Snowflake partners with Mistral AI, taking its open LLMs to the data cloud

Today, data cloud major Snowflake announced it has signed a multi-year agreement with Mistral, the Paris-based AI startup that raised Europe’s largest-ever seed round in June 2023 and has since become a rising star in the global AI domain.

Under the partnership, Snowflake said it plans to bring all open large language models built by Mistral into its data cloud, making them directly available to customers looking to build LLM apps. The company also confirmed it is investing in the nine-month-old startup through its corporate venture capital arm. However, the exact amount has not been disclosed.

The deal marks another significant vote of confidence for Mistral, which has been moving forward aggressively by teaming with big tech majors. The company has also been expanding its model line-up at a rapid pace, with its newest Mistral Large model delivering a performance comparable to offerings from OpenAI, Anthropic and Google.

What to expect from Snowflake-Mistral deal?

Since its inception, Snowflake has heavily focused on building the data infrastructure of choice for enterprises, now referred to as the Data Cloud, and enabling a wide range of downstream use cases, including AI and analytics. The company has continually evolved its product to meet the growing demands of customers. Consequently, when customers needed a solution to develop generative AI-powered applications using the data stored on the platform, it launched Cortex, a fully managed service tailored for building LLM apps.

Announced during last year’s Snowday event, Snowflake Cortex provides enterprises using data cloud with a suite of AI building blocks, including open-source LLMs, to analyze data they have on the platform – with security, privacy and governance of Snowflake – and build applications targeting different business-specific use cases. The company started with LLMs for specialized tasks such as sentiment analysis, translation, and summarization as well as Meta’s popular Llama 2. 

Now, with the partnership and investment in Mistral AI, the company is adding another open and highly performant family of models into Cortex for LLM app development. This not only includes the all-new Mistral Large, which sits just behind GPT-4 and outperforms Claude 2, Gemini Pro and GPT 3.5 with native proficiency across five languages and a context window of 32K tokens, but also Mixtral 8x7B and Mistral 7B models.

“By partnering with Mistral AI, Snowflake is putting one of the most powerful LLMs on the market directly in the hands of our customers, empowering every user to build cutting-edge, AI-powered apps with simplicity and scale,” Sridhar Ramaswamy, who recently took over as the CEO of Snowflake, said in a press statement. 

“With Snowflake as the trusted data foundation, we’re transforming how enterprises harness the power of LLMs through Snowflake Cortex so they can cost-effectively address new AI use cases within the security and privacy boundaries of the Data Cloud,” he added.

When reached out by VentureBeat, Snowflake refused to share the amount invested in Mistral or the names of early customers testing the French startup’s models – which are now available in public preview as part of Cortex.

“The models were just made available…so it is too early to drop names but Mistral Large, Mistral AI’s most powerful model, is one that many of our customers that wanted more performant models running in Snowflake were really looking forward to,” Baris Gultekin, Head of Product for AI at Snowflake, told VentureBeat.

Gultekin also confirmed that Google’s all-new Gemma 7B open model is also coming to Cortex alongside Mistral 7B. The goal, he said, is to give customers flexibility with curated access to the top-performing conversational models within their targeted size category. The company will add more models across sizes to build on this work.

“Snowflake is all about flexibility without complexity to really help organizations deliver value quickly and securely with generative AI. To deliver on that vision, we’ll continue to evaluate and update the options that will help us provide our customers with top-performing open and commercial LLMs as serverless functions inside Snowflake Cortex,” he noted. 

That said, while the current focus remains on offering LLMs to customers looking to build gen AI apps, the product head did indicate that the company’s internal engineering teams are considering using LLMs in other areas. He refused to share the specifics of this work though.

Mistral continues to shine with industry partnerships

The deal with Snowflake marks another major win for Mistral, expanding the startup’s reach and cementing its position as a vendor-trusted player in the AI category dominated by OpenAI, Anthropic and Google. 

Just last week, the company secured $16 million investment from Microsoft as part of a partnership that brings its models on the Azure cloud platform. The deal made Mistral only the second company after OpenAI to offer its models on the Microsoft platform. 

The same week also saw Mistral’s separate partnerships with IBM, making the Mistral 8x7B available on WatsonX, and with Perpexplity and Amazon. However, the Jeff Bezos-led technology giant is yet to roll the models on its Bedrock platform.

With Snowflake already on board today, it will be interesting to see what other partnerships can the startup secure to grow its prominence and drive AI use cases across sectors. For context, Databricks, which competes with Snowflake in the data ecosystem, also offers open-source, commercially usable language models to build generative AI apps. The includes models from MosaicML, which the Ali Ghodsi-led data major acquired for $1.3 billion last year.