H Company said Nov. 16, 2026, that its computer-use AI platform is now available on the AWS European Sovereign Cloud, giving European organizations a new location for agent workflows.

According to H, its agents interact with existing enterprise software by clicking, typing and navigating screens.

In its announcement, H says EU customer inference runs on NVIDIA-powered instances and that customer content and customer-created metadata remain in the EU. Model training continues on AWS infrastructure globally.

A separate operating boundary

AWS’s Sept. 16 architecture guide describes the service as a distinct cloud partition, with its own identity system, billing, control plane and endpoints. Its first region is in Brandenburg, Germany.

The guide says credentials from ordinary AWS partitions cannot be used across that boundary. Connections between the two environments instead use network or application interfaces with separately managed credentials. A deployment there therefore involves an operating environment separate from an organization’s existing commercial-AWS accounts.

An earlier infrastructure test

AWS’s Nov. 1 account describes an Oct. 24 exercise in which it disabled its global-network connection and the system managing controlled transfers of operational data. Traffic rerouted over the public internet.

AWS reports that customer workloads continued and service availability was unaffected during the several-hour test. That account concerns infrastructure operation; H’s new announcement describes where its application inference and customer data run.