Sovereignty (data / model)

One-line definition

The principle that AI inference data should stay in the user's chosen jurisdiction. Open weights make sovereignty technically possible — closed APIs cannot.

Full explainer

Sovereignty in the AI context is the principle that the data flowing through a model — prompts, retrieved context, outputs — should remain in a jurisdiction the user controls, rather than routing through a foreign vendor's infrastructure. Closed APIs cannot satisfy this requirement regardless of contractual assurances: the data physically crosses the vendor's network and is subject to that vendor's legal regime. Open weights make sovereignty technically achievable, because the model can be deployed inside the user's own environment — on-premises, in a domestic cloud, or on a desktop client like Switch. The value of DeepSeek's open-weight releases is therefore not only their benchmark performance but the fact that adopters of comparable capability are no longer forced into U.S. infrastructure dependencies. This is the underlying logic behind the Switch desktop client's value proposition: same DeepSeek-grade capability without routing your data through api.deepseek.com.