AI Statistics

What the latest research says about AI adoption, value, governance and skills, from McKinsey, BCG, Deloitte, PwC and more, alongside DoubleTrack's own research. Search it, or filter by theme and type of evidence.

Last updated November 2026 · 37 statistics

Planning an AI investment? See how we de-risk it

Everyone Is Using AI

Adoption's high in larger organizations, but smaller elsewhere; the number you read changes based on the research, but our analysis of Census Bureau data shows closer to 18% adoption across the United States.

88%

of organizations use AI regularly in at least one business function, up from 78% a year earlier.

McKinsey, The State of AI in 2025

Survey

Sample and caution

Sample: 1,993 respondents in 105 countries, fielded June 25 to July 29, 2025. 38% work at organizations above $1B in revenue.

Caution: The sample skews toward large organizations.

53%

of the population used generative AI within three years, faster than the PC or the internet reached that level.

Stanford HAI, AI Index 2026

Measured

Sample and caution

Sample: Population-level adoption across countries.

~70%

of firms use AI: 69% across four countries and 78% in the US.

NBER Working Paper 34836, 2026

Survey

Sample and caution

Sample: Nearly 6,000 CEOs, CFOs and senior executives in the US, UK, Germany and Australia, fielded November 2025 to January 2026.

Caution: A working paper, not yet peer reviewed.

$285.9B

of US private AI investment in 2025, against $20.9B in Europe and $12.4B in China.

Stanford HAI, AI Index 2026

Measured

Sample and caution

Sample: Investment totals by region.

18.6%

of US businesses use AI, or 23.3% across the largest metros. Counting every business, including the smallest, gives a lower number than surveys of large firms.

DoubleTrack, The AI Talent Gap, 2026

MeasuredDoubleTrack research

Sample and caution

Sample: Census Bureau Business Trends and Outlook Survey (BTOS) combined with Bureau of Labor Statistics data.

21%

of US workers say at least some of their work is done with AI, up from 16% in 2024. 65% still say they use AI little or not at all in their job.

Pew Research Center, About 1 in 5 U.S. Workers Now Use AI in Their Job, 2025

Survey

Sample and caution

Sample: A survey of US workers fielded in September 2025.

Caution: This counts workers, not businesses, so it is not directly comparable to the business adoption figures on this page.

Most Organizations Are Not Seeing the Value

Most organizations report little financial return so far. A small group is pulling ahead.

39%

of organizations report any enterprise-level EBIT impact from AI, and most of those say it is under 5% of EBIT. About 6% qualify as "high performers."

McKinsey, The State of AI in 2025

Survey

Sample and caution

Sample: The same 1,993-respondent survey.

Caution: The sample skews toward large organizations.

5%

of companies achieve AI value at scale. 35% are scaling and starting to see value. 60% report minimal revenue and cost gains despite substantial investment.

BCG, The Widening AI Value Gap, 2025

Survey

Sample and caution

Sample: 1,250 senior executives across nine industries. BCG scores maturity on 41 dimensions.

Caution: BCG defines "future-built" and sells transformation work.

Eric Heine, Chief Executive Officer at DoubleTrack

There’s a group of companies that are reluctant to approach AI because they’ve heard the horror stories of companies who spent millions of dollars and didn’t get any tangible business benefit from it. And then you’ve got some who are not even trying… frankly, they’re going to be crushed by the companies that are becoming successful with AI, and you have to be all-in.

Eric HeineChief Executive Officer, DoubleTrack

1.7×

the revenue growth and 1.6 times the EBIT margin: how BCG’s "future-built" 5% compare with laggards.

BCG, The Widening AI Value Gap, 2025

Survey

Sample and caution

Sample: The same 1,250-executive sample.

Caution: BCG’s own comparison of its own categories.

89%

of executives report no AI effect on labor productivity over the past three years, and more than 90% report none on employment. They forecast +1.4% productivity and -0.7% employment over the next three years.

NBER Working Paper 34836, 2026

Survey

Sample and caution

Sample: Nearly 6,000 executives, surveyed by the Atlanta Fed, Bank of England, Bundesbank and Macquarie University.

Caution: A working paper, not yet peer reviewed. Press coverage says productivity was measured as sales per employee.

56%

of CEOs report neither revenue nor cost benefits from AI. Only 12% report both.

PwC, 2026 Global CEO Survey

Survey

Sample and caution

Sample: 4,454 CEOs in 95 countries and territories, surveyed September 30 to November 10, 2025.

$617B

is what dirty data costs US businesses each year across about 139.8 million employees. The baseline is $4,912 per employee, and the industry-adjusted national average is about $4,400.

DoubleTrack, The Hidden Cost of Dirty Data, 2026

EstimateDoubleTrack research

Sample and caution

Sample: Built from Gartner’s $12.9M baseline, Census Bureau employment data and industry IT-spend multipliers.

Caution: A modeled estimate meant to show scale, not a measurement of any one company.

Annual cost of dirty data per employee, by industryIndustry rates range from $2,995 to $12,161 per employee. The industry-adjusted US average is about $4,400.
  • Information$12,161
  • Finance & Insurance$5,991
  • 13 industries at the baseline*$4,912
  • Professional, Scientific, Technical$4,014
  • Retail Trade$3,714
  • Transportation & Warehousing$3,175
  • Healthcare & Social Assistance$2,995
  • Manufacturing$2,995

*Thirteen of the 20 industries use the $4,912 baseline because no IT-spend data exists for them. Information covers software, tech hosting, media and telecom.

Source: DoubleTrack, The Hidden Cost of Dirty Data, 2026 Estimate DoubleTrack research

Where It Stalls: Data and Foundations

The projects that stall usually stall below the model, in the data and the architecture beneath it.

63%

of organizations do not have, or are not sure they have, the right data management practices for AI.

Gartner press release, 2025

Survey

Sample and caution

Sample: A survey of data management leaders in the second half of 2024. Gartner’s own materials report different sample sizes, so none is shown here.

Caution: Published in February 2025, slightly earlier than the rest of this page, and still the most direct data-readiness measure available.

60%

of AI projects that lack AI-ready data will be abandoned through 2026, according to Gartner.

Gartner press release, 2025

Forecast

Sample and caution

Sample: A Gartner forecast.

Caution: A prediction, not a measurement.

40%+

of agentic AI projects will be canceled by the end of 2027 because of cost, unclear value or weak risk controls, according to Gartner.

Gartner press release, 2025

Forecast

Sample and caution

Sample: A Gartner forecast.

Caution: A prediction, not a measurement.

Andy Boettcher, Chief Innovation Officer at DoubleTrack

Building AI agents isn’t the problem. I’m seeing these get shelved because solid outcomes weren’t put together ahead of time and clear data paths that pass the four Rs test aren’t laid in advance to make sure the right data gets in the right way.

Andy BoettcherChief Innovation Officer, DoubleTrackThe four Rs: relevant, reliable, revealing, reusable. See the approach →

Top 5%

BCG’s playbook for the top 5% of companies includes building a fit-for-purpose technology architecture and a data foundation.

BCG, The Widening AI Value Gap, 2025

Survey

Sample and caution

Sample: Qualitative finding from the 1,250-executive study.

46%

more AI roles are open than the data infrastructure roles needed to support them: 111,296 AI and machine learning postings against 76,271 data infrastructure postings.

DoubleTrack, Hiring Robots Before Mechanics, 2026

MeasuredDoubleTrack research

Sample and caution

Sample: Analysis of US job postings, with AI specialists commanding about $15,680 higher average salaries.

80% → 90%

A quoting agent launched at about 80% answer accuracy. After the team fixed outdated support documents it reached 90% within a month, and usage tripled.

DoubleTrack client work, The Four Ps of AI Readiness, 2026

MeasuredDoubleTrack research

Sample and caution

Sample: One anonymous client engagement.

Caution: A single case, shown as an illustration, not a benchmark.

Who is hiring AI before data foundations?How far one type of role outnumbers the other in open job postings, by sector. Regulated sectors lean toward data roles.

More data roles than AI roles

  • Hospitality295%
  • Manufacturing273%
  • Finance240%
  • Retail188%
  • Healthcare164%

More AI roles than data roles

  • Sales232%
  • Legal181%
  • Engineering168%
  • Marketing54%
  • Technology31%

Source: DoubleTrack, Hiring Robots Before Mechanics, 2026 Measured DoubleTrack research

Governance Is Behind Adoption

AI is being adopted faster than it is being governed, and the breach data shows the cost.

43%

of the breached organizations IBM studied had a shadow AI incident, up from 20% a year earlier. Those incidents averaged $5.39M, against $4.99M for all breaches.

IBM, Cost of a Data Breach 2026

Survey

Sample and caution

Sample: 602 breached organizations in 16 countries and 17 industries, studied by Ponemon. Breaches occurred March 2025 to February 2026.

Caution: IBM sells security products. Figures are as reported by IBM and corroborated across multiple outlets. Shadow AI means employees using AI tools nobody approved.

21%

of organizations had an AI-related breach, up from 13% a year earlier.

IBM, Cost of a Data Breach 2026

Survey

Sample and caution

Sample: The same 602-organization sample.

Caution: IBM sells security products.

97%

of organizations that reported a breach of their AI models or apps lacked proper AI access controls. 63% of breached organizations had no AI governance policy.

IBM, Cost of a Data Breach 2025

Survey

Sample and caution

Sample: 600 breached organizations, studied by Ponemon.

Caution: This is the 2025 edition of the report.

21%

of organizations have a mature governance model for autonomous agents, while nearly three-quarters (74%) plan to deploy agents within two years.

Deloitte, State of AI in the Enterprise 2026

Survey

Sample and caution

Sample: 3,235 business and IT leaders across 24 countries.

50%

of US adults say the increased use of AI in daily life makes them more concerned than excited. Just 10% say they are more excited, and 38% feel both equally. In 2021, 37% were more concerned.

Pew Research Center, How Americans View AI and Its Impact on People and Society, 2025

Survey

Sample and caution

Sample: A June 2025 survey of US adults.

Caution: This measures public sentiment, not business results.

Agents Are Still The Current Battleground

Most organizations continue to experiment with building agents, despite the fast rate of adoption. It's also still a bottleneck.

62%

of organizations are at least experimenting with AI agents, and 23% are scaling one somewhere. In any single function, no more than 10% are scaling agents.

McKinsey, The State of AI in 2025

Survey

Sample and caution

Sample: The 1,993-respondent survey.

Caution: The sample skews toward large organizations.

23% → 74%

of organizations use agentic AI at least moderately today. Within two years, 74% expect to, including 23% using it extensively and 5% fully integrating it.

Deloitte, State of AI in the Enterprise 2026

Survey

Sample and caution

Sample: 3,235 business and IT leaders across 24 countries.

Caution: The two-year figure is an expectation, not a measurement.

~130

of the thousands of vendors selling "agentic AI" are real, by Gartner’s estimate. The rest are rebranding assistants, chatbots and automation, which Gartner calls "agent washing."

Gartner press release, 2025

Estimate

Sample and caution

Sample: A Gartner estimate.

66.3%

is how accurately the best AI agents completed real computer tasks on the OSWorld benchmark in 2025, still below the 72.35% human baseline.

Stanford HAI, AI Index 2026

Measured

Sample and caution

Sample: OSWorld benchmark results, as compiled in the AI Index.

Caution: A test-environment benchmark, not a measure of results inside a business.

People and Skills

Buying AI is easy. Staffing it is the hard part.

62%

is the average wage premium for AI skills, up from 57% a year earlier.

PwC, 2026 Global AI Jobs Barometer

Measured

Sample and caution

Sample: More than one billion job ads in 27 countries.

Caution: Compares advertised wages in postings, not any one person’s pay change.

~20%

is how far employment for US software developers ages 22 to 25 had fallen from its 2022 peak by September 2025, while headcount for older developers kept growing.

Stanford HAI, AI Index 2026

Measured

Sample and caution

Sample: ADP payroll data analyzed by Brynjolfsson and colleagues, as reported in the AI Index.

Caution: The AI Index summary says "from 2024" in one place and "from its 2022 peak" in its detailed text. This card uses the detailed text. It is observational payroll data, and the report says the evidence does not point to broad, uniform job displacement.

1.3%

of US businesses have hired anyone trained to run AI, even though 18.6% use it.

DoubleTrack, The AI Talent Gap, 2026

MeasuredDoubleTrack research

Sample and caution

Sample: Census Bureau BTOS combined with Bureau of Labor Statistics data.

1 in 14

respondents say a lack of skilled AI workers is the problem. 61.6% say AI is not applicable to their business.

DoubleTrack, The AI Talent Gap, 2026

MeasuredDoubleTrack research

Sample and caution

Sample: Census Bureau BTOS responses.

17.6% → 30.6%

AI use rises with company size, from the smallest firms to the largest. By sector it ranges from 37.6% in Information to 5.1% in Agriculture.

DoubleTrack, The AI Talent Gap, 2026

MeasuredDoubleTrack research

Sample and caution

Sample: Census Bureau BTOS by company size and sector.

$130K

is the median pay for AI-compatible roles, roughly 2.6 times the typical US worker.

DoubleTrack, The AI Talent Gap, 2026

MeasuredDoubleTrack research

Sample and caution

Sample: Ten AI-compatible job titles, weighted by employment.

What businesses changed to adopt AIAmong businesses that adopted AI, nearly two-thirds made no real changes. Only 1.3% hired anyone trained to run it.
  • Made no changes64.3%
  • Developed new workflows15.4%
  • Trained existing staff15.0%
  • Used vendors or consultants3.8%
  • Hired AI-trained staff1.3%

Source: DoubleTrack, The AI Talent Gap, from Census Bureau BTOS data, 2026 Measured DoubleTrack research

About These Statistics

Last updated November 2026. This page collects research from 2025 and 2026 on how organizations are adopting AI and what is getting in the way.

Where the numbers come from. Each statistic links to the publisher's own report, announcement or page. Statistics are paraphrased rather than quoted, and each one shows its source, year, sample and any caution about how to read it.

Evidence types. Survey statistics reflect what respondents reported. Measured statistics come from observed data, such as job postings, incident counts or Census Bureau responses. Forecasts are predictions, not measurements. Estimates are modeled or analyst estimates.

Predictions and working papers. Gartner's figures about abandoned and canceled projects are forecasts and are labeled that way. The NBER findings come from a working paper that has not been peer reviewed.

Who published what. Several publishers sell services related to their research, including BCG and IBM. Where that matters, the caution on the statistic says so.

DoubleTrack research. Figures marked DoubleTrack research come from our published studies: The Hidden Cost of Dirty Data, Hiring Robots Before Mechanics and The AI Talent Gap. The dirty data total is a modeled estimate. The quoting-agent result comes from one anonymous client engagement, described in The Four Ps of AI Readiness.

Spot something out of date?Tell us and we will check it.

Cite this page. DoubleTrack. (2026). AI statistics. https://www.doubletrack.com/ai-statistics

See Where Your Data Stands In 3 Mins

Try It For Free

Architecture. Before activation.

© 2026 DoubleTrack. All rights reserved.

Company

(844) 4DATAVALUE
sales@doubletrack.com
AboutAI InformationClientsInsightsPartnersPrivacy Policy

Services

DealHub certified implementation partner badgeSalesforce Partner badge, partner since 2018