A new report released by MIT’s NANDA project highlights the stark difference between the promise of generative AI and its real-world financial performance within corporate settings. The GenAI Divide: State of AI in Business 2025 finds that 95% of enterprise AI pilot programs fail to generate measurable financial returns, an indictment of how companies are deploying AI technologies.
Based on in-depth interviews with 150 executives, surveys of over 350 employees, and analysis of 300 AI deployments across sectors, the study identifies a systemic “learning gap.”
While the hype around generative AI continues to swell in 2025, most firms are struggling to turn AI capabilities into cost savings or new revenue.
The report contends that the major barriers are organizational, not technological.
Why It Matters: As generative AI redefines digital transformation strategies, the gulf between early adopters and successful implementers widens. Companies that fail to learn from current missteps risk wasting millions in AI investments without achieving meaningful returns. Understanding what separates AI experiments from real financial outcomes is essential for entire workforces adapting to AI-augmented operations.
- Stalled Innovation: 95% of AI Pilots Fail to Scale: The study reveals that 95% of corporate generative AI pilots fall short of delivering any meaningful financial impact. These initiatives often remain stuck in the testing or prototype phase. Meanwhile, the successful outliers typically share a theme of tight integration between AI solutions and the business processes they are meant to improve.
- Misplaced Spending: Companies are investing the bulk of their generative AI budgets into sales and marketing functions. Yet the report identifies back-office functions like customer service automation and HR operations as delivering higher returns through cost reduction and efficiency gains. This disconnect shows a broader issue with corporate AI strategies often favoring visibility over value.
- External Tools Outperform Internal AI Efforts: AI systems sourced from specialized external vendors show a 67% success rate, more than double the performance of internally built tools. Despite this, many companies continue investing in proprietary development, particularly in highly regulated sectors such as finance and healthcare. MIT’s analysis suggests that these internal efforts often get bogged down by a lack of coordination with extended development cycles and misalignment with actual user needs.
- Rise of Shadow AI and the Agentic Future: Employees are adopting “shadow AI” tools such as ChatGPT and other consumer-grade assistants without formal oversight or IT governance. While these tools often improve individual productivity, they raise security and quality concerns. On the horizon, the report spotlights “agentic AI” as the next frontier of enterprise automation. These tools could reshape decision-making workflows but require robust internal infrastructure and cultural adaptation to scale responsibly.
- Subtle Workforce Impacts, Not Mass Layoffs: Contrary to common fears, the report finds no evidence of mass layoffs tied directly to generative AI. Instead, a quieter transformation is underway. Firms are choosing not to replace positions in administrative and customer service areas as they become vacant. This “soft attrition” model reshapes the workforce over time, reducing headcount in a less disruptive manner.
MIT Report Finds 95% of Generative AI Pilots Fail to Deliver Financial Impact – AInvest
Trusted insights for technology leaders
Our readers are CIOs, CTOs, and senior IT executives who rely on The National CIO Review for smart, curated takes on the trends shaping the enterprise, from GenAI to cybersecurity and beyond.
Subscribe to our 4x a week newsletter to keep up with the insights that matter.
