Most firms delay AI projects amid governance fears

Cloudera has published a global survey showing that 95% of enterprises have delayed or cancelled AI projects because of infrastructure and governance constraints. The findings point to a broad rethink of corporate data architecture.

The survey drew responses from 1,500 enterprise architects, cloud infrastructure leads and data architects across the Americas, EMEA and Asia Pacific. It found that 77% of organisations are actively using AI, yet 72% said their current data architecture needs significant change to meet future AI requirements.

That tension is emerging as companies move beyond AI pilots and try to deploy systems across more of their operations. Three quarters of respondents said AI integrations had changed their organisation's data storage and architecture practices, while 84% reported higher infrastructure costs linked to AI workloads.

Governance pressure

Governance, compliance and regulation emerged as the main reasons projects are being held back. Nearly three quarters of respondents, or 73%, said AI had made data governance more complex.

More than half of organisations worldwide said they had delayed or cancelled more than six AI projects over the past year because of governance, compliance or regulatory issues. The research also found that 97% of respondents move data between environments at least once a month, making it harder to apply consistent controls across systems.

Sergio Gago, Chief Technology Officer at Cloudera, said the findings reflect a broader shift in how companies are approaching their technology foundations.

"This current era of AI is forcing organizations to rethink the foundations of their technology infrastructure," said Sergio Gago, Chief Technology Officer at Cloudera. "Many enterprises are discovering that the architectures built for traditional analytics weren't designed for the scale, governance, and flexibility AI demands today. Success will depend on building a data foundation that gives organizations the freedom to run AI wherever it makes the most sense, without compromising control or security."

Hybrid shift

The data also suggests businesses are moving away from a single-cloud approach. Two thirds of respondents globally said they had shifted AI workloads from public cloud environments back to private cloud or on-premises infrastructure during the past year.

That figure stood at 66% globally and 64% across Asia Pacific, indicating a wider move towards hybrid setups that combine public cloud, private cloud, on-premises systems and edge environments. One quarter of respondents said they plan to prioritise a hybrid-first architecture over the next two years.

In Asia Pacific, 68% said their existing architecture requires significant change. The report found that 92% of organisations in the region had delayed or cancelled at least one AI project, while 61% said AI had made data governance more complex.

Security, governance and compliance requirements were cited by 46% of Asia Pacific respondents as the main driver of infrastructure change. The figures suggest companies in the region are under similar pressure to peers elsewhere as they try to match older systems with newer AI demands.

Singapore focus

Singapore stood out in the regional figures for reassessing where AI workloads should run. The survey found that 43% of organisations in Singapore had already moved AI workloads out of the public cloud, while a further 36% were evaluating such a shift.

Edge infrastructure is also drawing attention in the city-state, with 35% of local respondents expecting to increase spending in that area over the next two years. Another 41% said data security, governance and compliance were the main reasons for changing AI infrastructure.

Remus Lim, Senior Vice President for Asia Pacific & Japan at Cloudera, said the regional focus is shifting from simple adoption to deciding which environment best suits each workload.

"Across the Asia Pacific region, the conversation is transitioning from merely selecting an AI host to determining which environment yields the superior business results for specific workloads," said Remus Lim, Senior Vice President for Asia Pacific & Japan at Cloudera. "Singapore reflects this evolution as enterprises collectively evaluate latency, performance, expenditure, and governance when defining their AI deployment strategies. To scale AI breakthroughs effectively, businesses must possess the agility to transition data and AI seamlessly among cloud, edge, and on-premises setups while maintaining strict governance and operational command."

The survey covered nine markets across three regions: the US, Canada, Brazil, South Africa, Spain, the UK, Singapore, India and Japan. Respondents were drawn from organisations with at least 1,000 employees in most markets, with a lower threshold in Spain, Singapore and South Africa.

The findings underline how quickly AI deployment has outpaced many existing enterprise systems. Even as adoption becomes more common, the report indicates that data movement, governance demands and the cost of running AI are forcing companies to redesign the infrastructure beneath their projects.