AI Power Needs: Who Pays the Bill?

The debate about artificial intelligence often focuses on chips, software and security. A less discussed but crucial element is electricity. Data centers that run AI models need huge amounts of power, and building the physical infrastructure to deliver it can shape a nation’s industrial future. The main question is whether this power will boost domestic manufacturing or increase reliance on foreign suppliers. In Loudoun County, Virginia, data centers have grown so fast that the local electric grid faces demands of up to 1, 000 megawatts—an order of magnitude higher than it was designed for. These spikes are not theoretical; they have already strained utilities and threatened the local environment, with farms turning into solar fields while residents see higher bills. The choice of power source matters. Nuclear power can provide the steady base load that data centers need, but its construction takes years of permitting and financing. In the meantime, solar panels can be installed quickly and help bridge the gap. Solar is not a permanent solution, but it offers an immediate response while the long‑term infrastructure develops. Importantly, the question is who will build and profit from this new solar capacity. American companies such as Suniva and NextPower are already expanding domestic production of solar cells and tracking systems. The talent and factories exist; what is missing is political support to keep the manufacturing chain in the United States. If foreign firms dominate, American workers will bear the cost without reaping the benefits.The same logic applies to large data‑center operators. These private businesses generate massive profits and demand vast amounts of electricity. Instead of spreading the cost across all ratepayers, it makes sense for the companies themselves to fund the necessary infrastructure. This principle has already been adopted by regulators in Pennsylvania, where customers driving new capacity are required to cover the associated costs. Balancing local environmental concerns with economic realities is essential. Farmland disappearing for solar and rising utility bills are real worries, but they do not justify inaction. They call for a strategy that builds power quickly, keeps production domestic, and ensures that those who create the demand pay for it. Policymakers at all levels—federal, state and local—should adopt three clear principles: 1) large‑scale power demand should be paid for by its users; 2) new infrastructure should favor American manufacturing; and 3) emergency expansions must avoid increasing import dependence. These steps will help the AI economy grow while strengthening national industry. The AI revolution is already underway. The power infrastructure that supports it must be built here, fast, and with a fair distribution of costs. That is the path to avoiding another cycle of outsourcing and ensuring that American workers benefit from this new era.