AI Unite
Suggest a report
MIT NANDA · July 2025

The GenAI Divide: State of AI in Business 2025

MIT NANDA's The GenAI Divide: State of AI in Business 2025 (July 2025; n=153 survey responses + 52 organization interviews + 300+ public initiatives reviewed, fielded Jan–Jun 2025) found that 95% of organizations are getting zero return from GenAI, per the report's executive summary.

Read the original report ↗Cite2 min read · Summary updated
Only 5% of organizations got enterprise AI tools into productionPercent of organizations, embedded or task-specific GenAI tools, by stage reached, 2025
Investigated
60%
Piloted
20%
Successfully implemented
5%
Source: MIT NANDA, The GenAI Divide: State of AI in Business 2025, 2025, p. 6.

Key findings

  1. 01
    Just 5% of integrated AI pilots are extracting millions in value; most show no measurable P&L impact.
  2. 02
    Over 80% of organizations explored or piloted tools like ChatGPT/Copilot; only 5% of custom enterprise AI tools reached production.
  3. 03
    In the interview sample, externally partnered tools reached deployment about 67% of the time versus about 33% for internally built tools.

By the numbers

95%of organizations are getting zero return from GenAI, per the report's executive summary
5%of organizations got custom or vendor-sold enterprise AI tools into production, after 60% evaluated them
90%of surveyed companies had workers regularly using personal AI tools for work; only 40% had bought an official LLM subscription
Bought AI tools reached deployment twice as often as internal buildsPercent of deployments reaching production, by approach (interview sample), 2025
Strategic partnerships (buy)
66%
Internal development (build)
33%
Source: MIT NANDA, The GenAI Divide: State of AI in Business 2025, 2025, p. 19.

What it means for you Draft

For executives at $10–100M companies

The widely quoted '95% fail' number rests on a small, self-selected sample, so we would not treat it as a precise failure rate. The more useful lesson is directional: pilots that never get wired into a real workflow rarely pay off. Before starting another pilot, it is worth naming the process it will change and how we will measure it.

Limitations

Low trust.Thin, conference-sourced sample (153 surveys, 52 interviews); authors call figures directional; success definition loose. Publisher does not sell AI.

The headline figure measures the “Pays off” stage of adoption (measurable profit impact).Why adoption numbers disagree

Cite the original

MIT NANDA. "The GenAI Divide: State of AI in Business 2025." July 2025. https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf

Related reports

McKinsey · Aug 2026The state of AI in 2026: On the road to ROI80% say AI improved their individual productivity, yet only about 6% are AI high performers, flat since 2025.RSM · Jul 2026RSM Middle Market AI Survey 2026: U.S. and Canada86% have partially or fully integrated AI into operations, and 97% say they are satisfied with the business value AI delivers.Deltek · May 202647th Annual Deltek Clarity Architecture & Engineering Industry StudyAI adoption among A&E firms rose from 53% to 70% year over year; generative AI use rose from 64% to 78%.