WEBINAR
Improving the economics of agentic AI: A guide for business leaders
Spending on AI continues to climb, but most organizations have yet to see significant results.
As AI continues to get better and faster, the conversation has turned from the technology’s capabilities to its costs. Although 80 percent of respondents to the latest McKinsey Global Survey on the state of AI say the technology has added to their productivity, only 37 percent say AI has meaningfully improved their organization’s bottom line. What’s more, just 6 percent of organizations qualify as true AI high performers. How can the other 94 percent translate spending into outcomes?
In a McKinsey Live event, Senior Partners Lari Hämäläinen and Tanguy Catlin explored the gap between spending and enterprise-level payoff and discussed how companies can bridge that gap.
The gap between adoption and impact
Companies are broadening their use of AI across functions and scaling it. But most are still not capturing enterprise value from it.
The experience of AI high performers—those that can show an EBIT impact of 5 percent or more—offers several lessons. Rather than simply using more AI, high performers deploy AI in areas that offer operating leverage for value creation. They also reimagine the end-to-end workflows around AI (versus plugging AI into existing workflows); pursue growth, innovation, and efficiency; and maintain stronger leadership and measurement discipline.
The economics of AI spending
As AI investment rises, many leaders discover that AI spend behaves differently from a traditional software budget. One in five survey respondents says AI-related operating costs are creating constraints. At the same time, 60 percent of respondents plan to increase AI investment next year.
High performers tend to invest more in AI than their peers do, particularly in the software development life cycle, and they are more likely to report cost constraints in software coding agents. They were also the first to measure the business outcomes from token consumption and start optimizing.
The technology shifts every business leader should be watching
October 15, 2026, 9:30 a.m. ET / 3:30 p.m. CET
Join McKinsey experts Michael Chui and Roger Roberts for a discussion on tech trends, including where momentum is building, how the major trends are converging, and what these shifts mean for business leaders, workers, and society.
The levers that can optimize AI spend
Once leaders understand the economics of AI, they may be tempted to deploy levers such as cheaper models, tighter limits, and fewer tokens. But no single cost lever can move the needle. A McKinsey analysis has shown that there are about 40 levers that generate the best outcomes in optimizing spend, spanning forecasting, optimization, sourcing, and governance.
These levers can be broken down into three priorities: visibility and attribution, optimized workflow, and demand and sourcing discipline. The goal is to spend on intelligence that delivers high returns, not to cut costs indiscriminately.
The shift from AI tools to agentic factories
Deploying chatbots and other tools can boost employee competency and digital literacy, but the largest gains from AI come from redesigning end-to-end workflows that are critical to the operating performance of the organization. The next frontier is “agentic factories,” in which employees supervise a much larger virtual workforce of agents, enabling organizations to unlock value at scale.
What CEOs can do now to build the agentic operating model
CEOs can take several actions today to turn AI into a scalable source of value:
- Identify two to three lighthouses to rebuild with AI. Choose areas of the business that truly matter, and then start experimenting.
- Own the contextual data. From meeting notes to process documentation, data is highly valuable, especially as agents become part of the process.
- Build agentic operating capability as a discipline. Redesign workflows end to end rather than inserting agents into legacy processes.
- Reshape the boundary of the firm. Think strategically about which processes to outsource and which to keep in-house.
- Control costs tightly. Measure outcomes and manage technology budgets accordingly.
