Duke/Fed NBER Paper: CFOs Claim 1.8% AI Productivity Gains But Revenue Data Shows Much Smaller Impact — Solow's Paradox Returns

Fortune·high signal

A new NBER working paper from Duke University's Fuqua School of Business and the Federal Reserve Banks of Richmond and Atlanta surveyed nearly 750 U.S. executives and found CFOs reported average AI-driven productivity gains of 1.8% in 2025, but actual revenue and employment data show 'much smaller' gains across all major industries. Co-author John Graham told Fortune: 'It's not really hitting the top line yet in full force.' The pattern mirrors economist Robert Solow's 1987 computer productivity paradox, suggesting a one-year implementation lag before revenue benefits materialize.