Enterprise AI adoption is running into structural limits as organisations struggle to reconcile the data mobility that AI systems require with tightening privacy regulations and sovereignty mandates, according to new research published by NTT DATA.

The 2026 Global AI Report, which surveyed nearly 5,000 senior decision-makers across more than 30 markets and five regions, reveals a significant disconnect between awareness and action. More than 95 per cent of respondents said private and sovereign AI are important to their organisations, yet only 29 per cent are prioritising sovereign AI in a concrete, near-term way.

For years, enterprise architecture has been designed to move data across systems, clouds, and borders with speed and efficiency. That model is now showing its limits.

AI systems depend on continuous access to and movement of data. But sensitive data must be protected, workloads must run within defined jurisdictions, and models must operate under tighter governance controls. The result, the report argues, is that data jurisdiction has become a core architectural constraint — not a secondary compliance consideration.

“The constraint is no longer model performance alone,” the report states. Enterprises that built their infrastructure for centralised, borderless data flows are now finding those foundations misaligned with what modern enterprise AI actually requires.

Also Read: China builds robot armies while the West chases robot brains

Leaders and laggards are diverging

The research identifies a measurable split between organisations redesigning their AI infrastructure proactively and those layering AI onto environments that were never built to support it.

Roughly 35 per cent of Chief AI Officers identify building and managing complex AI models in private or sovereign environments as their primary barrier to adoption. Nearly 60 per cent of AI leaders cite cross-border data restrictions as a major challenge, and only 38 per cent report high confidence in their cloud security posture — a foundational requirement for both private and sovereign AI.

Abhijit Dubey, CEO and Chief AI Officer at NTT DATA, said organisations that are succeeding are treating architecture, infrastructure and governance as strategic requirements rather than compliance obligations. “They are building the operating foundation for AI that can perform across markets, jurisdictions and business environments,” he said.

The report draws a distinction between two related but separate concepts. Private AI focuses on protecting sensitive enterprise data, controlling access and limiting exposure. Sovereign AI addresses whether AI systems, data and operating environments meet national, regional or jurisdictional regulatory requirements.

Both are increasingly intertwined. More than half of organisations surveyed cite integration complexity as their top challenge, underlining that greater control does not mean greater simplicity. In practice, private and sovereign AI rely on tightly coordinated ecosystems of partners, platforms and providers.

The report’s central warning is straightforward: organisations that delay redesigning their enterprise AI architecture risk falling behind in regulated, distributed and data-sensitive markets. Those moving decisively — aligning infrastructure, governance and operating models early — are better positioned to scale AI from pilot programmes into durable, production-grade deployments.

The NTT DATA research is part of a broader global series examining strategies that differentiate AI leaders from the wider market.

The post Enterprise AI hits barriers as privacy, sovereignty demands grow appeared first on e27.