McKinsey said in an Oct. 14 analysis that cost-effective AI use across selected workplace tasks could represent $2.52 trillion in annual global value, while higher operating costs could substantially narrow that opportunity.
With AI costs set 30% higher for the exercise, the annual value estimate is $1.67 trillion, $850 billion, or 34%, below baseline. McKinsey uses that input to test the model’s sensitivity; it does not predict that AI prices will rise by that amount.
The figure is a modeled annual value pool from cost-effective AI deployment across selected tasks, not a forecast of realized GDP growth. McKinsey’s model analyzes region–industry–occupation combinations spanning 831 Standard Occupational Classification occupations and compares labor costs, including benefits and employer overhead, with AI usage and the costs of human review, validation and rework. The analysis assesses tasks within occupations rather than concluding that entire jobs can be automated.
Financial services has the largest sector estimate, at $530 billion a year. Education follows at $410 billion, technology, media and telecommunications at $360 billion, and health care at $310 billion.
McKinsey also finds strong per-hour economics in legal; health care practitioners and technical occupations; management; and computer and mathematical occupations. In these fields, AI could support professionals’ work.
The cost question matters to the scale of the buildout: McKinsey estimates that AI infrastructure investment could reach $7 trillion by 2030, requiring about $760 billion in annual profit pools to justify it. The report points to possible pressures including specialized-chip supply, data-center power, computing needs for more capable models and hardware trade restrictions; it does not present them as proof of a 30% cost increase.
The modeled pool leaves out implementation spending on infrastructure, platforms, training and organizational change, which McKinsey says would reduce the net value captured. The analysis does not estimate how gains will be divided among companies, workers and customers, or how many jobs may be created or eliminated.
