Cloudera and Mistral partner to run AI models directly on enterprise data
Key Points
- Cloudera will integrate Mistral's AI models and tools with its hybrid data and AI platform, supporting deployments across cloud, on-premises, edge and air-gapped environments.
- Enterprises will be able to run private inference and customize Mistral models with proprietary data while keeping workloads within their existing governance boundaries.
- The partnership will include Mistral Forge and span reasoning, coding, document intelligence, chat and voice models, with the companies also planning work on edge AI.
Cloudera and Mistral announced a partnership on September 10, 2026, to integrate Mistral's AI models with Cloudera's hybrid data platform. The deal lets enterprises run Mistral's models, including reasoning, chat, coding, document intelligence, and voice tools, against their own governed data without moving that data to an external service. According to Mistral SVP of Partnerships Kamal Brar, the integration will roll out over the coming weeks.
What materially changed
Enterprises using Cloudera's platform can now deploy and customize Mistral's models across cloud, on-premises, edge, sovereign, and fully air-gapped environments, keeping sensitive data inside their existing security perimeter. Through Mistral Forge, a system for training models on proprietary enterprise data, Cloudera customers will be able to fine-tune Mistral's models against Cloudera's data estate, described in the announcement as roughly 30 exabytes. One exabyte equals one billion gigabytes. This is a shift from consuming Mistral's models purely through public APIs to running and training them inside a customer's own controlled infrastructure.
Context
The partnership follows a funding round led by Samsung Electronics that pushed Mistral's valuation above $24 billion. It also follows Mistral's expanded collaboration with Microsoft last month and its May partnership with legal AI startup Harvey. Cloudera, founded in 2008 and operating in nearly 100 countries, has said that 65 to 70 percent of its annual recurring revenue comes from regulated industries such as financial services, healthcare, life sciences, and the public sector.
Why it matters
For enterprise AI buyers in regulated industries, the deal adds an option to run frontier models without sending proprietary data outside their own environment, a requirement that has slowed AI adoption in sectors like banking and healthcare. For Mistral, it's a distribution move: "We're getting access to 30 exabytes of really differentiated, proprietary data," said Kamal Brar, and access to Cloudera's regulated-industry customer base gives Mistral a revenue channel beyond direct API sales. The deal also reflects a broader pattern of AI labs partnering with data-infrastructure providers to reach enterprises that need to keep data under their own control, rather than competing solely on public model access.
What happens next
Cloudera and Mistral said the integration will begin rolling out within weeks. The two companies also said they intend to collaborate on deploying AI models at the edge, aimed at environments with limited connectivity or centralized cloud access. Neither company disclosed financial terms of the partnership.