Tech execs are getting wise about ROI from AI
Most organizations have been largely unable to measure financial returns from AI, but analysts say new ways to calculate return on investment are emerging. “There’s a delay between the development of technology, even the investment in the technology, and the value that an organization can capture from it,” said Michael Chui, a senior fellow at McKinsey. But more executives are asking questions.…
Many companies have struggled to quantify the financial benefits of artificial intelligence, but experts say novel methods to calculate return on investment are emerging. McKinsey's Michael Chui explained that it takes time for organizations to reap the rewards of technology investments. CFOs, investors, and CEOs are increasingly pressing technology leaders to prove the ROI of AI.
According to McKinsey's "State of AI" survey, around 80% of respondents reported AI boosting productivity, yet only 37% attributed measurable profit gains, a figure similar to the previous year. Only 6% claimed AI delivered significant value, accounting for at least 5% of operating profit. The disparity lies between individual worker gains and organizational returns, according to Chui.
McKinsey's Technology Trends Outlook suggests the greatest gains will arise from redesigning workflows where humans collaborate with AI agents. Merely adding agents to existing processes isn't sufficient. Chui emphasized that high-performing companies redesign entire workflows with AI integration. Controlling costs is crucial for better returns, and managing token costs while selecting the appropriate model for each task is essential.
Gartner's Gareth Herschel warned that a lack of transparency regarding token usage can lead to surprise costs. In fact, 60% of IT leaders expressed concerns about unexpected AI costs, with token expenses for coding assistance being higher than human developer wages in some cases. As more agents collaborate, financial risks escalate.
Robert Thanaraj from Gartner likened the situation to giving a teenager a credit card, highlighting the potential for financial risks. To mitigate these risks, companies must track costs during prototyping, such as determining the cost per completed task for each agent. This approach helps evaluate various large language models or opt for more affordable alternatives like smaller language models or open-source models.
While financial ROI remains a priority, analysts stress the importance of a broader perspective. McKinsey's Chui pointed out that challenges like unexpected costs, poor data quality, scaling issues, and low user adoption contribute to the difficulty in calculating AI ROI. However, executives are becoming more adept at tracking AI expenses and returns.
In 2025, the likelihood of an AI initiative achieving ROI stands at just one in five. Beyond financial metrics, companies should link AI projects to both financial and non-financial outcomes, a concept Gartner refers to as "return on intelligence." "We need to shift the emphasis from cost to value," added Gartner's Robert Thanaraj.
A strong foundation consisting of context, infrastructure, and governance is vital for achieving AI ROI. Data quality and a well-designed AI system are critical; without them, large language models function only as guesswork. Gartner's Jack Gold emphasized the need for robust networking infrastructure to support agent-to-agent interactions.
Organizations are increasingly establishing AI harnesses, software layers that govern models, tools, and workflows. A recent KPMG survey revealed that 55% of organizations have a formal AI harness, with 86% of those reporting established ROI. Companies that foster accountability, coordinated governance, resilience, and reliable value measurement are more likely to transform widespread AI adoption into sustained performance.
Written by urgent.news from Computerworld's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.