Cloudera brings Mistral AI’s frontier models into its secure hybrid data environments

Big data company Cloudera Inc. is pushing to become the go-to partner for enterprise-grade sovereign artificial intelligence workloads after announcing a massive, nine-figure strategic partnership with the French AI model maker Mistral AI SAS.

The partnership is aimed at bringing secure, sovereign AI to wherever enterprises’ data lives, be it in the cloud, on-premises or in hybrid environments, Cloudera said at its EVOLVE26 event in São Paulo today. Mistral will integrate its powerful large language models directly with Cloudera’s hybrid data and AI platform to enable this.

Cloudera says it’s trying to address a major problem for enterprises that operate in highly regulated industries, which have to go above and beyond when it comes to safeguarding their data. Though those enterprises really want to explore how AI can help them, they find it challenging to do so.

To take full advantage of AI, they need to be able to feed their most sensitive and proprietary information to external AI models, but doing this is extremely risky, because it opens up a minefield of security, compliance and data privacy issues. There’s also the issue of costs, which can quickly spiral out of control when relying on public application programming interfaces to access the best AI models.

Cloudera believes it has hit upon a solution, and it’s Mistral that’s helping to make it possible. By bringing Mistral’s powerful frontier models that can do everything from reason and chat to coding and understanding documents and more, Cloudera is giving organizations the option to deploy AI in any environment – be it the public cloud, private servers and more.

Cloudera’s hybrid data platform runs in any environment, and by bringing Mistral’s AI models inside it, it’s giving enterprises a way to run AI on their most sensitive information, without having to send this data to another location. It also eliminates the need to access AI models via an external API. It means companies can run inference, generative AI and agentic workflows in exactly the same place as where their data is, within the confines of Cloudera’s robust security and governance perimeter.

Cloudera Chief Business Officer Abhas Ricky, who also serves as its general manager of Applied AI, said the next step for enterprises is to unlock “specialized intelligence” that can only be created by training models on their own data. “Together with Mistral, we are giving enterprises the ability to run AI where their data lives, customize it with their own intellectual property and maintain control over their data, infrastructure and economics,” he said.

With this partnership, what Cloudera is really providing is data sovereignty, which is hugely important to enterprises when it comes to their most sensitive and valuable data, said NAND Research Chief Analyst Steve McDowell. He said organizations are naturally very worried about uploading their most secretive information to AI providers, because those companies tend to be very opaque about what they’re doing with it.

“Will that data be used to train or fine-tune results that might benefit competitors? Or will there be a damaging information leak?” the analyst asked. “It’s making CISOs very uncomfortable and part of what is driving the adoption of open-weight and locally-hosted models.”

Cloudera, McDowell added, is giving organizations an alternative by letting them use frontier models directly on the data they already trust it with. “It provides a more comforting degree of control for enterprises,” he said. “The collaboration marries Mistral’s capable foundation models with Cloudera’s framework to give enterprises a turnkey solution for AI inference. It’s a strong story.”

Improved economics is a big part of this partnership too, Ricky stressed. When organizations can run inference workloads within their own environments, it means they have full control over the infrastructure those models run on. In addition, this approach paves the way for more AI workloads to run at the network edge for low-latency applications where it’s not practical to use centralized cloud servers.

With regard to Ricky’s concept of proprietary intelligence, the key factor here is Mistral Forge. That’s the model maker’s fine-tuning platform that enables enterprises to enhance its frontier model’s capabilities by training them on their own data. Companies will be able to feed Mistral’s models with petabytes of data, safe in the knowledge that it will never escape the confines of their environment. It means that decades of institutional knowledge and domain expertise can finally be leveraged by AI agents to perform more complex work in highly regulated industries, without ever exposing an organization’s trade secrets, Ricky said.

“Mistral Forge is the most compelling part of this announcement,” McDowell said. “Being able to use enterprise data to create and fine-tune a Mistral model without moving that data is extremely powerful. It brings a known model backed by a known entity, which can appeal to enterprises concerned about the origins of open-weight models, most of which are developed in China. It’s a more flexible approach than we see with OpenAI, Anthropic and Google, which all require you to bring your data to them.”

Mistral’s senior vice president of partnerships and alliances, Kamal Brar, acknowledged this, saying his company wants to give enterprises a way to access its most powerful frontier capabilities without having to risk losing control of their data or IP. Until now, the risk of data leaks has simply been too great for many companies to countenance such an idea, he said. “Cloudera manages some of the world’s most valuable enterprise data estates, making this partnership a powerful opportunity to bring our technology directly to where that data lives,” Brar said. “Together, we can give customers the ability to build AI that reflects their own data and expertise and deploy it securely wherever their business requires.”

Image: SiliconANGLE/Gemini