Cloudera survey finds AI projects stalled by governance

Cloudera has released survey findings showing that 95% of enterprises have delayed or cancelled AI projects because of infrastructure and governance constraints. The study points to a broad redesign of corporate data architecture around hybrid environments.

The report is based on responses from 1,500 enterprise architects, cloud infrastructure leads and data architects across nine markets. It found that 77% of organisations already use AI, but many are struggling to expand those deployments because older systems were not built for current governance, compliance and operational demands.

Nearly three-quarters of respondents, or 72%, said their existing data architecture needs a significant overhaul to meet future AI requirements. The finding suggests many businesses now see AI less as an application layer and more as a force driving a wider rebuild of data systems.

Growing strain

AI use is also reshaping the economics of enterprise infrastructure. Three-quarters of respondents said AI integrations have changed their organisation's data storage and architecture practices, while 84% reported higher infrastructure costs linked to AI workloads.

The problem is not limited to computing expense. Data governance is emerging as a main reason projects stall, with 73% saying AI has made governance more complex.

More than half of respondents, or 55%, said they had delayed or cancelled more than six AI projects over the past year because of governance, compliance or regulatory issues. That points to a gap between enthusiasm for AI adoption and the practical ability to run systems under existing internal controls.

Distributed data is adding to that pressure. Almost all respondents, 97%, said they move data between environments at least once a month, underlining the difficulty of applying consistent rules across public cloud, private cloud, on-premises systems and edge locations.

For large organisations, that challenge is closely tied to trust in the data feeding AI systems. When information is spread across multiple environments and subject to different controls, scaling AI tools across a business becomes harder and more expensive.

Sergio Gago, Chief Technology Officer at Cloudera, said the findings show companies are being pushed to revisit long-standing technology assumptions.

"This current era of AI is forcing organisations to rethink the foundations of their technology infrastructure," Gago said.

"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 organisations the freedom to run AI wherever it makes the most sense, without compromising control or security," he added.

Hybrid shift

One of the clearest findings in the survey is the move towards hybrid infrastructure. Two-thirds of respondents, or 66%, said they had shifted AI workloads from public cloud back to private cloud or on-premises infrastructure over the past year.

That suggests many companies no longer treat public cloud as the default destination for AI. Instead, they appear to be making more selective decisions based on cost, governance requirements, system performance and data control.

A quarter of respondents said they plan to prioritise a hybrid-first architecture over the next two years. Rather than relying on a single model, businesses are spreading AI workloads across cloud, on-premises and edge environments.

This marks a shift in the enterprise technology debate. Early AI adoption often focused on getting models into production quickly, but the survey suggests companies are now concentrating on where those systems should run and under what governance rules.

Broader implications

The results also reflect a wider shift in how large organisations approach modernisation. Instead of replacing old systems simply to improve efficiency, many are now being driven by the need to make AI usable at scale without creating new compliance risks.

Cloudera's survey covered respondents in the United States, Canada, Brazil, South Africa, Spain, the UK, Singapore, India and Japan. Participants worked at companies with at least 1,000 employees in most markets, with a lower threshold in a small number of countries.

The emphasis on senior architecture and infrastructure roles means the findings reflect the views of decision-makers directly responsible for data design, cloud operations and enterprise AI deployment. That makes the report a measure of operational pressure inside large organisations rather than a survey of general business sentiment.

The figures suggest AI has reached a stage where adoption alone is no longer the main benchmark. For many enterprises, the harder task is reorganising the underlying data estate so projects can move beyond pilots without hitting governance, cost or infrastructure barriers.

Among those surveyed, 75% said AI integrations had already changed storage and architecture practices, while 72% said a significant overhaul is still needed.