The incoming US Congress, set to take office in January, faces mounting pressure to address the rapid expansion of AI data centers and their impact on electricity costs for ordinary households and small businesses, according to a new analysis from the Brookings Institution.

Bipartisan public concern over the facilities has become a major political issue, with Representative Greg Casar (D-Texas), chair of the Congressional Progressive Caucus, describing the backlash as "one of the top issues of the 2026 election," Brookings reported. Democrats won control of both the House and the Senate in that election.

The analysis, authored by Brookings nonresident senior fellow Mark MacCarthy and independent consultant David Klaus, argues that the current boom in data center construction resembles a speculative bubble. Investors are pouring money into new facilities that may ultimately exceed demand for AI computing power, a pattern seen before with railroads and internet infrastructure, the authors wrote. If many planned data centers become obsolete, the electricity infrastructure built to serve them could be left stranded, with costs potentially falling on other ratepayers.

Ratepayer protection mechanisms

The authors argue that policymakers already know how to shield residential and small business customers from these costs. The approach has two main components: ensuring that ordinary ratepayers do not pay for infrastructure built specifically for AI data centers, and protecting them from stranded investment costs when data centers require less power than anticipated.

Proper cost allocation studies can identify which generation, transmission, and distribution investments are made to serve data centers, Brookings reported. An equitable rate structure can then require data centers to pay those costs, leaving residential and small business customers unaffected. State regulators bear primary responsibility for designing these structures, though the Federal Energy Regulatory Commission (FERC) can encourage the process under current law.

To address stranded investment risk, the authors recommend long-term contracts that obligate data centers to pay even if they later need less energy. This could involve upfront payments from data centers to utilities, combined with regulatory requirements guaranteeing payment, to protect against the possibility that data center developers go out of business.

The AI industry has signed onto voluntary ratepayer protection pledges under the current administration, Brookings noted, but the authors stressed that voluntary agreements require effective implementation and enforcement to be meaningful.

Legislative action underway

The current Congress has already begun engaging on the issue. In early September, the House passed the Ratepayer Protection Act of 2026 by a bipartisan vote of 417–3, but the Senate rejected it on the grounds that it only required state regulators to "consider" protections, making it effectively optional, according to Brookings.

A separate bill, the Bipartisan American Affordability and Jobs Act of 2026, which aims to streamline permitting for new energy projects, contains mandatory ratepayer protections. Section 2107 of that bill would instruct FERC to require public utilities to charge AI data center customers both the embedded and incremental costs of transmission service, and would prohibit allocating any portion of those incremental costs to other customers. It also provides that cost recovery from AI data centers would continue even if a data center stops purchasing power.

Unlike current law and the earlier Ratepayer Protection Act, these provisions would give federal force to ratepayer protections that otherwise depend largely on state regulatory agencies. The bill is expected to be considered in the Senate during the lame-duck session and might pass, Brookings reported, though House consideration is less likely now that Democrats have taken control of that chamber. The bill faces opposition from a coalition of environmental, consumer protection, and public interest groups, which weakens its chances. However, the authors suggested it could serve as a starting point for congressional action in the next Congress if introduced as a standalone bill.

Unresolved challenges

Brookings identified four issues that remain outside the current consensus on ratepayer protections.

First, some AI data centers want to bring their own power, independent of the electricity grid serving other customers. While this approach seems to protect ratepayers — if companies overbuild, they pay the price — the authors noted that such "behind the meter" projects are rarely completely off grid. Most still require a grid connection for backup, peak loads, and stability, meaning other ratepayers could end up paying for that standby capacity. Additionally, locking up energy supplies for off-grid projects can increase wholesale energy prices that ultimately filter through to other ratepayers.

Second, because AI data centers need power immediately, they tend to rely on the most readily available source — often natural gas — rather than cleaner but slower-to-build renewable energy. Ireland and other European countries now require new data centers to use renewable power, Brookings reported, but similar policies in the US would delay data center construction and slow the rollout of AI technology.

Third, the ratepayer protections described may not prevent rate increases caused by the rising cost of generating additional power. Electricity bills include charges for energy consumption, transmission, and local distribution. While the proposed protections can shield ratepayers from transmission and distribution costs incurred for AI data centers, the enormous demand for new power generation drives up wholesale energy prices, which in turn raises retail energy usage charges for everyone. Brookings cited evidence from a Federal Reserve study and an analysis by the PJM market monitor suggesting this is already happening. No consensus exists on how to address this issue.

Fourth, some lawmakers are calling for a moratorium on data center construction to buy time for implementing protections. Senator Bernie Sanders (I–Vt.) and Representative Alexandria Ocasio-Cortez (D–N.Y.) support this approach, and an increasing number of states are imposing or considering moratoria, according to Brookings. While a moratorium could complicate efforts to meet current compute capacity shortages and slow the diffusion of AI benefits, the authors noted that getting ratepayer protections right — along with addressing noise, pollution, and water concerns — will take time. If construction proceeds unchecked, it may be too late to put needed protections in place.

Macroeconomic risks

A recent Brookings report estimated that AI investment will total $10.3 trillion between 2025 and 2032, averaging 3.63% of US gross domestic product per year — larger than major historical investment booms in canals, railroads, electrification, highways, and telecommunications. Gartner, the research firm, projects the industry will spend $7 trillion on infrastructure through 2030 but generate only $3.6 trillion in related revenue over the same period, a shortfall that Brookings described as a potential measure of the coming economic crash.

The authors acknowledged that policymakers lack the tools to prevent this contraction or mitigate its widespread macroeconomic consequences. However, they argued that policymakers should not ignore the damage already occurring to residential and small business electricity customers, and that ratepayer protection measures represent an essential element of a comprehensive response to public concern over AI data centers.