Where Your Data Should Live: Sovereign Cloud Without the Panic
For about a decade, the answer to "where should this run?" was easy: a hyperscaler region near your customers, with a recovery copy somewhere sensible. That answer has not become wrong, but it no longer satisfies the board. Directors now ask who could be compelled to hand over the data, what happens if a supplier relationship turns politically awkward, and how long it would take to leave if the business had to.
Gartner has given the shift a name. In its Top Strategic Technology Trends for 2026, published in October 2025, it listed geopatriation: moving company data and applications out of global public clouds into local options such as sovereign clouds, regional providers or an organisation's own data centres, because of perceived geopolitical risk. Gartner forecast that by 2030 more than 75% of European and Middle Eastern enterprises will geopatriate their virtual workloads into solutions designed to reduce geopolitical risk, up from less than 5% in 2025.
That is a striking forecast, and it is still a forecast. For most leadership teams the useful question is not whether to follow a trend but which workloads, if any, genuinely belong somewhere else.
Residency, localisation and sovereignty are three different problems
The terms get used interchangeably, which is how organisations solve the wrong one.
- Data residency is about where data is physically stored and processed. It is often a contractual demand as much as a legal one.
- Data localisation is a legal mandate that certain data stays in a country. India's payment rules are the textbook case. The Reserve Bank of India's circular of 6 April 2018 requires payment system providers to store the entire data relating to the payment systems they operate only in India. According to the RBI's own FAQs, processing can happen abroad, but the data must then be deleted overseas and brought back within one business day or 24 hours, whichever is earlier.
- Sovereignty goes further. It asks who operates the infrastructure, which governments and courts can reach it, and whether you could keep running if a foreign supplier withdrew or was instructed to.
A workload can pass a residency test and fail a sovereignty one: data sitting in Frankfurt on a platform administered from another jurisdiction, say. Plenty of workloads have no such requirement at all, and treating them as though they do is expensive.
What the rules actually say
Much of the pressure is perception rather than law. The GDPR does not require personal data to stay in the EU. Chapter V restricts transfers to countries outside the EEA and requires a lawful route, such as an adequacy decision or standard contractual clauses. For transfers to the United States, the EU-US Data Privacy Framework remains available to certified organisations. The EU General Court dismissed a challenge to it on 3 September 2025, but the claimant lodged an appeal with the Court of Justice in October 2025, so its long-term footing is not settled.
India takes a negative list approach. Under the Digital Personal Data Protection Act, transfers abroad are permitted unless the central government restricts a destination, and stricter sector rules such as the RBI's continue to apply. The DPDP Rules were notified in November 2025, with most obligations, including the cross-border rule, phasing in over 18 months. Our note on what the DPDP Rules mean for Indian businesses covers the detail.
One EU law pushes in the opposite direction, by making exits cheaper. Under the EU Data Act, cloud providers may currently charge for switching only at cost, and from 12 January 2027 they are generally no longer allowed to charge for switching at all, including data egress for the switch. Egress charges for running services in parallel across providers remain permitted.
None of this is static, and lawyers read parts of it differently. Treat what follows as a way of organising the decision, not as legal advice, and confirm the specifics with counsel.
The supply side has moved quickly
Gartner forecast in February 2026 that worldwide spending on sovereign cloud infrastructure as a service would reach $80 billion in 2026, up 35.6% on the previous year, and that this would shift 20% of existing workloads from global public clouds to local providers. AWS opened its European Sovereign Cloud in January 2026, with a first region in Brandenburg run separately from its other regions. Public buyers now purchase on explicit sovereignty criteria: in April 2026 the European Commission awarded four contracts, worth up to EUR 180 million over six years, assessed against its own Cloud Sovereignty Framework and its eight sovereignty objectives.
So real options exist now. What they rarely offer is feature parity with the main public regions on day one. Ask which managed services, AI models and support arrangements actually sit inside the sovereign boundary.
Sovereignty is a property of specific workloads, not of a whole estate. Price it workload by workload.
A decision framework that starts with the data
We would work through five steps, in this order, and resist the urge to start at step five.
1. Classify data and workloads. Tag what each system holds (personal, payment, health, defence-related, commercially sensitive, public) and how critical it is to operations. Most estates have a small sensitive core and a long tail of ordinary systems. A sound data strategy should already hold this inventory. 2. Map legal and contractual obligations. List the laws that bite on each class, then the contractual promises sales teams have made to customers about location and access. The contracts are often stricter than the law. 3. Assess concentration and geopolitical risk. Ask what happens to each critical workload if one provider, one region or one jurisdiction becomes unavailable or hostile. Score it as part of your business continuity planning: likelihood, impact and time to recover. 4. Weigh cost and capability. Sovereign and local options can mean fewer services, a smaller partner ecosystem and migration effort that stalls other work. Model the full cost, including the people time, as you would for any cloud cost decision. 5. Choose the lightest control that works. For some workloads that means moving. For many it means customer-held encryption keys, tighter administrative access, a documented exit plan, or simply keeping data in a regional deployment you already have.
Avoid the knee-jerk repatriation
Picture a mid-sized insurer that reads the headlines and decides to pull everything back into a local data centre. Eighteen months later it has spent its transformation budget on migration, lost access to managed analytics and AI services its competitors still use, and swapped dependence on one large supplier for dependence on a smaller one.
A few principles help avoid that outcome:
- Separate the sensitive core from the rest. Move or harden what genuinely needs it; leave the long tail where it runs well.
- Build the exit before you need it. Portable data formats, infrastructure defined as code, and contracts with clear exit terms give you much of the protection of moving, usually for far less effort. The Data Act's switching rules make this more realistic for EU customers from 2027.
- Revisit annually. Rules, court cases and offerings are all moving. A placement decision made in 2026 should be reviewed, not framed.
Common questions
What is geopatriation?
Gartner uses the term for moving data and applications out of global public clouds into local options, such as sovereign clouds, regional providers or an organisation's own data centres, because of perceived geopolitical risk. It named geopatriation as one of its top strategic technology trends for 2026. For most organisations the sensible scope is a set of selected workloads rather than the entire estate.
Does GDPR require data to be stored in the EU?
No. The GDPR restricts transfers of personal data outside the EEA and requires a lawful transfer route, such as an adequacy decision or standard contractual clauses. Sector rules, national laws or customer contracts may still impose location requirements.
Is a sovereign cloud more expensive than a standard public cloud region?
It can be, but the bigger trade-off is often capability: fewer managed services, newer AI features arriving later and a smaller partner ecosystem. Compare total cost, including migration effort and what you would lose, workload by workload rather than across the whole estate.
Where Ganexa can help
Ganexa helps leadership teams decide where data and workloads should live: classifying the estate, mapping obligations with your legal advisers, scoring concentration risk and building exit plans that hold up. Our modern data strategy work covers the inventory and governance that make these decisions defensible. If sovereignty questions are reaching your board, book a free consultation.