Gartner warns of 40% agentic AI failure by 2027 - Industry leaders push back

Time of India

09-07-2025

In a recent report, Gartner has predicted that more than 40% of Agentic AI projects would be cancelled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls. Though making significant predictions, the report does not outline the nature of the study, the parameters or the methodology involved in reaching such conclusions.'Most agentic AI projects right now are early stage experiments or proof of concepts that are mostly driven by hype and are often misapplied,' says Anushree Verma, Senior Director Analyst, Gartner. 'This can blind organizations to the real cost and complexity of deploying AI agents at scale, stalling projects from moving into production. They need to cut through the hype to make careful, strategic decisions about where and how they apply this emerging technology.'
"If we don't invest in the invisible guardrails that Agentic AI warrants or if we design Agentic AI by overlooking overall system design, costs can escalate and repatriation can start. Left shift is a must as far as security, performance and costs are concerned. These should be the considerations at the architecture stage and not at the monitoring stage. Not everything needs to become agentic and not everything needs to go on cloud," says Anjali Satam, Head of Engineering and Technology Director at IKS Health.
Assessing Gartner's initial claims
Gartner's initial prediction about agentic AI adoption—prior to June 2025—was highly optimistic, especially regarding customer service and enterprise automation. They also stated that agentic AI would autonomously resolve common customer service issues without human intervention, leading to a reduction in operational costs.
"I agree with the current Gartner point of view. Because, the ability to provide flawless access to data for agentic AI to truly function as intended is a major challenge, because most data is not ready and the majority of organizations have not invested in the computing infrastructure needed to deliver on the real promise of agentic AI," says Rajendra Deshpande, former CIO at Intelenet Global Services and a technology consultant.
Implementations should be driven by business objectives
A few technology experts are of the view that the current agentic AI models don't have the maturity and agency to autonomously achieve complex business goals or follow nuanced instructions over time and may not be able to autonomously achieve complex business goals or follow nuanced instructions over time and therefore could lack value or return on investment (ROI). Moreover, some use cases positioned as agentic AI may not require agentic implementations to be successful.
'Most agentic AI propositions lack significant value or return on investment (ROI), as current models don't have the maturity and agency to autonomously achieve complex business goals or follow nuanced instructions over time. Many use cases positioned as agentic today don't require agentic implementations,' adds Verma.
Understanding Gartner's hype cycle
Such overarching statements are not unusual. Most technologies follow the hype cycle. The hype cycle is a method to assess how technology evolves along an S-curve. Gartner categorizes current trends into five buckets, each representing a different stage of maturity. This process involves combining technologies and grouping them according to their maturity by a team of experts helps them evaluate the status of each technology on the S-curve.
Such observations have usually held true for almost all emerging technologies. Almost 40% of projects fail in the first 1000 days, whether it is blockchain, drones or generative AI. The reasons could be diverse, including factors such as maturity of use cases, readiness of the specific industry, the cost of technology acquisition and implementation, challenges of integration with the existing legacy systems, etc. Gartner explains this phenomenon through its concept of the hype cycle, which is a graphical tool that plots the maturity, adoption, and social application of emergent technologies, providing insights into their potential risks and benefits. It includes five phases: the innovation trigger, peak of inflated expectations, trough of disillusionment, slope of enlightenment, and plateau of productivity.
"I am not sure if they use any statistical models, but a team of experts definitely reviews and evaluates the status of each technology on the S-curve. This means there is a level of subjectivity embedded in the assessment for sure. Again, it would be interesting to see how many of their predictions have actually come true. I think the hype cycle is more of a framework to assess the level of maturity of technologies, rather than a precise prediction method. It would be even more interesting to know if Gartner has any audited data about the accuracy of their predictions," observes Deshpande.
A major driver of the current hype around agentic AI is the widespread rebranding of existing products by many suppliers, often without delivering the true, full capabilities that define agentic AI.
"In 2000, dot-coms rose fast and fell faster. In 2010, many IT services projects grew big but couldn't scale well. In 2020, SaaS exploded, but too much of it lacked clear value. Every hype wave leaves behind good ideas that failed because of weak execution or no real go-to-market!" points out Kingshuk Hazra, Founder, LeadStrategus.