Reporting / Energy & power

Investigation · Permitting & influence

AI data centers: What did the Senate actually vote for?

A brightly lit aisle between rows of black computer cabinets at Oak Ridge National Laboratory
Frontier supercomputer, Oak Ridge National Laboratory, June 2022. File photo for context, not a facility covered by the investigation.
Photo: OLCF at ORNL · CC BY 2.0 · Resized and cropped for display.

In a September 2026 Quinnipiac poll of 1,202 U.S. adults nationwide, 72% said they would oppose an AI data center in their own community. Weeks later, senators advanced a broader permitting bill that could change who pays for grid upgrades and how long some permits can be challenged.

Politics Interpreted · Reporting through November 18, 2026
This is a dated investigation, not a live legislative-status tracker.

Read the analysisInspect the sourcesUpdates & corrections

The essential distinction

Three numbers. Not one verdict.

The debate over data-center development combines questions about local control, electricity costs, union jobs and political influence. The records below illuminate different parts of that debate. They do not, on their own, explain why any particular senator voted yes.

What people said

An AI data center in your community?

Oppose72%
Support21%
No opinion7%

Quinnipiac University poll · 1,202 U.S. adults nationwide, Sept. 24–27, 2026. Not a poll on the bill.

What the Senate voted on

End debate on the amendment?

Yes 70

49 R · 20 D · 1 I

No 27

2 R · 24 D · 1 I

Republicans (R)Democrats (D)Independents (I)

One dot per senator who voted. Not voting: 2 R, 1 D. Roll call 264 · Nov. 16, 2026. Not final passage.

Look up how each senator voted

Cropped excerpt of the Data Center Coalition’s filed lobbying report showing $480,000 in lobbying expenses

What the coalition reported

$480,000 in quarterly lobbying

The Data Center Coalition reported this total for all its federal lobbying in Q3 2026, not spending on this bill alone.

Original LD-2 filing · Cropped to show the expense line.

What did the public actually answer?

Quinnipiac University surveyed 1,202 U.S. adults nationwide from September 24 to 27, 2026. Asked about an AI data center in their own community, 72% opposed it, 21% supported it and 7% gave no opinion. The poll had a margin of error of ±3.5 percentage points. It measured views on a hypothetical nearby project, not on a specific site or on the Bipartisan American Affordability and Jobs Act, or BAAJA.

That distinction does not make public concern irrelevant. It tells us the question policymakers still have to address: how are local concerns being weighed in a national permitting deal?

What did the Senate actually do?

The November 16 roll call records a 70–27 vote on cloture for the substitute amendment. Cloture limits debate and moves consideration forward. It is not the same as voting to pass the bill.

On November 18, the Senate adopted the substitute amendment, 69–27, then voted 65–31 to limit debate on the amended bill. Neither vote was final passage.

This is a snapshot of what the Senate had done through November 18, not a claim that the bill became law.

Voters and their representatives

Voters were wary. The Senate moved ahead.

Quinnipiac’s registered-voter results put that tension in sharper focus. Of 1,032 self-identified registered voters surveyed nationwide September 24 to 27, 2026, 73% opposed an AI data center in their community and 21% supported one. On November 16, 70 senators voted to end debate on the permitting amendment and 27 voted against ending debate.

Pennsylvania puts names to the vote

In a July 2026 Quinnipiac poll of 895 Pennsylvania registered voters, 74% opposed a nearby AI data center and 16% supported one. Both of the state’s senators, Democrat John Fetterman and Republican Dave McCormick, voted yes on the November 16 procedural motion.

Those are different questions. The polls did not ask voters whether they supported this bill, and the Senate did not vote to approve a specific data center. Still, both Pennsylvania senators helped advance a deal that changes how projects are reviewed. They should explain what weight they gave to the concerns voters expressed about projects close to home.

What does the money establish?

The Data Center Coalition’s third-quarter federal lobbying disclosure reports $480,000 in total expenses and names BAAJA among multiple issues. The amount covers the reporting period’s lobbying work, not a separately priced campaign for this bill.

The filing lists government bodies contacted. It does not identify individual senators, disclose what each meeting covered, or establish a connection between a particular request and a particular vote. Lobbying expenses are also different from campaign contributions.

Why the trade-offs deserve scrutiny

IBEW makes the case that permitting reform can advance energy infrastructure and employment. The bill’s authors also point to protections intended to keep certain new data-center grid costs from being shifted to other customers. The substitute’s Section 2107 would require utilities to recover certain incremental costs of serving qualifying new data centers from those customers. Supporters argue that this protects other ratepayers. Whether it would do so in practice depends on how those costs are measured and recovered.

NCAI warns that the substitute leaves threats to Tribal sovereignty, sacred places and judicial remedies unresolved. For example, Section 1134 would shorten the window for certain court challenges involving projects covered by the federal FAST-41 permitting process from two years to 150 days. That is a specific change to how some permits can be challenged, not blanket approval for data centers. The relevant question is not simply whether a process is faster, but what safeguards remain.

Read the authors’ release and linked substitute text alongside the union and Tribal statements in the source ledger. The bill language and the advocacy for or against it are different kinds of evidence.

Who did you meet with, what did they ask for, and what changed in the bill because of it?

That is our editorial demand for transparency. It is not an allegation that a contribution bought a vote. The documents cited here do not establish that.

Separate the record from the inference

What we know.
What we don’t.

Documented

The official roll call records the procedural vote. The coalition’s disclosure names BAAJA and lists the Senate, House and Department of Energy among bodies contacted on its energy issues.

IBEW supports the legislation, citing energy infrastructure and work for its members. NCAI opposes it, warning of weakened Tribal sovereignty and cultural protections.

Not established by these records

The lobbying filing does not identify which senators were contacted or allocate spending to BAAJA. It does not establish why a senator voted yes.

A local-opposition poll cannot be presented as a national referendum on this legislation. Lobbying expenses are not campaign contributions.

Why supporters and opponents see different stakes

Supporters argue that faster permitting can move power projects forward and support union jobs. The Democratic authors also describe provisions intended to keep certain data-center grid costs off household bills. Opponents focus on environmental review, access to courts, and Tribal participation. These are competing arguments about the trade-offs, not interchangeable claims. Read each organization’s own statement below.

Original documents · Annotated

The source ledger.

Original sources, with a note on what each does and does not prove.

National public opinion

Quinnipiac poll

September 30, 2026 · Question 22

Establishes the survey’s community-level data-center question and 72% opposition among adults. It does not measure support for the permitting bill.

Read the poll release · Open original results PDF
Original Quinnipiac question 22: support or oppose building an AI data center in your community. Total adults: support 21%, oppose 72%, don't know/no answer 7%.
Authentic crop from the results PDF, question 22. Open the original for sample details, methodology and the full table.
Official voting record

Senate roll call 264

November 16, 2026

Records 70 yeas and 27 nays on cloture for Capito amendment 6883 to H.R. 2347. A vote to end debate is distinct from final passage.

Inspect the official roll call
Federal lobbying disclosure

Data Center Coalition LD-2

Q3 2026 · Filed October 18

Reports $480,000 in overall lobbying expense and names BAAJA among its issues. The filing lists government bodies, not individual senators contacted.

Read the original disclosure
Legislation & authors’ position

The substitute amendment

Released November 7, 2026

The authors’ announcement links the amendment text and their explanation of its changes. Their affordability and grid-cost arguments should be read as the authors’ case for the legislation.

Read the release and linked bill text
Support / Labor

IBEW endorsement

November 6, 2026

The electrical workers’ union argues that permitting reform can unlock energy development and protect members’ livelihoods. This establishes its position, not guaranteed economic outcomes.

Read IBEW’s statement
Opposition / Tribal sovereignty

NCAI’s objections

November 8, 2026

NCAI says the substitute fails to resolve concerns about Tribal consultation, sacred places and judicial remedies. Read its specific objections rather than treating all opposition as one argument.

Read NCAI’s statement
Politics Interpreted owl emblem

Forthcoming · Video

The 90-second breakdown

The visual edit is prepared. Narration is in production. The finished pilot will accompany the investigation here.

Updates & corrections

The record of this story.

November 19, 2026 · Source and context note

Prepared from the investigation’s November 18 reporting snapshot. The visual pilot is prepared; narration is pending. No subsequent legislative status is asserted here.

No correction entries have been added. Material corrections will identify what changed, why, and the supporting source. New developments will be dated separately.

Read our corrections standard