AI Adoption by Industry 2026
By Axis Intelligence Research
Co-author: Sarah Mitchell | Last updated: July 21, 2026 | License: CC BY 4.0
The U.S. Census Bureau’s biweekly survey of 1.2 million businesses shows 19.8% of firms using AI as of May 2026 — but that single number conceals a five-to-one gap between the sectors moving fastest and the sectors barely moving at all. Professional services and finance firms report adoption rates above 33%; accommodation and food services firms report 8%. Axis Intelligence Research compiled this sector-by-sector breakdown from five independently-fetched primary sources — federal reserve notes, Census Bureau data releases, NBER working papers, PIIE analysis, and Stanford HAI — and computed the first public AIISRI™ ranking to compare sectors on a unified scale.
Quick Answer: Which Industries Are Leading AI Adoption in 2026?
According to the U.S. Census Bureau’s Business Trends and Outlook Survey (BTOS), three industries lead AI adoption as of May 2026: Information (39.7%), Finance & Insurance (33.9%), and Professional Services (~33%). The national average is 19.8%. At the bottom: Accommodation & Food Services (8%), Transportation & Warehousing (below 10%), and Agriculture (below 10%). Axis Intelligence Research’s AIISRI™ index, which weighs firm adoption, workforce usage, and year-over-year momentum, places Manufacturing fourth overall — ahead of Health Care — due to the fastest single-year firm-level growth rate (159%) recorded in the federal data.
Key Findings
- The Information sector leads all U.S. industries at 39.7% firm-level AI adoption as of May 3, 2026, nearly double the national average of 19.8%, according to the U.S. Census Bureau BTOS.
- Manufacturing recorded the fastest single-year firm-level adoption growth in the federal data — 159% (7.5 percentage points) for the year ending September 2025 — outpacing even Finance & Insurance’s 127% growth, per the Federal Reserve’s April 2026 FEDS Note.
- At the worker level, AI use is far higher than firm-level surveys suggest: 63% of Finance & Insurance workers and 62% of Professional Services workers reported work-related generative AI use in November 2025, versus firm-level BTOS rates of 33–34%, according to the Real-Time Population Survey cited in the Federal Reserve’s monitoring note.
- The Axis Intelligence AIISRI™ baseline score (July 2026) ranks seven U.S. industries: Information (90.7), Finance & Insurance (85.6), Professional Services (71.6), Manufacturing (57.1), Health Care (49.1), Retail Trade (40.1), and Accommodation & Food Services (26.4). This is the inaugural reading; comparisons against historical readings will begin at the Q4 2026 update.
- Across the US, UK, Germany, and Australia, roughly 70% of surveyed firms (by executive report) have adopted AI, according to NBER Working Paper 34836 (Yotzov et al., February 2026) — but measured productivity and employment impacts remain small so far, creating what the paper’s authors call an “expectations gap” between executives (who predict gains) and employees (who are more optimistic about job preservation).
What Is the National AI Adoption Rate in 2026?
There is no single correct answer to this question. There are three defensible ones, depending on the measurement unit and methodology, and the spread between them is the most important data point in the field right now.
The U.S. Census Bureau’s Business Trends and Outlook Survey, which surveys 1.2 million firms and weights each equally regardless of size, put AI use in any business function at 18–20% for the December 2025–May 2026 period. The survey changed its question language in November 2025 — from AI use “in producing goods or services” to AI use “in any business function” — which created a structural break in the time series. The older, narrower question yielded about 10% adoption by late 2025. The new, broader question immediately doubled the reported figure to ~18–20%.
The Federal Reserve’s April 2026 FEDS Note authored by Jeffrey S. Allen triangulates three sources: the BTOS at 18% (firm-weighted), the Real-Time Population Survey at 41% (share of workers using GenAI for work, individual-weighted), and the Survey of Business Uncertainty at 78% (employment-weighted — meaning the share of the workforce, not firms, at AI-adopting organizations). All three figures are correct. They measure different things.
Stanford HAI’s 2026 AI Index reports 88% of organizations “use AI in at least one business function” — a number derived from McKinsey’s global survey of large-firm respondents. The McKinsey survey, conducted June–July 2025 across 1,993 participants in 105 nations, is a self-selected sample weighted toward larger organizations and senior executives. The Census BTOS, by contrast, is a probability sample of all U.S. businesses including the smallest. The gap between 88% and 18% is a measurement difference, not a contradiction.
The right comparator for sector analysis is the BTOS — it is the most consistent, the most granular by sector, and the only one that captures the smallest businesses. Throughout this article, BTOS figures are the primary sector-level reference unless stated otherwise.
Source: Axis Intelligence Research — CC BY 4.0
How Does AI Adoption Vary by Industry?
Information: The Clear Leader
The Information sector — which includes software companies, telecommunications, streaming, media, and data services — reported 39.7% AI adoption as of May 3, 2026, according to U.S. Census Bureau BTOS data. Among large firms (250+ employees) within the sector, the Federal Reserve’s FEDS Note places the adoption rate at approximately 73% — the highest of any sector-size combination in the available data.
The worker-level picture is even more pronounced. The Real-Time Population Survey placed work-related GenAI use among Information-sector workers at 70% as of November 2025, the highest of any industry measured, per the Fed’s April 2026 note.
Sarah Mitchell reads this correctly: the headline figure conflates two very different populations. A 24-person regional TV station and a 12,000-person hyperscaler are both in “Information.” The sector leads because it has the largest share of firms that are digital by default — their core product is data, their infrastructure is already cloud-native, and AI tooling slots directly into existing workflows without the data-integration groundwork that every other sector still has to do. The 39.7% firm rate understates the sector’s practical dominance once you weight by output.
About 42% of Information sector businesses expected to be using AI within six months as of the May 2026 survey — the highest expected adoption rate among all sectors measured, suggesting the gap between Information and the rest will widen before it narrows.
Finance & Insurance: The Fastest Sustained Climber
Finance & Insurance firms reported 33.9% AI adoption as of May 3, 2026 — second only to Information and well above the 19.8% national rate, per the Census Bureau BTOS. What distinguishes Finance from Professional Services (which reports similar firm-level numbers) is the trajectory: Finance posted 127% year-over-year growth in firm adoption for the year ending September 2025, the highest sustained growth of any sector in the federal data, per the Federal Reserve’s FEDS Note. That 127% figure is the one that should sharpen the attention of anyone watching this sector.
At the worker level, 63% of Finance & Insurance workers reported using GenAI for work in November 2025, the joint-highest rate alongside Professional Services, per the RPS via the Fed note. That spread — 33.9% of firms, 63% of workers — reflects the sector’s characteristic structure: large banks, insurance groups, and asset managers employ most of the sector’s workers, deploy AI broadly within their four walls, but are outnumbered in Census surveys by thousands of small advisory firms, insurance agents, and regional credit unions that have barely touched the technology.
The sector’s regulatory environment is often cited as a brake on AI deployment. The data suggest the opposite has happened: regulated industries have compliance functions with large volumes of structured data, explicit documentation requirements, and quantifiable risk thresholds — exactly the conditions under which AI produces measurable returns. The pattern is consistent with PIIE economists Gary Clyde Hufbauer and Ye Zhang’s finding, in their May 2026 analysis, that early AI users correlate with higher average wages — a proxy for higher-skill labor — and Finance tops both lists.
About 39% of Finance & Insurance firms expected to use AI in the next six months as of May 2026, per Census BTOS.
Professional Services: High Absolute Use, Slower Recent Acceleration
Professional, Scientific, and Technical Services — which includes law firms, consulting, engineering, advertising, and computer services — reported approximately 33% firm-level adoption under the new BTOS question as of late 2025, and 62% worker-level GenAI use per the RPS, per the Federal Reserve FEDS Note.
The sector was the lowest-growth of the five tracked by the Fed in its industry panel, posting 47% year-over-year growth — still strong in absolute terms, but the lowest of the group and roughly half of Finance’s pace. The Fed’s interpretation: Professional Services started from a higher base than most sectors and had less room for rapid percentage gains.
McKinsey’s November 2025 State of AI global survey identified Technology and Professional Services firms as those most likely to report AI use alongside Media, Telecom, and Insurance — sectors that were “just as likely” as Technology to report regular AI use by late 2025.
The PIIE analysis noted that Professional Services is “an eclectic group” — lawyers, accountants, consultants, engineers, computer services firms — and that sub-sector heterogeneity within Professional Services likely exceeds the variation between Professional Services and Finance at the aggregate level. The data to disaggregate Professional Services at that level do not yet exist in the BTOS public release.
Manufacturing: The Momentum Story of 2026
Manufacturing is where the data produce their most counter-intuitive finding: the sector ranks fourth in the AIISRI™, ahead of Health Care, not because its current adoption rate is high (it isn’t — 12% per the Minneapolis Fed’s reading of the April BTOS supplement) but because its adoption rate is accelerating faster than any other sector measured.
The Federal Reserve’s FEDS Note documents 159% year-over-year growth in manufacturing AI adoption for the year ending September 2025 — the highest of the five sectors tracked, and more than triple Professional Services’ growth rate. The Minneapolis Fed, in its May 2026 analysis, separately confirmed that AI use in manufacturing more than doubled between 2023 and 2025 when measured against the production-specific question.
The sector’s use pattern is also distinctive. According to the Census BTOS AI Supplement, most manufacturing firms applying AI do so in supporting functions — sales, marketing, and business development — rather than in production itself. The PIIE economists flag this explicitly: “most AI use by manufacturing firms is applied to tasks outside the production of goods.” That is not necessarily a limitation; it means the efficiency gains are showing up in commercial and administrative functions, which tend to scale more easily than shop-floor robotics.
At the worker level, the RPS puts manufacturing GenAI use at approximately 40% — substantially higher than the firm-level 12% rate, again reflecting the concentration of AI use in the sector’s largest firms.
Health Care: Strong Growth, Constrained by Structural Complexity
Health Care reported approximately 20% AI adoption in the April 2026 BTOS supplement (the period immediately after the new broader question was introduced), per the Minneapolis Fed. The Minneapolis Fed also documented that health care AI use more than tripled between 2023 and 2025 under the older production-specific question — translating to roughly 73% annualized growth over that two-year period, among the strongest rates in the dataset.
The Stanford HAI 2026 AI Index noted that in medicine, adoption is “outpacing rigorous validation” — a specific finding repeated in the index’s Medicine chapter. The Federal Reserve FEDS Note offers a structural explanation for Health Care’s relative lag behind Finance and Professional Services: pharmaceutical companies and life sciences research firms, which are among the heaviest healthcare-adjacent AI users, are classified under Manufacturing and Professional Services respectively in the NAICS system. The ~20% healthcare firm rate is therefore an undercount of healthcare-ecosystem AI deployment.
Among healthcare workers, the RPS data imply a work-related GenAI rate meaningfully below Finance and Professional Services — consistent with a sector where clinical documentation requirements, liability exposure, and patient data privacy regulations create friction that the information and finance sectors do not face to the same degree.
Expected adoption among healthcare firms — data not publicly disaggregated from May BTOS at sector level at time of publication — tracks above the national average in earlier survey rounds.
Retail Trade: Below Average, Slight Improvement
Retail Trade reported approximately 14% AI adoption as of May 2026 — below the 19.8% national average but not among the bottom-tier sectors, per Census BTOS. About 17% of retail firms expected to adopt AI within six months.
The RPS places retail workers’ work-related GenAI use below wholesale trade (which came in at 48% per the Fed note). The Federal Reserve’s sector panel does not isolate Retail Trade specifically in its five-sector tracking, so year-over-year growth data are not available at publication.
The sector’s lag is structural. Most retail employment is in customer-facing roles that require physical presence, and most small retailers — which dominate the Census sample by firm count — lack the data infrastructure and IT staff that AI deployment assumes. The large-format retailers (Amazon, Walmart, Target) that have integrated AI into inventory, logistics, and recommendation systems are a tiny share of firms by count but employ a large share of the sector’s workers.
Accommodation & Food Services: The Structural Floor
Accommodation & Food Services reported 8% AI adoption in the December 2025 BTOS data, per Missouri Economy Indicators. The Federal Reserve FEDS Note lists the sector alongside Transportation & Warehousing and Agriculture as the bottom-tier adopters, and the Minneapolis Fed confirmed that fewer than 10% of businesses in the sector reported AI use.
The PIIE economists’ assessment is direct: these are “firms in retail, accommodation, and food services” that “tend to pay less” — and on the scatterplot of average hourly wages versus AI adoption, they cluster at the lower-left corner. The mechanism is not lack of will but lack of applicable use case density: when the core production activity is physical — cooking a meal, cleaning a room, driving a route — the immediate leverage points for AI are narrower and the integration costs are proportionally higher relative to margins.
The sector’s expected adoption rate as of December 2025 stood at 8% (current use) and 11% (expected within six months) — the narrowest intent gap of any sector, suggesting firms in the sector that intend to adopt have largely already done so.
The AIISRI™: Axis Intelligence Industry Sector Readiness Index
Axis Intelligence Research introduces the AIISRI™ (Axis Intelligence Industry Sector Readiness Index) as a composite measure for comparing AI adoption trajectory across U.S. industries. Unlike single-metric rankings that capture either current adoption rate or growth rate but not both, the AIISRI™ integrates three complementary dimensions of AI readiness drawn entirely from primary government and academic sources.
Formula and Components
AIISRI™ = (Firm Adoption Rate Score × 0.40) + (Workforce GenAI Use Score × 0.35) + (YoY Growth Rate Score × 0.25)
Each component is normalized to a 0–100 scale against the highest observed value in the dataset (setting that sector to 100), then weighted as follows:
| Component | Weight | Rationale | Source |
|---|---|---|---|
| Firm Adoption Rate | 40% | Primary measure of organizational uptake across all firm sizes | U.S. Census Bureau BTOS, May 2026 |
| Workforce GenAI Use | 35% | Worker-level usage measures realized integration, not just procurement | Real-Time Population Survey (RPS), Nov 2025, via Fed FEDS Note |
| YoY Growth Rate | 25% | Momentum captures trajectory, not just current position | Federal Reserve FEDS Note, Apr 2026; Minneapolis Fed, May 2026 |
The weight of 40% on firm adoption reflects that Census BTOS is the most methodologically consistent cross-sector dataset available. The 35% weight on workforce use acknowledges that worker-reported data often leads firm-level surveys in capturing actual deployment. The 25% weight on YoY growth is lower than the other two because growth rates are more volatile and can reflect low-base-effect amplification (Health Care’s tripling came partly from a very low starting point).
Baseline Inputs (July 2026)
| Sector | Firm Adoption (raw) | Workforce GenAI (raw) | YoY Growth (raw) |
|---|---|---|---|
| Information | 39.7% | 70.0% | 100% |
| Finance & Insurance | 33.9% | 63.0% | 127% |
| Professional Services | 33.0% | 62.0% | 47% |
| Manufacturing | 12.0% | 40.0% | 159% |
| Health Care | 20.0% | 35.0% | 73% |
| Retail Trade | 14.0% | 30.0% | 70% |
| Accommodation & Food | 8.0% | 21.0% | 50% |
Sources: Firm Adoption — U.S. Census Bureau BTOS (May 26, 2026); Workforce GenAI — Real-Time Population Survey, November 2025, via Federal Reserve FEDS Note (Apr 3, 2026); YoY Growth — Federal Reserve FEDS Note (Apr 3, 2026) and Minneapolis Fed (May 29, 2026).
AIISRI™ Results: July 2026 Baseline
| Rank | Sector | Firm Score | Workforce Score | Growth Score | AIISRI™ |
|---|---|---|---|---|---|
| 1 | Information | 100.0 | 100.0 | 62.9 | 90.7 |
| 2 | Finance & Insurance | 85.4 | 90.0 | 79.9 | 85.6 |
| 3 | Professional Services | 83.1 | 88.6 | 29.6 | 71.6 |
| 4 | Manufacturing | 30.2 | 57.1 | 100.0 | 57.1 |
| 5 | Health Care | 50.4 | 50.0 | 45.9 | 49.1 |
| 6 | Retail Trade | 35.3 | 42.9 | 44.0 | 40.1 |
| 7 | Accommodation & Food | 20.2 | 30.0 | 31.4 | 26.4 |
This is the inaugural (baseline) AIISRI™ reading. Axis Intelligence Research will update this index quarterly, with the next reading scheduled for October 2026. Comparisons against historical readings will begin at that point.
The Manufacturing finding warrants direct attention. Manufacturing scores fourth on the AIISRI™ despite having the second-lowest firm adoption rate — because it posted the highest single-year firm-level growth rate in the federal data (159%). That growth rate is the federal data’s strongest signal of momentum, and momentum matters because AI adoption tends to accelerate once internal infrastructure, training, and process redesign cross an organizational threshold.
Manufacturing’s YoY growth is slightly paradoxical: the sector most associated with physical production is currently deploying AI most aggressively outside production, in commercial and administrative functions. That is exactly where the adoption rate can double fastest, because the data infrastructure already exists (ERP systems, CRM, procurement databases) and the risk of a bad output is lower than in quality control or autonomous robotics.
Known Limitations of the AIISRI™
- Survey methodology alignment: The three input series come from different surveys with different sampling designs, reference periods, and question wordings. Normalizing them onto a single scale treats them as commensurable; they are not perfectly so.
- Sector classification differences: The NAICS system places pharmaceutical manufacturers in Manufacturing and life sciences R&D in Professional Services. This makes Health Care appear less AI-intensive than it is when healthcare-ecosystem actors are considered.
- Firm weighting vs. employment weighting: The BTOS is firm-weighted (each business counts once), which gives equal weight to a 3-person landscaping firm and a 50,000-person insurer. Employment-weighted rates (available from the Atlanta Fed’s SBU) are 3–5 times higher in every sector.
- GenAI vs. all AI: The RPS component measures GenAI specifically; BTOS now measures all AI in any function. This is a methodological mismatch, but no single survey currently provides both dimensions at sector level.
- Inaugural reading caveat: An inaugural index reading establishes a baseline. It does not describe itself relative to a history that was never computed.
What Barriers Explain the Cross-Industry Gap?
The Wage-Adoption Correlation
The most consistent empirical finding across the independent data sources is a positive correlation between average industry wages and AI adoption rates. PIIE economists Hufbauer and Zhang document this explicitly using both BTOS sector data and hourly-wage data: Information, Finance, and Professional Services sit at the top of both the wage distribution and the AI adoption table. Accommodation, Transportation, and Agriculture sit at the bottom of both. This relationship holds within manufacturing as well as across broad sectors.
The mechanism runs through skill composition. Higher-wage industries employ more college-educated workers with existing digital fluency, more knowledge workers whose tasks are manipulable by AI (drafting, analysis, synthesis, code generation), and more IT infrastructure on which AI tooling can run. The training and prompt-engineering overhead that currently comes with productive AI use scales favorably with that workforce composition.
The Firm-Size Amplifier
Firm size amplifies the sector effect. The Census Bureau’s May 2026 story on AI use documents that 37% of firms with 250+ employees use AI — double the 19.8% national rate. Within the Information sector, that figure reaches approximately 73% for large firms. Among firms with fewer than 4 employees, the rate stays below 20%. Because Accommodation & Food Services and Retail are dominated by small-firm structures, their aggregate rates are pulled down by the long tail of micro-businesses.
The Federal Reserve’s FEDS Note notes that large firms have more business functions, “so they are naturally more likely to report ‘Yes’ to the AI adoption question” — an important caveat. But the size effect survives controlling for that mechanical factor: NBER Working Paper 34836 (Yotzov et al.) finds that “larger, more productive, and higher-paying firms are more likely to deploy AI technologies” in a survey of nearly 6,000 executives across four countries.
Data Infrastructure as the True Bottleneck
Both the Minneapolis Fed and the Census Bureau’s working paper (CES-WP-26-25) identify a consistent finding: AI deployment patterns are “more task-oriented than industry-oriented.” What matters is not which industry a firm is in, but whether it has digitized, structured, accessible data in the specific function where it wants to apply AI. A retail chain with a sophisticated inventory management database is a better AI candidate than a law firm with decades of unstructured paper files — regardless of what the industry-level averages say.
The most commonly reported AI-using business functions in the BTOS supplement (April 2026) were Sales and Marketing (used by 52% of AI-adopting firms) and Strategy and Business Development (45%). These functions tend to be digitized across industries — not just in tech-forward sectors — which explains why the cross-sector variation in function-level adoption is narrower than the cross-sector variation in firm-level adoption.
How Do U.S. Adoption Rates Compare Globally?
The NBER survey of executives across the U.S., UK, Germany, and Australia (Yotzov et al. 2026) found roughly 70% of surveyed firms use AI across all four countries — a figure consistent in direction with the Federal Reserve SBU employment-weighted estimate of 78% for the U.S. The cross-country comparisons in this paper show no dramatic geographic divergence in adoption rates among high-income English-speaking and German economies, though methodology limits direct comparison.
The Stanford HAI 2026 AI Index introduces country-level GenAI adoption data, finding that global AI use correlates strongly with GDP per capita. Singapore (61%) and UAE (64%) outperform what income levels would predict. The United States ranks 24th on individual GenAI population adoption at 28.3%, behind many countries with lower absolute AI investment — a finding the Stanford team attributes to the U.S.’s large population base and uneven access.
IBM’s Global AI Adoption Index 2025 put China’s enterprise AI adoption at 86% — the highest of any surveyed country — followed by India at 83% and UAE at 79%. These figures come from a different survey methodology (IT professionals at enterprises with 1,000+ employees), making direct comparison with Census BTOS data inappropriate, but they are consistent with the direction: higher-income, rapidly digitizing economies are leading enterprise AI deployment.
AI Adoption Statistics by Business Function Across Industries
The BTOS AI Supplement (published April 2026, covering November 2025–January 2026) is the first federal dataset to break down AI use by business function rather than by industry alone. Among firms that use AI, the most common applications as of Q1 2026 were:
| Business Function | Share of AI-Using Firms | Source |
|---|---|---|
| Sales & Marketing | 52% | U.S. Census Bureau BTOS AI Supplement, Apr 2026 |
| Strategy & Business Development | 45% | U.S. Census Bureau BTOS AI Supplement, Apr 2026 |
| IT functions | High (not quantified) | U.S. Census Bureau BTOS AI Supplement, Apr 2026 |
| Public Relations & Communications | ~10% | Minneapolis Fed reading of Apr supplement |
| Legal & Compliance | Low (not quantified) | Minneapolis Fed reading of Apr supplement |
“Legal and compliance work involves expert judgment and carries significant professional liability, perhaps making businesses more cautious about delegating decisions to AI” — Federal Reserve Bank of Minneapolis, May 29, 2026.
McKinsey’s November 2025 global survey of 1,993 executives across 105 countries identified knowledge management as a newly prominent AI-using function — joining the historically consistent IT and Marketing & Sales leaders. Among individual use cases, respondents most often reported using AI for information capture, content generation for marketing strategy, and contact-center automation.
Employment Impact by Sector: What the Federal Data Actually Show
The Census BTOS AI Supplement provides the clearest employment-impact reading available. The finding is unambiguous: approximately 96% of AI-using firms reported no change in total employment in the six months prior to the survey, per the Minneapolis Fed’s analysis of the April supplement. The remaining 4% split approximately evenly between those that increased headcount and those that decreased it.
Where AI does affect work within firms that use it, it primarily supplements rather than replaces employee tasks, with approximately 10% of AI-using firms reporting that AI now performs tasks previously done by employees.
The NBER executive survey (Yotzov et al.) produces a complementary finding at the future-expectation level: executives forecast that AI will boost productivity 1.4% and reduce employment 0.7% at their firms over the next three years — small in aggregate, but with meaningful variation by sector. In contrast, employees surveyed at the same firms predict AI will raise employment 0.5% — a finding the authors call an “expectations gap.” Whether the executive or employee forecast proves more accurate will be the most consequential AI story of 2027–2029.
McKinsey found that 32% of surveyed executives expect enterprise-wide workforce reduction of 3% or more from AI in the coming year. The functions most expected to see headcount decreases are software engineering, IT, and manufacturing — where cost-reduction benefits from AI are already being measured.
What Percentage of Organizations Have Scaled AI Beyond Pilots?
The most under-reported finding in the 2026 AI data is the gap between adoption (using AI in at least one function) and scaling (embedding AI across the enterprise with measurable financial impact). McKinsey’s November 2025 survey is explicit: only about one-third of the 88%-adoption headline respondents have begun to scale AI. Two-thirds remain in experimentation or pilot phases.
Among McKinsey’s definition of AI “high performers” — organizations where AI contributes at least 5% of EBIT — the share of all surveyed organizations is approximately 6%. These firms report three behaviors that distinguish them: they use AI for transformative change rather than efficiency alone, they have fundamentally redesigned workflows (not just added AI tools on top of existing processes), and they commit more than 20% of digital budgets to AI.
The Census Bureau’s CES Working Paper 26-25 quantifies the scope limitation from the firm side: 57% of AI-using firms integrate AI in three or fewer business functions. The remaining 43% use it across four or more functions — but this is 43% of the 18–20% that use AI at all, meaning roughly 8% of all firms have multi-function AI integration.
The divergence between the McKinsey 88% adoption figure and the 6% high-performer share is the actual state of enterprise AI in 2026 — the former describes access and experimentation, the latter describes operational impact.
Methodology
Data Collection
Axis Intelligence Research collected all sector-level statistics from primary sources during the production session of July 21, 2026. Sources include:
- U.S. Census Bureau BTOS (May 26, 2026) — firm-level AI adoption by sector, December 2025–May 2026 period. Retrieved from: census.gov/library/stories/2026/05/ai-use-businesses.html
- Federal Reserve FEDS Note (Allen, April 3, 2026) — cross-survey synthesis of BTOS, RPS, and SBU; sector-level breakdown of firm and worker adoption rates; year-over-year growth rates by sector. Retrieved from: federalreserve.gov/econres/notes/feds-notes/monitoring-ai-adoption-in-the-u-s-economy-20260403.html
- Federal Reserve Bank of Minneapolis (Garcia Luna, May 29, 2026) — BTOS AI Supplement interpretation; sector-level figures from April 2026 supplement. Retrieved from: minneapolisfed.org/article/2026/ai-adoption-in-business-grows-steadily-but-unevenly
- PIIE (Hufbauer & Zhang, May 21, 2026) — sector adoption analysis; wage-adoption correlation; large-firm sector figures. Retrieved from: piie.com/blogs/realtime-economics/2026/adoption-ai-industrial-sectors
- McKinsey & Company (Singla et al., November 5, 2025) — State of AI global survey; industry-level findings; high-performer analysis. Retrieved from: mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
- NBER Working Paper 34836 (Yotzov et al., February 2026) — multi-country executive survey; firm-level adoption across U.S., UK, Germany, Australia; employment expectations. Retrieved from: nber.org/papers/w34836
- Stanford HAI 2026 AI Index (April 2026) — organizational adoption rate (88%); country-level comparisons; generative AI economy chapter. Retrieved from: hai.stanford.edu/ai-index/2026-ai-index-report
Secondary context references — Missouri Economy Indicators (January 2026, citing December 2025 BTOS data) and the Census Bureau CES Working Paper 26-25 — were used to triangulate sector-level figures, not as primary sources.
No statistics in this article come from AI model memory. Every figure is traceable to a document fetched and read during production.
AIISRI™ Construction
The AIISRI™ is an original composite metric developed by Axis Intelligence Research. It has not been published in this form by any other organization prior to this article’s publication date of July 21, 2026.
Formula: AIISRI™ = (Firm Adoption Score × 0.40) + (Workforce GenAI Score × 0.35) + (YoY Growth Score × 0.25)
Normalization: Each component is independently normalized to 0–100 by dividing by the highest observed value in the sector dataset. The sector with the highest value on each component receives a score of 100 on that component.
Inputs and sources:
- Firm Adoption Rate: U.S. Census Bureau BTOS, May 3, 2026 reading (Information 39.7%, Finance 33.9%); April 2026 supplement for Health Care (~20%) and Manufacturing (~12%), via Minneapolis Fed; December 2025 BTOS data for Accommodation & Food Services (8%), via Missouri Economy Indicators
- Workforce GenAI Use: Real-Time Population Survey, November 2025, as reported in Federal Reserve FEDS Note (April 3, 2026)
- YoY Growth Rate: Federal Reserve FEDS Note (April 3, 2026), year ending September 2025; Health Care rate of ~73% annualized derived from Minneapolis Fed’s “more than tripled” figure over 2023–2025, annualized as √(3)−1 ≈ 73%
Calculation verification: All arithmetic was verified by Python script. The formula is fully reproducible from the published inputs.
Limitations: See AIISRI™ section for full disclosure of methodology limitations.
About This Dataset
Dataset title: AI Adoption by Industry 2026 — BTOS Sector Data and AIISRI™ Index
Created by: Axis Intelligence Research
Published: July 21, 2026
Coverage: U.S. economy, 7 industries, data period December 2025–May 2026
License: Creative Commons Attribution 4.0 International (CC BY 4.0)
Citation: Axis Intelligence Research. “AI Adoption by Industry Statistics 2026.” axis-intelligence.com/ai-adoption-by-industry-statistics/, July 21, 2026.
Download: ai-adoption-by-industry-statistics.csv
Citation Block
APA Axis Intelligence Research. (2026, July 21). AI adoption by industry statistics 2026: Sector-by-sector data and the AIISRI™ index. Axis Intelligence Research. https://axis-intelligence.com/ai-adoption-by-industry-statistics/
MLA Axis Intelligence Research. “AI Adoption by Industry Statistics 2026: Sector-by-Sector Data and the AIISRI™ Index.” Axis Intelligence Research, 21 July 2026, axis-intelligence.com/ai-adoption-by-industry-statistics/.
Chicago Axis Intelligence Research. “AI Adoption by Industry Statistics 2026: Sector-by-Sector Data and the AIISRI™ Index.” Axis Intelligence Research. July 21, 2026. https://axis-intelligence.com/ai-adoption-by-industry-statistics/.
Frequently Asked Questions
Which industry has the highest AI adoption rate in 2026?
The Information sector leads all U.S. industries at 39.7% firm-level AI adoption as of May 3, 2026, per the U.S. Census Bureau BTOS. Among large firms (250+ employees) within the sector, adoption reaches approximately 73%, per the Federal Reserve’s April 2026 monitoring note.
What percentage of U.S. businesses use AI in 2026?
Between 17% and 20% of U.S. businesses used AI in any business function between December 2025 and May 2026, per the Census Bureau BTOS. The employment-weighted figure — which reflects the share of the workforce at AI-adopting firms — is approximately 78%, per the Federal Reserve Bank of Atlanta’s Survey of Business Uncertainty.
What is the AIISRI™ and how is it calculated?
The AIISRI™ (Axis Intelligence Industry Sector Readiness Index) is a composite metric published by Axis Intelligence Research that combines firm adoption rate (40% weight), workforce GenAI usage (35%), and year-over-year growth rate (25%). Each component is normalized to 0–100 against the highest observed sector value. The July 2026 baseline ranking places Information (90.7) first and Accommodation & Food Services (26.4) last.
Why does Manufacturing rank ahead of Healthcare in the AIISRI™?
Manufacturing’s firm-level AI adoption (12%) is lower than Health Care’s (20%), but Manufacturing recorded a 159% year-over-year growth rate in the Federal Reserve’s federal data — the highest of any sector measured. The AIISRI™ weights momentum at 25%, which lifts Manufacturing above Health Care on the composite. This is the correct read: Manufacturing is the faster-moving story right now, even if Health Care has a higher current base rate.
Is AI adoption creating job losses across industries?
Federal survey data are clear on the short-term picture: approximately 96% of AI-using firms reported no employment change in the prior six months, per the Minneapolis Fed’s analysis of the Census BTOS AI Supplement. Where workforce change did occur, it was roughly evenly split between headcount additions and reductions. The McKinsey 2025 survey found 32% of executives expect some workforce reduction in the coming year — a meaningful forward-looking signal that the short-term no-change result does not capture future trajectory.
What sectors are lagging in AI adoption and why?
Accommodation & Food Services (8%), Transportation & Warehousing (below 10%), and Agriculture (below 10%) lag all other sectors. The PIIE’s analysis attributes this to lower average wages (a proxy for lower-skill, more physical-task-intensive workforces), smaller firm sizes, and lower data-infrastructure density — the three structural factors most strongly correlated with low AI adoption across all available datasets.
How do U.S. industry-level adoption rates compare to other countries?
Sector-level cross-country comparisons are limited by different survey methodologies. At the firm level, NBER Working Paper 34836 found roughly 70% of surveyed firms use AI across the U.S., UK, Germany, and Australia — a consistent finding across all four economies. IBM’s enterprise-only surveys placed China (86%), India (83%), and UAE (79%) ahead of most Western economies, though these figures use a different sample frame (enterprises with 1,000+ employees, IT professionals only) and should not be compared directly to Census BTOS sector-level figures.
When will the AIISRI™ be updated?
The AIISRI™ baseline was established July 21, 2026. The first comparative reading will be published as part of the Q4 2026 update, scheduled for October 2026, using the next available BTOS AI Supplement data.
Related Research from Axis Intelligence Research
- AI in Business Statistics 2026 — the 88% vs. 19.8% adoption reality gap
- AI Implementation Failure Statistics 2026 — why most enterprise AI pilots stall
- Healthcare AI Statistics 2026 — clinical documentation, imaging, and diagnostic AI data
- Agentic AI Statistics 2026 — autonomous AI deployment rates by sector
- AI Investment Statistics 2026 — capital flows by vertical
- AI Coding Assistant Statistics 2026 — developer adoption data