Regional Endpoints

Large online services don’t run on a single computer. Behind them stand huge buildings full of machines, so-called data centers. These buildings are spread across the world, for instance in Ireland, in Frankfurt, in Virginia, or in Singapore. A regional endpoint is the address through which a program specifically addresses one of these data centers. Instead of simply calling “the service,” you call the version at a particular location. One example would be an address containing the word “europe-west4”: it leads to machines in the Netherlands.

Why the location of the machines matters

The first reason is legal. In the EU, the General Data Protection Regulation, or GDPR, applies. It prescribes how personal data may be handled. For many companies, it’s much simpler if this data never leaves EU territory at all. A regional endpoint in Frankfurt is the technical implementation of that. Hospitals, banks, and government agencies often explicitly demand this in contracts.

The second reason is speed. Data travels through fiber-optic cables, but not infinitely fast. From Europe to California and back can easily take 150 milliseconds, purely due to distance. With a single search query, you barely notice this. With a voice assistant that is continuously communicating back and forth, the delay feels disruptive. A closer endpoint saves this time.

The third reason is fault tolerance. If a data center fails due to a power outage or software error, this initially only affects one region. Whoever distributes their service across multiple regions still remains reachable. That’s exactly why, in outage reports from major providers, the affected region is always named.

How the assignment to a region comes about

Technically, an endpoint is simply an internet address that a program addresses. With a global endpoint, the provider itself decides which data center handles the request. Usually it chooses whichever is currently cheapest or least utilized. With a regional endpoint, on the other hand, the region is fixed in the name of the address. The developer makes the decision, not the provider.

You can imagine this like branches of a bakery chain. The central phone number connects you to wherever someone happens to pick up. If you call the branch on your street directly, you know exactly who answers the phone. The price for this is commitment: that branch must be open, otherwise you reach no one.

A common misconception is that a regional endpoint automatically keeps everything within the region. This applies to the actual processing and storage. Billing data, logs, or abuse checks can still run elsewhere. What exactly stays where is stated in the provider’s terms and differs from service to service.

Where regions show up in everyday life and in the news

Regions are most visible with cloud providers, meaning companies that rent out computing power. Amazon Web Services, Microsoft Azure, and Google Cloud each operate dozens of regions worldwide. Anyone who creates an account there must select a region for almost every service. This choice also affects the price, since electricity and land don’t cost the same everywhere.

The topic has also arrived in AI models. Providers of language models are now explicitly advertising with European endpoints. The background is the dispute over whether European users' data is safe with US companies. For business customers, an endpoint in the EU is often the condition for signing a contract at all.

As a private individual, you rarely notice this directly. Indirectly, you encounter it when a service launches in Europe later than in the US. Often the appropriate region is simply still missing. In business news, you read about the same thing under the keyword digital sovereignty.