The market is now three years past the introduction of ChatGPT, with discussions surrounding generative AI increasingly framing it as a “bubble,” questioning the viability of the technology beyond a few leading providers. Recent findings from the MIT NANDA report highlighted that 95% of AI pilot projects fail to scale or demonstrate clear return on investment (ROI). This sentiment is echoed by McKinsey, which suggests that agentic AI could be pivotal for realizing operational benefits. During a recent Technology Council Summit hosted by The Wall Street Journal, AI leaders advised Chief Information Officers (CIOs) to reconsider concerns about measuring AI ROI, as these metrics can often be misleading.
This situation raises questions for technology leaders who rely on established tech stacks. Introducing new technologies may seem risky, particularly when existing systems are already effective and stable. Historically, organizations have hesitated to replace components of their tech stacks out of fear that doing so could disrupt essential business workflows.
To realize value from new technological investments, organizations must consider their data as a critical asset. Effective deployment of AI often involves integrating proprietary business data into AI models. This requires navigating concerns around confidentiality and negotiating terms with model vendors who rely on access to non-public data.
Additionally, the pace of AI innovation has resulted in a cluttered marketplace, with many models quickly becoming obsolete. Successful AI implementations tend to focus on specific business challenges, adding value without necessitating constant updates to the latest versions.
Enterprises should align new technology strategies with their operational capabilities rather than adhering strictly to vendor recommendations. By designing systems to match the actual consumption patterns of users, organizations can avoid costly pitfalls and enhance efficiency. The emphasis should remain on practical applications that minimize disruption while ensuring that AI deployments genuinely leverage business data to drive value.
Source: https://www.technologyreview.com/2025/10/28/1126693/finding-return-on-ai-investments-across-industries/