Return on AI: Weekly Digest — September 28, 2026
This edition covers 16 findings across workforce economics, spending benchmarks, legal services, supply chain operations, and AI deployment outcomes. Visier finds half of workers are faking AI use as job loss fears rise, which raises a measurement problem that sits directly upstream of the ROI shortfalls Gartner and BCG are tracking from different angles: Gartner urges CFOs to set ROI criteria before full rollouts, while BCG puts measured GenAI and AI agent returns at 13.8 percent in 2026 for organizations that do. The EY model requires a 16 percent white-collar labor cut to fund AI investment by 2031, the Federal Reserve puts 2026 enterprise AI spend at $2,068 per worker, and the IMF finds AI could lift worker output 3.8 percent over the long term. Separately, Forrester warns that agentic supply chain AI surfaces bad master data before it can fix anything, silent agent failures in banking carry the highest customer cost, in-house legal teams are using AI to redirect attorneys toward higher-impact work while rival providers use it to undercut law firm hourly rates, and US enterprises are shifting toward low-cost open-source models as Walmart rules out AI-driven personalized pricing.
Key Takeaway of the Week
CFOs who reinvest AI efficiency gains rather than pocket them as one-time savings are generating compounding cost returns that widen competitive separation each quarter. BCG's measured 13.8 percent ROI from GenAI and AI agents in 2026 confirms that returns are real and quantifiable, but only for organizations that define the payoff equation before deployment. The leaders pulling ahead are setting ROI criteria upfront, monitoring agent failures before they reach customers, and treating workforce redesign as a financial decision, not an HR one.
Sponsored by Jetson Collective