You can also access this resource, and many others, in Issue 364 of my Talent Edge Weekly newsletter.
McKinsey’s new 31-page State of AI in 2026 report includes several useful findings from its global survey of 1,719 participants across 97 countries. One data point that caught my attention is the economics of scaling AI. About 20% of respondents say AI-related operating costs, including token costs (the units of data processed when using an AI model), have caused some organizations to limit how much they use AI. The issue becomes even more pronounced among AI high performers, which McKinsey defines as organizations that attribute at least 5% of EBIT to AI use and describe its impact as significant. This reinforces an important point: running AI at scale is not free or unlimited, and workforce plans need to account for the actual economics of scaling it. Another related data point: 39% of respondents expect AI to decrease their organization’s overall head count over the next year. From my standpoint, layering in the token and operating cost constraints noted above, the 39% headcount expectation raises an important second-order consideration: if headcount is reduced based partly on the assumption that AI will take on more of the workload, but AI use later becomes limited by cost or other constraints, workload demand could outpace the organization’s capacity to deliver. AI’s capabilities are tremendous and will increasingly need to be part of workforce models. At the same time, we need to be intentional about these decisions so we can create ROI from AI’s benefits while reducing unintended consequences.
