Sovereignty boosts enterprise AI returns, study finds
Sat, 16th May 2026 (Today)
MIT Technology Review Insights has published research with EnterpriseDB linking AI and data sovereignty to stronger returns from enterprise AI projects. Organisations with the strongest commitment to sovereignty report five times the return on investment of their peers.
The findings are based on a global survey of more than 2,000 senior executives across 13 countries, more than 15,000 simulations covering over 500 variables, and interviews with specialists in AI, cyber security, data governance and infrastructure. The analysis found a 0.93 correlation between commitment to sovereignty and positive AI outcomes, describing it as the strongest single driver identified in the study.
More than half of surveyed organisations already have AI agents in production. The research also found that 95% plan to establish their own AI and data platforms within three years.
The report argues that the spread of autonomous systems is shifting the enterprise AI debate away from simple adoption and towards control over data, infrastructure, models and governance. In this view, sovereignty is no longer just a compliance issue, but a core operational requirement for companies that want AI systems to make decisions on live business data.
The control question
Security and resilience emerged as the main reason for sovereignty efforts, cited by 85% of respondents. Data localisation followed at 74%, while ownership and control were cited by 72%.
The research points to hybrid environments as the leading operating model, with companies trying to balance access to AI tools and innovation with tighter control over where data sits, how models are used and which rules apply in different jurisdictions.
Laurel Ruma, Global Director of Custom Content at MIT Technology Review Insights, said the findings reflect a broader reassessment of how much control companies want over AI systems and the structures around them.
"As enterprises continue evaluating AI adoption strategies, the report explores how sovereignty may involve a shift from simply consuming AI services toward developing a clearer understanding of the architecture and systems that govern them. It also reflects how many organizations may be reassessing questions of control and long-term AI strategy as autonomous systems continue to evolve," Ruma said.
The report draws a distinction between control over the underlying information and control over the systems acting on that information. That distinction becomes more important as AI agents take on more tasks with limited human intervention.
"Sovereignty defines which agents can touch the data, in which region, under which policies, and how all of that is monitored and audited. Rather than being a brake on agentic AI, sovereignty sets the safe operating boundaries that allow organizations to scale with confidence," Pratt said.
Pratt added: "Sovereign data governs the information; sovereign AI governs the systems that act on that information."
Economic pressure
The financial argument is central to the report's conclusions. It says companies that exert tighter control over their AI and data environments are seeing materially better outcomes from generative and agentic AI initiatives than organisations that rely more heavily on external systems without the same level of governance.
Kevin Dallas, Chief Executive Officer at EDB, framed the issue around intellectual property and competitive risk.
"Data is really a new currency; it's the IP for many companies. The big concern is, if you're deploying an AI-infused application with a cloud-based large language model, are you losing your IP? Are you losing your competitive position?" Dallas said.
The report also argues that conventional governance approaches built for relatively static data estates are under strain. In environments where models, agents and multi-agent workflows are updated frequently, sovereignty must be treated as an ongoing discipline rather than a one-off policy exercise.
Another issue it raises is a leadership gap, with many senior executives still lagging on AI itself as well as the policy, ownership and governance questions tied to it.
"The C-suite is quite behind in terms of understanding AI in general, let alone the nuances of data sovereignty and ownership and policy and driving those policy guidelines. Sovereignty is not isolation. It comes from having discipline of control, evidence, and choices," Hoque said.
Operational models
Beyond the survey data, the report examines the technology choices companies are using to build what it calls sovereign AI environments. These include confidential computing, edge AI, post-quantum cryptography, converged data platforms and open-source infrastructure, including PostgreSQL.
It also outlines a short-term operating framework for companies that want to review their exposure. The framework includes mapping sensitive data, setting governance guardrails, creating secure environments for AI testing and rolling out sovereign AI projects in phases tied to practical use cases.
The overall message is that as AI agents move deeper into day-to-day operations, businesses are placing greater weight on where systems run, who controls them and how their actions are monitored. For organisations already deploying autonomous tools into production, those questions are moving from the margins of policy to the centre of operating strategy.
"We're in an AI-first, sovereign-first world. Build your own sovereign data and AI factories. Lead from the front. Disrupt or be disrupted," Dallas said.