Mistral AI and Cloudera Partner on Sovereign Enterprise AI

Mistral AI and Cloudera are partnering to bring customizable AI models to enterprise data, as detailed in a Mistral blog post. The integration will let customers run inference through Cloudera's hybrid data platform while keeping models and data within infrastructure they control.

Companies will be able to deploy Mistral models across private clouds, public clouds, on premises and fully air gapped environments. The partnership also covers training custom models against proprietary data inside controlled environments.

Abhas Ricky, Cloudera's chief business officer and general manager for applied AI, said general purpose models are the starting point rather than the finish line, and that the real advantage comes from models trained on decades of proprietary data, the loan decisions, production runs and network telemetry that no one else has. That is the substitution the partnership is selling: owning a specialised model rather than renting a general one.

Kamal Brar, Mistral's SVP of partnerships and alliances, puts the scale of what the deal reaches at 30 exabytes of customer managed data running on Cloudera's platform.

Customers can choose where training and inference run, including the infrastructure and the jurisdiction, and can adapt and own models on open weights while managing governance and updates in their own environments. Mistral's announcement puts no value on the partnership.