Only 5% of generative AI pilots at companies are delivering meaningful results, according to a new report from MIT’s NANDA initiative, which warns that corporate enthusiasm for AI has outpaced real-world success.
The report — The GenAI Divide: State of AI in Business 2025 — analysed 300 public AI deployments, 150 executive interviews, and survey data from 350 employees. It found that while AI tools promise faster growth and efficiency, most enterprise projects fail to generate measurable impact on profit and loss statements.
“Some large companies’ pilots and younger startups are really excelling with generative AI,” said Aditya Challapally, the report’s lead author and a research contributor to MIT’s project NANDA.
“Startups led by 19- or 20-year-olds have seen revenues jump from zero to $20 million in a year. It’s because they pick one pain point, execute well, and partner smartly with companies who use their tools.”
The “GenAI Divide”: startups surge, enterprises stall
The report identifies a widening gap between nimble startups and larger corporations — a phenomenon MIT calls the “GenAI Divide.”
While small, focused teams are translating generative AI into clear commercial wins, 95% of corporate pilots stall, producing little or no productivity uplift. MIT attributes this to poor integration, not model quality.
“The issue isn’t regulation or model performance,” Challapally explained. “It’s that enterprise systems aren’t learning from their own workflows. Generic tools like ChatGPT excel for individuals because they’re flexible — but in businesses, they don’t adapt, they don’t integrate, and so they stall.”
This “learning gap” between tools and organisations, the study argues, is the single biggest drag on enterprise AI performance.
MIT’s findings also highlight a mismatch in corporate AI investment. More than half of enterprise GenAI budgets are currently spent on sales and marketing applications, even though the highest return on investment comes from back-office automation — areas such as document processing, compliance, and finance operations.
According to the research, companies that used AI to replace business process outsourcing (BPO), cut agency costs, or streamline internal workflows saw the strongest returns, while those deploying AI for content generation or chatbots struggled to show value.
“Executives want fast wins in visible areas like sales,” Challapally noted. “But the real value is hiding in the unglamorous operational work where AI can quietly save millions.”
Among the 5% of AI pilots that succeeded, MIT found three common factors:
• Clear, narrow use cases tied to a measurable outcome.
• Deep collaboration between AI teams and end-users.
• A focus on data integration before deployment, not after.
These pilots often achieved revenue acceleration of 15-25%, validating the technology’s potential when applied with precision.
The report urges enterprises to “move from experimentation to operationalisation” — integrating generative AI into existing systems, rather than treating it as a separate innovation silo.
The findings come as companies across industries race to embed AI into workflows following the explosive adoption of tools such as ChatGPT, Claude, and Google Gemini. Analysts estimate that corporate spending on generative AI exceeded $40 billion globally in 2024, but measurable returns remain elusive.
“We’re seeing an extraordinary gap between expectation and execution,” Challapally said. “The winners will be those who stop chasing buzzwords and start solving specific problems — one workflow at a time.”
