Summary

• Leaseweb's survey of 800 executives found that 89% did not expect to move critical workloads from their largest cloud or infrastructure provider within three weeks.
• Some 79% expect AI to increase dependence on a small number of providers over the next three years, rising to 83% among UK respondents.
• Nearly three-quarters would pay at least 11% more for greater digital independence, although the responses measure preferences and expectations rather than completed migrations.

Nearly nine in ten businesses do not expect to move their critical workloads away from their largest cloud or infrastructure provider within three weeks, according to research showing how confidence in data control can exceed an organisation’s practical ability to change suppliers.

The Digital Independence Survey 2026, commissioned by Leaseweb, found that 76% of executives believed they had complete or substantial control over moving their data to another provider or environment. Yet only 11% expected their organisation to complete a migration of business-critical workloads within three weeks.

More than half, 55%, anticipated that moving would take longer than three months or was not currently possible. The survey covered 800 C-suite executives at organisations with between 500 and 5,000 employees, with 200 respondents each in the UK, Germany, the Netherlands, and the United States.

Conducted by Censuswide between 22 and 30 September, the research captures executives’ expectations about infrastructure flexibility rather than the results of independently observed migration exercises. Leaseweb, which sells cloud and infrastructure services, also has a commercial interest in encouraging organisations to reconsider their supplier dependencies.

Data access is not workload portability

The contrast between perceived control and migration times reflects a distinction between being able to retrieve information and being able to operate a complete technology environment elsewhere.

A business may own its data, have contractual rights to export it, and maintain copies outside its primary provider while depending on that provider’s databases, identity services, networking tools, and operational procedures. Transferring information does not automatically recreate the applications that process it.

Migration can involve modifying software, replacing managed services, testing security arrangements, and checking that operational processes continue to work. Applications that depend on proprietary interfaces may require more extensive engineering than systems built around widely supported technologies.

Availability is another constraint. A critical application cannot always be switched off while its replacement is prepared, particularly where customer transactions, production systems, or other services must continue operating. Moving a workload can therefore require parallel environments and a carefully managed transition.

Respondents in the Netherlands reported the longest expected migration times. Some 65% anticipated a move taking more than three months, compared with 59% in the UK, 55% in the US, and 40% in Germany. Dutch executives estimated an average migration period of 4.7 months.

Those differences should not be treated as evidence that infrastructure in one country is inherently less portable. The figures reflect respondents’ assessments and may be influenced by the complexity of their workloads, the services they use, and the assumptions made about the scale of a migration.

AI introduces additional switching costs

The survey also found that 79% of executives expect artificial intelligence to increase dependence on a small number of technology providers during the next three years. Expectations were highest in the US at 89%, followed by the UK at 83%, Germany at 76%, and the Netherlands at 67%.

AI deployments can add new dependencies to an existing cloud environment. Applications may rely on a particular model provider, hosted vector databases, orchestration services, specialised processors, or identity and monitoring arrangements integrated into the cloud platform.

Changing infrastructure suppliers can consequently require changes to the application itself. A different model may produce responses with different characteristics, while a substitute managed service may require changes to security controls, data processing arrangements, or the way information is retrieved.

Specialised computing capacity presents a further limitation. Organisations using high-performance processors for AI training or inference may find that equivalent capacity is not immediately available from an alternative provider at the price, location, or scale they require.

Among the possible barriers to digital independence examined in the survey, dependence on AI technology and providers was the most frequently selected overall, cited by 28%. German respondents instead identified legacy-system complexity most often, at 30%, showing that long-established infrastructure can remain a more immediate obstacle than newer AI services.

The findings do not mean organisations should avoid managed AI or cloud services. External platforms can reduce the cost and technical effort involved in deploying sophisticated applications, although the financial benefits need to be considered alongside the ability to change arrangements when circumstances require it.

Whether a dependency is acceptable depends partly on its commercial terms and operational consequences. A difficult migration may be manageable for a non-critical application, but a similar constraint affecting essential services can limit an organisation’s options during a prolonged outage or supplier dispute.

Independence carries a stated price

Almost all respondents indicated some willingness to pay more for greater control, reduced provider dependence, or more freedom to migrate. Some 73% were prepared to pay at least 11% more, with the average stated premium reaching approximately 16%.

The Netherlands recorded the strongest willingness to accept a premium of at least 11%, at 81%, while 26% of Dutch respondents were prepared to pay at least 21% more. Germany followed, with 79% willing to accept an increase of at least 11%.

These figures represent stated preferences, not expenditure commitments. A business may indicate that greater independence is valuable while choosing a less portable service if the immediate cost, performance, or functionality is more attractive.

Actual purchasing decisions also depend on the price of operating alternative infrastructure and the investment required to make existing applications portable. Maintaining several complete environments can increase expenditure, particularly when workloads require substantial computing and storage resources.

Geopolitical concerns are already influencing supplier decisions. Leaseweb found that 78% of respondents had taken action to reduce dependence on a particular provider in response to international political or trade pressures, including 87% in the UK and 91% in the US. Another 45% overall planned further action.

These measures could include supplier diversification or changes to infrastructure design, although the survey’s headline results do not establish how much those actions have reduced actual switching costs.

Compliance creates another constraint

Only 28% of respondents identified regulatory and legal compliance as a non-negotiable cloud procurement requirement, with the same rounded proportion selecting security and data protection. Meanwhile, 67% indicated that they would accept at least a minor compromise on compliance.

Those responses describe purchasing preferences in a survey; they do not demonstrate that participating organisations are failing to meet legal obligations. Businesses remain responsible for complying with applicable requirements regardless of the trade-offs they would consider when comparing suppliers.

National priorities also differed. UK respondents most frequently selected provider reputation and trust as a non-negotiable consideration, at 34%, while German respondents placed security and data protection first, at 30%. In the Netherlands, regulatory compliance and geopolitical exposure were tied at 27%.

The variation suggests that decisions about infrastructure independence involve several competing priorities, rather than a single requirement for locally based services. A provider’s jurisdiction, contractual obligations, technical architecture, and operational arrangements can each influence the degree of control a customer retains.

Migration capability needs operational evidence

Testing portability requires an understanding of which services support critical workloads and how those services could be reproduced or replaced. Organisations can examine export formats, application dependencies, network connectivity, identity controls, and the engineering work needed to establish an alternative environment.

Recovery exercises can also expose assumptions that would otherwise remain untested. A business may discover that its data can be transferred quickly but that rebuilding an application takes considerably longer because the software relies on platform-specific databases or configuration tools.

Open standards and portable deployment methods can reduce some dependencies, although they cannot guarantee straightforward migration where applications rely on specialised services or large quantities of data.

Lex Boost, co-CEO at Leaseweb, said: “Digital independence comes down to having real choices, and our research shows many businesses have fewer than they think.”

The survey offers evidence of how executives perceive their options rather than proof that individual cloud platforms prevent customers from leaving. Its central finding remains a substantial difference between reported confidence in moving data and the time respondents believe would be needed to move the applications their organisations depend on.