
A moving target: Data sovereignty in the age of AI

As organisations take steps to secure their digital futures – amid rising geopolitical tensions and growing dependence on a small number of dominant infrastructure providers – they face hard questions. How to balance security with agility in a globalised economy? How to turn sovereignty from a compliance burden into a source of trust and competitive advantage?
Those questions shaped the agenda for Data Sovereignty in the Digital Age, peer-to-peer discussions held in London and Singapore in November and December 2025. They built on survey research conducted by University of Technology Sydney and commissioned by Pure Storage, which found near- universal concern among data leaders about service disruption, exposure to foreign legal regimes and the erosion of customer trust, but far less consensus on how organisations should respond in practice. Against this backdrop, leaders from healthcare, financial services, technology and the public sector convened to compare strategies.
A matter of control
Data sovereignty is not a new concern for leaders. Organisations in the private and public sectors alike have long faced requirements to process certain sensitive data domestically. But participants agreed that the scope and complexity of the challenge has expanded dramatically.
Sovereignty now encompasses not just where data sits and is processed (its “residency”), but also its legal status. “Who controls the data, who has access to the data, the ownership of the company that is managing the data – those are all additional dimensions that are more important than physical location,” observed one participant, the CTO of a global IT services company.

Who controls the data, who has access to the data, the ownership of the company that is managing the data – those are all additional dimensions that are more important than physical location.
CTO, Global IT services

This creates two distinct challenges. The first is managing risk: the concentration of critical infrastructure in a handful of providers; exposure to foreign legal jurisdictions; and the flow of sensitive data into externally controlled AI systems.
The second is responding effectively: navigating complex regulation, without sacrificing the agility and access to global technology that competitiveness requires.
At the heart of both these challenges is the question of trust and customer confidence that data is handled responsibly.
Concentration and its drawbacks
In Europe, policymakers are actively pushing for more sovereign digital infrastructure and reduced reliance on US hyperscalers, while seeking to encourage domestic cloud and AI providers.
One participant from a UK high-street bank noted how her organisation was caught up in media scrutiny following a high-profile cloud outage, despite not being a direct customer of the provider, but because a critical supplier relied upon it.
“We had a dependency, it was in our supply chain, and it was incredibly embarrassing,” she said. This incident has prompted a strategic shift toward more on-premises data infrastructure.
Commentators framed the outage as a wake-up call, citing not just the risks of service disruption, but also the systemic risk of concentrating critical workloads in infrastructure governed by foreign law.
Many agreed that dependence on any single provider is increasingly hard to justify, despite the costs and complexity of diversification.
Many agreed that dependence on any single provider is increasingly hard to justify, despite the costs and complexity of diversification. This echoes the Pure Storage survey findings, in which global data leaders cited service disruption as the top data sovereignty risk; in the study, fully 100% of respondents said data sovereignty risks, including the possibility of service disruption, have caused organisations to consider where their data is located.
However, few believed it is realistic to abandon large cloud providers altogether. Even among organisations handling more sensitive data – in healthcare and financial services – there was recognition that hyperscalers often offer capabilities that individual organisations cannot match.
The question, they argued, is not whether to use hyperscalers, but how to govern their use, limit concentration risk and retain operational control. “There's nothing wrong with [any particular provider] as long as you've got something you can fail over to, especially when you run a business that is critical to people’s lives and people’s businesses,” the UK bank leader insisted.
Others cautioned that failover capability addresses only part of the risk. Several participants noted that foreign legal frameworks can enable government access to data held by their domestic providers, regardless of where it is physically stored. One pointed to a recent case in which a major platform was fined not for storing data improperly, but because staff in another jurisdiction had ongoing access to it. This serves as a reminder that sovereignty concerns extend beyond physical location to questions of access and control.
These concerns prompted support among participants in both London and Singapore for hybrid and multi-cloud strategies, alongside sovereign or private-cloud fallback capacity. Although some welcomed the shift towards greater control, others warned that pushing sovereignty too far risks eroding competitiveness. Additional layers of compliance, bespoke infrastructure and restricted access to global platforms can slow deployment and raise costs.
The discussion exposed a clear contrast between London and Singapore. Participants in London put more emphasis on control and risk mitigation, whereas those in Singapore emphasised interoperability, portability and competitive agility. Participants based in Asia were more inclined to see sovereignty as strategic positioning, rather than as insulation from foreign providers.
The costs of compliance
Strong policy responses to these risks were widely seen as necessary and good. But those responses are creating challenges of their own.
Firms face overlapping national requirements, sector-specific rules and evolving regional initiatives. In London, participants expressed a sense of regulatory fatigue, noting that inconsistencies persist even within the European single market, complicating compliance and undermining efficiency, but expressed optimism for convergence. In Singapore, in contrast, participants anticipated only continued divergence. “Regulators today are not co-ordinating and framing the laws,” said one participant. “What works for one country might not work for another.”
Several argued that the answer lies less in harmonising regulation than in harmonising standards and operating models, particularly for cloud and AI interoperability. Open-source technologies were suggested as one possible path, reducing dependency on foreign vendors, while improving transparency and trust.
In London, participants expressed a sense of regulatory fatigue, noting that inconsistencies persist even within the European single market, complicating compliance and undermining efficiency.
Beneath the legal and technical debates sits the question of trust. Pure Storage’s survey findings highlight that foreign access to sensitive data is seen as a major risk to customer and citizen trust, particularly in sectors exposed to geopolitical tension. Service disruption compounds financial and reputational damage. More than nine in ten respondents (92%) said that failure to deal adequately with data sovereignty concerns can lead to reputational damage, and 85% of respondents said it would result in loss of customer trust.
Roundtable participants differed on how much everyday users truly care about sovereignty as such. Some argued that convenience and free services still trump concerns over privacy and security; others suggested that the next generation is more privacy-conscious than ever. All agreed that trust must be earned. “Our customers tell us, as long as you use my data to help me in my daily life – and I know I can trust you – I’m OK,” one London participant said, while acknowledging the heavy responsibility this places on firms.

Agile solutions for a fast-changing environment
A further complication is the growing role of AI as the primary consumer of sensitive data. As organisations adopt foundation models and generative AI, data is no longer simply stored or analysed locally; it is increasingly fed into large, externally developed models. Some participants expressed concern over how these are owned and operated by US-based firms and function as black boxes, with limited transparency. For many others, however, the central concern was less who owned these models than the practical reality that, in their current form, using them often requires data to leave national borders for inference before results are returned, thereby reintroducing jurisdictional and governance risks even where storage remains local.
This has prompted interest in open-source models, and in sovereign or regional AI initiatives. But building competitive models, compute capacity and talent pipelines remains costly and problematic, especially for smaller markets.
As one senior healthcare leader observed, any technical control put in place today is likely to be outdated by the time it is fully implemented.
In this environment, shaped by fast-moving technology and shifting geopolitics, the dominance of established cloud providers is facing new pressures. As organisations weigh sovereignty risks more seriously, many are actively exploring alternatives in sovereign cloud arrangements, on-premises infrastructure, open-source platforms and multi-cloud strategies designed to reduce dependence on any single provider. Hybrid and agile approaches are increasingly seen as the only viable path forward, combining strong local governance with selective use of external capabilities, continuous reassessment of risk and acceptance that sovereignty in the age of AI is not a fixed state, but a moving target.
