The House Voted 417-3 to Make Data Centers Pay Their Own Grid Costs. The Word Doing the Work Is “Consider.”
H.R. 9340 passed the House by the widest possible bipartisan margin on September 16th. Six days later a Senate Democrat blocked it outright, and the reason why is buried in a single verb the bill uses forty times.
Four hundred seventeen to three. That's how the House voted on September 16th to pass the Ratepayer Protection Act, H.R. 9340, and in a Congress that can't agree on what day it is, a margin like that usually means the bill does something everyone already wanted and nobody had to give anything up for. The three no votes — Summer Lee, Delia Ramirez, and Rashida Tlaib, all progressive Democrats who argued the bill doesn't go far enough — tell you more about its actual reach than the 417 yes votes do.
The bill, sponsored by Rep. Gabe Evans, a Colorado Republican, and Rep. Kathy Castor, a Florida Democrat, is aimed at a real and well-documented problem: utilities building new transmission, substations, and generation to serve a single 300-megawatt data center campus, then spreading the depreciation and financing cost of that build across every residential and small-business ratepayer on the system, whether or not the data center ever shows up to use the power. It's the same dynamic Texas regulators were reacting to when they adopted their large-load interconnection deposit rule this month — cost risk sitting with the utility and its captive customers instead of the developer who created it. H.R. 9340 tries to do at the federal level what Texas and a handful of other states have started doing on their own.
What the bill actually requires
For any new large load of 100 megawatts or more, the bill directs state utility commissions to adopt standards requiring the load to cover the full, incremental cost of whatever generation, transmission, or distribution upgrade its interconnection requires — not a blended, averaged cost, the actual cost its own connection caused. It requires financial assurances from the customer before the utility starts construction, and it requires guaranteed cost recovery for the utility if that customer walks away from its power contract early, so a canceled or downsized project doesn't leave stranded infrastructure on the ratepayer base. Separately, it directs FERC to write its own rules on large-load cost allocation, financial security, and cost-estimate disclosure for loads that interconnect directly at the transmission level, a distinct track from the state-level retail rules.
Read only that paragraph and the bill sounds like a mandate. It isn't one. Every one of those requirements is something state commissions are directed to “consider” adopting, not something they're ordered to put in place. The bill's actual command is procedural: states that don't already have comparable cost-allocation standards on the books have two years to hold a hearing and make a determination one way or the other. A state can hold that hearing, weigh the evidence, and decide against adopting any of it, and be in full compliance with federal law. States that can show they've already got something equivalent — and several already do — don't have to hold the hearing at all.
The industry split nobody expected
The version that passed committee in July isn't the version Evans and Castor introduced. The original bill applied its cost-allocation standard to any large load, full stop — crypto mining operations, reshored semiconductor and battery plants, hydrogen electrolyzers, data centers, whatever cleared the 100-megawatt threshold. The Energy and Commerce Committee narrowed that scope to single out data centers specifically, and the rewrite split an industry coalition that's normally been unified against this kind of bill. The Data Center Coalition, whose members include Amazon, Google, and Microsoft, came out against the narrowed version, arguing it singles out one industry while leaving households exposed to the same cost-shifting risk from every other kind of large load. The American Iron and Steel Institute and the Industrial Energy Consumers of America, by contrast, both endorsed the narrower bill — because it means their members' new plants and expansions won't face the same disclosure and collateral requirements data centers now will.
That's an unusual alignment. Heavy industrial users and data center operators have mostly stood on the same side of large-load interconnection fights, both wanting fewer strings attached to getting connected fast. Here, the industrial trade groups got what they wanted by letting data centers take the hit alone, and the Data Center Coalition is now making the same fairness argument — don't single us out — that ratepayer advocates have made about the industry for two years.
Why the Senate stopped it cold
Sen. Jon Husted, an Ohio Republican, tried to fast-track the bill through the Senate by unanimous consent on September 18th, pointing to the 417-3 House margin as proof it had the votes to pass through regular order anyway. Sen. Martin Heinrich, a New Mexico Democrat, objected and blocked it. His stated reason is the same gap described above: “It's not enough for us to tell states to consider making data centers pay for grid updates,” Heinrich said. “Rather than voluntary pledges or suggestions to states, Congress needs to pass real legislation with real teeth.” Husted's response was to point back at the vote count, which is true and also beside the point — 417 members can agree to ask states to think about something without agreeing on what happens if a state thinks about it and says no.
That leaves H.R. 9340 sitting in the Senate with no clear path: strong enough support to pass a floor vote eventually, not strong enough consensus to skip the process, and now an objection on record that a rewrite would have to answer. Bills with that profile can sit for a long time.
What this means for diligence, whether or not it passes
Here's the part that matters for anyone actually siting a load, independent of what the Senate does with this bill: the two-year hearing trigger only applies to states that don't already have comparable standards, and that exemption is doing real work right now. Colorado's Xcel Energy already carries collateral and minimum-bill terms in its large-load tariff filings. Texas just adopted its $50,000-per-megawatt refundable deposit rule. States that get there first set the terms their commissions will point to as already-adequate when H.R. 9340, or whatever replaces it, comes due. States that haven't started are the ones facing two years of open question marks on who eats the interconnection cost if a rewrite dies.
We treat this as evidence, not politics. In a Kill Test, the commercial-readiness domain already asks whether a state's utility has a published, applied cost-allocation and collateral framework for large loads, or whether that gets negotiated deal by deal with no public benchmark. A federal bill that may or may not pass doesn't change that question. It just confirms which answer regulators everywhere are converging on — and it's a good reason to check, before you option a parcel, whether the state you're siting in is one of the ones that's already written the rule down or one of the ones about to spend two years arguing over whether to.
Evaluating a specific parcel for a 50–300 MW load? GridPermit's Preliminary Site Kill Test grades the power, zoning, water, and commercial-readiness evidence before you buy the land.
Start a Qualification ReviewSee state-by-state coverage · Read the methodology · See a sample report