In the summer of 2025, electricity customers in Washington, D.C., experienced a notable increase in their power bills, with an average rise of about $21 per month for Pepco customers. This hike was largely attributed to increased capacity costs in the PJM Interconnection, the regional grid operator serving 13 states in the eastern United States, according to the District of Columbia Office of the People’s Counsel. A significant contributing factor was the rapid expansion of data centers, which many residential consumers had not expected to influence their electricity expenses directly.

Data centers have become major industrial electricity consumers, with large facilities demanding hundreds of megawatts and some proposed complexes projected to require up to a gigawatt of power. The Electric Power Research Institute has estimated that by 2030, U.S. data centers could consume between 9% and 17% of the nation’s electricity—a substantial rise from the current 4% to 5%. This surge is driven in part by artificial intelligence applications, which demand extensive and reliable power infrastructure including generation capacity, substations, transformers, and transmission lines.

The growing power demand associated with data centers raises a critical question: who should bear the costs of upgrading and expanding the electrical grid to meet this demand? Historically, regulators and grid operators have adhered to the principle that new demand requiring new infrastructure should be responsible for funding those investments. Reflecting this approach, PJM’s capacity costs jumped from $28.92 per megawatt-day in 2024-25 to $269.92 in 2025-26, with further increases projected in subsequent years. Data centers alone contributed roughly $9.3 billion to PJM’s 2025-26 capacity auction.

A key challenge arises because grid infrastructure must often be expanded based on forecasts made before the actual demand materializes. Such uncertainties expose utilities and ratepayers to financial risks if anticipated demand fails to develop as expected. Addressing this, several major technology firms—including Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI—committed in March 2026 to a White House Ratepayer Protection Pledge. This agreement obligates these companies to finance power generation and infrastructure costs associated with their data centers and cover expenses even if the electricity is not ultimately used. Moreover, the Federal Energy Regulatory Commission (FERC) initiated regulatory reviews of tariffs applied to large loads across regional grid operators, and the U.S. House of Representatives passed the Ratepayer Protection Act in September by a wide margin. The legislation would require state regulators and utilities to establish standards ensuring that large data centers contribute appropriately to power grid costs.

China faces similar challenges amid a rapidly growing electricity sector. In 2025, the country became the first to consume over 10 trillion kilowatt-hours of electricity, with internet and related services expanding by over 30%. To meet this demand, China’s State Grid announced plans to invest approximately 4 trillion yuan ($596 billion) between 2026 and 2030—an increase of around 40% from the previous five years—while China Southern Power Grid is expected to contribute an additional trillion yuan. These investments support broader objectives including renewable energy integration, electrification, electric vehicle support, manufacturing growth, and improving grid reliability. This complexity underscores the need for clear policies in cost allocation, particularly differentiating regional infrastructure benefiting many customers from dedicated facilities built primarily for large individual consumers.

Experts suggest four guiding principles for allocating costs associated with large electricity users like data centers. First, categorize very large loads as a separate customer class with long-term contracts that recover specific infrastructure investments, including provisions for deposits or minimum payments. Second, distinguish between dedicated infrastructure—funded by the individual customer—and shared infrastructure, which can be jointly financed under transparent rules. Third, recognize the value of flexibility in demand, rewarding data centers that reduce consumption during system stress, while accounting for those requiring reliable maximum power. Finally, ensure transparency in demand forecasting, requiring developers planning substantial capacity to submit financial commitments and allowing regulators to evaluate forecast assumptions.

Data centers are critical to modern commerce, communications, scientific research, and the development of artificial intelligence. Proper planning and flexible consumption strategies can integrate their demand into grid management, potentially enhancing system resilience. Ultimately, power systems must balance physical and financial constraints, with the prevailing consensus that the costs of new infrastructure should align with the sources of new demand.