A significant portion of enterprise artificial intelligence (AI) projects are failing to deliver tangible value, with 42% of companies abandoning most of their AI initiatives in 2025, a sharp increase from 17% the previous year. This trend highlights a growing disconnect between ambitious AI investments and the ability to demonstrate clear returns on investment (ROI).
The challenge stems from the inherent difficulty in quantifying AI's benefits, which are often distributed, non-linear, and extend beyond traditional financial metrics. "Asking for the ROI of enterprise AI is like asking for the ROI of electricity," stated Russell Fradin, a prominent voice in the AI space, in a recent social media post. He noted that despite this complexity, executives often pressure teams to produce a single, simple ROI percentage, leading to fabricated numbers and market punishment for this short-sighted approach.
According to S&P Global's 2025 data, the primary reason for this high abandonment rate is "unclear value." Further emphasizing the struggle, a Boston Consulting Group (BCG) study found that only 5% of organizations are generating meaningful value from AI at scale. Fradin pointed out that a mere 21% of companies are actively measuring the impact of their AI investments, suggesting that many are proceeding without a clear understanding of their initiatives' effectiveness.
Experts suggest that for AI projects to succeed, organizations must move beyond traditional ROI models. Fradin advocates for a "multi-dimensional measurement system" that tracks metrics such as Capacity Reallocation Value and the Cost of Delay. His firm has published a framework designed to translate operational AI metrics into a financial narrative that resonates with executive boards, aiming to provide a more comprehensive view of AI's impact. The rising failure rates underscore the urgent need for more sophisticated evaluation methods to unlock the true potential of enterprise AI.
