Most companies say they are committed to AI. Most of those same companies have also quietly shelved AI projects in the past year because their data infrastructure couldn’t keep up.

That’s the central finding from Cloudera’s new “Great AI Re-Architecture” report, released August 11, 2026. The company surveyed 1,500 Enterprise Architects, Cloud Infrastructure Leads, and Data Architects worldwide and found that 95% of enterprises have delayed or cancelled AI initiatives because of data governance, compliance, or regulatory challenges.

The 5% who haven’t? Either they got very lucky with their existing setup, or they haven’t tried building anything ambitious yet.

The Gap Between AI Ambition and Data Reality

Here’s the detail that makes the 95% number worse: 55% of organizations have delayed or cancelled more than six AI projects in the past twelve months alone. This isn’t a one-off problem caused by a tricky compliance edge case. Enterprises are repeatedly hitting the same infrastructure wall.

The survey found 77% of organizations are actively using AI in some capacity, which sounds encouraging. But 72% say their current data architecture requires a significant overhaul before it can actually meet their AI requirements. That’s a large majority of companies trying to run AI on foundations they already know are wrong for the job.

Infrastructure costs are rising alongside these challenges. 84% of respondents say AI workloads have increased their infrastructure spending. That’s a hard pill to swallow when those same workloads are getting delayed by governance and compliance issues.

Why Governance Became the Bottleneck

For years, enterprise data teams focused on making data accessible. Now the conversation has shifted to making data trustworthy, traceable, and compliant — and it turns out those are much harder problems.

AI workloads amplify every data quality issue. A dashboard built on messy data might still give a useful rough picture. An AI agent built on messy data will confidently do the wrong thing at scale. Regulatory risk doesn’t just double when you add AI — it multiplies, because AI systems make decisions continuously and at speed.

Cloudera’s report names security, governance, and compliance as now outranking performance as the leading driver of architectural change in enterprise IT. That’s a fundamental inversion of how enterprise data teams traditionally thought about infrastructure priorities.

Only 25% of respondents say they plan to prioritize hybrid-first architecture over the next two years, suggesting most organizations are still working out where their data should actually live before they can properly govern it.

What This Means for Business

If you have been trying to roll out AI and running into unexplained friction, it’s probably not the models. The models are good enough. The problem is almost certainly in the data layer underneath them.

The companies pulling ahead with AI right now have done the unglamorous work of cleaning up their data architecture first. They know which data is authoritative, who can access what, and how to audit what the AI actually did. That foundation makes every AI deployment faster, cheaper, and lower-risk.

The companies still struggling are treating AI as a technology project dropped on top of their existing setup. It won’t work. You can’t bolt governance onto infrastructure that was never designed for it.

For business leaders, the re-architecture question deserves the same urgency as the AI question itself. A $50k AI project sitting on $2 of data infrastructure will fail. A $50k AI project sitting on a solid, governed data foundation has a real shot.

This is exactly the problem Enterprise DNA was built to help teams navigate. Understanding your data architecture — what you have, what you need, and how to close that gap — is the first practical step in any serious AI strategy. Our data skills training for teams builds the internal capability to assess and improve your data foundation, so AI projects have somewhere solid to land.

And for organizations that need strategic direction beyond training, Omni Advisory works with leadership to map the data and AI readiness gaps holding back their highest-priority initiatives.

The AI opportunity is real. The infrastructure prerequisite is real too. Cloudera’s 95% figure is a useful prompt to check which side of that divide your organization is currently on.