ISG Report: How AI is Pushing UK Data Centres to Sovereignty

A new ISG report says that UK enterprises are prioritising sovereign infrastructure as they balance AI performance with energy efficiency and operational resilience.
The report, titled ISG Provider Lens Private/Hybrid Cloud – Data Center Services, finds that sovereignty has become a defining factor in infrastructure decisions.
The findings come alongside the growing need for facilities capable of supporting highly demanding AI workloads.
AI is changing infrastructure priorities
The report says enterprises are not looking to simply maintain legacy environments.
Instead, they are modernising infrastructure into containerised, AI-ready platforms that can support data-intensive applications while reducing technical debt.
That transformation is also reshaping the data centre itself.
Providers are upgrading existing facilities and developing AI-native sites capable of supporting more than 50kW per rack, compared with the 5–10kW that was considered standard only a few years ago.
Meeting those power densities requires new approaches to cooling, power delivery and operational management.
“AI is fundamentally changing how enterprises evaluate cloud infrastructure,” says Mathew Hannon, Director and UK Public Sector Lead at ISG.
“Organisations increasingly expect providers to combine operational resilience, sustainable infrastructure and AI-enabled automation to support long-term business priorities.”
Why sovereignty matters
ISG finds that sovereignty is now among the top three considerations for many UK organisations when selecting infrastructure providers.
As enterprises deploy private AI models trained on proprietary data, they are more often choosing infrastructure that keeps both data and AI models within UK jurisdiction.
This helps organisations comply with the Data Protection and Digital Information framework while reducing concerns around cross-border data transfers.
Geopolitical uncertainty is reinforcing that trend, with enterprises placing greater emphasis on jurisdictional control alongside performance and governance.
Sustainability meets AI
The report also highlights the growing relationship between AI expansion and sustainability requirements.
The expansion of the Energy Savings Opportunity Scheme (ESOS) and wider sustainable industry policies has shifted carbon reporting from an environmental target to an operational requirement.
- ESOS is a mandatory energy assessment scheme for organisations in the UK that meet the qualification criteria.
- It is overseen by the Environment Agency, requiring audits of energy consumption across buildings, transport and processes every four years.
Data centre providers are therefore expected to demonstrate both energy efficiency and the ability to support increasingly dense GPU deployments.
Meanwhile, some organisations are reassessing the economics of public cloud.
ISG says slowing hyperscale cost efficiencies are prompting certain enterprises to move high-performance computing workloads into private infrastructure or colocation facilities to improve cost predictability.
The next phase of infrastructure
Alongside sovereign infrastructure, enterprises are adopting FinOps practices and moving from traditional disaster recovery strategies towards continuous availability across hybrid environments.
Providers are responding by embedding AIOps and MLOps into managed services while investing in infrastructure designed for AI.
“Organisations are evaluating infrastructure providers on their ability to support long-term transformation rather than isolated technology upgrades,” says Meenakshi Srivastava, Lead Analyst at ISG Provider Lens Research.
“AI-ready capabilities, sovereign operations and sustainable infrastructure have become essential considerations in complex hybrid environments.”
ISG expects these changes to continue as edge computing expands with 5G and IoT deployments and agentic IT becomes more common.
According to the report, the success of data centres now depends on delivering high-density AI capacity, sovereign hosting and sustainable infrastructure that can meet the demands of enterprise AI.


