- The “The GenAI Divide: State of AI in Business 2025” report from MIT NANDA shows that 95% of generative AI pilot programs at companies are not achieving significant revenue growth.
- The research was conducted through:
- 150 interviews with business leaders
- A survey of 350 employees
- An analysis of 300 public AI projects
- The main reasons for failure include:
- A knowledge gap between the AI tool and the organization using it.
- AI is poorly integrated into business processes, failing to learn from or adapt to actual workflows.
- Companies often build in-house solutions, but the success rate is only 33%, while buying from external vendors succeeds up to 67% of the time.
- Only 5% of companies are achieving clear results, typically young, leader-led startups that target specific pain points, collaborate effectively, and scale quickly (growing from $0 to $20 million in revenue in one year).
- The majority of AI budgets (over 50%) are spent on sales and marketing departments, whereas MIT suggests the highest ROI comes from logistics automation and reducing outsourcing costs.
- Other issues:
- Concerns about copyright and data ownership when using AI.
- “Shadow AI”: Employees secretly using ChatGPT and other unapproved tools.
- Difficulty in measuring the true impact of AI on profits or productivity.
- Advanced organizations are experimenting with “AI agentic” systems – AI systems that can learn, remember, and act independently within set boundaries.
📌 A new MIT report warns that 95% of corporate generative AI projects are failing, despite high expectations. The main reasons: poor integration, wrong tool selection, and misplaced investment focus. Meanwhile, small startups are achieving success through targeted implementation. Companies need to focus on logistics automation and collaborate with specialized vendors to effectively leverage AI.
