McKinsey published the most recent snapshot of AI adoption in business, in a survey with nearly 2,000 respondents across more than 100 countries: 88% of organizations already use AI in at least one function — up from 78% the year before. That’s the number every board deck highlights. The number almost no board deck highlights is this: only 7% of those companies have managed to scale AI across the entire operation, and only about a third have moved past the pilot stage. Adoption has become commonplace. Scale remains rare.
The funnel narrows even further when the criterion is financial impact. Only 39% of companies report any measurable impact on operating profit (EBIT) from AI — and most of those attribute less than 5% of that result to it. Even among the few that already see some effect on the bottom line, the effect tends to be marginal, not structural.
Adoption isn’t the same as scale — and the difference isn’t technical
Adopting an AI tool usually costs the company’s structure very little: a team picks up a copilot to draft reports, a support area plugs a chatbot into the first line of service, an analyst uses a model to summarize contracts. The scope stays contained within a single team, the data involved already belonged to it, and nobody has to negotiate anything with anyone. That’s exactly why 88% of companies get this far.
Scaling is a different operation. For that same copilot to generate value across the whole company, it needs to connect to other departments’ data, feed decisions that currently belong to a different manager, and eventually redesign roles that have existed for years. None of that was in the original pilot’s scope — and that’s exactly where the project stalls. The symptom that shows up in the status report is usually “data team is short on time” or “waiting on prioritization.” The real cause, more often than not, is different: nobody gave the project a mandate to touch processes that weren’t its own.
According to João Paulo Batistella, the barrier is political — not technological
João Paulo Batistella, an innovation executive and organizational transformation specialist, names this barrier directly: the biggest difficulty in scaling AI inside a company isn’t technical, it’s political — a more efficient operation threatens departments, processes, and roles that exist today, and the organization’s own survival instinct resists that. The pilot threatens no one because it stays contained within one team; scale does, because it redistributes decision-making power.
Batistella also points out why this kind of project rarely gets resolved with more technology investment alone: when AI is treated purely as an IT tool, the company risks optimizing a process that maybe shouldn’t keep existing in that form at all — and the decision about what changes structurally belongs to the CEO and the board, not an isolated functional area. An AI project without sponsorship at that level can still adopt the tool successfully. It just doesn’t scale.
Three questions to know if your company is stuck in the pilot
- Does the team responsible for the AI project have a mandate to propose changes in other departments’ processes — or can it only touch what was already its own from the start?
- Is there someone at CEO or board level tracking this project specifically, or does it only get reported inside the IT structure?
- If the project scaled tomorrow, which role, process, or department would visibly change? If the answer is “none,” the project was probably never designed to scale — only to automate one specific task.
None of this makes the pilot useless — a well-designed pilot remains the cheapest way to learn what works before committing the whole company. The mistake is treating it as the final destination instead of the first test of a bigger change. McKinsey’s 88% adoption figure proves the barrier to entering AI has practically disappeared. The 7% scale figure proves the next barrier was never about the tool.
Follow Eleva Tecnologia for more analysis on technology applied to business: follow @ElevaTechno on X or @elevatechnologies on Instagram, or learn more about the group at elevatec.net/about.
