McKinsey: AI Agents Drive Costs Up by Up to 30 Times

Key Takeaways

1The costs of AI agents can vary by up to a factor of 30 according to McKinsey

2One third of organizations allocate more than 10% of their tech and communications budget to AI

360% of respondents plan to increase their AI spending next year

Why it matters : Companies need to assess whether the productivity gains from AI agents offset the highly variable and rising operational costs.

Organizations are already tightening the use of AI in response to rising bills, as McKinsey warns that agent costs can vary by up to 30 times. While these tools can significantly reduce the time spent on certain tasks, the firm urges companies to assess their real value before accelerating their deployment.

Companies Adjust AI Usage and Exercise Caution Over 12 Months

After a year in which companies encouraged their employees to use AI more, some are reverting to more regulated practices. Amazon has shut down an internal leaderboard that tracked AI token usage after employees completed tasks to climb the rankings. Other companies, such as Coinbase and Salesforce, have begun to limit AI usage as bills increase. According to Tanguy Catlin, senior partner and director of the McKinsey Global Institute, AI spending is now becoming significant and visible. Lari Hämäläinen, senior partner at McKinsey, notes that many aspects remain to be discovered and believes this issue will gain particular importance in the next 12 months.

Budgets: One Third Beyond 10% and 60% Plan to Increase

According to a McKinsey survey, about one third of organizations are already allocating more than 10% of their technology and communication budgets to AI. Furthermore, 60% of respondents plan to increase these expenditures next year. Approximately one in five indicates that AI spending is starting to weigh on their operating costs.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

Agent Costs: More Expensive Than Text and Highly Volatile

McKinsey emphasizes that the profitability of AI agents varies and their use can lead to expenses significantly higher than those of text-based AI solutions, as they actually perform tasks. These often require multiple successive phases, and there are various ways to achieve the same goal, leading to significant cost fluctuations. According to the firm, the final cost associated with different agents or various executions can differ by a ratio of up to 30. Lari Hämäläinen illustrates this point by mentioning an operation whose cost could multiply by 30 from one day to the next. This phenomenon is particularly pronounced in development teams that automate coding, a highly token-consuming activity.

Announced Productivity and the Need to Prove Value

McKinsey reports that AI agents can reduce the human time spent on certain tasks by 35% to 40% for transformation work, and sometimes by 70% or more. However, the firm recommends that companies ensure the value generated by these agents justifies their operating costs. It also anticipates that AI-related expenditures could continue to rise as agents become more widespread.