For most organisations, AI is no longer optional. According to Stanford’s 2025 AI Index, 78% of organisations used AI in at least one business function in 2024, up from 55% in 2023. AI has moved from pilot projects into production, but using it does not mean it pays off. So the question for executives is now where and how AI will produce measurable business value.
Many companies adopt AI without redesigning the workflows around it, and that is where the value is lost. McKinsey’s 2025 State of AI report found that while adoption is widespread, only 39% of companies saw AI make any difference to EBIT across the whole business. Most of those said AI accounted for less than 5% of their EBIT. Deploying AI without restructuring operations is like installing a new engine without changing the drivetrain. The result is activity without acceleration, with plenty of AI in use and little change in profit.
High-performing firms treat AI as a driver of margin, so they target core processes, redesign roles and focus on outcomes. That approach puts AI alongside labor and capital as a factor of production, judged by what it does to the business. Each AI project then has to show what productivity lift it will generate, how it shifts the cost structure and where it changes margins. Answering those questions means moving past one-off pilots and building for scale, technically, operationally and financially.
Execution separates the leaders from the laggards. McKinsey found that top-performing firms share three traits. Along with the workflow redesign many companies skip, they assign senior leaders to AI governance and integration and invest in talent, data infrastructure and coordination across departments. Firms that scatter AI projects across separate teams, with no one accountable, struggle to turn spending into results.
Despite the hype, results remain uneven. Even among companies using AI in service operations, the same Stanford report shows only 49% reporting cost savings, and most of those savings were under 10%. Double-digit gains are possible, but they require operational maturity and careful targeting of workflows where AI can affect more than 5% of a company’s costs or revenue.
