In June 2025, Gartner predicted that more than 40% of agentic AI projects will be cancelled by the end of 2027. It named three causes: rising costs, unclear business value and insufficient risk controls. Gartner also estimated that only around 130 vendors are genuinely agentic out of the thousands claiming to be, and called the relabelling of chatbots and automation as agents 'agent washing'.
What the forecast does not say
It does not say agents don't work. The same research expects a growing share of everyday work decisions and enterprise applications to involve agentic AI by 2028. The message is narrower: many projects are started for the wrong reasons and fail on cost, value or control before they ever fail on technology.
Map each cause to a decision you can make now
- Rising costs — price the run, not just the build: model calls, review time and maintenance, expressed per completed task.
- Unclear value — pick one workflow with a number attached today (cost, delay, error rate) and agree in advance what improvement counts as success.
- Insufficient risk controls — decide what the agent may never do, enforce it in the tool layer, and log every action so it can be reviewed.
- Agent washing — ask the vendor to show the loop: what it plans, which tools it calls, how it recovers from a failed step.
A forecast of cancellations is also a forecast of which projects survive: the ones with a measured baseline, a bounded scope and a named owner.
A useful habit
Write the cancellation criteria before the project starts. If review time exceeds the target for two months, or the measured saving is below the cost per task, you stop or change scope. Teams that define the exit in advance spend less on the projects that deserved to end, and have more left for the ones that didn't.