Cloud & Compute Infrastructure

French sovereign-cloud operator S3NS made managed model serving and inference generally available on PREMI3NS, supporting open and customer-provided models.

· ComputeLabs Research · from the November 17, 2026 edition

Google’s sovereign-AI announcement states that managed model serving and inference are generally available on PREMI3NS, the S3NS cloud built on Google Cloud Dedicated. This is an available managed service, not simply a planned model offering.

Customers can deploy open models from providers including Mistral and Google through Model Garden, or bring their own models. S3NS operates the environment, which is designed to keep deployed models and prompts inside PREMI3NS under S3NS control.

Google distinguishes this partner-operated, isolated-infrastructure model from Google Cloud Data Boundary on its public cloud and Google Distributed Cloud in customer data centers. Its separate announcement that Gemini is “coming to” Google Cloud Dedicated should not be confused with the already generally available PREMI3NS serving capability.

  • S3NS
  • PREMI3NS

All 20 stories from November 17, 2026