A House subcommittee is expected to advance legislation that would require tech companies to bear the electricity costs of running the data centers that power their artificial intelligence operations. The bill represents one of the first direct congressional attempts to attach an energy price tag to AI infrastructure — and to decide who, specifically, should pay it.
What the Bill Would Do
A data center is, at its most basic level, a warehouse full of computers drawing continuous power. AI workloads — training large models and running inference for users — are among the most electricity-intensive tasks those computers perform. Today, that energy cost is distributed across utilities, grid operators, and in some cases ratepayers in the regions where those facilities sit. The legislation moving through the House subcommittee would redirect that financial responsibility to the tech companies operating the data centers themselves.
The distinction matters because it changes the incentive structure. When a company does not pay directly for the electricity its infrastructure consumes, it has less reason to optimize for efficiency or site its facilities where the grid can absorb the load most cleanly. Making the operator pay shifts the calculation.
Why This Is Moving Now
The push comes as AI demand has driven a measurable surge in data center construction and power consumption. Grid operators in multiple regions have flagged capacity concerns tied directly to the expansion of large-scale compute facilities. Congress is responding to a practical question that utilities and regulators have been unable to resolve on their own: who is responsible for the energy a data center uses?
The House subcommittee advancing the bill is the body that will determine whether this question gets a legislative answer or remains a patchwork of local utility negotiations.
What Comes Next
Subcommittee advancement is an early step. Legislation must pass the full committee, then clear both chambers, before it becomes law. Tech companies — which have been among the largest builders of new data center capacity — would face a direct operating cost increase if the bill reaches the president's desk. How that cost gets reflected in pricing for AI services, or in decisions about where and how to build future infrastructure, depends on details the source does not yet specify.
The bill's trajectory will be worth watching closely, both for what it signals about congressional appetite to regulate AI's physical footprint and for how the industry responds to the prospect of paying what the grid already knows it owes.