78% use GenAI, 80% see no value
- McKinsey’s 2025 AI research sharpened the enterprise AI story: 78% of companies now use AI somewhere, but most still cannot show clear business gains. - The telling split is 71% using generative AI in at least one function, while about 80% report no significant topline or bottom-line impact. - The gap matters because the winners are redesigning workflows and leadership habits, not just adding copilots or more model access.
Enterprise AI has a weird problem. The tools are everywhere, budgets are real, and usage is now mainstream inside big companies. But the promised payoff still mostly hasn’t shown up. That’s the core of the latest McKinsey picture: adoption has surged, yet measurable business value is still concentrated in a small minority of organizations. (hai.stanford.edu) ### Where do the 78% and 80% numbers come from? They’re two different but related stats that keep getting mashed together. McKinsey’s 2025 state-of-AI survey says 78% of respondents report AI use by their organizations in at least one business function. A separate McKinsey framing of the “gen AI paradox” says nearly 80% of companies report using the latest gen(hai.stanford.edu) bottom-line gains. So the broad point is real — use is widespread, value is not — but the exact numbers come from adjacent McKinsey materials, not one single sentence in one chart. (hai.stanford.edu) ### So is this about AI in general or generative AI? Mostly both — and that’s part of the confusion. The Stanford AI Index, pulling from McKinsey survey data, says 78% of organizations used AI in 2024, while 71% used generative AI in at least one business function, up from 33% in 2023. That means classic predictive AI and newer generative tools are now blended(hai.stanford.edu)mplifying a messier reality: 78% is the broader AI-adoption figure, while 71% is the cleaner generative-AI number. (hai.stanford.edu) ### Why isn’t usage turning into value? Because most deployments are still thin. McKinsey’s own explanation is basically that lots of companies rolled out horizontal tools — chatbots, copilots, generic assistants — that save a little time but don’t reliably move revenue, margin, or EBIT. The 2025 survey also says most organizations have not scaled AI deeply, a(hai.stanford.edu)e using AI every day, but the company P&L often barely notices. (mckinsey.com) ### What are the winners doing differently? They’re changing work, not just software. McKinsey’s 2025 survey says workflow redesign is one of the clearest markers of stronger results, and high performers are more likely to pair efficiency goals with growth and innovation goals. That matters because a copilot layered on(mckinsey.com)transformative. (mckinsey.com) ### Is leadership part of the problem? Yes — probably more than the models are. McKinsey’s workplace report says almost all companies invest in AI, but only 1% describe themselves as mature, and the biggest barrier to scaling is leaders not moving f(mckinsey.com)ow, usage stays cosmetic. (mckinsey.com) ### Does this mean the AI boom was fake? No. It means adoption came first and operating discipline came second. McKinsey still argues the upside is huge — trillions in annual economic potential, with much of the value concentrated in customer operations, marketing and sales, software engineering, and R&D. But potential value and captured value are not the same thing. Most firms are still stuck in the gap between the demo and the redesign. (mckinsey.com) ### What should readers take from this? The real story is not “AI failed.” It’s that enterprise AI has entered its boring phase — the part where strategy, workflow design, governance, and measurement matter more than novelty. The companies getting returns are treating AI like an operating model change. Everyone else is still counting logins. (mckinsey.com) ### Bottom line AI use is now common enough to stop being the headline. Value is the headline. And right now, value is still rare. (hai.stanford.edu)
