The Uptime report

Uptime Report Research · Survey Reconciler

Small Business AI Adoption Is About 1 in 4, Not 3 in 4: 23.8% of Firms Use AI, Census Data Shows

Small business AI adoption is real but far lower than the surveys that dominate search results suggest. In the Census Bureau’s Business Trends and Outlook Survey, 23.8% of US employer businesses said they used AI in any business function in the two weeks before Sep 21, 2026, and 23.9% of firms with 1 to 4 employees did. Private surveys put the figure at 58% to 89%.

23.8%of US employer businesses used AI in the last two weeks (Sep 21 to Oct 4, 2026 survey)
23.9%of firms with 1 to 4 employees, the smallest size class
40% vs 23%250+ employee firms vs 1 to 4 employee firms, six latest estimates averaged
3.1%of AI-using firms with 1 to 4 employees used a vendor or consultant to install or integrate it

One question, eight answers

Headline AI adoption numbers from the pages that rank for this search, against the Census Bureau’s probability-sample rate (red). The two sources that cover a broad population, Census and JPMorganChase payment records, sit in the same range, and for the end of 2025 they match almost exactly (17.7% and 17.8%). The opt-in surveys sit far above both.

0%20%40%60%80%100%JPMorganChase Institute: 17.7%JPMorganChase 17.7%Reimagine Main Street (PPSI, with PayPal): 25%Reimagine Main St. 25%Orion Policy Institute: 35%Orion (IBM) 35%U.S. Chamber of Commerce / Teneo: 58%US Chamber 58%Digital Applied: 68%Digital Applied 68%Goldman Sachs 10,000 Small Businesses Voices: 76%Goldman Sachs 76%ICIC: 89%ICIC 89%Census BTOS: 23.8%Census BTOS 23.8%

We downloaded every Business Trends and Outlook Survey (BTOS) release of the Census Bureau’s AI questions since September 2023: national, firm size, sector, state, and sector by firm size, plus the 2026 AI supplement. The survey goes to about 200,000 businesses every two weeks, drawn from an annual probability sample of about 1.2 million employer businesses. Below we test three things: how big the gap between small and large firms really is, whether it is closing, and why private surveys report numbers two to four times higher.

What the Census question actually measures

The BTOS asks: “In the last two weeks, did this business use Artificial Intelligence (AI) in any of its business functions?” with machine learning, natural language processing, virtual agents and voice recognition as examples. Until the November 17, 2025 collection it asked about AI used “in producing goods or services”. The Census Bureau changed the wording after cognitive testing found owners answering no while using AI for hiring, accounting or project management, and it started a new time series because the level jumped. The last estimate on the old wording was 10.0% (Sep 22 to Oct 5, 2025); the first on the new wording was 17.3% (Nov 17 to Nov 30, 2025). That 7.3 point jump is mostly the wording: in the last quarter of old-wording data, the rate was rising about 0.07 points per survey, so six weeks of normal growth would explain well under 1 point of it. We never splice the two series together.

The answer is a share of businesses, not of workers, and it covers only the past two weeks. A firm that tried ChatGPT once in the spring and has not touched it since counts as a no.

AI use by firm size, 2023 to 2026

1 to 4 employees 250 or more other size classes all businesses

Questionreworded0%10%20%30%40%202420252026All businesses (biweekly)All businesses (biweekly)1 to 4 employees1 to 4 employees5 to 9 employees5 to 9 employees10 to 19 employees10 to 19 employees20 to 49 employees20 to 49 employees50 to 99 employees50 to 99 employees100 to 249 employees100 to 249 employees250+ employees250+ employees1 to 4 (23.8%)5 to 9 (21.2%)10 to 19 (22.1%)20 to 49 (24.2%)50 to 99 (28.1%)100 to 249 (35.0%)250+ (39.6%)Old wording: AI “in producing goods or services”New wording: “any of its business functions”

Share of employer businesses answering yes, three-survey trailing average for each size class; the dashed line is the published biweekly rate for all businesses. No data were collected for October 6 to November 16, 2025 (federal funding lapse). Source: U.S. Census Bureau BTOS.

The biggest firms use AI most: 39.5% of businesses with 250 or more employees in the six surveys from Jul 13 to Oct 4, 2026, against 22.8% of businesses with 1 to 4 employees. But the smallest firms are not the laggards. Businesses with 1 to 4 employees use AI at least as often as those with 5 to 49 employees (20.9% for 5 to 9, 21.6% for 10 to 19, 23.2% for 20 to 49). The rate only climbs clearly from 50 employees up.

The size gap is a back-office gap

The old wording tells a different story. When Census asked about AI used “in producing goods or services”, firms with 1 to 4 employees (10.5% in the six surveys to October 5, 2025) were almost level with firms of 250 or more (11.8%) and ahead of every class in between. The gap between the smallest and largest firms was just 0.4 points in late 2023 and 1.4 points in mid-2025. It opened to 12.9 points only when the question widened to “any of its business functions”. The 2026 AI supplement shows where that difference sits. In delivering their own services or products, small and large firms use AI at similar rates (7.0% of 1 to 4 employee firms, 8.9% of 250+ firms), and the same holds for research and development (11.9% and 11.9%). The gap is in internal functions: information technology (11.2% against 25.9%), finance and accounting (5.9% against 20.5%) and human resources (2.9% against 13.6%). Large firms have those departments; a business with one to four people mostly does not.

Is the gap closing?

Not in percentage points. Between the first six surveys on the new wording (Nov 17, 2025 to Feb 8, 2026) and the latest six, firms with 1 to 4 employees gained 5.1 points and firms with 250 or more gained 9.0. The gap widened from 12.9 to 16.8 points. The 3.9 point widening has a standard error of about 1.3, so it is statistically significant at the 90% level. In relative terms nothing changed: both groups grew by about 29%, and the largest firms remain 1.7 times as likely to use AI as the smallest. Gains rise with firm size across the table below.

Employees Old wording, Sep to Dec 2023 Old wording, Jul to Oct 2025 New wording, Nov 17, 2025 to Feb 8, 2026 New wording, Jul 13 to Oct 4 2026 Change, new wording (points) Expect to use in next 6 months (latest)
1 to 4 4.5% 10.5% 17.7% 22.8% +5.1 26.1%
5 to 9 3.2% 8.0% 17.1% 20.9% +3.8 24.9%
10 to 19 3.3% 7.6% 17.2% 21.6% +4.3 25.4%
20 to 49 2.9% 7.3% 18.5% 23.2% +4.7 28.1%
50 to 99 2.9% 7.7% 21.3% 27.8% +6.5 33.1%
100 to 249 3.4% 8.8% 25.4% 33.2% +7.8 38.9%
250+ 5.0% 11.8% 30.5% 39.5% +9.0 46.7%
All businesses 4.0% 9.5% 17.8% 22.8% +4.9 26.4%

Each cell is the simple mean of six published biweekly estimates. The 250+ class has about 450 to 500 responses per survey, so its biweekly estimates swing by several points; averaging six surveys narrows that.

Which sectors lead

use AI now (latest six surveys) ◆ expect to use in six months ○ use AI, Nov 17, 2025 to Feb 8, 2026

0%10%20%30%40%50%60%InformationInformation: 45.9% using AI, 49.4% expect to in six months45.9%Professional, scientific, technical servicesProfessional, scientific, technical services: 41.9% using AI, 46.5% expect to in six months41.9%Finance and insuranceFinance and insurance: 38.3% using AI, 45.1% expect to in six months38.3%Educational servicesEducational services: 35.4% using AI, 39.5% expect to in six months35.4%Real estate and rentalReal estate and rental: 29.1% using AI, 33.8% expect to in six months29.1%Management of companiesManagement of companies: 27.7% using AI, 27.4% expect to in six months27.7%Health care and social assistanceHealth care and social assistance: 24.1% using AI, 28.7% expect to in six months24.1%Administrative and support servicesAdministrative and support services: 21.0% using AI, 25.1% expect to in six months21.0%ManufacturingManufacturing: 20.6% using AI, 24.3% expect to in six months20.6%Wholesale tradeWholesale trade: 19.5% using AI, 23.5% expect to in six months19.5%Arts, entertainment, recreationArts, entertainment, recreation: 18.8% using AI, 22.3% expect to in six months18.8%UtilitiesUtilities: 16.0% using AI, 19.6% expect to in six months16.0%Retail tradeRetail trade: 15.8% using AI, 19.1% expect to in six months15.8%ConstructionConstruction: 15.2% using AI, 17.4% expect to in six months15.2%Other servicesOther services: 12.8% using AI, 15.0% expect to in six months12.8%Transportation and warehousingTransportation and warehousing: 10.6% using AI, 13.5% expect to in six months10.6%Accommodation and food servicesAccommodation and food services: 9.3% using AI, 12.2% expect to in six months9.3%Mining, oil and gasMining, oil and gas: 8.3% using AI, 9.1% expect to in six months8.3%

Sector separates AI users more sharply than size. Information (45.9%), professional, scientific, technical services (41.9%) and finance and insurance (38.3%) lead; accommodation and food services (9.3%) trails among the large sectors. The biggest gains since the new question began were in information (37.7% to 45.9%) and manufacturing (12.4% to 20.6%); in relative terms construction (9.2% to 15.2%) and manufacturing grew fastest, each by about 66%. The sectors leading on AI are also the heaviest IT spenders: our analysis of IT spending by industry shows information and finance putting the largest shares of output into IT.

Size within sector

The heatmap answers the question most small business owners actually have: how do firms like mine compare? A professional services firm with 1 to 4 employees (40.7%) uses AI at a higher rate than a manufacturer (36.4%) or a health care organization (31.1%) with 250 or more employees. In construction, size matters a lot more: 13.7% of 1 to 4 employee firms against 46.6% of the largest.

Sector 1 to 4 5 to 9 10 to 19 20 to 49 50 to 99 100 to 249 250+
Information 44 45 50 53 60 65 72
Professional, scientific, technical services 41 43 45 47 55 58 55
Finance and insurance 36 45 49 40 51 51 59
Educational services 37 30 31 37 35 49 53
Real estate and rental 29 28 30 31 35 40 65
Health care and social assistance 23 23 25 26 26 27 31
Administrative and support services 20 19 22 24 28 32 36
Manufacturing 19 19 19 20 30 35 36
Wholesale trade 17 19 21 26 30 34 41
Arts, entertainment, recreation 17 24 22 21 23 26 49
Retail trade 15 15 17 21 33 32 39
Construction 14 15 18 22 24 34 47
Other services 12 13 18 21 23 S S
Transportation and warehousing 9 12 15 18 29 31 S
Accommodation and food services 9 7 9 10 13 20 17

Percent of businesses using AI, by sector and employees, mean of the six latest biweekly estimates (Jul 13 to Oct 4, 2026). S: suppressed by Census in three or more of the six surveys. Sectors suppressed at every size (utilities, mining, management of companies) are omitted.

Sector Use AI now Nov 17, 2025 to Feb 8, 2026 Change (points) Expect to use in 6 months 1 to 4 employees 250+ employees
Information 45.9% 37.7% +8.3 49.4% 43.9% 72.3%
Professional, scientific, technical services 41.9% 34.2% +7.7 46.5% 40.7% 55.5%
Finance and insurance 38.3% 30.4% +7.9 45.1% 35.6% 59.3%
Educational services 35.4% 30.6% +4.7 39.5% 36.9% 53.3%
Real estate and rental 29.1% 23.5% +5.6 33.8% 29.1% 64.5%
Management of companies 27.7% S 27.4% S S
Health care and social assistance 24.1% 20.7% +3.4 28.7% 23.4% 31.1%
Administrative and support services 21.0% 16.6% +4.4 25.1% 20.3% 35.9%
Manufacturing 20.6% 12.4% +8.2 24.3% 19.4% 36.4%
Wholesale trade 19.5% 13.6% +6.0 23.5% 17.3% 40.7%
Arts, entertainment, recreation 18.8% 17.5% +1.3 22.3% 16.6% 49.0%
Utilities 16.0% 13.4% +2.6 19.6% S S
Retail trade 15.8% 12.6% +3.2 19.1% 14.8% 39.1%
Construction 15.2% 9.2% +6.1 17.4% 13.7% 46.6%
Other services 12.8% 9.8% +3.0 15.0% 11.6% S
Transportation and warehousing 10.6% 7.6% +3.0 13.5% 8.5% S
Accommodation and food services 9.3% 8.0% +1.3 12.2% 8.8% 17.0%
Mining, oil and gas 8.3% 8.5% -0.3 9.1% S S

Sector is the 2-digit NAICS code of the business. Multi-sector companies (Census code XX) are excluded. “Use AI now” is the mean of the six latest biweekly estimates.

AI use by state

Alaska: 17.3% of businesses used AI (average of six biweekly estimates, standard error about 2.1 points)AK17%Maine: 19.6% of businesses used AI (average of six biweekly estimates, standard error about 1.5 points)ME20%Vermont: 19.2% of businesses used AI (average of six biweekly estimates, standard error about 1.9 points)VT19%New Hampshire: 20.3% of businesses used AI (average of six biweekly estimates, standard error about 1.7 points)NH20%Washington: 26.2% of businesses used AI (average of six biweekly estimates, standard error about 0.8 points)WA26%Idaho: 23.3% of businesses used AI (average of six biweekly estimates, standard error about 1.5 points)ID23%Montana: 21.2% of businesses used AI (average of six biweekly estimates, standard error about 1.7 points)MT21%North Dakota: 17.6% of businesses used AI (average of six biweekly estimates, standard error about 2.0 points)ND18%Minnesota: 24.8% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)MN25%Illinois: 21.5% of businesses used AI (average of six biweekly estimates, standard error about 0.5 points)IL21%Wisconsin: 19.5% of businesses used AI (average of six biweekly estimates, standard error about 0.7 points)WI19%Michigan: 21.2% of businesses used AI (average of six biweekly estimates, standard error about 0.7 points)MI21%New York: 17.0% of businesses used AI (average of six biweekly estimates, standard error about 0.3 points)NY17%Rhode Island: 21.5% of businesses used AI (average of six biweekly estimates, standard error about 2.4 points)RI22%Massachusetts: 23.1% of businesses used AI (average of six biweekly estimates, standard error about 0.9 points)MA23%Oregon: 22.9% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)OR23%Nevada: 25.2% of businesses used AI (average of six biweekly estimates, standard error about 1.1 points)NV25%Wyoming: 26.0% of businesses used AI (average of six biweekly estimates, standard error about 2.2 points)WY26%South Dakota: 22.4% of businesses used AI (average of six biweekly estimates, standard error about 1.9 points)SD22%Iowa: 18.6% of businesses used AI (average of six biweekly estimates, standard error about 1.1 points)IA19%Indiana: 20.2% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)IN20%Ohio: 21.4% of businesses used AI (average of six biweekly estimates, standard error about 0.6 points)OH21%Pennsylvania: 19.8% of businesses used AI (average of six biweekly estimates, standard error about 0.6 points)PA20%New Jersey: 18.7% of businesses used AI (average of six biweekly estimates, standard error about 0.6 points)NJ19%Connecticut: 19.2% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)CT19%California: 24.2% of businesses used AI (average of six biweekly estimates, standard error about 0.3 points)CA24%Utah: 30.1% of businesses used AI (average of six biweekly estimates, standard error about 1.3 points)UT30%Colorado: 27.4% of businesses used AI (average of six biweekly estimates, standard error about 0.7 points)CO27%Nebraska: 20.2% of businesses used AI (average of six biweekly estimates, standard error about 1.3 points)NE20%Missouri: 21.1% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)MO21%Kentucky: 21.9% of businesses used AI (average of six biweekly estimates, standard error about 1.1 points)KY22%West Virginia: 13.2% of businesses used AI (average of six biweekly estimates, standard error about 1.7 points)WV13%Virginia: 24.2% of businesses used AI (average of six biweekly estimates, standard error about 0.7 points)VA24%Maryland: 22.3% of businesses used AI (average of six biweekly estimates, standard error about 0.9 points)MD22%Delaware: 28.0% of businesses used AI (average of six biweekly estimates, standard error about 1.9 points)DE28%Arizona: 26.6% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)AZ27%New Mexico: 19.7% of businesses used AI (average of six biweekly estimates, standard error about 1.2 points)NM20%Kansas: 20.7% of businesses used AI (average of six biweekly estimates, standard error about 1.2 points)KS21%Arkansas: 20.5% of businesses used AI (average of six biweekly estimates, standard error about 1.5 points)AR21%Tennessee: 21.3% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)TN21%North Carolina: 23.6% of businesses used AI (average of six biweekly estimates, standard error about 0.7 points)NC24%South Carolina: 23.0% of businesses used AI (average of six biweekly estimates, standard error about 1.2 points)SC23%District of Columbia: 33.6% of businesses used AI (average of six biweekly estimates, standard error about 2.9 points)DC34%Oklahoma: 19.0% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)OK19%Louisiana: 19.0% of businesses used AI (average of six biweekly estimates, standard error about 1.1 points)LA19%Mississippi: 16.0% of businesses used AI (average of six biweekly estimates, standard error about 1.5 points)MS16%Alabama: 18.7% of businesses used AI (average of six biweekly estimates, standard error about 1.3 points)AL19%Georgia: 25.4% of businesses used AI (average of six biweekly estimates, standard error about 0.9 points)GA25%Hawaii: 21.0% of businesses used AI (average of six biweekly estimates, standard error about 1.9 points)HI21%Texas: 24.5% of businesses used AI (average of six biweekly estimates, standard error about 0.5 points)TX25%Florida: 26.5% of businesses used AI (average of six biweekly estimates, standard error about 0.6 points)FL26%Puerto Rico: 13.5% of businesses used AI (average of six biweekly estimates, standard error about 1.0 points)PR13%

District of Columbia (33.6%) has the highest state rate in the latest six surveys and West Virginia (13.2%) the lowest. State estimates are noisy: the typical margin of error on a six-survey average is about ±1.7 points at 90% confidence, so neighboring ranks are usually not different from each other. The spread between the top and bottom of the table is still wide: 2.5 to 1. For comparison, the U.S. Chamber’s 2025 survey reported state rates of small businesses using an AI platform from 31% (West Virginia, which is also last in the Census data) to 77% (Maine).

# State Use AI 90% MOE
1 District of Columbia 33.6% ±4.8
2 Utah 30.1% ±2.2
3 Delaware 28.0% ±3.2
4 Colorado 27.4% ±1.2
5 Arizona 26.6% ±1.6
6 Florida 26.5% ±0.9
7 Washington 26.2% ±1.3
8 Wyoming 26.0% ±3.7
9 Georgia 25.4% ±1.4
10 Nevada 25.2% ±1.8
11 Minnesota 24.8% ±1.7
12 Texas 24.5% ±0.8
13 California 24.2% ±0.5
14 Virginia 24.2% ±1.1
15 North Carolina 23.6% ±1.1
16 Idaho 23.3% ±2.4
17 Massachusetts 23.1% ±1.5
18 South Carolina 23.0% ±2.0
19 Oregon 22.9% ±1.6
20 South Dakota 22.4% ±3.2
21 Maryland 22.3% ±1.4
22 Kentucky 21.9% ±1.8
23 Rhode Island 21.5% ±3.9
24 Illinois 21.5% ±0.8
25 Ohio 21.4% ±1.0
26 Tennessee 21.3% ±1.6
27 Michigan 21.2% ±1.2
28 Montana 21.2% ±2.8
29 Missouri 21.1% ±1.6
30 Hawaii 21.0% ±3.2
31 Kansas 20.7% ±2.0
32 Arkansas 20.5% ±2.4
33 New Hampshire 20.3% ±2.8
34 Nebraska 20.2% ±2.2
35 Indiana 20.2% ±1.6
36 Pennsylvania 19.8% ±1.0
37 New Mexico 19.7% ±2.0
38 Maine 19.6% ±2.5
39 Wisconsin 19.5% ±1.2
40 Connecticut 19.2% ±1.7
41 Vermont 19.2% ±3.1
42 Louisiana 19.0% ±1.7
43 Oklahoma 19.0% ±1.7
44 New Jersey 18.7% ±0.9
45 Alabama 18.7% ±2.1
46 Iowa 18.6% ±1.8
47 North Dakota 17.6% ±3.3
48 Alaska 17.3% ±3.4
49 New York 17.0% ±0.6
50 Mississippi 16.0% ±2.4
51 Puerto Rico 13.5% ±1.7
52 West Virginia 13.2% ±2.8

Expected use is a good forecast

The BTOS also asks whether the business expects to use AI in the next six months. That number is useful only if it predicts something, so we lined up each survey’s expected-use rate with the actual-use rate 13 surveys (26 weeks) later, within each wording. Across 51 such pairs, actual use landed within 0.9 points of the expectation on average (actual minus expected averaged -0.7 points). In the latest six surveys, 26.4% of businesses expected to use AI in six months, against 22.8% using it; in the most recent survey alone the figures were 27.7% and 23.8%. If that relationship holds, roughly 27% of businesses will be using AI by spring 2027. That is a projection from the survey’s own track record, not a Census forecast.

used AI in the last two weeks share that expected to use AI, plotted six months after it was reported

0%10%20%30%Expected in Nov 30, 2025: 21.1%. Actual in May 31, 2026: 20.6%Expected in Dec 14, 2025: 20.9%. Actual in Jun 14, 2026: 20.6%Expected in Dec 28, 2025: 21.3%. Actual in Jun 28, 2026: 20.6%Expected in Jan 11, 2026: 22.4%. Actual in Jul 12, 2026: 21.7%Expected in Jan 25, 2026: 22.1%. Actual in Jul 26, 2026: 21.5%Expected in Feb 8, 2026: 22.1%. Actual in Aug 9, 2026: 21.8%Expected in Feb 22, 2026: 21.0%. Actual in Aug 23, 2026: 22.4%Expected in Mar 8, 2026: 21.8%. Actual in Sep 6, 2026: 23.2%Expected in Mar 22, 2026: 22.3%. Actual in Sep 20, 2026: 23.8%Expected in Apr 5, 2026: 22.5%. Actual in Oct 4, 2026: 23.8%Expected in Sep 24, 2023: 6.3%. Actual in Mar 24, 2024: 4.4%Expected in Oct 8, 2023: 6.3%. Actual in Apr 7, 2024: 4.2%Expected in Oct 22, 2023: 6.5%. Actual in Apr 21, 2024: 4.8%Expected in Nov 5, 2023: 6.5%. Actual in May 5, 2024: 4.7%Expected in Nov 19, 2023: 6.9%. Actual in May 19, 2024: 4.8%Expected in Dec 3, 2023: 7.3%. Actual in Jun 2, 2024: 4.6%Expected in Dec 17, 2023: 6.5%. Actual in Jun 16, 2024: 5.0%Expected in Dec 31, 2023: 6.7%. Actual in Jun 30, 2024: 4.7%Expected in Jan 14, 2024: 6.4%. Actual in Jul 14, 2024: 5.0%Expected in Jan 28, 2024: 6.5%. Actual in Jul 28, 2024: 5.1%Expected in Feb 11, 2024: 6.6%. Actual in Aug 11, 2024: 5.1%Expected in Feb 25, 2024: 6.6%. Actual in Aug 25, 2024: 5.5%Expected in Mar 10, 2024: 6.4%. Actual in Sep 8, 2024: 5.7%Expected in Mar 24, 2024: 6.7%. Actual in Sep 22, 2024: 5.9%Expected in Apr 7, 2024: 6.6%. Actual in Oct 6, 2024: 6.0%Expected in Apr 21, 2024: 6.8%. Actual in Oct 20, 2024: 6.0%Expected in May 5, 2024: 7.0%. Actual in Nov 3, 2024: 6.0%Expected in May 19, 2024: 6.9%. Actual in Nov 17, 2024: 6.1%Expected in Jun 2, 2024: 6.6%. Actual in Dec 1, 2024: 6.1%Expected in Jun 16, 2024: 6.8%. Actual in Dec 15, 2024: 6.1%Expected in Jun 30, 2024: 6.9%. Actual in Dec 29, 2024: 6.3%Expected in Jul 14, 2024: 6.9%. Actual in Jan 12, 2025: 6.0%Expected in Jul 28, 2024: 7.2%. Actual in Jan 26, 2025: 6.7%Expected in Aug 11, 2024: 7.0%. Actual in Feb 9, 2025: 6.8%Expected in Aug 25, 2024: 8.1%. Actual in Feb 23, 2025: 7.3%Expected in Sep 8, 2024: 8.5%. Actual in Mar 9, 2025: 7.4%Expected in Sep 22, 2024: 8.7%. Actual in Mar 23, 2025: 7.8%Expected in Oct 6, 2024: 9.1%. Actual in Apr 6, 2025: 8.0%Expected in Oct 20, 2024: 9.0%. Actual in Apr 20, 2025: 8.3%Expected in Nov 3, 2024: 9.3%. Actual in May 4, 2025: 8.7%Expected in Nov 17, 2024: 8.7%. Actual in May 18, 2025: 8.7%Expected in Dec 1, 2024: 8.9%. Actual in Jun 1, 2025: 9.2%Expected in Dec 15, 2024: 8.8%. Actual in Jun 15, 2025: 9.2%Expected in Dec 29, 2024: 9.2%. Actual in Jun 29, 2025: 9.3%Expected in Jan 12, 2025: 9.5%. Actual in Jul 13, 2025: 9.3%Expected in Jan 26, 2025: 9.4%. Actual in Jul 27, 2025: 9.4%Expected in Feb 9, 2025: 9.3%. Actual in Aug 10, 2025: 8.8%Expected in Feb 23, 2025: 9.8%. Actual in Aug 24, 2025: 9.7%Expected in Mar 9, 2025: 9.9%. Actual in Sep 7, 2025: 9.1%Expected in Mar 23, 2025: 10.5%. Actual in Sep 21, 2025: 9.9%Expected in Apr 6, 2025: 10.5%. Actual in Oct 5, 2025: 10.0%202420252026

All employer businesses. Each wording is compared only with itself, so no pair crosses the November 2025 question change. Source: U.S. Census Bureau BTOS.

Why the surveys disagree

Every page in the top results for this search quotes a number. Here is what each one measured, with the sample it reports. We list them to show the range, not to endorse any of them.

Source Headline Sample and method
U.S. Chamber of Commerce / Teneo 58% of small businesses say they use generative AI 3,870 businesses with fewer than 250 employees, online, June 6 to 26, 2025
Goldman Sachs 10,000 Small Businesses Voices 76% of small businesses currently use AI 1,256 participants in the Goldman Sachs 10,000 Small Businesses program; Babson College and David Binder Research, Jan 27 to Feb 4, 2026
ICIC Nearly 89% of surveyed small businesses use AI tools More than 3,700 small business owners plus five focus groups
Digital Applied 68% of small businesses use AI regularly No survey of its own; credits a 2025 Chamber and Teneo survey, whose published figure is 58%
Reimagine Main Street (PPSI, with PayPal) 1 in 4 use AI in day-to-day operations; 76% are using or exploring it Nearly 1,000 small business owners, June 2025
Orion Policy Institute 35% of businesses use AI significantly (an IBM figure) Cites an IBM study of businesses of all sizes, not a small business sample
JPMorganChase Institute 17.7% of small businesses had paid for an AI service by end of 2025 (26.1% of employer firms) Card and account payment records of 4.6 million small businesses, 2019 to 2025
Gene Marks (Forbes column, on Medium) “Don’t believe it”: owners are dabbling with chatbots, not adopting AI No survey; the author’s 600-client accounting firm and conference conversations
U.S. Census Bureau BTOS 23.8% of employer businesses used AI in a business function in the last two weeks About 24,000 responses in the latest survey, from a probability sample of about 1.2 million employer businesses a year, weighted to all employer businesses
Who answers. The opt-in surveys poll owners who chose to respond, belong to a program, or follow the sponsor. The two sources that cover a broad population line up: JPMorganChase’s payment records show 17.7% of small businesses paying for an AI service by the end of 2025, and the Institute itself notes the BTOS figure for the same period was 17.8%.
What counts as use. Several surveys count any generative AI use. We can test how much that matters inside the Census data: in the 2026 AI supplement, 20.8% of businesses said employees used generative AI for any work task in the past six months, and 22.6% said employees used AI of any kind. A broader question moves the Census number by a few points, not by 40.
Wording. The Census Bureau’s own wording change is a natural experiment. Moving from “producing goods or services” to “any of its business functions” raised the rate from 10.0% to 17.3%. That is the size of a wording effect on a probability sample: about 7 points.
Firms or workers. BTOS counts businesses, and most businesses are small. A Federal Reserve note found 78% of the labor force works at firms using AI in the Atlanta Fed’s Survey of Business Uncertainty, an employment-weighted figure. That describes where workers sit, not what a typical small business does.

None of this means the private surveys are fabricated. It means a 76% headline from 1,256 program participants describes engaged owners, and a 23.8% rate from a federal probability sample describes the typical employer business. The Census number is also not perfect: its response rate in the latest survey was 12% (8% among 1 to 4 employee firms), and Census corrects for that with weighting, not new responses.

How small firms use AI, and why most do not

The Census AI supplement, collected from Nov 17, 2025 to February 8, 2026 across the full annual sample, asked follow-up questions. Firms with 1 to 4 employees use AI most often for sales and marketing (13.1% of all such firms) and strategy (12.0%), and almost never bring in outside help to do it.

Business functions using AI (share of all firms)

Function, last six months 1 to 4 250+
Sales and marketing 13.1% 19.0%
Strategy and business development 12.0% 18.2%
Information technology 11.2% 25.9%
Finance and accounting 5.9% 20.5%
Customer service 7.1% 14.2%
Human resources 2.9% 13.6%

Changes made to use AI (share of AI-using firms)

Change 1 to 4 20 to 49 250+
Used vendors or consulting services to install or integrate AI 3.1% 4.5% 10.3%
Trained current staff to use AI 13.4% 16.9% 31.4%
Developed new workflows 15.0% 15.3% 21.1%
Purchased cloud services or cloud storage 8.5% 8.6% 12.7%
Changed data collection or data management practices 6.7% 7.0% 11.3%
Made no changes to use AI 65.5% 62.9% 48.5%

Why firms do not plan to use AI (share of non-planners)

Reason (multiple allowed) 1 to 4 20 to 49 250+
AI is not applicable to this business 63.3% 58.1% 42.2%
Lack of knowledge on the capabilities of AI 21.2% 23.4% 20.0%
Concerns about privacy/security 20.1% 21.6% 29.7%
AI is not a mature enough technology yet 12.6% 13.5% 19.4%
Too expensive 6.7% 7.0% 9.0%
Lack of skilled workforce 6.4% 8.6% 8.1%

61.6% of businesses with no plans to use AI say it is not applicable to them. The next two reasons, lack of knowledge about what AI can do (22.0%) and privacy or security concerns (20.7%), are things an IT provider can address. Cost is cited by only 6.9%.

What this means for MSPs

Most of your small clients are not using AI yet. About three in four employer businesses said no or did not know in the latest Census survey. Plan AI readiness work around the realistic base: in a book of 100 clients with 1 to 9 employees, expect about 21 to 23 to be active users today, and the industry mix moves that more than headcount does. A client list heavy in professional services or finance will run near 36% to 41% among the smallest firms; one heavy in construction, retail or restaurants will run at 9% to 15%.

The ones using AI are doing it alone. Only 3.1% of AI-using firms with 1 to 4 employees used a vendor or consultant to install or integrate it, and 65.5% made no changes at all: no training, no new workflows, no data-handling changes. That is the gap a readiness offer fills. A practical package covers an inventory of which AI tools staff already use, a written acceptable-use policy, data-loss settings in the client’s Microsoft 365 or Google Workspace tenant, and identity controls on the AI apps themselves. Our MFA statistics show which sign-in methods online services actually support.

Lead with security and knowledge, not price. Among small firms that do not plan to use AI, privacy and security concerns (20.1%) and lack of knowledge (21.2%) each outrank cost (6.7%) by about three to one. The pitch that matches the data is “here is how to use it without leaking client data,” not “here is a cheaper tool.” AI tools also push data into cloud services; our cloud computing statistics track how fast the hyperscalers behind most of those tools are growing.

Use the expectation number for timing. Nationally, the share expecting to use AI has tracked actual use six months later. With 26.1% of 1 to 4 employee firms and 28.1% of 20 to 49 employee firms expecting to use AI soon, the next two quarters are when new users will be setting habits, with or without a policy.

Frequently asked questions

What percent of small businesses use AI?

In the Census Bureau’s Business Trends and Outlook Survey, 23.9% of businesses with 1 to 4 employees and 21.5% of those with 5 to 9 employees said they used AI in a business function in the two weeks before Sep 21, 2026. Across all employer businesses the rate was 23.8%.

Is AI being adopted by small businesses?

Yes, and faster than before. The Census rate for all businesses rose from 17.3% in late November 2025 to 23.8% in the Sep 21, 2026 to Oct 4, 2026 survey. Firms with 1 to 4 employees gained 5.1 points between the Nov 17, 2025 to Feb 8, 2026 and Jul 13 to Oct 4, 2026 survey windows.

Why do small business AI surveys give such different numbers?

They ask different people different questions. Private surveys mostly poll owners who opt in or belong to a program, and many count any use of a chatbot. The Census survey draws a probability sample of about 1.2 million employer businesses a year and weights the answers to all employer businesses. Even its broad question on employees using generative AI for work tasks gets 20.8%, not 58% or more.

Which industries use AI the most?

Information (45.9%), professional, scientific, technical services (41.9%) and finance and insurance (38.3%) lead in the six latest Census biweekly estimates. Accommodation and food services is lowest among large sectors at 9.3%.

What companies are adopting AI the most?

Large ones, by share. 39.5% of businesses with 250 or more employees used AI in the latest six Census estimates, against 22.8% of firms with 1 to 4 employees. Sector matters as much as size: professional services firms with 1 to 4 employees (40.7%) use AI at a higher rate than manufacturers with 250 or more employees (36.4%).

How many small businesses plan to use AI?

26.1% of firms with 1 to 4 employees expect to use AI in the next six months, against 22.8% using it now. In the Census data, the share expecting to use AI has been a good forecast: actual use six months later landed within 0.9 points of it on average.

Download the data. small-business-ai-adoption-census-btos.csv (CSV). Licensed under CC BY 4.0: free to reuse with attribution to The Uptime Report.

Methodology

Source. U.S. Census Bureau, Business Trends and Outlook Survey (BTOS), an experimental data product. We downloaded the current national, employment size class, sector, state, and sector by employment size class workbooks, the “AI Core Questions (Original)” workbook with the old-wording series, the 2026 AI supplement table, and the response rate workbook from the BTOS data downloads page on October 15, 2026. The latest survey in the data is the Sep 21 to Oct 4, 2026 collection.

Definitions. “Uses AI” means the business answered yes to: “In the last two weeks, did this business use Artificial Intelligence (AI) in any of its business functions?” From September 11, 2023 to October 5, 2025 the question read “in producing goods or services.” The two wordings are separate series and are never combined. “Expects to use” is the parallel six-month question. Rates are shares of businesses, weighted by Census to all employer businesses in scope (nonfarm employer businesses, excluding a short list of industries such as public administration and religious organizations).

Our calculations. Period averages are simple means of six consecutive published biweekly estimates, with a standard error approximated as the square root of the summed squared published standard errors divided by six (each survey uses a different panel of businesses). Sector, state, and sector by size averages require at least four of six unsuppressed estimates. The expectation test pairs each survey’s expected-use rate with the actual-use rate 13 surveys later within the same wording. The chart by firm size uses three-survey trailing averages.

Coverage and gaps. No data exist for October 6 to November 16, 2025 because collection stopped during the federal funding lapse. The latest survey had about 24,000 responses and a 12% unit response rate. Nonemployer businesses (no paid staff) are outside the BTOS. The download holds every published “yes” estimate (and the national “do not know” share) for both AI questions and both wordings; estimates Census suppressed are left out. Survey headline numbers in the comparison table were read from each publisher’s page on October 15, 2026; the Goldman Sachs sample details come from coverage of its March 17, 2026 release.

Not advice. This is an analysis of public survey data, not a recommendation for any business.

Cite as:The Uptime Report. "Small Business AI Adoption Is About 1 in 4, Not 3 in 4: 23.8% of Firms Use AI, Census Data Shows." October 2026. https://theuptimereport.com/small-business-ai-adoption/

References

  1. U.S. Census Bureau, Business Trends and Outlook Survey. Accessed October 2026.
  2. U.S. Census Bureau, BTOS data downloads (national, size, sector, state, sector by size, AI Core Questions original, AI supplement 2026, URR). Accessed October 2026.
  3. U.S. Census Bureau, BTOS AI Core Question Updates (December 3, 2025). Accessed October 2026.
  4. U.S. Census Bureau, BTOS Methodology. Accessed October 2026.
  5. U.S. Census Bureau, “AI Use at U.S. Businesses” (May 26, 2026). Accessed October 2026.
  6. Federal Reserve Board, “Monitoring AI Adoption in the U.S. Economy,” FEDS Notes (April 3, 2026). Accessed October 2026.
  7. U.S. Chamber of Commerce, Empowering Small Business 2025 and full report (methodology, p. 37). Accessed October 2026.
  8. JPMorganChase Institute, “Understanding the use of AI among small businesses”. Accessed October 2026.
  9. Goldman Sachs 10,000 Small Businesses Voices, AI survey (March 17, 2026) and SCX Change coverage with sample details. Accessed October 2026.
  10. ICIC, AI in Business. Accessed October 2026.
  11. Reimagine Main Street, AI and Small Business Survey and NSBA summary. Accessed October 2026.
  12. Orion Policy Institute, “Empowering Small Businesses”. Accessed October 2026.
  13. Digital Applied, “Small Business AI Adoption”. Accessed October 2026.
  14. Gene Marks, “Small Businesses Adopting AI? Don’t Believe It.” Accessed October 2026.

Research library

More original data reports from The Uptime Report. Every report includes its methodology and a free CSV download. Browse all research

Threats and breaches

Infrastructure and uptime

Regulation and disclosure

Workforce and pay

MSP market and spending

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *

More posts