The MIT NANDA Report provides one of the most comprehensive examinations of enterprise adoption and impact of Generative AI (GenAI) to date. Despite the rapid pace of investment-estimated between $30–40 billion annually-the overwhelming majority of organizations are failing to capture measurable business value. The study reveals a stark divide between experimentation and scaled success, highlighting both the promise of AI-driven transformation and the structural barriers preventing meaningful returns.
This summary outlines the key statistics, patterns, and implications uncovered in the report. It shows where organizations are finding success, where they are stalling, and what differentiates the 5% of companies that are achieving real P&L impact.
Headline Statistics
- 95% of organizations report zero ROI from GenAI, despite massive investment (p.3)
- Only 5% of AI pilots deliver millions in measurable value; the rest show no P&L impact (p.3)
- 80%+ have piloted ChatGPT or Copilot, and nearly 40% have deployed them (p.3)
-
Of those exploring enterprise-grade AI:
-
60% evaluated tools
- 20% reached pilot stage
- 5% achieved production (p.3)
Emerging Patterns
- Limited disruption – only Tech and Media show structural change
- Enterprise paradox – large firms lead in pilots but lag in scaling
- Investment bias – budgets favor front-office use cases over back-office
- Implementation advantage – external builds succeed 2× more than internal projects (p.3)
Adoption and ROI
- Top-performing firms cut costs by reducing reliance on BPO and agencies (p.4)
- Only Tech and Media show systemic workforce disruption (p.5)
- Just 5% of custom enterprise AI tools reach production (p.6)
- Chatbots scale widely (~83%) but fail to add value in critical workflows (p.6–7)
- 40% of companies purchased official licenses, yet 90% of employees still use personal AI tools (p.8)
- 50% of budgets target Sales & Marketing, though back-office automation yields higher ROI (p.9)
Barriers to Scaling
- Employee resistance is the top challenge (p.11)
- Other barriers: poor model performance, UX friction, weak executive sponsorship
Workforce and Usage Preferences
- For high-stakes work, 90% of employees prefer humans
-
For basic tasks, AI is the preferred tool:
-
70% for emails
- 65% for analysis (p.13)
Enterprise Expectations of AI Vendors
Organizations want partners who are:
- Trusted and transparent
- Deeply embedded in workflows
- Minimally disruptive
- Capable of continuous improvement (p.15)
ROI Examples
-
Front-office:
-
40% faster lead qualification
- 10% improvement in customer retention (p.21)
-
Back-office:
-
$2–10M annual savings from BPO elimination
- 30% reduction in agency costs
- $1M annual savings in financial risk checks (p.21)
These gains typically reduce vendor spend, not workforce size.
Labor Market Impact
- Selective displacement (5–20%) in support/admin roles
- 80%+ of Tech/Media executives expect reduced hiring within two years (pp.21–22)
-
Project Iceberg analysis:
-
Only 2.27% of U.S. labor value currently automatable
- Yet $2.3T in latent exposure, affecting 39M jobs (p.22)
