AI Adoption Statistics 2026: Firms, Workers and Industries

About one US business in four uses AI, and about four workers in ten use it for their job. The Census Bureau's survey found 24.6% of firms used AI in the two weeks before October 5 to 18, 2026, and the St. Louis Fed's survey put work use at 40.5% of employed adults in August 2026. Higher figures you see elsewhere count something different, not more adoption.

That gap is the most misread number in AI coverage. A firm-level survey counts a two-person landscaper the same as a bank, while a spending index or an employee poll leans toward bigger, office-based employers. This page lines the main sources up side by side, says who each one counts, and reports the latest figure from each one's own release.

AIpresso data: AIpresso readers click links in the newsletter at 1.7% on average, against 1% for Techpresso, our general tech brief, in the Beehiiv audience file refreshed November 8, 2026. Readers of an AI tools brief act on what they read more often than general tech readers, which is why adoption data gets its own page here.

AI adoption statistics at a glance

Statistic	Figure	Period	Source
US firms that used AI in the last two weeks	24.6%	Collected Oct 5 to 18, 2026	Census BTOS national data
US firms expecting to use AI in the next six months	28.6%	Same cycle	Same file
Firms with 250 or more employees using AI	43.6%	Same cycle	Census BTOS size-class data
Employed adults using generative AI for work	40.5%	August 2026	GenAI Adoption Tracker
Adults 18 to 64 using generative AI for any purpose	60.8%	August 2026	Same tracker
Employees who use AI in their role at least a few times a year	52%	May 6 to 20, 2026	Gallup workforce study
Employees who say their organization has integrated AI tools	47%	May 6 to 20, 2026	Same Gallup study
US businesses on Ramp paying for AI	55.9%	September 2026	Ramp AI Index
US working-age population using AI	More than 30%	Report of May 28, 2026	Microsoft US AI diffusion post

All figures were read on the publishers' own pages and data files on October 28, 2026. None of these surveys is wrong; each answers a narrower question than its headline suggests.

Why do AI adoption numbers disagree so much?

The denominator decides the answer. The same economy produces the Census rate and a rate more than twice as high because one counts every employer firm by survey and the other counts card and bill-pay customers of a fintech by what they bought. Read the "who is counted" column before you quote any number.

Source	Who is counted	What counts as adoption	Latest rate
Census Bureau BTOS	A national sample of US employer businesses, every size, every nonfarm sector	The firm says it used AI in any business function in the last two weeks	24.6% of firms
Ramp AI Index	More than 85,000 US businesses that use Ramp's corporate card and bill pay	Any positive payment to an AI product or service that month	55.9% of businesses
Gallup Panel	Employed US adults, full and part time (n of 22,573)	The worker uses AI in their role a few times a year or more	52% of employees
St. Louis Fed Real-Time Population Survey	Employed adults aged 18 to 64	"Do you use Generative AI for your job?"	40.5% of workers
Microsoft AI Economy Institute	The whole working-age population	Share of working-age people using AI	More than 30% of people

Three mechanics explain most of the spread:

• The Census rate weights every firm equally. A five-person shop counts as much as a 5,000-person employer, and small firms adopt less (see the size table below), so the national average sits near the small-firm rate.
• Spending data skews to software buyers. A Ramp customer already runs its expenses through a modern finance tool, and one $20 seat on a chatbot is enough to count the whole company.
• Worker surveys weight by people, not firms. A large employer with thousands of staff counts thousands of times, so worker rates sit closer to the big-firm rate.

If you see a figure near or above 80%, check the denominator first: a survey limited to large companies, or one that counts a firm as an adopter when any single team uses AI, will land far above the Census rate. For a sense of what those tools cost per seat, our AI subscription prices 2026 index tracks the plans most teams buy.

How many US businesses use AI?

24.6% of US businesses used AI in a business function in the two-week reference period of the Census cycle collected October 5 to 18, 2026, and 28.6% expected to use it in the next six months. Both are the highest readings since the Census Bureau widened the question in November 2025.

Census collection period	Used AI in the last two weeks	Expect to use AI in six months
Nov 17 to 30, 2025	17.3%	21.1%
Jan 26 to Feb 8, 2026	18.9%	22.1%
Apr 20 to May 3, 2026	19.8%	22.8%
Jun 29 to Jul 12, 2026	21.7%	24.5%
Aug 24 to Sep 6, 2026	23.2%	27.3%
Oct 5 to 18, 2026	24.6%	28.6%

Source: Census BTOS national file, published October 22, 2026.

That is a rise of 7.3 percentage points in eleven months, or about 42% in relative terms, by our arithmetic. The pace picked up after the spring: the rate sat near 19% to 20% from February to May, as the Census Bureau's May 2026 story on AI use described, then climbed every few cycles through October.

The earlier, lower figures you may remember (3.8% in late 2023) used a narrower question about AI "in producing goods or services". In November 2025 the Census Bureau switched to "any business function", so the two series do not line up. The current questionnaire also labels the technology "Super Intelligence or SI (formerly referred to as Artificial Intelligence or AI)", with the same examples of business use.

Does company size change AI adoption?

Firms with 250 or more employees use AI at 43.6%, against 24.3% for firms with one to four employees, in the October 2026 Census cycle. The biggest firms adopt at roughly 1.8 times the rate of the smallest, our arithmetic.

Firm size	Used AI, Oct 5 to 18, 2026	Apr 20 to May 3, 2026	Nov 17 to 30, 2025
1 to 4 employees	24.3%	19.0%	17.0%
5 to 9	22.6%	19.5%	15.5%
10 to 19	24.6%	19.4%	17.8%
20 to 49	26.0%	21.3%	18.7%
50 to 99	29.0%	25.8%	20.7%
100 to 249	35.3%	32.0%	23.7%
250 or more	43.6%	37.3%	30.5%

Source: Census BTOS size-class file.

The detail worth noticing is the smallest band. In May the Census Bureau reported no significant change among firms under 20 employees, yet by October micro firms had gained about five points, so the catch-up at the bottom is recent. Ramp's spending data shows the same ladder at a higher level: 65.7% of large businesses, 62.5% of medium-sized ones and 50.0% of small ones paid for AI in September 2026.

Which industries use AI the most?

Information firms lead at 46.4%, followed by professional, scientific and technical services at 43.8% and finance and insurance at 40.5%, in the October 2026 Census cycle. Agriculture sits at 6.0%, the lowest of any sector.

Sector (NAICS)	Used AI, Oct 5 to 18, 2026	Apr 20 to May 3, 2026
Information	46.4%	39.7%
Professional, scientific and technical services	43.8%	37.0%
Finance and insurance	40.5%	33.9%
Management of companies	36.4%	Not published
Educational services	33.4%	34.7%
Real estate	29.0%	24.4%
Manufacturing	22.9%	16.4%
Retail trade	18.5%	14.3%
Transportation and warehousing	13.5%	7.6%
Accommodation and food services	10.7%	8.4%
Agriculture, forestry, fishing and hunting	6.0%	Not published

Source: Census BTOS sector file. "Not published" marks cells the Census Bureau suppressed for that cycle.

The ranking follows screens. Sectors where most work happens at a desk, on documents and data, adopt first; sectors built on physical service adopt last. Ramp's index ranks the same sectors in nearly the same order but much higher, with information at 79.9% and finance and insurance at 73.7% of its customers paying for AI in September 2026.

For a buyer, the sector rate is a benchmark, not a target. An accounting or consulting firm without AI is still with the majority of its sector, but that majority shrank from 63% to 56% between May and October, our arithmetic; a restaurant group without any AI tool is still with roughly nine peers in ten.

How many US workers use AI?

40.5% of employed adults used generative AI for their job in August 2026, according to the Real-Time Population Survey run by economists at the St. Louis Fed, Vanderbilt and Harvard. Use for any purpose, at work or at home, reached 60.8% of adults aged 18 to 64.

Quarter	Workers using generative AI for their job
Q3 2025	37.4%
Q4 2025	40.6%
Q1 2026	42.8%
Q2 2026	45.2%
Q3 2026	40.5%

Source: FRED series RPSGENAIUSAGESHAREWORK, updated October 20, 2026; quarterly values rounded to one decimal by us.

The third quarter broke the upward run with a drop of about 4.7 points from the second, our arithmetic. One quarterly reading is not a trend, so wait for the next wave before calling a decline. The longer view is clear: the St. Louis Fed's September 2026 analysis reports work use rising from 33% in August 2024 to 45% in May 2026.

Gallup's quarterly panel lands higher because its bar is lower. In May 2026, 52% of US employees said they use AI in their role at least a few times a year, 30% used it a few times a week or more and 15% used it daily. In the second quarter of 2023 those figures were 21%, 11% and 4%.

Usage intensity is still modest. The adoption tracker estimates that 5.3% of all work hours involved generative AI in August 2026 and that AI saved 2.1% of total work hours. Most workers who use AI use it for a slice of the day, not the whole job.

Which jobs use AI the most?

Computer and information research scientists lead at 87.3% adoption, while animal caretakers sit at 5.3%, in the St. Louis Fed's occupation index built from nearly 14,000 workers surveyed between August 2025 and May 2026.

• Widespread: in more than 80% of occupations, at least one worker in five uses AI on the job.
• Shallow: only 40% of occupations have adoption above 50%, and just 16% are above 70%.
• Task-level ceiling: fewer than 3% of job tasks have adoption above 50%, and none exceed 70%.
• Top tasks: reading documents for technical information (61.3%), preparing research reports (60.7%) and analyzing data for trends (57.5%).

The surprises are where exposure scores miss. Medical secretaries use AI at 16.8%, against a predicted 61%, because privacy rules and the cost of a mistake hold them back. Special education teachers (70.9%) adopt at roughly double the predicted rate. If you sell or roll out AI tools, the lesson is that "could use AI" and "does use AI" diverge most where records are sensitive.

Gallup's data points the same way on tasks. Writing and editing (51%) and search or research (49%) were the top uses among AI users in May 2026, while coding help and automation were each cited by 16%, yet those two drew the strongest productivity ratings (77%). Our AI agents ranking and AI coding assistants ranking cover the tools behind those task-specific uses.

What businesses pay for: the spending view

55.9% of US businesses on Ramp paid for an AI product or service in September 2026, up 12.1 points from a year earlier, according to the Ramp AI Index. Anthropic was paid by 43.8% of those businesses and OpenAI by 40.6%.

Spending per head is still small for most. The median firm paying for AI spent $15.70 per employee per month in September 2026, while the top 1% of paying firms spent a median of $9,567.53 per employee. AI made up 5.77% of business spend excluding payroll.

This is the best real-money check on the survey figures, with one limit: Ramp's customers are not a random sample of US firms. Treat the level as an upper bound for software-forward companies and the month-to-month direction as the useful signal. If your firm sits near that median, compare what you pay with the plans in our ChatGPT pricing and Claude pricing guides. Both vendors also appear on Toolradar, the software directory we run: ChatGPT and Claude.

Where the US stands globally

More than 30% of the US working-age population uses AI, which ranks the country 21st in the world, according to Microsoft's US AI diffusion report released May 28, 2026. That was three points higher than at the end of 2025.

The same report found a sharp urban and rural split: usage averaged 32.9% in metropolitan counties against 16.2% in rural ones. Counties with more residents aged 18 to 24 ran at 28.6%, compared with 20.3% elsewhere, which points to colleges as local adoption engines.

Microsoft's figure sits below the worker surveys partly because it covers everyone of working age, including people outside the labor force.

What these statistics do not tell you

• Use is not depth. A firm that ran one AI query in two weeks counts the same as one that rebuilt its support desk around agents. Our explainer on what agentic AI is covers that deeper end.
• Self-reports drift with wording. The Census series jumped when the question widened in November 2025, and Gallup's "few times a year" bar is far looser than a daily-use measure.
• Periods differ. The Census figure is from October, the worker survey from August, Gallup from May and Microsoft from the first quarter. Compare trends within one source, not levels across sources.

For category-wide figures on the tools themselves, see our AI tools statistics, ChatGPT statistics and AI image generator statistics. If you are choosing a first tool for a small team, our best AI tools for business guide starts there.

AIpresso sends a short daily brief on AI tools, prices and adoption data like this. Join at aipresso.com.

How we compiled these statistics

We read the Census Bureau's BTOS national, sector and size-class data files (published October 22, 2026), the St. Louis Fed's September 2026 analysis, the FRED series and the GenAI Adoption Tracker, Gallup's July 20, 2026 workforce study, the Ramp AI Index page and Microsoft's May 28, 2026 post, all on October 28, 2026. Each figure is quoted as the publisher reported it, with its period. Point changes, ratios and relative growth are our arithmetic and are labeled as such wherever they appear. We used no analyst estimates or vendor-commissioned surveys; Census BTOS is biweekly, so the next cycle may move these figures by a point.

FAQ

What share of US businesses use AI in 2026?

24.6% of US businesses said they used AI in a business function in the last two weeks, in the Census Bureau survey cycle collected October 5 to 18, 2026. Another measure, Ramp's spending index, found 55.9% of its business customers paid for AI in September 2026, because it counts a narrower, more software-heavy group of firms.

What percentage of workers use AI at work?

40.5% of employed US adults used generative AI for their job in August 2026, according to the St. Louis Fed's Real-Time Population Survey. Gallup's May 2026 panel found 52% of employees use AI in their role at least a few times a year, and 15% use it daily.

Which industry uses AI the most?

The information sector leads, with 46.4% of firms using AI in the October 2026 Census cycle, ahead of professional, scientific and technical services and finance and insurance, both above 40%. Agriculture is lowest at 6.0%, so the gap between sectors is wider than the gap between firm sizes.

Do large companies use AI more than small ones?

Yes. 43.6% of firms with 250 or more employees used AI in October 2026, against 24.3% of firms with one to four employees, per the Census Bureau. Small firms gained about five points between May and October 2026.

Is AI adoption still growing in 2026?

At the firm level, yes: the Census rate rose from 17.3% in November 2025 to 24.6% in October 2026. At the worker level, the St. Louis Fed series dipped between the second and third quarters of 2026, a single reading that the next wave will confirm or reverse, after rising for a year before that.

Why do some reports say almost every company uses AI?

Check who they count. A survey limited to large companies, or one that counts a company as using AI when any one team does, lands far above the Census Bureau, which weights every employer firm equally and asks about the last two weeks only. Even among firms with 250 or more employees, the Census rate was 43.6% in October 2026.

Cite this: AIpresso, "AI Adoption Statistics 2026: Firms, Workers and Industries", November 2026.

Some offers on this page may be paid placements or contain affiliate links.