Gartner expects more than 40% of agentic projects to be cancelled by 2027
The stated reasons are escalating costs, unclear business value, and inadequate risk controls. Only one of those three is a technical problem, and it is the one nobody is working on.
Gartner's forecast is that more than 40% of agentic AI projects will be cancelled by the end of 2027, citing escalating costs, unclear business value, and inadequate risk controls.
It is worth separating those. Escalating costs are partly technical — agents call models repeatedly, and a loop that retries is a loop that bills. Unclear business value is a scoping failure that predates AI by decades; organisations have always funded pilots without deciding what success looked like.
Inadequate risk controls is the interesting one, because it is the only item on the list that the field could build its way out of and largely has not. An agent that acts needs a record of what it did and why, a boundary on what it may do unsupervised, and a way to stop it. Those are engineering problems with known shapes, and they are consistently the last thing built rather than the first.
Set the forecast against the same period's adoption numbers and it stops sounding pessimistic. If 88% of pilots never ship, a 40% cancellation rate among the ones that do is not a collapse — it is a market discovering which applications were real. The failures will cluster where the accountability is highest and the tolerance for a wrong answer is lowest.
The forecast is a projection, not an observation, and Gartner's hype-cycle calls have a mixed record. The mechanism it names, though, matches what the deployment data already shows.
Joget — AI Agent Adoption 2026: What the Analysts' Data Shows → · Azumo — 60+ AI Agent Statistics for 2026: Adoption, ROI & Market Growth →