Less than half of U.S. CFOs plan any AI-related job cuts at all. That’s the finding from an NBER working paper cited in a Fortune report published last month, based on a survey of 750 chief financial officers. When the study’s authors projected those CFO intentions across the broader economy, the result was approximately 0.4% of the workforce, a figure that carries real-world weight, but one that sits well below the displacement narratives that have dominated AI coverage.
The article’s own framing makes the point explicitly. Fortune’s headline characterizes the projected AI layoff rate as “9x higher than last year, and still a fraction of ‘doomsday’ predictions.” That’s a specific editorial thesis: the rate of AI-related job cuts is accelerating in relative terms, and the absolute number is still modest. Both things are simultaneously true, and the tension between them explains why this data tends to generate very different reactions depending on which half you emphasize.
On the 9x figure: this describes a relative increase over the prior year’s baseline. It does not mean the absolute number of AI-related cuts has reached a level that dominates the labor market. A rate that is 9x higher than a small base can still produce a small absolute number. That’s precisely what the 0.4% figure suggests.
The 44% participation rate deserves equal attention. If only 44% of CFOs plan AI-related cuts, that means 56%, a majority, are not. This tells us something important about where AI-driven automation is and isn’t landing across the economy right now. AI adoption is uneven. Companies and sectors where AI tools have clear operational applications are making workforce decisions on that basis. Companies where the technology’s fit with existing workflows is still unclear are not. The CFO survey captures that unevenness more accurately than aggregate disruption narratives do.
Two important caveats. First, this data was published in late March and reflects survey responses gathered before that point. The AI adoption timeline is moving fast enough that CFO intentions reported in early 2026 may not reflect decisions made in late 2026. Second, the Fortune article body is only partially accessible in the source review for this brief, meaning some numerical details in the full NBER methodology aren’t confirmed in the text available. The headline figures, 750 CFOs, 44%, 0.4%, and 9x, are visible in accessible text. The full methodology and margin of error aren’t.
What to watch: The NBER paper, when its full text is accessible, will be the primary source for understanding what the 9x baseline actually is, how “AI-related” job cuts are defined in the methodology, and whether the 0.4% figure accounts for second-order effects like hiring freezes or role consolidation that don’t register as layoffs. Those methodological details are the difference between a data point and an argument.
For context on specific company-level AI displacement events, Oracle’s reported 30,000 cuts and the attribution question they raise, see today’s Oracle brief and the TJS displacement brief.