Pressured to Adopt AI? How to Turn Board Mandates into Real Results

CEO’s are putting a lot of pressure on their organizations to “adopt artificial intelligence.” Often it stems from a board mandate. But few know what it means to “adopt” and recent data from MIT proves this point.

The MIT NANDA Report (published in July 2025) provides one of the most comprehensive examinations of enterprise adoption and impact of Generative AI (GenAI) to date. The report serves as a proxy for companies large and small.

Estimated between $30-40 billion annually, investment remains brisk. However, most organizations fail to capture measurable business value. The study reveals a stark divide between experimentation and scaled success. It highlights 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. By now, many of you have downloaded and read the report. Here is a refresher.

Most AI Initiatives End in Failure

• 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)
• Top-performing firms cut costs by reducing reliance on Business Process Outsourcing (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)

Four Patterns Emerged from the Data

1. Limited disruption – only Tech and Media show structural change from AI usage. This could be because there are more early adopters in their industries. More traditional industries are in a wait-and-see mode hoping AI’s flaws get ironed out. For high-stakes work, 90% of employees prefer humans over AI.
2. Enterprise paradox – large firms lead in AI pilots but lag in scaling the capabilities. There are simply more human points of failure with access to greater budgets. Executive sponsorship weakens the further you get away from the innovative vision. Without a clear mandate or day-to-day benefit, employee resistance has become the top challenge (p.11)
3. Investment bias – budgets favor front-office use cases over back-office readiness. Lack of investment in maintaining quality data combined with ill-defined processes cause front-end systems failure. By then it’s too late and the initiatives are discontinued.
4. Implementation advantage – external builds succeed 2× more than internal projects (p.3). It is a mistake to simply “vibe code” a way into a solution by “offshoring” development to AI. There is still the need for experts that know how to make the code work. Most firms lack that internal know-how and it hurts progress.

Workforce and Usage Preferences Focus on Basic Tasks

• 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)

Solve the Right Problem and the ROI Will Follow

Our firm works with Private Equity backed companies on an ambitious growth plan. Executives across multiple portfolio companies have shared real-world implementations of AI to enhance customer support, knowledge management, agent efficiency, and internal governance.

One of our PE clients implemented several AI solutions to enhance their CS team’s performance. What we have learned is that success requires the coordination of several solutions. This Portco is using NiCE and SymTrain that enabled improvements in two use cases:

1. AI-powered guidance during live calls, case note summarization, and automation of disposition logging.
2. Simulated training environments for agents using real-life scenarios.

The outcome was 20% reduction in handle time, $15M+ in savings, and better ramp time for new agents. Once the savings were realized (both in time and dollars) adoption snowballed.

Successful Implementations Enhance Human Capability

The wave of job loss predicted by the collective social consciousness has yet to materialize on a catastrophic scale. It’s not to say it won’t happen in the future, but it certainly hasn’t happened yet. Job displacement is selective.

• 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)

Align AI initiatives with strategy, governance, and execution discipline

In the rush to “adopt artificial intelligence,” CEOs are applying immense pressure on their organizations. This is often in response to board mandates rather than strategic clarity. Yet, as the latest MIT NANDA Report (July 2025) makes clear, few leaders truly understand what meaningful adoption entails.

Despite $30–40 billion in annual investment, most companies still struggle to translate AI enthusiasm into measurable business value. The data reveals a widening gap between experimentation and scalable success.

Ultimately, the MIT NANDA report exposes a hard truth: while nearly every enterprise is experimenting with AI, only about 5% are realizing tangible P&L impact. For those companies, success comes not from chasing hype, but from aligning AI initiatives with strategy, governance, and execution discipline. As the rest of the field works to bridge this gap, the MIT findings offer a vital roadmap.

True AI adoption is less about technology and more about human transformation