A new forecast from Gartner warns that more than 40% of agentic AI projects will be canceled by the end of 2027, as organizations grapple with rising costs, unclear returns, and poor risk management.

The prediction, released ahead of the Gartner IT Symposium/Xpo, reflects growing skepticism around the real-world viability of agentic AI — systems designed to make autonomous decisions and act on behalf of users.

“Most agentic AI projects right now are hype-driven experiments or proof of concepts,” says Anushree Verma, Senior Director Analyst at Gartner. “Without a clear strategy or understanding of the technology’s complexity, many organizations will struggle to scale beyond the prototype stage.”

A Gartner poll conducted in January 2025 found that just 19% of companies had made significant investments in agentic AI, while 42% remained cautious, and nearly a third were undecided or inactive.

The report also highlights a trend of “agent washing”, where vendors rebrand older technologies like chatbots and RPA tools as agentic AI without delivering true autonomous capabilities. Gartner estimates that only 130 vendors worldwide offer genuine agentic AI products.

Despite early pitfalls, Gartner sees long-term promise in the field. By 2028, the firm predicts 15% of routine work decisions will be handled by agentic AI, up from virtually none today, and that one-third of enterprise software will embed agentic capabilities.

To succeed, Gartner advises organizations to focus on areas where agentic AI clearly boosts enterprise productivity, rather than retrofitting it into outdated systems. “Real value comes from applying AI where it drives measurable improvements in cost, speed, quality, or scalability,” Verma says.

As agentic AI matures, Gartner expects it to become a transformational force, but only for businesses that adopt it with discipline, not hype.