Senator Adam Schiff has filed legislation to make large data centers pay for the power and grid upgrades they require, a proposal that could change the economics of AI infrastructure in the places already feeling pressure from rising electricity demand.
Schiff's Energy Cost Fairness and Reliability Act puts a simple political question in front of the AI buildout: if a hyperscale data center needs a new supply of electricity, who should pay for it? The bill would apply to data centers over 50 megawatts and require those facilities to secure their own power and cover the grid upgrades needed to serve them, rather than letting those costs spill into ordinary utility bills.
That distinction matters because data centers already pay for the electricity they consume. The dispute is over the infrastructure around that consumption: new generation, transmission lines, interconnection work, and reliability costs that can be spread across a broader customer base when utilities expand to serve big new loads. According to Bloomberg Law, Schiff's proposal is designed to turn the White House's March ratepayer protection pledge into a binding legal requirement for large facilities.
The bill arrives at a moment when data center politics have moved from local zoning fights into federal energy policy. AI companies and cloud providers are racing to secure sites near power, fiber, water, and tax incentives, while regulators are being asked to decide whether that growth creates public benefits large enough to justify shared costs. Schiff's answer is clear: growth can continue, but the bill should follow the demand.
The Pressure Is Moving From Compute To Power
For hyperscalers, the practical effect would be a higher up-front cost for new capacity in constrained markets. A facility that can once lean on a utility's broader capital plan may now need dedicated power purchase agreements, on-site generation, private transmission arrangements, or a more expensive bespoke tariff. That does not make AI infrastructure impossible. It does make cheap grid access harder to treat as a background assumption.
This is why the bill matters beyond Washington. Site selection for AI data centers is already becoming an energy strategy as much as a real estate strategy. Operators that can find regions with spare generation, faster interconnection queues, or utilities willing to structure large-load contracts will have an advantage over rivals trying to build in crowded markets where households and small businesses are already sensitive to rate increases.
The politics are not one-sided. Supporters argue that the rule is basic fairness: the customer creating the incremental demand should fund the incremental supply. Critics will warn that tougher cost allocation rules could slow domestic AI deployment, raise the cost of cloud services, or push investment toward countries that move faster on energy permitting. Both arguments have weight, which is why the bill will likely face heavy scrutiny from utilities, developers, state regulators, and data center trade groups.
The measure also comes without Republican co-sponsors for now, although the broader idea has bipartisan traction. President Donald Trump's Ratepayer Protection Pledge, signed in March by major technology and AI companies, called on hyperscalers to build, bring, or buy the power needed for new data centers and to pay for related grid infrastructure. Schiff's legislation is an attempt to move that promise from a voluntary commitment into federal law.
States Are Already Testing The Model
The federal bill is not arriving in a vacuum. Ohio has been wrestling with data center tariffs after AEP Ohio won approval for a structure requiring large customers to make stronger commitments tied to projected demand. Colorado lawmakers introduced large-load data center legislation aimed at making operators cover electricity and grid investment costs, while Xcel Energy has proposed safeguards to keep existing customers from subsidizing new high-demand users.
Virginia, the country's most important data center hub, shows why the debate is difficult. The state has benefited from years of investment, tax revenue, and dense infrastructure around Northern Virginia, but it has also faced mounting pressure over transmission needs and long-term electricity demand. State regulators have moved toward separate large-load treatment so data centers pay rates more closely tied to their specific costs.
Utilities have their own problem to solve. Clearer cost allocation can protect residential customers and reduce political blowback, but it also makes planning more complex. A utility that knows a massive customer must pay for dedicated upgrades has less risk of stranded costs, but it still has to coordinate generation, transmission, reliability planning, and local approvals in a grid that was not built for sudden clusters of 24-hour compute demand.
Industry groups are already pushing back on the broader narrative that data centers are the main cause of rising electricity bills. The Data Center Coalition has pointed to research arguing that retail rates have largely tracked inflation and that cost pressures vary widely by region. That critique will matter in hearings because a national rule can look blunt when electricity markets, utility structures, and local grid conditions differ so much from state to state.
The near-term outcome is likely to be more negotiation, not an immediate halt to construction. Hyperscalers will try to lock in long-term contracts, utilities will design more specialized tariffs, and states will keep experimenting with rules that protect ratepayers without chasing away investment. For investors and operators, the message is already clear. AI infrastructure is no longer only about chips, land, and fiber. Power access, and who pays for it, is becoming one of the defining constraints on the next phase of the market.
Also read: Figure AI turns a robot sorting demo into a test of labor economics • ByteDance's Lance Puts Open, Efficient Multimodal AI Within Reach • Alibaba's Qwen team pushes forward with Qwen 3.7 release amid export-control headwinds
