Expanding AI Data Centers Means Higher Electricity Bills for Americans

Forecast of the U.S. data-center power demand to 2035. Courtesy of  BloombergNEF.

At the end of 2025, Baltimore resident Nicole Pastore realized that her electricity bill jumped 50% in comparison to what she paid the year prior as well as the previous 17 years in her home. The state of Maryland has recently been the locus of much debate regarding the surge of companies seeking to build artificial intelligence data centers in the state, with 41 of these centers already existing in Maryland: 14 of which are located in the Baltimore metropolitan area alone. Government officials claim these data centers create necessary infrastructure and provide job opportunities that stimulate the economy, while some critics cite this expansion as the source of a gratuitous fiscal burden on the American people.

Data centers are essential for the functioning of artificial intelligence applications. They create the capacity for AI programs to store and process data that is used to inform decision-making and interactions that occur in user-facing models, such as ChatGPT. While data centers have been used to facilitate technological capabilities of storage databases for decades, AI data centers require advanced levels of processing and functionality due to their high traffic and computational demands. In turn, their energy consumption is much greater than that of traditional data centers, and they yield a much higher overhead cost. Their increased popularity necessitates the installation of technology like Graphics Processing Units (GPUs) while traditional centers could opt for Central Processing Units (CPUs). CPUs are unable to process simultaneous computing at the rate that is required, thus the more expensive GPUs are required. 

The global economy has seen accelerated stimulation from investments into artificial intelligence, with the amount of private investment into generative AI having increased exponentially in the last five years. President Trump has continued to subsidize this sector in the United States, with significant financial commitments being made to continue developing AI technology and generating continuous sectoral growth. Simultaneously, prices for electricity have skyrocketed nationwide, and data analysis conducted by Bloomberg revealed that costs are up to 267% more expensive than five years ago in areas with significant data center activity. Companies like PJM - one of the most popular electrical companies in the country, serving over 65 million Americans - have declared rising prices are a result of increased supply and demand. However, when speaking to their recent financial decisions, they attributed $5.9 billion in project investments largely to the expansion of data centers.

Despite the unpopularity of AI data center construction in many states including Maryland, Virginia, and Illinois, taxpayer dollars are being funneled into this venture alongside internal allocations. In its Independent Market Monitor for PJM, Monitoring Analytics cited data center induced stress as the main reason for the increased prices and strain on the electrical grid. They criticized PJM’s statements in their analysis published in June, stating that “it is misleading to assert that the capacity market results are simply just a reflection of supply and demand,” as that is unsupported by both historical data and future projections. Electricity bills in Illinois have increased 16% this year alongside a state-wide rise of 12% in Virginia, and taxpayers in Maryland are projected to see a spike of up to 24% by the end of the year. 

Companies such as PJM have been utilizing regulations set in a time before the AI boom to pass raising costs off on average consumers. Electricity regulations stipulate that costs are to be spread out amongst consumers in the case of new infrastructure developments being warranted, including scenarios in which systems need to be updated, and new power supplies installed. New “large customers” have registered energy demands under singular titles to allow for their energy interests to be viewed under this category, therefore shifting massive fees into the communal bill pool. By this method, a single large customer is able to request access to 300W (the amount of energy used by an entire city) at the earliest possible accessibility. 

Trump has supported AI development in terms of direct federal subsidies as well as calling for the interconnection of new data centers to grid makeups. This has been a point of contention for governmental representatives including former chair of the Federal Energy Regulatory Committee, Mark Christie. Christie stated his concerns regarding the policies on the grounds that they run into legal challenges in their disregard for state sovereignty in the matter. Enacting federal legislation regarding AI data centers’ creation and grid connections within individual states violates precedents set in Supreme Court cases including West Virginia v. EPA, where this circumstance would require congressional approval to move forward with implementing policy decisions. 

The addition of costs to electricity bills across the nation has raised red flags for governmental involvement in the energy sector. Trump is known to have close ties with wealthy individuals who stand to gain from the promotion of artificial intelligence and easing its access to the populace, and there is some question as to whether this pursuit has taken priority over the democratic voice of the people. Polling on these issues demonstrates that many Americans have concerns with the expansion of this infrastructure, and less than 25% believe that these policies are worth the additional cost to consumers.