of GenAI projects were forecast to be abandoned after proof of concept by the end of 2025.
Forecast: Gartner, July 2024Source evidence brief · From proof to production
A successful AI proof of concept is not yet an enterprise capability.
In July 2024, Gartner forecast that at least 30% of generative-AI projects would be abandoned after proof of concept by the end of 2025. The warning focused on four conditions that separate an impressive demonstration from an initiative the enterprise can justify, govern and scale.
Primary source: Gartner. Gartner Predicts 30% of Generative AI Projects Will Be Abandoned After Proof of Concept By End of 2025, featuring Rita Sallam, Distinguished VP Analyst at Gartner, published July 29, 2024. View the original publication.
The forecast
The difficult transition begins after the demonstration works.
major abandonment drivers were named: data quality, risk controls, escalating costs and unclear business value.
Source: Gartner, 2024business leaders participated in Gartner’s separate value survey conducted from September through November 2023.
Survey source: Gartner“There is no one size fits all with GenAI.”Rita Sallam, Distinguished VP Analyst, Gartner
Why projects stop
Four problems can turn a promising pilot into an abandoned investment.
Poor data quality
A capable model cannot create dependable enterprise results when the information supplied to it is incomplete, obsolete, inconsistent or unsuitable for the use case.
Inadequate risk controls
Production use introduces privacy, intellectual-property, security, compliance and governance requirements that a contained demonstration may not expose.
Escalating costs
Model consumption is only one cost. Integration, data preparation, infrastructure, evaluation, governance, change and ongoing operation determine the real economic burden.
Unclear business value
Technical success does not automatically become financial value. The use case, workflow, beneficiaries, measurement method and expected outcome must be explicit.
Value is possible
The warning is about execution, not a claim that GenAI cannot produce value.
Gartner also reported positive outcomes from a separate survey of 822 business leaders. Respondents reported average improvements that varied by use case, job type and workforce skill. Gartner cautioned that benefits are company-specific and may take time to appear.
average reported revenue increase among survey respondents.
Gartner survey, 2023average reported cost savings among survey respondents.
Gartner survey, 2023average reported productivity improvement among survey respondents.
Gartner survey, 2023These are averages reported by Gartner survey respondents, not guaranteed outcomes for an individual organization or a claim about Trilivy.AI.
The economic question
Cost, ambition and value must be evaluated together.
Gartner described GenAI deployment costs as variable and difficult to predict. The total depends on what the organization is trying to accomplish, which deployment approach it selects and how broadly the capability is intended to operate.
$5M–$20M
Deployment approaches cited by Gartner
Gartner reported that GenAI deployment approaches aimed at transforming business models and creating new opportunities can involve costs ranging from $5 million to $20 million. The figure illustrates why an attractive proof of concept can become a much more consequential investment when it moves toward enterprise production.
The amount is not a universal price for every GenAI project. Scope, use case, deployment model, risk profile and strategic ambition materially change the cost.
View Gartner’s original Figure 1: Costs Incurred in Different GenAI Deployment Approaches →
Define the ambition.
Productivity assistance, extending an existing process and reinventing a business model carry different levels of cost, risk and strategic impact.
Measure the full system.
Evaluate business outcomes against total cost, including data, integration, governance, production operations and organizational change.
Allow for delayed value.
Some benefits emerge indirectly or over time. Immediate ROI is not the only decision measure, but expected future value still requires disciplined evidence.
Expand from proven outcomes.
When results meet expectations, the enterprise has a basis for extending the capability to more users, processes or business units.
The Trilivy interpretation
Move from an isolated proof to a reusable enterprise foundation.
The four abandonment drivers are larger than any one technology. Platform-First addresses their architectural dimension: prepare and govern company knowledge, make retrieval reusable, establish access boundaries, connect applications and agents through stable interfaces, and define ownership before separate initiatives multiply.
Jumpstart is designed to remove foundational assembly work and create deployment momentum. It cannot replace a real business objective, C-level ownership, a senior AI Data Expert, risk decisions or accountable execution. Those commitments are what turn a working platform into a working enterprise capability.
This interpretation is Trilivy’s analysis. Gartner did not evaluate or endorse Trilivy.AI, Jumpstart or the Platform-First doctrine.
Source record
Original publication details.
| Publisher | Gartner, Inc. |
|---|---|
| Title | Gartner Predicts 30% of Generative AI Projects Will Be Abandoned After Proof of Concept By End of 2025 |
| Featured analyst | Rita Sallam, Distinguished VP Analyst at Gartner |
| Publication date | July 29, 2024 |
| Location | Sydney, Australia |
| Supporting survey | Gartner survey of 822 business leaders conducted between September and November 2023 |
| Original source | Read the Gartner publication |
This page is an independent summary and commentary prepared by Trilivy.AI. It is not a reproduction of the Gartner publication. Gartner and its publication titles are the property of their respective owner.