Seven per cent. That’s how many enterprises say their data is genuinely ready for artificial intelligence, according to research published by Harvard Business Review Analytic Services and Cloudera on 25th March. The figure exposes a widening chasm between corporate AI ambition and operational reality.
The gap is most pronounced in the Middle East.
Across the Gulf states, governments have poured billions into national AI strategies and digital transformation initiatives. Yet the Harvard study—which surveyed 230 executives involved in AI data decisions during October 2025—found that 27 per cent of organisations admit their data isn’t ready at all. Another 66 per cent fall somewhere in between: experimenting with AI tools whilst their underlying data infrastructure remains fragmented, ungoverned, or simply inaccessible.
Seventy-three per cent of respondents acknowledged their organisations should prioritise AI data quality far more than they currently do. An identical proportion reported that preparing data for AI has proved challenging in practice. The obstacles aren’t abstract. Fifty-six per cent cited siloed data and difficulties integrating sources. Forty-four per cent pointed to the absence of a clear data strategy. Forty-one per cent flagged data quality and bias issues, whilst 34 per cent noted regulatory constraints on data use.
These aren’t teething problems.
They’re foundational deficits that prevent companies from moving AI projects beyond pilot stages. Enterprises can deploy machine learning models and experiment with generative tools, but without unified, governed access to data across hybrid environments—spanning public clouds, private data centres, and edge locations—those initiatives stall. Data sovereignty requirements, compliance frameworks, and security concerns mean mission-critical information often can’t be moved to where AI models operate. The result: expensive experimentation that rarely scales.
“AI is only as powerful as the data behind it,” said Sergio Gago, Chief Technology Officer at Cloudera. “To move from pilots to production, organizations need secure access to 100% of their data, anywhere it resides. Bringing AI to data instead of moving data to your AI is what separates experimentation from enterprise-scale impact.”
Leadership teams have noticed. Whilst only 23 per cent of surveyed organisations have established a formal data strategy for AI adoption, another 53 per cent are actively developing one. The priorities within those strategies reveal where pain points bite hardest: 59 per cent rank protecting sensitive data and privacy as critical, 46 per cent emphasise data quality, and 41 per cent highlight governance.
By the time most firms formalise those strategies, the AI landscape will have shifted again.
Sixty-five per cent of respondents expect many of their business processes to be augmented or replaced by agentic AI within two years. These aren’t chatbots or recommendation engines—agentic systems autonomously execute tasks, make decisions, and interact with other software without human intervention. They demand real-time access to accurate, contextually rich data across distributed environments. Forty-seven per cent of executives believe agentic AI might solve their data quality issues, though that optimism may prove misplaced if the underlying infrastructure remains broken.
The Middle East’s predicament mirrors broader global trends but carries higher stakes. National visions—Saudi Arabia’s Vision 2030, the UAE’s AI Strategy 2031—hinge on rapid adoption. Ahmad Shakora, Group Vice President for South-META at Cloudera Middle East, framed the challenge bluntly.
“Across the Middle East, organizations are accelerating AI adoption as part of broader national digital transformation agendas. However, the challenge many enterprises face today lies in ensuring their data environments are ready to support AI at scale. Fragmented systems, evolving governance frameworks, and the complexity of operating across hybrid and multi-cloud environments continue to slow progress. As AI capabilities advance, particularly with the emergence of agentic systems, organizations will need to prioritize trusted data, strong governance, and modern data architectures to translate AI ambition into measurable business impact.”
The report, titled *Taming the Complexity of AI Data Readiness*, underscores a paradox: organisations invest heavily in AI capabilities whilst neglecting the data foundations those capabilities require. Innovation captures executive attention and headlines. Governance, data integration, and hybrid architecture design don’t—until projects fail.
Cloudera’s research arrives as enterprises confront the limitations of cloud-first strategies. Many discovered that moving vast datasets to public cloud environments incurs prohibitive costs, introduces latency, and conflicts with data residency regulations. The alternative—bringing compute power and AI models to wherever data already resides—requires architectures that function consistently across on-premises data centres, multiple cloud providers, and edge deployments.
Cloudera itself operates across more than 25 exabytes of enterprise data globally, built on open-source foundations designed for hybrid environments. The company argues that converging public cloud capabilities with on-premises infrastructure delivers the flexibility enterprises need without forcing data movement or surrendering control. Whether that approach becomes standard practice depends partly on how quickly competitors—hyperscalers like Amazon Web Services, Microsoft Azure, and Google Cloud—adapt their offerings.
What’s certain is that the 93 per cent of companies not yet ready face mounting pressure. Agentic AI won’t wait for data strategies to mature. Competitors won’t pause whilst governance frameworks get formalised. Regulatory bodies won’t relax compliance requirements because integration proves difficult.
The executives surveyed in October 2025 recognised these realities. Their responses suggest awareness of the problem has spread widely across industries. What remains unclear is whether that awareness will translate into the architectural overhauls, strategic investments, and organisational changes required to close the readiness gap.
For now, the figures tell a stark story. Seven per cent are ready. Twenty-seven per cent aren’t close. The rest are somewhere in the middle, racing to prepare data infrastructures before the AI wave they’ve already committed to crashes over unprepared foundations. The timeline for catching up is measured in months, not years. The consequences of failure will be measured in competitive disadvantage, wasted investment, and strategic opportunities lost to better-prepared rivals.
The full report is available on Cloudera’s website.