Policy
Congress Urged to Shield Ratepayers from AI Data Center Costs
Brookings scholars propose federal mandates to prevent residential customers from subsidizing infrastructure built for compute facilities that may become stranded assets.
Key takeaways
- AI data center expansion drove voter backlash in 2026 midterms, with concerns centered on electricity rate increases for residential and business customers.
- Projected AI infrastructure investment of $10.3 trillion through 2032 may exceed actual demand by trillions, creating risk of stranded assets that ratepayers could be forced to cover.
- Consensus ratepayer protections exist—requiring data centers to pay their own transmission and distribution costs—but lack federal enforcement mechanisms.
- The House passed weak protections 417-3, while stronger provisions in a separate bill face opposition due to unrelated energy permitting provisions.
- Unresolved issues include treatment of off-grid projects, renewable energy requirements, wholesale price impacts, and whether construction moratoria are warranted.
As the new Congress prepares to convene in January 2027, lawmakers face mounting pressure to address how AI data centers impact electricity costs for residential and small business customers. The issue gained prominence during the 2026 midterm elections, where Democrats won control of both chambers amid voter concerns about rate increases tied to data center expansion.
According to analysis published by the Brookings Institution, the core challenge stems from a familiar pattern: the AI industry appears to be experiencing an investment bubble. Brookings scholars Mark MacCarthy and David Klaus argue that current compute capacity shortages are likely temporary, with more data centers being built than future demand will support. This dynamic mirrors historical overbuilding in railroads and telecommunications infrastructure.
The scholars note that AI investment is projected to reach $10.3 trillion between 2025 and 2032—averaging 3.63% of U.S. GDP annually and exceeding major historical infrastructure booms. Yet Gartner estimates the industry will spend $7 trillion on infrastructure through 2030 while generating only $3.6 trillion in related revenue, suggesting a substantial overcapacity problem ahead.
The ratepayer protection framework
MacCarthy and Klaus outline a two-part policy approach that they say is already understood and feasible. First, proper cost allocation studies can identify infrastructure investments made specifically to serve AI data centers, with rate structures requiring those facilities to pay their own transmission and distribution costs rather than passing them to other customers.
Second, addressing stranded investment risk requires long-term contracts with payment obligations that persist even if data centers reduce their energy consumption. The scholars recommend combinations of upfront payments and regulatory guarantees to protect against scenarios where data center operators go out of business.
State regulators hold primary responsibility for implementing these protections, though the Federal Energy Regulatory Commission can encourage the process. The AI industry itself has endorsed these principles through voluntary pledges, but enforcement remains uncertain.
Congressional action underway
The House passed the Ratepayer Protection Act of 2026 by a 417-3 vote in September, though the Senate rejected it as too weak because it only required state regulators to "consider" protections rather than mandating them.
A separate measure, the Bipartisan American Affordability and Jobs Act of 2026, contains stronger provisions. Section 2107 would require FERC to mandate that AI data centers pay both embedded and incremental transmission costs, explicitly prohibiting cost allocation to other customers. The provision also requires continued cost recovery even if data centers stop purchasing power.
That bill faced Senate consideration during the lame-duck session but drew opposition from environmental and consumer groups due to other provisions in the legislation.
Unresolved questions
Four issues remain outside the consensus framework, according to the Brookings analysis. These include how to treat "behind the meter" projects that bring their own power but may still rely on grid backup, whether to require renewable energy sources despite construction delays, how to address wholesale price increases that affect all ratepayers, and whether temporary moratoria make sense while protections are designed.
Sen. Bernie Sanders and Rep. Alexandria Ocasio-Cortez support moratoria, and several states are considering or have implemented them. The scholars acknowledge this approach buys time for proper planning but could slow technology deployment.
Why it matters
While policymakers lack tools to prevent a potential AI investment crash or mitigate its macroeconomic consequences, they can act now to protect ordinary electricity customers from subsidizing infrastructure that may become obsolete. Without federal intervention, residential and small business ratepayers risk paying twice: first for overbuilt capacity and again for stranded assets when the bubble deflates. The incoming Congress has an immediate opportunity to establish mandatory protections before more infrastructure is built.
These details were first reported by Mark MacCarthy and David Klaus in commentary published by the Brookings Institution.