The House of Representatives passed the Ratepayer Protection Act by 417 votes to 3 on September 16, one of the most lopsided margins any bill has drawn in this Congress, and it targets the exact question the AI-power trade is built on: who bears data center power costs. The bill aims to stop large data centers from shifting the cost of new grid and generation infrastructure onto ordinary households, as the House Energy and Commerce Committee announced. For anyone holding independent power producers or nuclear-adjacent names, a near-unanimous vote converts a diffuse political risk into a specific one, with a 100-megawatt threshold and a named mechanism.

The bill does not force states to make data centers pay; it requires state utility commissions to consider adopting standards that would. That gap between a loud political signal and a soft legal requirement, on a bill the Senate has not taken up, is the whole story for investors trying to price what actually changed about who pays data center power costs.

The vote lifted reactor developers Oklo and NuScale far more than the broad uranium complex or the market, on the logic that data centers bearing their own power costs strengthens the case for dedicated generation. Data: 24/7 Wall St., market data · Chart: FinanceFeeds.

The Bill Codifies the White House Ratepayer Pledge Through PURPA

The bill works through an existing federal framework rather than creating new authority, drawing on Section 111(d) of the Public Utility Regulatory Policies Act of 1978. Under it, state commissions would have to consider a standard requiring data centers above the threshold to cover the full incremental cost of the generation, transmission and distribution built to serve them, and to post financial assurances before a utility starts the work, according to the sponsor’s office.

The measure codifies parts of the voluntary Ratepayer Protection Pledge that Amazon, Google, Meta, Microsoft, OpenAI, Oracle and xAI signed at the White House in March, under which the companies agreed to cover the data center power costs their facilities create.

The three no votes were progressive Democrats Summer Lee, Delia Ramirez and Rashida Tlaib, and Republican leaders moved the bill under suspension, a fast-track procedure requiring two-thirds support, which is part of why the tally ran so high. An earlier amendment narrowed the bill to apply only to data centers rather than every large electricity user, a change the Center for Data Innovation criticized for treating one industry differently from other 100-megawatt loads.

The 100-Megawatt Threshold and Who Pays Data Center Power Costs

The threshold is what gives the bill teeth for the AI buildout specifically. A single hyperscale data center routinely draws well past 100 megawatts, and the largest campuses run into the hundreds or thousands, so the standard reaches essentially every facility driving the current surge in electricity demand. That demand is not hypothetical, with one analysis putting the amount already added to consumer bills from data-center growth at roughly $23 billion, and Reuters reporting that households in the PJM grid region could face rate increases of up to 60% over five years as the buildout accelerates.

Data centers have become a live issue heading into the November midterms, with a cross-partisan backlash forming in communities from Virginia to Arizona, as CNBC reported. A 417-3 vote is what that backlash looks like once it reaches the House floor, and it explains why lawmakers fast-tracked a measure that, on its own, changes less about data center power costs than the margin implies.

Why “Consider” Is Not the Same as “Adopt”

The distinction between considering a standard and adopting it is where the market read gets complicated. PURPA obliges state commissions to take up and rule on the federal standard, but it leaves them free to reject it, so the bill guarantees a proceeding, not an outcome. Rate design would still be set state by state, which means the result could look very different across vertically integrated utilities, competitive wholesale markets and states with retail choice.

That softness is compounded by two things. Analysts told Utility Dive the bill “would largely reinforce” a shift already underway, since most states are adding large-load tariffs on their own, and that it is unlikely to clear the Senate before the midterms. The measure is a political milestone rather than a legal one, and pricing it as though it forces a nationwide change in data center power costs would overstate what a 417-3 vote on a “consider” mandate actually delivers.

Investor Takeaway

The 417-3 margin is a political signal, not a legal mandate, because the bill only requires states to consider the standard and leaves them free to reject it, so the near-unanimity overstates how much changes on the ground.

Oklo and NuScale Popped, Then Faded as Uranium Sat Out

The market response was concentrated and revealing. Reactor developers Oklo and NuScale jumped about 13% and 10% on the vote, far ahead of the uranium complex, on the logic that a data center facing the full cost of grid upgrades has a stronger reason to buy dedicated generation, and small modular reactor firms sell exactly that, as 24/7 Wall St. detailed.

The move faded quickly, with Oklo slipping back to about $37.82 the next session, and both names remain pre-revenue and down 36% to 49% for the year, so the jump reads as a policy-headline bounce rather than a business inflection. Fermi, another pre-revenue power-and-data-center play, sat near $4.93 and barely moved.

The independent power producers are the more fundamentally exposed group. Vistra, trading around $141.58, monetizes data-center demand through capacity auctions and long-dated hyperscaler contracts, including 4 gigawatts of nuclear under 20-year deals with Amazon and Meta, so a rule making hyperscalers bear their own data center power costs cuts in a nuanced way, raising the incentive to sign dedicated supply while capping the ratepayer-funded upside.

The spread of outcomes shows up across the scenario ranges, from Oklo’s $72 bull versus $22 bear case to Vistra’s $225 bull, $95 bear range and the wide band on Fermi’s $12 bull against a $2.50 bear, a company with no revenue and a $486 million loss.

Investor Takeaway

The rally was a sentiment move concentrated in reactor developers, since Oklo and NuScale popped while uranium miners barely budged, which tells you the market priced a policy narrative, not a change in earnings.