Only 2% of Companies Let CFOs Run AI—That’s Why Productivity Gains Miss Revenue

CFOs keep claiming AI boosts productivity, yet the dollars stay stuck in slide decks. A fresh Fortune analysis finally cracks the puzzle: only 2% of companies let the CFO actually control AI initiatives even though 76% of firms that do see “great value” from the technology. That disconnect, published 10 days after the Duke/Fed working paper, reframes every earlier number.

The original survey—750 U.S. CFOs fielded by Duke’s Fuqua School and the Richmond and Atlanta Fed banks—found 44% planning AI-driven layoffs totaling 502,000 jobs this year. That’s a ninefold jump from last year’s 55,000 cuts. CFOs privately told researchers these reductions feel like “sandpaper, not a buzz saw,” smaller than the apocalyptic forecasts from Microsoft’s Mustafa Suleyman (office jobs gone in 18 months) or Anthropic’s Dario Amodei (half of entry-level roles vaporized).

But the missing revenue piece now has an owner: the chief AI officer—or whoever grabbed the AI budget first. Fortune’s new reporting shows finance leaders are sidelined in 98% of organizations, leaving them to watch costs fall without seeing the upside flow through revenue lines. When CFOs do run the program, three-quarters report measurable gains. The other 98%? They’re still waiting.

The half-million figure hasn’t changed, but its context has. Cognizant doubled its 2023 disruption forecast to 93% of jobs facing AI exposure worth $4.5 trillion in economic impact, yet the timeline gap between “could be automated” and “will be eliminated” remains cavernous. The Fortune data suggests part of that lag is bureaucratic: CFOs can green-light layoffs faster than they can green-light revenue experiments if they don’t own the roadmap.

White-collar roles—data entry, basic analysis, first-pass reporting—still top the cut list. CFOs describe the reductions as “routine pruning” rather than strategic restructuring. Without finance-chief oversight, however, the savings get booked as margin improvement while the upside (new products, faster sales cycles, premium pricing) stays locked in pilot purgatory.

The 2% who run AI from finance share three moves. First, they tie every AI project to a P&L line item before the vendor demo ends. Second, they sunset pilots that don’t hit break-even within two quarters. Third, they fund cross-training for staff whose tasks are automated, converting saved labor hours into revenue-generating work instead of head-count reductions.

Those steps sound obvious, yet 98% of organizations skip them. The result: CFOs can cut 502,000 jobs but still can’t point to a revenue spike because no one empowered them to chase one.