95% of GenAI Pilots Show No P&L Impact, MIT NANDA Reports

mclynd · x · 2026-07-25

The author pushes back against the idea that better models will make foundational deployment engineering (FDE) obsolete, arguing that as intelligence becomes commoditized, the competitive edge shifts entirely to deployment and implementation.

Citing data from MIT NANDA, he notes that approximately 95% of GenAI pilot programs fail to generate actual P&L impact. The real challenge lies in the "last mile" of engineering, not the model itself. Consequently, better models actually raise the ceiling for deployers.

Original post →

More from AGI Musings

AGI Musings channel →