A survey by Cloudera of over 1,500 IT architects reveals that 95% of enterprises have delayed or canceled AI projects due to infrastructure limitations.
The finding underscores a widespread need to replace legacy data architectures with flexible, hybrid systems that can support scalable and secure AI.
Organizations face mounting complexity in data governance and must fundamentally rearchitect their infrastructure to unlock AI capabilities.
What happened
A Cloudera survey of over 1,500 IT architects found that 95% of enterprises have delayed or canceled AI initiatives because their existing data infrastructure cannot support them. The report describes this challenge as "The Great AI Re-Architecture," reflecting the need to overhaul legacy systems to enable scalable and secure AI across hybrid environments.
Why it matters
Most enterprises lack the flexible, integrated data management systems required to deploy AI reliably. As organizations move toward hybrid data infrastructures, the gap between current capability and AI requirements is forcing costly delays and strategic reassessment of how data governance and infrastructure must evolve.
What to watch
The report emphasizes that enterprises must prioritize building systems capable of efficiently integrating AI while maintaining robust data governance—a foundational shift that will reshape technology spending and vendor selection in the coming months.
Over 1,500 IT architects participated in a Cloudera survey that paints a stark picture of enterprise AI adoption: 95% of organizations have delayed or canceled AI initiatives because their data infrastructure cannot support the requirements of modern AI deployment. The underlying challenge is not a shortage of AI talent or ambition, but rather the mismatch between legacy data architectures and the demands of scalable, secure, and hybrid AI.
The survey identifies what Cloudera terms "The Great AI Re-Architecture"—a fundamental shift in how enterprises must design their data infrastructure. Legacy systems were built for different workloads: batch analytics, transactional processing, or cloud-first strategies that do not account for hybrid deployments spanning on-premises, multiple cloud providers, and edge locations. AI workloads require continuous data movement, real-time feature engineering, governance that tracks data lineage across pipelines, and security policies that adapt to hybrid and multi-tenant environments. These demands expose the rigidity of systems designed years or decades ago.
The report emphasizes that overcoming this bottleneck requires not just new tools but a rethinking of data strategy. Organizations must build flexible systems that efficiently integrate AI while maintaining robust data management and governance. This means investing in platforms that can unify data across silos, provide consistent governance policies in hybrid environments, and enable rapid deployment of AI workloads without sacrificing security or compliance. The transition is complex and costly, but the survey's findings suggest it is unavoidable for enterprises serious about unlocking AI's value.
The Cloudera report captures a critical inflection point in enterprise AI adoption. While organizations recognize the strategic value of AI, the survey's 95% figure shows that infrastructure has become the binding constraint—not appetite or strategy. This gap reflects a structural problem: most enterprises built their data systems for transactional and analytical workloads, not for the continuous, low-latency data movement and governance that AI inference and training demand. The term "The Great AI Re-Architecture" signals that enterprises are not making incremental upgrades; they are facing the need to fundamentally redesign how data flows, is governed, and is secured across on-premises, cloud, and edge environments.
The shift toward hybrid data infrastructures represents both a technical and organizational challenge. Legacy architectures—whether monolithic data warehouses or siloed cloud deployments—cannot easily adapt to AI's requirements for real-time feature engineering, model retraining pipelines, and data lineage tracking. Enterprises must now invest in platforms that unify governance, integrate disparate data sources, and provide the flexibility to deploy AI workloads wherever latency, cost, or sovereignty constraints demand. This rearchitecture is not optional; the survey indicates it is a prerequisite for moving beyond the current impasse of delayed and canceled projects.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
As AI systems scale beyond single servers to rack-level, multi-rack, and pod-level deployments, the performanc…
Steve Eisman, the investor known for shorting subprime mortgages, told CNBC that OpenAI and Anthropic represen…
Space Exploration Technologies Corp (SpaceX) has entered the top five of Robinhood's Investor Index (ranked No
IBM announced a partnership with OpenAI and a multi-year US$240 million agreement with Together AI to deploy N…
OpenAI's annualized revenue run rate has passed $40 billion, roughly doubling from late 2025, according to Blo…
OpenAI's annualized revenue is set to top $40 billion, according to Bloomberg, positioning the ChatGPT maker f…
The AI news that matters, in one minute each morning.
Sign up free