Andy Goundry, Co-founder AI Value Diagnostics · 28 September 2026

A year of record AI spending. The value needle didn't move.

TL;DR. McKinsey's State of AI 2026 survey has landed. The share of organisations attributing any profit impact to AI is 37%, against 39% a year earlier, a difference McKinsey calls about the same. The share getting significant value is 6%, also unchanged. A year of record spending improved neither number. What did become clearer is what the 6% do differently, and none of it is technology.

Our homepage carries one claim above everything else. Nearly nine in ten organisations are adopting AI, and around 6% are getting real value from it. That claim rests on McKinsey's State of AI survey, so when the 2026 edition landed we went through it expecting to rewrite our numbers.

We barely had to. That is the story.

The numbers that didn't move

In the 2026 survey, 37% of respondents attribute at least some profit impact to AI, against 39% a year earlier, a difference McKinsey describes as about the same. The share McKinsey calls AI high performers, meaning organisations attributing 5% or more of EBIT to AI and describing its value as significant, is 6%. Also unchanged. In between those two surveys sits a year of record investment, frontier model releases and the loudest agent push the industry has ever run. None of it improved the value numbers. We've updated our homepage figures to the 2026 edition, and the update was mostly the year in the citation.

The tension that grew

One pair of numbers explains why the spending continues anyway. 80% of people using AI report improved individual productivity. Only 37% of organisations can attribute financial impact to it. McKinsey's own summary is that "conviction in AI is growing faster than the immediate financial returns" organisations can attribute. Individual gains are real. They just don't roll up into organisational value on their own. Getting from one to the other is placement and measurement work, not procurement.

The number that did move

Agents. Among organisations with over a billion dollars in revenue, the share scaling AI agents in one or more functions rose from 27% to 40% in a year. Smaller organisations stayed flat at around 22%. Scale is arriving whether the value numbers are ready or not, which is exactly why we wrote about what separates the operators from the embedders in AgentCore in the wild.

What the 6% do differently

The survey's most useful finding is what distinguishes the high performers. Nearly three quarters have fundamentally redesigned workflows because of AI, against a quarter of everyone else. Their senior leaders own the AI initiatives rather than sponsoring them from a distance. And they have defined processes for deciding when model outputs need human validation. None of that is technology. It is placement, measurement and honesty about risk, which is the gap our homepage has described since the day it went live.

The averages won't move first. Individual organisations will move first, and the survey says they move by doing unglamorous things deliberately. If you want to know which side of the 6% line a workflow of yours would land on, the first conversation is diagnostic, not a pitch.

Sources

• https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai (McKinsey, The State of AI: Global Survey 2026)