and xAI have reportedly agreed to cover the infrastructure costs associated with their AI operations rather than transferring those expenses to existing electricity customers.
That principle is correct.
Data centers require enormous quantities of power. Meeting that demand can involve new transmission lines, substations, generating facilities and grid upgrades. These investments may strengthen the electrical system, but they are frequently initiated to serve a small number of highly profitable corporate customers.
Residential customers and small businesses should not pay higher monthly bills so trillion-dollar technology companies can expand computing capacity.
The impact is already visible in industrial regions. Belden Brick Company in Ohio reported a 90 percent increase in electricity costs, driven largely by rising capacity charges. Manufacturers with energy-intensive operations face particular pressure because they cannot easily reduce consumption without cutting production.
This threatens to create a damaging contradiction in U.S. industrial policy. Washington wants to expand domestic manufacturing while also encouraging rapid AI infrastructure development. If data centers push electricity costs higher for factories, one strategic priority will undermine the other.
The Ratepayer Protection Pledge recognizes the danger, but voluntary commitments are not sufficient.
Utility pricing is complex, and costs can be shifted indirectly. A technology company might pay for a dedicated connection while leaving broader reliability expenses, reserve capacity or transmission upgrades to other customers. Contracts may also receive confidential treatment, making it difficult for the public to determine who is paying what.
State utility commissions should require transparent cost allocation for large data centers. Regulators should have the authority to demand long-term financial assurances, minimum payment commitments and protections from abandoned projects.
This is especially true as technology cycles can move quickly. A company may announce a huge data center based on today’s demand for AI, only to scale back or scrap the project if computing becomes more efficient or market conditions change. Local utilities could then be left with infrastructure built for a customer that no longer needs it.
Environmental consequences must also be considered. Dozens of proposed gas-powered facilities could be constructed specifically to serve data centers. Environmental groups estimate that planned projects could produce hundreds of millions of tons of greenhouse gas emissions each year. Some privately operated facilities may face less scrutiny than conventional utility projects.
The country should not pretend that digital services exist without a physical footprint. AI depends on land, water, chips, transmission infrastructure and vast amounts of electricity. Every chatbot request and automated process ultimately relies on industrial-scale systems.
There are legitimate public benefits to expanding AI capacity. The technology could improve medical research, logistics, education, scientific modeling and business productivity. It may also strengthen the United States in a strategic competition with China.
Those benefits, however, do not justify privatizing profits while socializing infrastructure costs.
Technology companies should pay for the generation, transmission and grid reinforcement their projects require. They should also provide financial security for communities asked to approve large developments based on projected jobs and tax revenue.
The proposed pledge establishes an important standard: AI expansion should not increase electricity bills for ordinary Americans. The next step is turning that principle into enforceable regulatory policy.
