Does model inference have to be in-region too?

If the requirement covers processing and not only storage, yes. A request sent to a model outside the region has left it.

How is this different from zero data retention?

Residency constrains where data goes. Zero data retention constrains how long it is kept. A deployment can satisfy one and not the other.

Data Residency

Data residency is the requirement that a company’s data is stored and processed in a particular country or region, and not moved outside it.

Residency is usually a legal or contractual obligation rather than a technical preference: a regulator, a customer contract or an internal policy names where data may live. For AI systems the requirement reaches further than storage, because a request that leaves the region to reach a model has moved the data even if nothing was written down outside it.

That is why residency and model choice are connected decisions. A platform can hold its database in-region and still fail a residency requirement if inference happens elsewhere. The question worth asking a vendor is not where the data is stored, but where every hop in a request is processed.

Frequently Asked Questions

Related but not identical. Residency is about where data physically sits and is processed. Sovereignty is about whose laws can compel access to it, which can differ even when the servers are local.