AI Market Growth Statistics Updated November 2026

Human Written & Fact Checked · Updated

100 AI Adoption Statistics for 2026

Before comparing AI adoption rates, ask who is being counted. A firm, an adult and a telemetry-adjusted user belong to different populations; a single headline percentage cannot represent all of them.

This hub separates recent US business evidence, Microsoft's international diffusion estimates and original US adult-survey observations. Observation periods are identified explicitly. International estimates are older than the latest business release and are presented as dated context, not current-quarter readings.

Headline figures

  • Census November release, reported business AI/SI use in the prior two weeks, yes: 24.2%. (U.S. Census Bureau, Nov 2026)
  • Census November release, expected business AI/SI use in the following six months, yes: 27.6%. (U.S. Census Bureau, Nov 2026)
  • Census November release, firms with 250 or more employees, current AI/SI use: 44.1%. (U.S. Census Bureau, Nov 2026)
  • Census November release, information, current business AI/SI use: 45.0%. (U.S. Census Bureau, Nov 2026)
  • United Arab Emirates, Microsoft estimated generative-AI user share in June 2026: 73.3%. Working-age population, ages 15–64. (Microsoft AI Economy Institute, Sep 2026)
  • France, Microsoft estimated generative-AI user share in June 2026: 49.6%. Working-age population, ages 15–64. (Microsoft AI Economy Institute, Sep 2026)
  • United States, Microsoft estimated generative-AI user share in June 2026: 33.0%. Working-age population, ages 15–64. (Microsoft AI Economy Institute, Sep 2026)
  • World, Q2 2026 estimated generative-AI user share: 18.8%. Working-age population, ages 15–64. (Microsoft AI Economy Institute, Sep 2026)
  • August 2026 any-purpose genAI use, any frequency: 60.8%. Among US adults aged 18–64. (Bick, Blandin, Deming / RPS, Sep 2026)
  • August 2026 genAI use for work, any frequency: 40.5%. Among eligible US workers aged 18–64. (Bick, Blandin, Deming / RPS, Sep 2026)
  • August 2026 share of weekly work hours actively using genAI: 5.3%. Among eligible US workers aged 18–64. (Bick, Blandin, Deming / RPS, Sep 2026)
  • August 2026 work genAI use among workers with a graduate degree: 58.2%. Among eligible US workers aged 18–64. (Bick, Blandin, Deming / RPS, Sep 2026)

How widely are US businesses using AI?

Business adoption needs a business denominator. Census measures firms within its employer-business coverage, so the results cannot tell you what share of employees use an assistant.

What do firms report about current use and future plans?

  • Reported business AI/SI use in the prior two weeks, no: 61.9%. (U.S. Census Bureau, Nov 2026)
  • Reported business AI/SI use in the prior two weeks, do not know: 13.9%. (U.S. Census Bureau, Nov 2026)
  • Expected business AI/SI use in the following six months, no: 46.2%. (U.S. Census Bureau, Nov 2026)
  • Expected business AI/SI use in the following six months, do not know: 26.2%. (U.S. Census Bureau, Nov 2026)

Census distinguishes reported recent use from expectations, with uncertainty recorded explicitly. Current use refers to the reference fortnight in October; expectations refer to the following half-year. A business answering that it does not know is not automatically a non-user. The two questions should therefore remain separate when estimating demand. Neither response measures the depth of integration, the number of licensed seats or the proportion of work completed with AI.

How does current use vary with employer size?

Reported use differs across employment-size classes. Each percentage is a share of businesses in that class, instead of a share of the national workforce. The latest observation uses the same reference fortnight across sizes. Small differences between adjacent categories should not be called statistically significant without considering the published standard errors. The largest-firm result appears in Main Takeaways; the remaining classes below complete the size comparison.

Which firms expect to use AI in the next six months?

  • Firms with 1–4 employees, expected next-six-month AI/SI use: 25.2%. (U.S. Census Bureau, Nov 2026)
  • Firms with 5–9 employees, expected next-six-month AI/SI use: 26.8%. (U.S. Census Bureau, Nov 2026)
  • Firms with 10–19 employees, expected next-six-month AI/SI use: 30.3%. (U.S. Census Bureau, Nov 2026)
  • Firms with 20–49 employees, expected next-six-month AI/SI use: 34.0%. (U.S. Census Bureau, Nov 2026)
  • Firms with 50–99 employees, expected next-six-month AI/SI use: 43.1%. (U.S. Census Bureau, Nov 2026)
  • Firms with 100–249 employees, expected next-six-month AI/SI use: 47.2%. (U.S. Census Bureau, Nov 2026)
  • Firms with 250 or more employees, expected next-six-month AI/SI use: 53.6%. (U.S. Census Bureau, Nov 2026)

Larger employment classes report higher expected use in this release. These are expectations collected from businesses, not an independently validated forecast of adoption. Plans may change, and an affirmative response can include firms already using the technology. The figures therefore cannot be interpreted as the proportion planning a new implementation.

Which business sectors report AI use?

Sector differences are meaningful only when the question and survey period match. These estimates share the same instrument and reference window; suppressed categories are excluded instead of treated as non-users.

What do production and trade businesses report?

Use is reported across production, distribution and retail activities, with different levels across sectors. These estimates describe whether a firm used the technology in any business function; a factory can qualify through marketing or administration instead of production-line automation. That broad question counts when interpreting the manufacturing result. The comparison does not identify which tasks were automated or how intensively tools were used.

How do information and business services compare?

  • Transportation and warehousing, current business AI/SI use: 12.6%. (U.S. Census Bureau, Nov 2026)
  • Finance and insurance, current business AI/SI use: 39.2%. (U.S. Census Bureau, Nov 2026)
  • Real estate, rental and leasing, current business AI/SI use: 28.2%. (U.S. Census Bureau, Nov 2026)
  • Professional, scientific and technical services, current business AI/SI use: 41.6%. (U.S. Census Bureau, Nov 2026)
  • Administrative, support and waste services, current business AI/SI use: 22.2%. (U.S. Census Bureau, Nov 2026)
  • Information-sector firms expecting AI/SI use in the following six months: 45.1%. (U.S. Census Bureau, Nov 2026)

Knowledge-intensive activities show substantial reported use, but each sector still has its own denominator. The information-sector current-use result is in Main Takeaways; its expectation measure below answers a different question. Transportation and administrative services remain separate sector groups. Avoid turning these differences into a technology-readiness ranking: this release does not measure implementation quality, profitability or access to specialist staff.

What do public-facing services and multi-sector firms report?

  • Educational services, current business AI/SI use: 39.5%. (U.S. Census Bureau, Nov 2026)
  • Health care and social assistance, current business AI/SI use: 24.6%. (U.S. Census Bureau, Nov 2026)
  • Arts, entertainment and recreation, current business AI/SI use: 22.0%. (U.S. Census Bureau, Nov 2026)
  • Accommodation and food services, current business AI/SI use: 12.1%. (U.S. Census Bureau, Nov 2026)
  • Other services, current business AI/SI use: 15.7%. (U.S. Census Bureau, Nov 2026)
  • Multi-sector firms (XX category), current business AI/SI use: 34.4%. (U.S. Census Bureau, Nov 2026)

Health, education, hospitality and other services cannot be treated as one adoption market. Census reports them separately, while its multi-sector category groups businesses spanning activities. The latter is a classification category instead of an additional conventional industry. These employer-business results do not cover all public institutions or every self-employed person. They also do not establish whether AI was used in customer-facing decisions.

Which countries lead Microsoft's AI diffusion measure?

A country ranking is a measurement model, not a census of every tool. Microsoft adjusts observed usage for device and operating-system market shares, internet penetration and population.

Which leading economies have the highest estimated user shares?

Microsoft's leading economies combine high estimated diffusion with very different population sizes. All country observations here describe the report's June measurement and working-age population, not November activity. The UAE and France observations appear in Main Takeaways. The remaining leaders below use the same model, making them more comparable with each other than with a differently designed national survey. No estimate identifies the proportion using AI at work specifically.

How do the next leading economies compare?

Country differences reflect estimated population reach, not subscription purchases or vendor revenue. Equal displayed percentages should be read as ties at the report's rounding precision. The methodology adjusts observed usage instead of contacting every resident. These are therefore model-based diffusion estimates, and the country labels should always travel with the reporting period.

What do the remaining top-ranked economies show?

These economies complete the leading group shown in the source chart. A smaller reported share does not establish that local businesses are less productive or that domestic AI suppliers are weaker. The measure concerns people using generative-AI products, adjusted to the working-age population. It is useful for comparing reach under a shared method; it is not a direct measure of how many businesses have deployed AI or how much they spend.

How broad is global AI diffusion?

Regional diffusion and commercial market size answer different questions. These population shares provide geographic context without implying spending, vendor revenue or enterprise deployment.

Where do the United States and nearby-ranked economies stand?

The US result in Main Takeaways belongs to the same country model as these observations. It should not be substituted for the Census business result or the adult survey later in this article. The samples, age coverage, adjustment methods and periods differ. Similar country values do not imply comparable absolute user totals, because population sizes differ. Keep the country measurement attached to its source instead of averaging across incompatible adoption surveys.

What do other covered European and Middle Eastern economies show?

The report provides additional comparable country estimates beyond its leading chart. These observations preserve the original country labels and the same June reference period. Their proximity in the ranking should not be overinterpreted as a precise ordering of national AI capability. The metric does not distinguish casual experimentation from organisational deployment. It also cannot tell you whether usage is supported by employer policy, personal accounts or paid subscriptions.

How has estimated diffusion changed by world region?

The report's regional series uses its own Global North and Global South groupings. These regional observations are population measures, not firm adoption rates. The latest world observation is in Main Takeaways; the earlier world values and regional sequence appear below. Changes are descriptive model estimates, not causal evidence about policy or investment.

How often do US adults actually use generative AI?

Trying a tool and making it a daily habit are different behaviours. The Real-Time Population Survey separates any use from recent and daily use, and separates work from personal activity.

How common are recent and daily uses among adults?

The adult survey separates any experience from use during the preceding week and daily use during that week. The broad any-frequency estimate appears in Main Takeaways. The two observations below narrow the behaviour without changing the adult denominator. Respondents may use tools for work, personal purposes or both, so purpose-specific groups overlap. These survey estimates describe US adults within the study's age coverage, not every resident and not all internet users. The observation month is August; the linked workbook carries a September revision timestamp instead of a separately verified first-publication date.

How frequent is workplace use and what do workers report saving?

Work-use frequency is measured among employed adults within the survey age range. The broad work-use result and active-hours measure appear in Main Takeaways. Recent and daily use remain distinct from the reported time-savings estimate. The latter averages workers' own assessments across the eligible worker population, including non-users; it is not an experimental productivity effect or a measured reduction in payroll costs. The workbook's standard errors are retained in the inventory to support uncertainty checks.

How common is non-work use?

Non-work use is measured among all adults in the covered age range, including people who are not employed. It can overlap with workplace use, so the two purpose shares must not be added together. The frequency categories also overlap: daily users can qualify as recent and any-frequency users. That nesting is useful for understanding engagement, but it does not identify how many separate products people use. The figures describe respondent-reported behaviour in August, while the linked workbook's revision date is used for the citation. They should not be described as November usage levels.

Which workers report generative-AI use?

Workplace adoption is uneven across education, earnings and job categories. These are worker reports, not employer deployment rates or evidence that AI caused the differences.

How does workplace use differ by education?

Workers with different education levels report different use rates. The graduate-degree observation is in Main Takeaways; the remaining education groups below complete the comparison. Less than a bachelor's degree includes people with some postsecondary education, so it should not be labelled as no college. These are descriptive conditional shares among workers, not causal estimates of the effect of education. Job mix, employer practices and access to suitable tasks can vary across groups. The survey alone cannot isolate those explanations. The chart preserves the source's longer history, while the written observations below use the latest August wave.

How does use differ across weekly earnings groups?

Reported workplace use rises across the displayed earnings groups. The groups are weighted quartiles within each survey wave, not fixed dollar-income bands held constant over time. The earnings analysis applies the source's restrictions on positive earnings, hours and employment type, including exclusion of self-employed and family-business workers. The inventory retains those definitions. The association does not establish that AI use raised pay or that high earnings caused adoption. It instead identifies differences among the included employee groups in the same observation month.

Which industries and occupations report high worker use?

Industry and occupation describe different dimensions of a worker's job. Information-sector workers are not the same population as people in computer and mathematical occupations, and management roles can appear across sectors. The observations below preserve those distinct groups instead of combining them into a single ranking. They also differ from the Census sector estimates earlier: these are people reporting generative-AI use at work, while Census counts businesses reporting technology use in any function. The reference months and survey questions differ as well. No direct gap between the two sources is calculated.

Glossary

AI/SI
The label used in the latest cited Census questionnaire; not a claim that measured systems possess technical superintelligence.
RPS
Real-Time Population Survey.
Diffusion
Estimated user share under Microsoft's published methodology.

Methodology and sources

Every figure on this page comes from the organisation that produced it: regulators, statistics offices, public bodies, research institutes, and companies reporting their own data or surveys. Each one links to the original report or dataset and carries its release date. Survey results keep their population and fieldwork period, and financial figures keep the measure the producer used. Figures from different producers are not added together or ranked against each other. The page is reviewed monthly and updated when a producer releases newer data.