Uncategorized

Why Gartner Expects 40% of Agentic AI Projects to Be Cancelled: What It Means for Marketing Budgets

Marketing teams are being offered agents for SEO, ads and reporting, usually with a promise of scale. Before signing, it helps to know how often such projects fail. Gartner’s June 2025 prediction that more than 40% of agentic AI projects will be cancelled by the end of 2027 names three causes: escalating costs, unclear business value and inadequate risk controls.

Key Takeaways
  • Gartner’s June 25, 2025 press release predicts over 40% of agentic AI projects will be cancelled by the end of 2027.
  • It attributes cancellations to escalating costs, unclear business value or inadequate risk controls.
  • Gartner also described “agent washing” and estimated only about 130 vendors out of thousands offer genuinely agentic products.
  • The mapping to marketing in this article is our interpretation, not Gartner’s guidance.

What Gartner actually predicted

Gartner’s June 25, 2025 press release said it expects over 40% of agentic AI projects to be cancelled by the end of 2027, citing escalating costs, unclear business value or inadequate risk controls. It advised pursuing agentic AI only where it delivers clear value or return on investment.

The release also coined a warning for buyers. Agent washing: vendors relabeling existing assistants, chatbots or robotic process automation (RPA) as agentic AI. Gartner estimated that only about 130 of the thousands of vendors claiming agentic capability actually deliver it.

This is a forecast, not a measured failure rate, and it covers agentic projects across industries rather than marketing alone. Its value to a buyer is as a checklist of how projects fail, which is easier to act on than a headline percentage.

How strong the underlying evidence is

Gartner based part of its commentary on a January 2025 poll of 3,412 webinar attendees. In that poll, 19% said their organizations had made significant investments in agentic AI and 42% said conservative ones.

A webinar audience is self-selected, so those percentages describe interested attendees rather than the market as a whole. We would not quote them as an adoption rate. The cancellation forecast is likewise an analyst judgment, and nobody has yet checked it against outcomes.

The implication for a founder is balanced. The prediction does not show that agents fail, and it does not show they succeed. It shows that a respected analyst firm expects budget, value and governance gaps to be the common reasons projects end.

Three failure causes in a marketing context

The table translates Gartner’s three causes into marketing terms. The middle and right columns are our interpretation of where each cause tends to appear in agent-run SEO, content or ad work.

Gartner’s causeHow it can appear in marketingControl to set before launch
Escalating costsUsage-based fees grow as an agent runs more actions or tests more variantsA monthly spend ceiling and a clear pricing model
Unclear business valueReports show activity, such as pages produced, but not leads or revenueA baseline and one agreed outcome metric
Inadequate risk controlsAn agent edits live pages or shifts ad spend without reviewApproval gates and a rollback owner, as in our note on approval gates for agentic ad buyers

None of these controls requires an advanced setup. They require someone to decide, in writing, what counts as success and who can stop the agent.

What to write down before launch

Gartner’s causes suggest a short pre-launch document. The steps below are our recommendation, not Gartner’s, and each one maps to a cause above. Teams that skip them tend to discover the gap in a quarterly review, when the agent has already run for months.

  1. Write the single outcome the agent is meant to move, such as qualified leads, and record its current level.
  2. Set a maximum monthly spend and ask the vendor in writing how charges scale with activity.
  3. List the actions the agent may take alone and the actions that need a human approval first.
  4. Name one person who can pause the agent and one who owns reversing its changes.
  5. Pick a review date at 90 days with a stop condition, following the structure in our guide to a 90-day pilot with exit criteria.

If a vendor cannot answer the first three items, that is informative in itself. Our checklist for evaluating an agentic marketing vendor covers what to ask next.

Frequently Asked Questions

What did Gartner predict about agentic AI projects?

In a June 25, 2025 press release, Gartner predicted that over 40% of agentic AI projects will be cancelled by the end of 2027 because of escalating costs, unclear business value or inadequate risk controls.

What is agent washing?

It is Gartner’s term for vendors rebranding existing assistants, chatbots or RPA tools as agentic AI. Gartner estimated that only about 130 vendors offer genuinely agentic products.

Does the prediction mean marketing teams should avoid AI agents?

No. It is a forecast of where projects fail, and Gartner’s advice was to pursue agentic AI where it delivers clear value. Setting cost, value and risk controls first is the practical response.

A forecast like this is most useful when it turns into a short list of decisions made before money is spent. At Grow with Hafsa we help US and UK teams define baselines, spend limits and approval rules for AI agents in SEO, ads and content. That is part of our AI marketing and performance marketing services, and it begins with deciding what success would look like.

Inadequate risk controls often come down to what an agent is allowed to touch. Our note on least-privilege access for AI marketing agents applies OWASP’s excessive-agency guidance to ads, CMS and CRM permissions.