Adoption and earnings impact are not the same curve. McKinsey’s early-2024 State of AI survey showed generative AI jumping into everyday use. Follow-on survey work later in 2024 still found that most organizations report no tangible enterprise-level EBIT from gen AI. Usage scaled. Bottom-line attribution mostly did not.
Use rose fast
In The state of AI in early 2024, McKinsey reports that 65% of respondents said their organizations regularly use gen AI—nearly double the share from the previous survey about ten months earlier. Overall AI use in at least one business function sat at 72% in that early-2024 wave.
Expectations stayed high: roughly three-quarters still expected significant or disruptive industry change from gen AI. The adoption story is real. The question is what shows up in financial results.
EBIT attribution stayed thin
Even in the early-2024 survey, only a small subset of respondents (46 of 876) said a meaningful share of organizational EBIT came from gen AI—defined there as more than 10%. Unit-level cost and revenue effects were more common than company-wide earnings claims.
McKinsey’s later State of AI survey (fieldwork in mid/late 2024; published 2025) reports that more than 80% of respondents say their organizations are not seeing a tangible impact on enterprise-level EBIT from gen AI—even as gen AI use continued to climb.
Why the gap persists
Gen AI often lands in marketing drafts, coding assistance, and support pilots before it owns a P&L line. Local time savings do not automatically become enterprise EBIT. Capture requires process change, measurement, and sometimes headcount or revenue redesign.
That matches the broader pattern in adoption research: pilots and seat licenses are easy; wiring outcomes into the operating model is hard.
How to read the surveys
Treat “regular use” as a diffusion metric and “EBIT impact” as a value metric. Celebrating the first without the second is how roadmaps stay busy and balance sheets stay flat.
If you report progress internally, separate seats deployed, tasks completed, and financial effect. McKinsey’s split is a reminder that those three numbers can diverge for years.