Enterprise AI ROI Reality
A new analysis warns many corporate AI pilots aren’t delivering measurable ROI and that messy internal data and organizational inertia are the main culprits — meaning pilots often die before producing business value. The piece bluntly calls current enterprise AI efforts a mirage and urges measurable integration and data readiness up front. (mkaplan.substack.com)
Michael Kaplan’s Substack called recent enterprise AI programs a “grift” and titled a March post “The Enterprise AI Grift: Why Your $100M Strategy is a Digital Lobotomy,” arguing leadership and governance failures—not model capabilities—explain why many pilots stall. (mkaplan.substack.com) A new MIT “GenAI Divide” analysis found roughly 95% of enterprise generative-AI pilots produce no measurable P&L impact after analyzing about 300 public implementations and hundreds of interviews, while industry investment in GenAI is estimated between $30–$40 billion. (mlq.ai) MIT and follow-up coverage attribute the gap to brittle workflows and misaligned implementation approaches rather than model quality, noting most pilots never reach the governed-data, repeatable processes that enterprise systems require. (aimagazine.com) Enterprise vendors and platforms repeatedly point to governance, integration and observability shortfalls—examples include compliance and data‑residency requirements, manual handoffs, and lack of foundational data plumbing that prevent models from being trusted in production. (blogs.oracle.com) Market surveys show the fallout: S&P Global reported 42% of companies abandoned most AI initiatives in 2025, and Deloitte’s 2025 survey of 1,854 executives found 85% increased AI spend in the prior 12 months despite most projects taking two to four years to show satisfactory ROI. (forbes.com) Sector examples underscore the divide—healthcare vendor emtelligent reported one project processing 5.1 billion clinical notes that yielded up to an 80% increase in structured biomarker extraction when the solution was tightly integrated with clinical workflows. (healthcareitnews.com)
