IBM study finds executives struggle with AI sovereignty

IBM has published a global study on AI sovereignty among large organisations, finding that most surveyed executives struggle to control or fully understand their AI dependencies.

Based on responses from 1,000 senior executives, the research found that 71% would find it difficult to switch their primary AI vendor or model. It also found that 68% said meeting data residency and sovereignty requirements across different geographies was challenging.

The findings point to a growing gap between the spread of AI in core business operations and the governance structures needed to manage it. The study found that 91% of respondents did not fully understand their organisation's dependencies across AI vendors, models and infrastructure.

That lack of visibility appears to have direct operational consequences. Leaders reported an average of six AI-related disruptions over the past two years, while 81% said a seven-day vendor outage would cause severe or critical disruption to operations.

Control gap

The study examined how companies structure control across data, models, infrastructure and applications, linking those decisions to resilience and operating performance. Only 7% of surveyed organisations had reached the most advanced level of AI control.

Those organisations reported lower AI downtime and stronger financial protection from disruption. Businesses with the most advanced AI control arrangements protected 55% more operating profit from AI-driven disruptions than their peers, according to the analysis.

Ana Paula Assis, Senior Vice President and Chair, EMEA and APAC, IBM, framed the issue as a business risk rather than a narrow technical question.

"AI has introduced new forms of dependency that evolve faster than traditional governance, procurement, or technology cycles were designed to handle. That is why AI sovereignty has become one of the most defining leadership issues of this moment. The stakes are no longer technical; they are economic. Any loss of control can translate directly into margin pressure, compliance exposure, or outright business disruption," said Assis.

The survey suggests many companies are willing to pay more for greater flexibility. Nearly three-quarters of respondents, or 72%, said they would accept a 20% cost increase in their AI vendor arrangements if it improved strategic flexibility.

Multi-vendor reality

Most organisations in the survey, 73%, described their AI environments as intentionally multi-vendor. Yet the results suggest this diversity often reflects operational fragmentation rather than a deliberate sourcing strategy.

Independent decisions by business units and geographic requirements were each cited by 69% of respondents as leading reasons for using multiple vendors. Legacy complexity, including systems and structures shaped by mergers, acquisitions and older technology choices, was cited by 57%.

The study also highlighted the range of changes that can disrupt AI operations even when systems remain available. Respondents pointed to price increases, usage restrictions, model deprecations and performance degradation across the AI supplier landscape.

That mix of external dependency and internal complexity has become harder to manage as organisations embed AI more deeply into mainstream operations. The report argues that the ability to adapt data, models and infrastructure as conditions change is now central to business continuity.

Broad sample

The research was conducted with Oxford Economics between February and April 2026 and covered 16 countries and 17 industries. Participants were senior executives responsible for AI, data, technology and related enterprise functions.

IBM's analysis segmented organisations by their level of control across the AI stack and compared those profiles with resilience, performance and operating economics. It found a widening divide between organisations building adaptable AI systems and those constrained by supplier lock-in and limited oversight.

Among the clearest signs of that divide was the finding that more than two-thirds of executives face difficulty meeting cross-border data requirements, while a similar share said changing their main AI supplier would be hard.

The survey's central message is that many organisations have expanded their use of AI faster than their ability to govern the underlying commercial and technical dependencies, leaving core operations vulnerable to disruption.