Enterprise data readiness for AI is not where most organizations think it is - and the gap between executive confidence and practitioner reality is now showing up directly in failed projects and abandoned pilots. A string of surveys conducted between early 2025 and early 2026 points to the same structural problem: organizations have invested heavily in data platforms but have not resolved the underlying quality, governance, and accessibility issues that AI workloads actually require. The consequences are measurable and accelerating.
Sources: Gartner Feb 2025 and Apr 2026; Stibo Systems/CIO Dive Nov 2025; Informatica CDO Insights 2025
| Value (%) | Share (%) |
|---|---|
| Orgs lacking AI data mgmt | 63% |
| AI projects at risk of abandonment | 60% |
| Leaders citing data quality barrier | 28% |
| AI I&O use cases meeting ROI | 28% |
| Orgs with AI-sufficient data quality | 12% |
What the research says about data maturity self-assessment
Gartner reported in February 2025, ahead of its Data and Analytics Summit, that 63% of organizations either lack or cannot confirm they have the right data management practices to support AI deployments. That figure alone would be alarming enough, but a companion finding from an Actian survey of more than 600 enterprise data professionals - reported by CIO Dive in August 2025 - adds a sharper edge. Executives consistently rate their own organization's data maturity 12 percentage points higher than the data engineers and practitioners who operate inside those environments on a daily basis.
That 12-point gap is not a rounding error. It is a structural misalignment between board-level confidence and ground-level reality, and it has direct consequences for how AI investments get scoped, funded, and evaluated. When leadership believes data infrastructure is solid, governance gaps do not make it onto the priority list - until a GenAI pilot grinds to a halt because the training data is incomplete, inconsistent, or simply inaccessible across siloed systems.
Only 12% of organizations report data of sufficient quality and accessibility for AI applications - yet nearly half of business leaders list AI products as a top priority. That gap does not close on its own.
CDOs are already naming data quality as their top obstacle
The Informatica 2025 CDO Insights survey quantifies the problem from a different angle. Data quality and readiness ranked as the single largest obstacle facing chief data officers, cited by 43% of respondents. More striking is what the survey found at the other end of the scale: only 12% of organizations reported data of sufficient quality and accessibility to actually support AI applications. That means the vast majority of enterprises are building AI strategies on a foundation that their own data leaders consider inadequate.
A November 2025 CIO Dive report on a Stibo Systems and Wakefield Research study reinforced the same tension. While 49% of business leaders identified AI-driven products as a top 2025 priority, 28% simultaneously acknowledged that poor data quality is the primary barrier actively preventing AI adoption. Organizations are not choosing between AI ambition and data readiness - they are pursuing both in parallel while the infrastructure gap widens underneath them.
AI project failure rates are already reflecting the shortfall
The most direct evidence of what this gap costs in practice came in April 2026, when Gartner published results from a survey of 782 infrastructure and operations leaders conducted in late 2025. Only 28% of AI use cases in that domain fully succeeded and met ROI expectations. Among leaders who reported setbacks, 38% attributed the failure directly to poor data quality or limited data availability - making it the leading cited cause of AI underperformance in the I and O space.
Gartner's broader 2025 forecast, corroborated by Dataversity analysis in February 2026, projects that 60% of AI projects not backed by AI-ready data foundations will be abandoned by 2026. These are not hypothetical risks. They describe a pattern already visible in organizations that launched GenAI pilots on enthusiasm, discovered their data pipelines were unfit for purpose, and quietly shelved the work rather than invest in fixing the underlying infrastructure.
What CIOs need to do differently
The perception gap identified by the Actian research is the strategic problem CIOs must close first, because it determines whether data quality work gets funded at all. If the C-suite believes data readiness is in good shape, there is no internal pressure to invest in metadata management, data quality automation, or AI-ready architecture. Closing that gap requires CIOs to put specific, practitioner-sourced data maturity metrics in front of leadership - not platform adoption numbers, but pipeline reliability rates, data quality scores by domain, and accessibility benchmarks across business units.
From there, the work is architectural. Organizations need automated data quality pipelines that catch and flag issues before they reach AI model inputs, not after a project stalls. Governance frameworks must be updated to treat AI readiness as an explicit data product requirement, not an afterthought. Siloed data environments - which show up in almost every survey as a compounding factor - require active investment in data mesh or data lakehouse strategies that make assets accessible across teams without sacrificing control. None of this is technically exotic, but it does require sustained investment priority, and that priority will not materialize while executives are overrating their current state by 12 points.
