Evidence on influence. Perspective on progress.Independent · Source-linked · AI-assisted research
AI PUBLIC OPINIONFollow the influence.
Menu

Public opinion · What the surveys found

More Americans oppose
data centers near home.

Local opposition rose from 49% in early 2026 to 61% by summer. A majority in both party groups opposed new projects in their community.

Local opposition increased.

61% opposed

Up from 49% earlier in the year—about six in ten respondents by summer.

Majorities in both parties opposed.

54%Republicans
69%Democrats

Summer results. Groups include independents who lean toward that party.

Views of AI overall were steady.

Little change

39% expected a negative impact from AI, compared with 42% earlier. Annenberg found no statistically meaningful change.

Annenberg surveys, February–March and June–July 2026. Read the findings · Original survey tables. These surveys do not tell us whether a campaign changed anyone's mind.

See how each group changed

Survey details ↓

Choose a question and a group to explore the results. The full tables are also available below.

Annenberg / SSRS · AI7 · 2026

Oppose local data centers

0–100% scale

Whether respondents support or oppose new data centers being built in their area. Question & survey details ↓

Somewhat or strongly oppose (%). February–March compared with June–July 2026. Party groups include leaners.

All respondents

Up 12 points
Feb–Mar49%
Jun–Jul61%

Republicans

Up 15 points
Feb–Mar39%
Jun–Jul54%

Democrats

Up 12 points
Feb–Mar57%
Jun–Jul69%

Independents

Up 2 points
Feb–Mar51%
Jun–Jul53%

Annenberg reports a statistically significant overall increase. Party-level changes shown here are descriptive estimates, not independently tested significance claims.

All groups · Somewhat or strongly oppose (%) · n is the unweighted group sample
GroupFeb–Mar 2026Jun–Jul 2026Change
All respondents49%n = 133061%n = 1320Up 12 points
Republicans39%n = 51054%n = 515Up 15 points
Democrats57%n = 65469%n = 643Up 12 points
Independents51%n = 16253%n = 157Up 2 points

Changes are in percentage points: 49% to 61% is an increase of 12 points. Party changes are descriptive; subgroup uncertainty is not shown.

Uncertainty & full-sample margins

Full-sample margins: ±3.5 pp (spring), ±3.48 pp (summer), at 95% confidence. These are not subgroup margins.

Published overall change: +11.8 pp; 95% confidence interval +6.9 pp to +16.6 pp. The publisher uses unrounded estimates; the displayed percentages are rounded. Change analysis · PDF p. 14 ↗

Annenberg / SSRS · AI7 · 2026

Support local data centers

0–100% scale

Whether respondents support or oppose new data centers being built in their area. Question & survey details ↓

Somewhat or strongly support (%). February–March compared with June–July 2026. Party groups include leaners.

All respondents

Down 7 points
Feb–Mar21%
Jun–Jul14%

Republicans

Down 9 points
Feb–Mar28%
Jun–Jul19%

Democrats

Down 6 points
Feb–Mar16%
Jun–Jul10%

Independents

Down 5 points
Feb–Mar19%
Jun–Jul14%

Support fell over the same two waves. Neutral responses remain in the denominator; support is not calculated as 100 minus opposition.

All groups · Somewhat or strongly support (%) · n is the unweighted group sample
GroupFeb–Mar 2026Jun–Jul 2026Change
All respondents21%n = 133014%n = 1320Down 7 points
Republicans28%n = 51019%n = 515Down 9 points
Democrats16%n = 65410%n = 643Down 6 points
Independents19%n = 16214%n = 157Down 5 points

Changes are in percentage points: 49% to 61% is an increase of 12 points. Party changes are descriptive; subgroup uncertainty is not shown.

Uncertainty & full-sample margins

Full-sample margins: ±3.5 pp (spring), ±3.48 pp (summer), at 95% confidence. These are not subgroup margins.

Published overall change: -6.9 pp; 95% confidence interval -10.8 pp to -3.1 pp. The publisher uses unrounded estimates; the displayed percentages are rounded. Change analysis · PDF p. 14 ↗

Annenberg / SSRS · AI3 · 2026

Expect a negative impact from AI

0–100% scale

Whether AI's effects on the United States over the coming decade will be positive, negative, or a mixture. Question & survey details ↓

Somewhat or very negative expected impact (%). February–March compared with June–July 2026. Party groups include leaners.

All respondents

Down 3 points
Feb–Mar42%
Jun–Jul39%

Republicans

Down 2 points
Feb–Mar38%
Jun–Jul36%

Democrats

Down 4 points
Feb–Mar45%
Jun–Jul41%

Independents

Up 4 points
Feb–Mar42%
Jun–Jul46%

Annenberg describes the overall 42% to 39% change as statistically unchanged. The local data-center opposition trend does not establish a simultaneous broad deterioration in AI attitudes.

All groups · Somewhat or very negative expected impact (%) · n is the unweighted group sample
GroupFeb–Mar 2026Jun–Jul 2026Change
All respondents42%n = 133039%n = 1320Down 3 points
Republicans38%n = 51036%n = 515Down 2 points
Democrats45%n = 65441%n = 643Down 4 points
Independents42%n = 16246%n = 157Up 4 points

Changes are in percentage points: 49% to 61% is an increase of 12 points. Party changes are descriptive; subgroup uncertainty is not shown.

Uncertainty & full-sample margins

Full-sample margins: ±3.5 pp (spring), ±3.48 pp (summer), at 95% confidence. These are not subgroup margins.

Annenberg / SSRS · AI3 · 2026

Expect a positive impact from AI

0–100% scale

Whether AI's effects on the United States over the coming decade will be positive, negative, or a mixture. Question & survey details ↓

Somewhat or very positive expected impact (%). February–March compared with June–July 2026. Party groups include leaners.

All respondents

Up 1 points
Feb–Mar17%
Jun–Jul18%

Republicans

Up 1 points
Feb–Mar21%
Jun–Jul22%

Democrats

Up 2 points
Feb–Mar13%
Jun–Jul15%

Independents

Down 8 points
Feb–Mar19%
Jun–Jul11%

Annenberg describes the overall outlook as stable. These are expectations about the coming decade, not measured benefits or harms.

All groups · Somewhat or very positive expected impact (%) · n is the unweighted group sample
GroupFeb–Mar 2026Jun–Jul 2026Change
All respondents17%n = 133018%n = 1320Up 1 points
Republicans21%n = 51022%n = 515Up 1 points
Democrats13%n = 65415%n = 643Up 2 points
Independents19%n = 16211%n = 157Down 8 points

Changes are in percentage points: 49% to 61% is an increase of 12 points. Party changes are descriptive; subgroup uncertainty is not shown.

Uncertainty & full-sample margins

Full-sample margins: ±3.5 pp (spring), ±3.48 pp (summer), at 95% confidence. These are not subgroup margins.

Download all survey data in this collection

All survey data · CSV ↓Data + provenance · JSON ↓

Gallup · A separate survey, not another trend point

A second lens on
local resistance.

71%somewhat or strongly opposed
local AI data-center construction

Fielded March 2–18, 2026 · n = 1,000 US adults · Telephone · Full-sample margin ±4 pp at 95% confidence

Gallup's listed question names AI and has no explicit neutral option. Annenberg offers a neutral response and uses a different population and mode. We do not join these percentages into one time series.

Intensity of opposition, by party

Strongly opposed (%) — not total opposition.

Democrats56%
Republicans39%
Independents48%

Gallup's Party I.D. categories. Subgroup sample sizes and margins are not supplied in this crosstab. Bars use a 0–100% scale.

Gallup QN51 crosstab · PDF p. 3 ↗

The record behind the chart

Checked September 7, 2026

Read the story behind the numbers ↗

Annenberg Public Policy Center / SSRS · AI7 · Local data-center construction
Question summary
Whether respondents support or oppose new data centers being built in their area. This is a paraphrase; see the original instrument.
Response choices
Strongly support; somewhat support; neither; somewhat oppose; strongly oppose; not sure. Nonresponse retained in totals.
Population / mode
U.S. citizens age 18+. Probability-based panel, primarily online with some phone interviews.
Fieldwork / sample / full-sample margin
  • February 17, 2026–March 20, 2026: n = 1330; ±3.5 pp at 95% confidence.
  • June 16, 2026–July 19, 2026: n = 1320; ±3.48 pp at 95% confidence.
Party definition
APPC's stable party groups include party-leaning independents. The Independent category also includes party nonresponse.
Denominator
All respondents asked the item, including neutral and uncertain responses; percentages are weighted, sample sizes are unweighted.
Comparability
Both waves come from the same August 2026 topline, question AI7. Most respondents participated in both waves; these are not independent samples. Two observations describe this interval, not every intervening month.
Rounding
Use the publisher's NET totals. They are rounded after combining underlying values and may differ from sums of rounded component cells. Chart changes use whole-number endpoints.
Source location
PDF pages 4–5, AI7; page 14, published change estimates
Sponsor / publisher
Annenberg Public Policy Center, University of Pennsylvania
Annenberg Public Policy Center / SSRS · AI3 · AI's expected impact on the United States
Question summary
Whether AI's effects on the United States over the coming decade will be positive, negative, or a mixture. This is a paraphrase; see the original instrument.
Response choices
Very or somewhat positive; equally positive and negative; somewhat or very negative; uncertainty/nonresponse retained in totals.
Population / mode
U.S. citizens age 18+. Probability-based panel, primarily online with some phone interviews.
Fieldwork / sample / full-sample margin
  • February 17, 2026–March 20, 2026: n = 1330; ±3.5 pp at 95% confidence.
  • June 16, 2026–July 19, 2026: n = 1320; ±3.48 pp at 95% confidence.
Party definition
APPC's stable party groups include party-leaning independents. The Independent category also includes party nonresponse.
Denominator
All respondents asked the item, including mixed and uncertain responses; percentages are weighted, sample sizes are unweighted.
Comparability
A different question from local construction. Do not interpret AI-outlook percentages as support or opposition to a data-center project.
Rounding
Publisher's rounded NET totals, not inferred complements.
Source location
PDF pages 3–4, AI3; August release, overall AI outlook
Sponsor / publisher
Annenberg Public Policy Center, University of Pennsylvania
Gallup · QN51 · Local construction to support AI
Question summary
Whether respondents favor or oppose a local data center supporting AI technology in the United States, and how strongly. This is a paraphrase; see the original instrument.
Response choices
Strongly or somewhat favor; somewhat or strongly oppose. The listed question has no explicit neutral option; no opinion is recorded separately.
Population / mode
U.S. adults age 18+, 50 states and Washington, D.C.. Random-sample landline and cellular telephone interviews.
Fieldwork / sample / full-sample margin
  • March 2, 2026–March 18, 2026: n = 1000; ±4 pp at 95% confidence.
Party definition
Gallup's published Party I.D. categories; kept separate from APPC's stable party groups with leaners.
Denominator
All respondents asked the item, including no opinion. Percentages are weighted.
Comparability
Gallup identifies March 2026 as its first measurement of this construction question. A snapshot cannot establish change over time. Different wording, modes, neutral options, and populations prevent joining it to the APPC line.
Rounding
Published rounded estimates. Missing subgroup totals are null, not zero.
Source location
PDF page 1, method; page 2, Q51 total; page 3, QN51 Party I.D. crosstab
Sponsor / publisher
Gallup Poll Social Series

APPC's summer sample includes 1,241 returning participants. Its published change test accounts conservatively for the overlapping panel; within-person results are provided as a robustness check. The linked panel methods report describes recruitment; wave-specific sizes and margins above come from the 2026 releases/topline.

Inspect the original research

Survey sources

  1. 1. Survey topline and crosstabs · Annenberg Public Policy Center

    Annenberg AI survey: August 2026 topline and two-wave comparisons

    Primary published estimates, not respondent microdata. Both waves use this release’s stable party grouping. NET totals use unrounded components; panel overlap and subgroup uncertainty matter.

    Publication: 2026-08-10 · Checked: 2026-09-07
    Location in source: Pages 1–2: method/party definitions; pages 3–5: AI3/AI7; page 14: change estimates

    Source record: appc-ai-august-topline

  2. 2. Survey research release · Annenberg Public Policy Center

    Annenberg reports increasing opposition to local data centers

    Publisher’s interpretation of its own survey, not independent corroboration. The overall construction shift is reported as significant; broader AI outlook is described as stable.

    Publication: 2026-08-11 · Checked: 2026-09-07
    Location in source: Local data-center opposition; overall AI outlook; survey methods

    Source record: appc-data-centers-august

  3. 3. Survey research release · Annenberg Public Policy Center

    Annenberg spring 2026 AI survey release

    Used for spring fieldwork and methods. The August comparative topline supplies both waves’ charted values so versions and party definitions are not mixed.

    Publication: 2026-05-13 · Checked: 2026-09-07
    Location in source: Survey methods: field dates, sample, mode, party leaners, full-sample margin

    Source record: appc-ai-may

  4. 4. Survey panel methodology · SSRS / Annenberg Public Policy Center

    SSRS methods for Annenberg’s Institutions of Democracy panel

    The linked report describes panel recruitment and earlier fieldwork. Its older sample size/margin must not be substituted for the 2026 wave-specific figures.

    Publication: 2024 panel methods, linked by 2026 releases · Checked: 2026-09-07
    Location in source: PDF pages 3–4: citizen population, probability-based recruitment, web/phone modes

    Source record: appc-panel-method

  5. 5. Survey research release · Gallup

    Gallup’s first local AI data-center construction poll

    A national snapshot, not a Gallup time trend. The article also discusses a different April follow-up; those responses are not used as March construction-opinion data.

    Publication: 2026-05-13 · Checked: 2026-09-07
    Location in source: March field dates; first measurement; party differences in strong opposition

    Source record: gallup-data-centers-may

  6. 6. Survey topline and crosstabs · Gallup

    Gallup March 2026 data-center question and party crosstabs

    The full-sample margin is not a party subgroup margin. Strong opposition is not total opposition. The local-AI wording and response options differ from Annenberg’s instrument.

    Publication: Released May 13, 2026; fielded March 2–18 · Checked: 2026-09-07
    Location in source: PDF page 1: method; page 2: Q51; page 3: QN51 Party I.D. columns

    Source record: gallup-data-centers-topline