Return on AI: Weekly Digest

This week's coverage cuts to the central tension in enterprise AI right now: a handful of companies, including Better.com, BMW, and ABB, are posting hard numbers on cost reduction and operational efficiency, while MIT research and CEO surveys confirm that 95% of generative AI deployments show no measurable P&L impact. The throughline across every story is the same finding, whether it comes from Davos or Uber's infrastructure bill: AI spending only converts to financial results when organizations change how work is actually done, not just what tools are running. If you are still measuring success by consumption or deployment volume, you are measuring the wrong thing.

Key Takeaway of the Week

Deploying AI without redesigning the work around it produces no measurable P&L impact, as MIT's finding that 95% of enterprise generative AI implementations fail to move financial results makes clear. Better.com's 41% reduction in loan origination costs stands as the week's sharpest counterexample, showing that outcome-focused implementation tied to a specific, high-cost process delivers real returns. The dominant mistake executives are making right now is funding AI consumption and calling it progress, when the only number that matters is whether unit economics improved.

Forward this to a colleague who should be tracking AI ROI. Subscribe at returnonai.com.