The Senate just stalled a bill that asks states to “consider” making Big Tech pay for the grid. The meter is already running, and the party that promised to protect the ratepayer has handed him a handshake.
By Michael Phillips | Riptide
On Wednesday, the Senate failed to take up a bill almost nobody in the House had been willing to vote against.
The Ratepayer Protection Act passed the House 417–3. In the Senate, it drew 57 votes on a procedural motion that needed 60, with four Democrats (Maggie Hassan, Amy Klobuchar, Jon Ossoff and Raphael Warnock) crossing over. It was the second time the chamber turned the bill back this month, after an objection to a unanimous-consent request from Ohio Republican Jon Husted.
Majority Leader John Thune called it “common sense” and accused Democrats of denying Republicans a win in an election year. Minority Leader Chuck Schumer called it “toothless.” Both men were describing the same bill accurately, and that is the problem.
“Both men were describing the same bill accurately, and that is the problem.”
What the bill actually does
The bill tells state utility regulators to consider adopting a federal standard that makes the largest power users, particularly hyperscale data centers and other “large-load” customers, carry the cost of the grid upgrades they trigger instead of spreading it across household bills. States would have to begin that process within a year and finish it within two. They would not have to adopt anything.
That restraint follows from how the system works. Under a 1978 federal law, state utility commissions set their own rates, so a bill that does not preempt them can only ask. It is the legislative twin of the White House’s Ratepayer Protection Pledge, announced in March, and it shares the pledge’s central weakness: neither one has an enforcement mechanism.
The pledge is not nothing. Seven AI companies (Amazon, Google, Meta, Microsoft, OpenAI, Oracle and xAI) signed on to build, bring or buy their own new generation and pay for the delivery infrastructure their facilities require. The White House text goes a step further than most corporate promises: the companies agree to pay for that power and infrastructure whether or not they use it. That take-or-pay structure is the right idea, because it moves the risk of a data center that never fills its racks from the household to the developer.
In its July expansion, the White House announced 55 utilities, 106 cooperatives, 28 developers and 23 governors, every one of them a Republican. The coalition has kept growing: the White House’s pledge page, updated September 30, now claims more than 300 organizations, including 250-plus utilities and cooperatives and 45 developers, with the governors still at 23. Democratic governors Wes Moore of Maryland and Josh Shapiro of Pennsylvania, who had been involved in an earlier White House agreement, were not on the list. Their offices said they were never invited to sign.
It remains voluntary. Enforcement belongs to state commissions, and one analyst who follows the issue says the “bring your own power” piece has been the hardest to put into practice, because utilities disagree about whether large customers can buy generation directly. The White House’s own pledge page now acknowledges that skeptics, including the Wall Street Journal, doubted it could be enforced.
The meter doesn’t wait
“While Washington debates verbs, the region that hosts the world’s largest data-center cluster has been paying.”
While Washington debates verbs, the region that hosts the world’s largest data-center cluster has been paying. PJM Interconnection, the grid operator serving roughly 67 million people in 13 states and D.C., sets the price of standby capacity in an annual auction. Here are the results:
| Delivery year | Capacity price ($/MW-day) |
|---|---|
| 2024/25 | $28.92 |
| 2025/26 | $269.92 (BGE zone in Maryland: $466.35) |
| 2026/27 | $329.17 |
| 2027/28 | $333.44 |
| 2028/29 | $325.00 |
“A rise from $28.92 to $325 per megawatt-day does not mean anyone’s bill rose elevenfold.”
The last three prices sit at a cap that grew out of a settlement with Gov. Shapiro. PJM’s independent market monitor, the grid’s own referee, attributes the run-up primarily to demand from new and proposed data centers. Monitor Joseph Bowring told Utility Dive that data centers account for $6.3 billion, or 38%, of the $16.4 billion in charges from the latest auction, and $29.4 billion, or 46%, of the $63.6 billion across the last four. PJM’s own report puts the latest auction’s total at $16.4 billion as well, with the caveat that it is not a bill: load hedged through self-supply or bilateral contracts is not exposed to the clearing price.
Capacity is also only one line on a household’s bill. A rise from $28.92 to $325 per megawatt-day does not mean anyone’s bill rose elevenfold. Utilities also buy energy and pay for transmission and distribution, and how much of a capacity increase reaches a given customer depends on the state, the utility, and how it procures power. In Illinois, the Citizens Utility Board says capacity makes up roughly 20% of what ComEd customers pay for power, and that their power price is about 50% higher than two years ago.
Bowring’s remedy is the cleanest statement of the conservative principle in this entire debate: take data-center load out of the shared capacity market and have it procure its own, so that data centers pay for the capacity they cause. He has also said the PJM board’s plan keeps that load in the shared auction.
Reliability is slipping alongside price. The 2028/29 auction cleared 6,831 megawatts short of PJM’s reliability requirement, the second shortfall in a row. PJM’s board projects roughly 70 gigawatts of new large-load demand by 2038 against about 15 gigawatts of generation retired since 2022. Its answer was a one-time “backstop” procurement of the missing 6.8 gigawatts on 15-year contracts, due to open on September 30. On September 29, FERC accepted the proposal but suspended it for five months, until February 28, 2027, and PJM pulled the auction. The next regular auction, for 2029/30, is set for early December and comes first.
Maryland’s governor announced on July 15 that the price cap “saved” ratepayers $13.3 billion. That figure is the administration’s own estimate of costs avoided, measured against what the auction would otherwise have cost. Nobody was mailed a check, and the release does not spell out the baseline. It also deserves a second look from the other direction. In its report on the same auction, PJM states that a price cap set below the level the market’s own demand curve would produce can reduce investment and therefore supply, which is the thing the region is short of.
How big is it, really?
The best sober estimate comes from the Federal Reserve Bank of Dallas, whose economists describe their work as the first attempt to quantify how the data-center build-out feeds through to retail power prices and then to inflation. It was posted in March and re-circulated through the San Francisco Fed’s research network on September 1.
The headline is modest: under plausible assumptions, data centers add between 0.04 and 0.13 percentage points to annual PCE inflation by 2030. In the authors’ mid-capacity scenario where use is concentrated in peak hours, the effect is 0.05 points this year and 0.13 by 2030. They call a scenario in which every proposed project connects and runs flat out, which adds more than a full point, “highly implausible,” and say they expect outcomes between their low and mid cases.
Two details matter more than the headline. First, the authors note that large-load connection fees tend to be modest, which leaves existing customers carrying part of the grid expansion. Second, the result is fragile on the supply side: if solar and onshore wind fall well short of what has been proposed, the effect nearly doubles, and beyond 2030 the authors expect it to climb sharply absent substantial new gas and nuclear generation.
So the macro number is small, and the political number is large. A tenth of a point of inflation probably does not decide a Fed meeting by itself. A utility bill that jumps every year can become a campaign issue, because voters see it every month.
“A tenth of a point of inflation probably does not decide a Fed meeting by itself. A utility bill that jumps every year can become a campaign issue, because voters see it every month.”
The party’s fault line
Voters have noticed. Gallup found in March that 71% of Americans oppose an AI data center in their area, 48% strongly. That is a higher opposition rate than the 53% who say they oppose a nuclear plant nearby. Republicans are less hostile than Democrats, 39% strongly opposed to 56%, but even among Republicans roughly six in ten oppose a local project. Among opponents who explained their reasons in a follow-up survey, 15% cited higher utility bills or energy costs, behind a larger cluster of concerns about water, power use and pollution.
Republican officeholders are repositioning accordingly, and not uniformly. Texas Gov. Greg Abbott has directed the state to audit tech companies’ use of the grid and said he will work with the legislature to repeal their tax breaks. Florida Gov. Ron DeSantis signed a law in May letting communities reject data-center development and barring costs from being passed to consumers. Ohio’s Vivek Ramaswamy, who once talked up a friendlier regulatory climate for data-center and crypto-mining investment, now promises an executive order halting new construction unless communities receive tax breaks and environmental protections. On September 16, Missouri Sen. Josh Hawley announced a bill to bar data centers from claiming Opportunity Zone tax incentives, calling their use of the program “corporate welfare.” With Democrat Richard Blumenthal, he also introduced the GRID Act in February, which would go much further than the Senate’s bill by requiring new data centers to run on power sources separate from the grid.
The president sits at the other end of that spectrum. He has said the industry “could use a little public relations help,” and that if he were a mayor or a governor he would want a data center badly. Husted, who tops CQ Roll Call’s list of the most vulnerable senators and faces Democrat Sherrod Brown in a race where data centers have become a flashpoint, has led the push for the Ratepayer Protection Act. The NRSC, in a private memo to AI companies obtained by Axios, says Democrats have made data centers central to the campaign against Husted and that it is working. Brown has spent millions on ads calling Husted “the face of data centers in Ohio.”
Run the swapped-jerseys test. Suppose a Democratic administration answered public anger over drug prices with a voluntary pledge signed mostly by friendly governors, and its congressional allies followed with a bill telling states to “consider” the matter. Republicans would have a word for it, and it would be the one Schumer used on Wednesday.
Now flip it back. Schumer’s critique is fair on substance, and Democrats do have a named alternative. New Mexico’s Martin Heinrich objected to fast-tracking the Republican bill, then sought consent to pass his own GRID Savings Act (S. 5199, introduced July 30), which takes a different jurisdictional route. Rather than asking state regulators to consider a standard, it would amend the Federal Power Act to put the interconnection of large new loads, 150 megawatts or more, to interstate transmission under FERC’s jurisdiction and direct the commission to write binding rules within a year. Those rules would assign the costs of interconnection and direct-assignment facilities entirely to the customer, allocate network-upgrade costs to large loads with credits against their transmission charges, and require financial security so that other customers are protected if a project fails. It would not apply in Texas’s ERCOT grid, and as a transmission bill it does not reach the capacity-auction costs described above. (It should not be confused with the Hawley–Blumenthal GRID Act, which is a separate-power mandate.) Each route has a price. Heinrich’s moves more of the fight into federal transmission regulation; the Republican bill leaves it with the state commissions that set retail rates but gives them nothing binding. The real fight is in the state commissions of PJM’s 13 states and D.C. and at the Federal Energy Regulatory Commission, and it is not clear that either party’s federal message has much to say to them. As Semafor’s David Weigel put it, both parties have turned populist on this issue mainly in the places where they are trying to take power rather than where they already hold it.
What a conservative answer looks like
A serious center-right position doesn’t require a moratorium or a miracle. It rests on principles the movement already holds.
Cost causation. The party that creates a cost pays it. That means separate large-load tariffs with take-or-pay terms, real connection charges instead of modest fees, and, as the PJM market monitor urges, dedicated procurement for data-center load so it never again sets the price households pay.
“The party that creates a cost pays it.”
Build, don’t just bill. Making data centers pay is only half the job, because supply is the binding constraint. Faster interconnection, permitting that treats gas and nuclear as the firm capacity they are, and keeping existing plants online do more for the 2030 inflation number than any pledge. This is also where the politics gets hard: Gallup finds 53% of Americans oppose a nuclear plant in their area, and Americans who resent data centers are not eager to host the power plants that would serve them.
“Making data centers pay is only half the job, because supply is the binding constraint.”
Separate true subsidies from ordinary tax treatment. The strongest case is against benefits designed for another purpose, like the Opportunity Zone eligibility Hawley is targeting, and against local property-tax abatements for a business that has already chosen the site. The Cato Institute, hardly an enemy of the industry, draws the line there and argues that sales-tax exemptions on equipment and full expensing are ordinary tax treatment that data centers share with other capital-intensive businesses, not subsidies. Democrats Ron Wyden and Mark Warner would deny the federal bonus depreciation anyway. Texas lawmakers are asking whether the state’s sales-tax exemption still serves its original purpose. The right is split on where the line falls, which makes it the most interesting plank.
Say the quiet part. Some of this will cost money. PJM wants a mandatory registry of large loads that can be curtailed in an emergency and says it will compensate them when they are, and the consumer advocates at Illinois’s Citizens Utility Board want data centers that don’t bring their own capacity to be interruptible. Voters can handle a candid explanation of the trade-off. What they will not forgive is an explanation delivered after the bill arrives.
The case against
“A pledge cannot decide it.”
The argument has fair critics. Data centers bring jobs and tax revenue, and 55% of the Gallup respondents who favor a local project cited job opportunities. A Lawrence Berkeley National Laboratory study with the Brattle Group found that from 2019 to 2024, states with the highest load growth saw real electricity prices fall, because fixed grid costs were spread over more kilowatt-hours. The lab itself cautions that it is unclear whether that relationship holds under much larger, nationwide load growth, and consultancy E3’s review attributes roughly half of PJM’s 2024/25 to 2025/26 capacity price jump to load growth, so PJM looks different from the national average. Consumer advocates, meanwhile, argue the opposite of the deregulators: that voluntary measures will not hold and that only binding rules and an end to regional price caps will protect households.
Both camps agree on one thing. A pledge cannot decide it. The test for the 119th Congress, the governors who signed the pledge, and the commissions that actually set retail rates is whether those promises become enforceable tariffs and cost-allocation rules before another round of data-center demand is spread across the grid.
Sources: Roll Call (Sept. 30, 2026) on the Senate cloture vote and the bill’s terms; NOTUS/Washington Sun (July 23, 2026) on the expanded Ratepayer Protection Pledge; White House fact sheet and pledge text (March 2026) and pledge page (updated Sept. 30, 2026); PJM, 2028/2029 Base Residual Auction Report (July 14, 2026); Citizens Utility Board of Illinois (July 15, 2026) on auction prices; Utility Dive reporting on PJM’s independent market monitor, Monitoring Analytics (July 2026 and later), on data-center cost attribution and on the backstop proposal; PJM Inside Lines and RTO Insider (Sept. 29–30, 2026) on FERC’s order; Office of Governor Wes Moore press release (July 15, 2026); Babst Calland summary of the PJM Board’s July 27, 2026 decisional letter; Dallas Fed Economics, “Data center boom expected to raise electricity component of PCE inflation” (Kay, Kilian and Taylor, March 5, 2026; reposted by the San Francisco Fed on Sept. 1, 2026); Gallup, “Americans Oppose AI Data Centers in Their Area” (May 13, 2026; March 2–18 survey of 1,000 adults, margin of error ±4 points); Semafor (Aug. 19, 2026) on Republican candidates’ shifting positions; Axios (Aug. 19, 2026) on the NRSC memo; text of S. 5199, the GRID Savings Act (GovInfo); Fox News (Sept. 16, 2026) on Sen. Hawley’s Opportunity Zone bill; Sen. Hawley’s Feb. 11, 2026 release on the GRID Act; Fox local-station coverage and Latitude Media on Sen. Heinrich’s objection and the GRID Savings Act; Cato Institute (Sept. 10, 2026) on data-center tax treatment; Texas Public Policy Foundation reporting on the Texas Senate hearing; American Public Power Association summary of the Lawrence Berkeley National Laboratory study; Brookings and The New Republic on the pledge’s enforceability and the bill’s structure.
Keep This Reporting Free
If this work matters to you, please consider supporting it.
Your contribution helps fund independent reporting across our entire network.
Discover more from RIPTIDE
Subscribe to get the latest posts sent to your email.
