What the research found
McKinsey surveyed 1,719 people in 97 countries between May and June 2026. Almost nine in ten say their company uses AI regularly in at least one business function, and 44% say the technology is already scaling across the organization (up from 38% a year earlier).
The problem shows up when you look at results. Only 37% attribute any EBIT effect to AI use, practically the same as in 2025. The group McKinsey calls high performers, companies that attribute at least 5% of EBIT to AI and describe the impact as significant, remains at about 6%.
At the same time, 80% of respondents say AI has improved their own productivity. And one in five reports that their company already limits use because of operating costs, including token consumption.
What the 6% do differently
The difference is not having more tools. Almost three quarters of high performers have truly redesigned workflows around AI, versus about a quarter of the rest. They also have visibly engaged leadership and defined processes to measure impact, both twice as often as other companies.
And in Brazil
The Panorama 2026 survey by Amcham Brasil and Humanizadas, released in September 2025 with 629 executives from mid-sized and large companies, points to the same mismatch: 84% of companies use AI, 77% allocate up to 2% of their budget to it and 61% have not yet seen a relevant business impact.
Our take
The individual gain is real, but it doesn't reach the income statement on its own. An analyst who writes reports in half the time doesn't change company results if the process around them stays the same, with the same queues, approvals and rework.
In practice, we see two common mistakes. The first is choosing the tool before choosing the process. The second is not defining, before starting, which indicator should change and who will measure it. Without that, the company ends the year with several pilots and no clear answer for the board.
The cost finding deserves attention. A cheap pilot can become a significant monthly expense once it scales, and it is worth running the numbers before expanding.
Questions to take to your next meeting
- Which processes have really changed because of AI, rather than just gaining a new tool?
- Which financial indicator should each initiative move, and who tracks that number?
- How much will it cost per month to run the solution at scale?
- Who approves new AI uses and is accountable for the risks?
Sources
- The state of AI in 2026: On the road to ROI, McKinsey & Company, August 25, 2026.
- IA lidera prioridades para 2026, mas 77% das empresas ainda investem pouco (AI tops priorities for 2026, but 77% of companies still invest little; Panorama 2026, Amcham Brasil and Humanizadas), Diário do Comércio, September 2025.
The figures come from the sources cited. The selection and analysis are by Mapa Consultoria Empresarial.
Want to discuss how this applies to your company?
