What is data residency? Explained with picking a warehouse's country
Data residency means choosing which country or region your data is stored and worked on in. Explained with a storage company that has warehouses in many countries, and lets you pick one.
- Author
Umair Mateen- Published
- Reading time
- 5 min
- Category
- Engineering
This week's news kept mentioning data residency: Anthropic now lets companies run Claude Managed Agents on their own servers, and Google lists data residency support for Gemini 4 Hydrogen on Google Cloud. Here is what the phrase means.
In one sentence: Data residency means choosing, and being able to show, which country or region your data is kept in and worked on, so it stays where your laws, contracts and customers expect it to be.
The everyday picture
Picture your business renting space from a storage company. The company has warehouses in many countries. When you sign up, you pick which country your boxes will live in.
Each warehouse has two parts. There are the shelves, where boxes sit and wait. And there is a work room, where staff open boxes and do jobs with what is inside.
Data residency is the promise about where both of those are. Your boxes sit on shelves in the country you chose. And when a box is opened to work on, that happens in a work room inside the same borders, not in a building abroad.
Your request enters through the EU door, is worked on in an EU work room and stored on EU shelves. Data leaves only by courier, with safeguards.
How it works, step by step
- You pick a region (the warehouse country). Cloud services let you choose a location such as the EU or the US when you set up an account or a project. Microsoft calls this your default geography.
- Your app sends work through a regional endpoint (the front door of that warehouse). An endpoint is the web address your software talks to. Google, for example, offers separate
euandusendpoints for its AI models. - The work happens in that region (the work room). Experts call this "processing" or "data in use". For an AI model, it is the moment the model reads your prompt and writes an answer. A regional endpoint keeps that work inside the chosen borders.
- Data is stored in that region (the shelves). Experts call this "data at rest": files, databases, saved chat history. Providers commit to keeping it in the location you chose.
- Anything leaving the region needs rules (a courier taking a box abroad). Under EU rules, personal data can leave Europe only with extra safeguards that travel with it.
Try it · Myth or fact
Where does your data really live: myth or fact?
Decide for yourself first, then tap each statement to check it.
0 of 4 checked
Where the picture stops
- A warehouse is one building. A cloud "region" is usually several data centres, and a "multi-region" such as the EU can spread across many countries inside it.
- Shelves and work room are always in the same building in our picture. In the cloud they can be split. Google's documentation says a global endpoint may process requests in any of its locations worldwide, even if your stored data stays in one region.
- Boxes are physical, so you can see where they are. Data is copied easily, so residency depends on written promises, settings and audits, not on what you can see.
Why it matters for your business
- Winning contracts: banks, hospitals and public bodies often write location rules into contracts. Being able to say exactly where data sits and is processed can decide whether you can bid at all.
- Lower legal risk: the European Commission's guidance says personal data sent outside the EU/EEA needs safeguards such as an adequacy decision or standard contractual clauses. Keeping data in-region avoids many of those questions.
- Using AI safely: many AI tools send each request to a model somewhere else. Anthropic's self-hosted Managed Agents keep agent history in your own database, but each model call still sends the conversation so far to the model provider you choose. Residency has to cover that step too.
- Trade-offs: Google notes that its global endpoint offers higher availability than a single region. Strict residency can mean fewer features or less spare capacity, so choose it where it is really needed.
Try it · What would you do?
Your client needs EU-only data. Which setup do you pick?
A client contract says their customer data must stay in the EU. Your team wants an AI tool to summarise those customers' emails. Stored data is already in an EU region.
Pick what you would do. You can change your mind.
An example
Imagine a 30-person recruitment agency in Germany that wants an AI assistant to summarise CVs. A client contract says candidate data must stay in the EU. The agency stores CVs in an EU cloud region, which covers the shelves. Then it checks the AI step: the tool it likes calls a model through a global endpoint, so the work room could be anywhere. It switches to a setup that uses the provider's EU endpoint, writes the choice into its records, and can now answer the client's question with a clear yes.
Frequently asked questions
Does GDPR say data must be stored in the EU?
Not on its own. The European Commission explains that personal data can leave the EEA when safeguards travel with it, such as an adequacy decision or standard contractual clauses. Some contracts, sector rules or national laws do require local storage, so check the ones that apply to you.
What is the difference between data at rest and data in processing?
Data at rest is stored: files, databases, saved history. Data in processing is being used right now, for example when an AI model reads your prompt. A provider can promise one location for storage and a different, wider one for processing, so ask about both.
Does picking an EU region guarantee my AI requests stay in the EU?
Not always. It depends on the endpoint and the model. Google's documentation says its EU multi-region endpoint keeps processing within EU member states, while its global endpoint gives no residency guarantee.
Is data residency the same as encryption?
No. Residency is about where data is. Encryption is about whether someone who gets the data can read it. Regulated businesses usually need both.
Related explainers
The DevDrift builds cloud platforms and AI features with region choices and approvals built in: see how.
Sources
Could you explain this to a colleague now?
Mark it and posts that mention it will show a tick. Saved in this browser only.