Demand
U.S. Business AI Use by Industry
Weighted share of U.S. employer businesses reporting AI use in the prior two weeks across six Census-selected NAICS sectors. The comparison shows where AI is diffusing into industry workflows, but survey uncertainty limits rankings and the November 2025 wording change creates a noncomparable break.
U.S. Business AI Use by Industry
- Information (NAICS 51)
- Finance and Insurance (NAICS 52)
- Professional, Scientific, and Technical Services (NAICS 54)
- Educational Services (NAICS 61)
- Manufacturing (NAICS 31–33)
- Retail Trade (NAICS 44–45)
In the latest BTOS period ending 2026-08-23, Information (NAICS 51) reported 46% current AI use. Compare sectors only with their sampling uncertainty and within the same question-wording era.
Methodology
The metric is the Census BTOS weighted percentage answering “Yes” to current AI use in the prior two weeks. Values and point-specific standard errors are copied exactly from the official period-by-sector API; no smoothing, interpolation, ranking, or estimation is applied.
The fixed display set is Information (51), Finance and Insurance (52), Professional, Scientific, and Technical Services (54), Educational Services (61), Manufacturing (31–33), and Retail Trade (44–45). These are the six sectors selected in the Census public BTOS AI visualization, retained in a fixed editorial order rather than sorted by estimate.
Periods 31–84 ask whether AI was used “in producing goods or services.” From period 88, the question asks about AI use “in any of its business functions.” Census observed a level shift and created a new time series; values across the break must not be interpreted as one continuous change.
BTOS is a voluntary probability survey of U.S. nonfarm employer businesses. Its roughly 1.2 million-business annual sample is split into six panels, with each panel surveyed once every 12 weeks. Weighted estimates remain subject to sampling error, nonresponse bias, coverage error, measurement error, and processing error.
The CSV note preserves Census’s standard error for every point. A 90% margin of error is 1.645 times that standard error; the standard chart schema has no uncertainty column, so users should consult the notes/source before treating sector differences or period changes as meaningful. Suppressed API estimates are omitted, and periods 85–87 are structural gaps because BTOS did not collect during the 2025 federal funding lapse.
Census’s official wording bulletin defines period 88 as the new-series boundary. The live API’s period-96 QUESTION text anomalously repeats the original phrase for all six selected sectors, while periods 88–95 and 97 onward use the broader phrase. Period 96 is retained in the official new series and its point notes flag this API metadata inconsistency.
Frequently asked questions
What exactly does this chart measure?
For each of six fixed 2022 NAICS sectors, it shows the Census-weighted percentage of U.S. employer businesses answering Yes to the BTOS current-AI-use question for the prior two weeks. It does not measure workers, spending, usage intensity, productivity, or the share of tasks performed by AI.
Why is this relevant to AI adoption and labor or business diffusion?
Sector differences indicate where AI is entering active business workflows and where diffusion may be slower. They can inform questions about organizational uptake and potential labor exposure, but the survey does not identify affected occupations, displaced work, use cases, investment, or productivity effects.
How should sector differences and the November 2025 break be interpreted?
Compare sectors within the same wording era and account for both estimates’ uncertainty. The highest plotted value is not automatically a statistically distinct leader. Do not compare across period 88: Census broadened the question from producing goods or services to any business function, observed a level shift, and began a new series.
How were the industries and values selected?
The chart keeps the six sectors used in Census’s public BTOS AI visualization—NAICS 51, 52, 54, 61, 31–33, and 44–45—in a fixed editorial order. Exact Yes percentages and standard errors come from the official period-by-NAICS API; suppressed cells are omitted, with no smoothing, interpolation, or ranking.
What survey uncertainty and classification limits apply?
BTOS is a voluntary probability survey of nonfarm employer businesses, so estimates face sampling, nonresponse, coverage, measurement, and processing error. NAICS groups can combine diverse businesses, and results do not cover nonemployers. Standard errors are preserved in CSV notes but are not drawn as bands.
Why are October and early November 2025 missing?
The Census Bureau states that BTOS did not collect periods 85–87 during the October 1–November 13, 2025 federal funding lapse and did not collect responses retroactively.