Reported from the source

Quick summary: Companies building AI infrastructure are facing growing backlash over electricity costs, with households and policymakers questioning if data centers are driving up power bills. U.S. residential electricity prices have risen over 36% since 2020. A SemiAnalysis report suggests that market design and policy decisions, such as specific pricing mechanisms in regional grids, play a greater role in these increases than AI infrastructure growth alone. While hyperscalers have made pledges to cover costs and use renewable energy, experts express skepticism due to the industry’s profitability challenges.

The rapid expansion of infrastructure for the artificial intelligence boom is drawing criticism over electricity costs, as communities and policymakers debate whether data centers are responsible for rising power bills. Residential electricity prices in the U.S. have increased by more than 36% since 2020, from 12.76 cents per kilowatt-hour to 17.44 cents in February 2026, with projections to reach 19.01 cents by September 2027, according to the U.S. Energy Information Administration (EIA). The EIA noted that retail electricity prices have outpaced inflation since 2022. However, a report from semiconductor research firm SemiAnalysis argues that while data center expansion contributes, market design and policy decisions are more significant factors in energy price increases. The report points to an obscure market pricing mechanism, the Base Residual Auction, in the PJM Interconnection area—a grid serving 13 eastern states with numerous hyperscaler data centers—as a primary driver of “runaway” energy prices. SemiAnalysis claims PJM’s forecasts often overestimate future demand, particularly given construction delays for planned data centers, leading to higher projected costs for consumers who pay two years in advance. In contrast, the Electric Reliability Council of Texas (ERCOT) has seen relatively stable prices since 2022 despite data center development. Other factors also contribute to rising consumer energy prices. Maeghan Rouch, a partner at Bain & Company, explained that investments in local grids, such as hardening and modernization, or overall inflation, can also weigh on households. She added that even without data center investment, some upward pressure on price growth would be expected. In response to concerns, major technology companies have made commitments. Microsoft outlined a five-point plan in January, including a pledge to cover additional electricity costs from its data centers. Anthropic made a similar commitment in February. Most recently, President Trump convened AI executives to affirm a “Ratepayer Protection Pledge,” aiming to ensure new AI data center expenses are not passed on to American consumers. While such commitments could help secure community support for projects, experts like Marc Einstein of Counterpoint Research question their legitimacy, citing the AI industry’s struggles with profitability. Einstein emphasized the need for hyperscalers to clarify their plans to address rising electricity costs. Tech companies have also committed to meeting data center needs through renewable energy sources. This is increasingly important given growing energy demand and long wait times for grid connections, which can be four to six years in primary markets and up to 10 years in some cities globally. However, skepticism toward renewable energy commitments within the current U.S. administration raises questions about the progress of such pledges domestically. Analysts suggest that delivering on these pledges is in the corporate interest of AI hyperscalers, improving public relations and potentially averting regulatory intervention.

Source: www.cnbc.com