AI project failure rates: evidence and interpretation
Our team compiled data from 10 unique sources to estimate AI project failure rates as of 2026 and found that more than 80% of AI projects fail to deliver their intended business value. Because each source had a different methodology for calculating failure, our model used a weighted average of all sources, with the weights based on the source’s longevity, credibility, and reputed accuracy. Further, we applied our model across industries, root causes, project stages, and company sizes to create a complete picture of where AI initiatives fall short.
AI project failure rate (2026)
The exact failure rate varies depending on how failure is defined, which institution conducted the research, and which industry is measured. The table below shows how leading research organisations measure AI project failure as of 2026.
| Institution | Failure Rates | Year |
| RAND Corporation | 80%+ of AI projects fail | 2024 |
| BCG | 60% of companies report no material AI value | 2025 |
| BCG | Only 5% of companies achieve substantial AI value | 2025 |
| KPMG | 95%+ of AI use cases fail to deliver measurable value | 2025 |
| McKinsey | Only 6% of organisations see real business value from AI | 2025 |
| Gartner | 60% of AI projects lacking AI-ready data will be abandoned through 2026 | 2025 |
The data reveals that while each institution measures failure differently, all converge on the same conclusion: fewer than 1 in 10 organisations achieve real business value from AI.
AI project abandonment rate by year (2022-2025)
The share of enterprises abandoning AI initiatives has risen sharply. The chart below tracks the percentage of companies that abandoned most of their AI initiatives each year.
| Year | Companies Abandoning Most AI Initiatives | Source |
| 2022 | ~8% | Directional Estimate |
| 2023 | ~12% | Directional Estimate |
| 2024 | ~17% | S&P Global Market Intelligence, 2025 |
| 2025 | ~42% | S&P Global Market Intelligence, 2025 |
2022 and 2023 figures are directional estimates based on trend interpolation.
The data reveals that the abandonment rate increased by 147% between 2024 and 2025 alone, and at least 50% of generative AI projects were abandoned after proof of concept in 2025, exceeding analyst predictions of 30%.
AI pilot-to-production rate (2026)
The table below shows how AI projects convert from proof of concept to full production deployment, across all AI types and specifically for generative AI.
| Metric | Rate |
| AI POCs that fail to reach wide-scale deployment | 88% |
| GenAI pilots that fail to scale to production | 95% |
| Custom enterprise AI tools that reach production | 5% |
| GenAI POCs abandoned after proof of concept (2025) | 50%+ |
| POCs launched per 4 that graduate to production | 33 |
The data reveals that between 88% and 95% of AI pilots fail to reach meaningful production deployment, with MIT Project NANDA finding that only 5% of custom enterprise AI tools ever reach production.
AI project failure rate by root cause (2026)
The chart below ranks the most common root causes of AI project failure by the percentage of organisations or projects affected.
| Root Cause | Finding | Source |
| Lack of AI-ready data management practices | Affects 63% of organisations | Gartner, 2025 |
| Leadership and strategy failures | Account for 84% of all AI project failures | RAND, 2024 |
| Employees not trained for AI transformation | Only 36% of employees have been trained for AI transformation | BCG, 2026 |
| No financial KPI tracking for AI initiatives | Only 5% of companies achieve substantial AI value | BCG, 2025 |
| Executive-employee misalignment on AI | 76% of executives feel positive about AI; only 31% of employees agree | BCG/HBR, 2025 |
The data reveals that data readiness and leadership decisions are the two dominant failure drivers, consistent across RAND, Gartner, and BCG research independently. 84% of AI project failures are primarily attributable to leadership decisions rather than technical limitations (RAND Corporation, 2024).
AI project failure rate by industry (2026)
Figures marked with an asterisk (*) are directional estimates derived from sector-specific regulatory, data, and adoption research, weighted against the global baseline. The five unmarked figures are drawn from primary research across 2,400+ enterprise AI initiatives.
| Industry | Failure Rate | Primary Failure Driver |
| Government and Public Sector | 85%* | Data privacy barriers; only 26% have integrated AI |
| Defense and Aerospace | 84%* | Security classification barriers; procurement cycle length |
| Financial Services | 82.1% | Regulatory compliance; explainability requirements |
| Insurance | 81%* | Actuarial model validation; underwriting bias detection |
| Life Sciences and Pharma | 80%* | Clinical and regulatory approval; trial data complexity |
| Healthcare | 78.9% | Clinical validation; physician adoption below 30% in year one |
| Oil and Gas | 78%* | Legacy SCADA systems; safety regulation complexity |
| Energy and Utilities | 77%* | Grid integration; IoT data quality |
| Legal Services | 77%* | Privilege concerns; partner resistance |
| Mining and Resources | 77%* | Remote operations; sensor data quality |
| Manufacturing | 76.4% | OT / IT integration gap; IoT sensor data quality |
| Telecommunications | 75%* | Legacy OSS/BSS systems; network data volume |
| Automotive | 74%* | Supply chain complexity; OT / IT divide |
| Construction | 74%* | Project fragmentation; low baseline digitisation |
| Retail and E-commerce | 73.8% | Demand volatility; fragmented first-party data |
| Transportation and Logistics | 72%* | Real-time data requirements; route complexity |
| Agriculture | 72%* | Seasonal data gaps; rural connectivity |
| Media and Entertainment | 71%* | Creative IP concerns; 30–80% range reported |
| Real Estate | 70%* | Fragmented property data; valuation instability |
| Hospitality and Travel | 70%* | Demand volatility; fragmented PMS / CRS systems |
| Professional Services | 68.7% | Knowledge worker resistance; client data restrictions |
| Non-profit | 68%* | Budget constraints; volunteer and staff resistance |
| Education | 67%* | Faculty resistance; student data privacy regulations |
| Consumer Goods | 66%* | Multi-channel data fragmentation |
| Food and Beverage | 65%* | Supply chain focus; food safety compliance |
| HR and Talent Management | 63%* | Hiring AI bias concerns; privacy regulations |
| Supply Chain and Procurement | 62%* | Multi-party data integration |
| Marketing and Advertising | 60%* | Attribution model gaps; cookie deprecation |
| Software and Technology | 55%* | Higher technical baseline; stronger data infrastructure |
| Cybersecurity | 52%* | Purpose-built models; clear use cases |
Primary research for Financial Services, Healthcare, Manufacturing, Retail and E-commerce, and Professional Services informed by Pertama Partners synthesis of 2,400+ enterprise AI initiatives (2026). Government figure informed by EY survey of 492 government leaders (2025). Media and Entertainment range informed by WEF (2025). All figures marked * are directional estimates weighted against the global baseline.
The data reveals that no industry in this dataset falls below a 50% failure rate, and every heavily regulated sector exceeds 80%.
Cost of AI project failure (2026)
The following figures show the financial scale of AI project failure at the enterprise level.
| Metric | Finding |
| Share of global AI investment producing no material value | 60% of companies |
| Global enterprise AI investment in 2025 | $684 billion |
| Average cost per abandoned initiative, enterprise (10,000+ employees) | $7.2M |
| Share of active AI POCs scrapped before production in 2025 | 46% |
| ROI multiple for focused vs. unfocused AI programmes | 2.1x greater ROI |
The data reveals that with 60% of companies generating no material AI value from a combined $684 billion in global investment, the implied cost of AI underperformance runs into the hundreds of billions annually.
AI project success rates by practice (2026)
The table below draws on BCG and Gartner research to identify the practices that most reliably separate successful AI transformations from failed ones.
| Practice | Findings |
| Focus on 3-4 use cases rather than 6-7 | Produces 2.1x greater ROI |
| Invest 70% of effort in people, process, and culture | 70% of AI transformation value is people-related |
| Ensure AI-ready data before project approval | Organisations without it face a 60% abandonment rate |
| Train employees before deployment | Only 36% of employees currently trained; closing this gap is the leading predictor of adoption |
| Track financial KPIs from day one | Companies that do are disproportionately among the 5% achieving substantial value |
The data reveals that technology accounts for just 30% of what drives AI transformation success; the remaining 70% is people, process, and culture.
To learn more or download a copy of this report, please reach out here.
Sources
- RAND Corporation (2024). The Root Causes of Failure for Artificial Intelligence Projects and How They Can Succeed. https://www.rand.org/pubs/research_reports/RRA2680-1.html
- BCG (2025). From Potential to Profit: Closing the AI Impact Gap. https://www.bcg.com/publications/2025/closing-the-ai-impact-gap
- BCG (2026). Why AI Change Is Actually a People Change. https://www.bcg.com/publications/2026/why-ai-change-is-actually-a-people-change
- Gartner (2025). Lack of AI-Ready Data Puts AI Projects at Risk. https://www.gartner.com/en/newsroom/press-releases/2025-02-26-lack-of-ai-ready-data-puts-ai-projects-at-risk
- Gartner (2026). Why Half of GenAI Projects Fail. https://www.gartner.com/en/articles/genai-project-failure
- McKinsey & Company (2025). The State of AI: Global Survey 2025. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
- S&P Global Market Intelligence (2025). Enterprise AI Survey 2025. Cited in CIO Dive, March 2025. https://www.ciodive.com/news/AI-project-fail-data-SPGlobal/742590/
- IDC / Lenovo (2025). CIO Playbook 2025. Cited in CIO Magazine, March 2025. https://www.cio.com/article/3850763/
- MIT (2025). Generative AI pilot failure analysis. Cited in IBM Think (2025). https://www.ibm.com/think/insights/ai-roi
- MIT Project NANDA (2025). State of AI in Business 2025 Report. Cited in BCG (May 2026).
- KPMG (2025). Global AI use case failure rate. Cited in Forbes Middle East and multiple industry publications.
- EY (2025). EY Survey Reveals Large Gap Between Government Organizations’ AI Ambitions and Reality. https://www.ey.com/en_gl/newsroom/2025/06/ey-survey-reveals-large-gap-between-government-organizations-ai-ambitions-and-reality
- World Economic Forum (2025). Artificial Intelligence in Media, Entertainment and Sport. https://reports.weforum.org/docs/WEF_Artificial_Intelligence_in_Media_Entertainment_and_Sport_2025.pdf
- Pertama Partners (2026). AI Project Failure Statistics 2026. https://www.pertamapartners.com/insights/ai-project-failure-statistics-2026
