Sovereign AI¶
UCloud features a built-in AI inference for running large language models on research data.
UCloud AI provides access to open-weight models through browser-based chat with a university login or third-party applications using an OpenAI-compatible API key.
Models, data, storage, and applications stay together in one UCloud project, where files can be used for inference without leaving the project. Interactive sessions and scheduled jobs use shared GPU capacity.
UCloud AI is a sovereign AI platform because the models it serves run on-premises on the Bitten AI supercomputer in Sønderborg, keeping inference in Denmark and under Danish law. Research data is never used for model training.