Databricks partners with OpenAI to boost AI development
GPT-5 is now the primary model for the data platform vendor's users. In addition, the partners will collaborate to improve model performance for real-world applications.
Databricks on Thursday revealed that it entered a multi-year, $100 million partnership with OpenAI to make OpenAI's models natively available within Databricks' Data Intelligence Platform and its Agent Bricks ecosystem for AI development.
In addition to making OpenAI's large language models (LLMs), such as GPT-5, available to Databricks users, the partnership includes a collaboration between the vendors to continually improve Open AI models for real-world enterprise applications. GPT-5, meanwhile, now becomes the primary model for Databricks users when developing AI tools.
While the technological integration will be important for Databricks customers, the partnership and collaboration aspects of the deal are perhaps more significant, according to Stephen Catanzano, an analyst at Enterprise Strategy Group, now part of Omdia.
"The partnership is notable since, according to [Databricks], it's OpenAI's first formal integration with a business-focused data platform vendor," he said. "The $100 million commitment suggests this goes beyond typical technical integration to potentially create differentiated AI experiences for Databricks users."
From a technical perspective, however, the integration between Databricks and OpenAI is not unique, Catanzano continued.
In addition to OpenAI's models, Databricks natively supports its own models as well as models from Anthropic, Google and Meta. OpenAI, meanwhile, is already natively supported on platforms including Snowflake, AWS, Google Cloud and Microsoft, among others.
"I wouldn't say it's very significant since OpenAI will partner with anyone, and Databricks partners with all LLMs," Catanzano said. "I can get OpenAI on AWS, Google and everywhere else and don't see this as unique except Databricks is now integrating it so its 20,000 users get easier access."
The partnership
Based in San Francisco, Databricks pioneered the data lakehouse format for data management.
The partnership is notable since, according to [Databricks], it's OpenAI's first formal integration with a business-focused data platform vendor. The $100 million commitment suggests this goes beyond typical technical integration.
Stephen CatanzanoAnalyst, Enterprise Strategy Group
After OpenAI's November 2022 launch of ChatGPT markedly improved generative AI (GenAI) technology, many enterprises increased their investments in AI development. In response, data management vendors such as Databricks and rival Snowflake -- along with tech giants including AWS and Microsoft as well as specialists such as Informatica and Qlik -- added ecosystems that streamline and simplify developing AI tools.
Since early 2024, much of the AI development focus has been on agents, which are applications that have reasoning capabilities and contextual awareness that enable them to act autonomously. Agent Bricks, first introduced in June, is the latest advancement in Databricks' AI development environment.
Given that the partnership between Databricks and OpenAI aims to help Databricks customers develop agents and other AI tools, it provides value, according to Kevin Petrie, an analyst at BARC U.S.
All AI initiatives require data operations, model operations and development operations -- which now includes agent operations -- to work together, he noted. Making OpenAI's models natively available in Databricks' development environment will aid that synchronization.
"This partnership makes it easier for data, AI and developer teams to collaboratively optimize their data pipelines, AI models and agentic applications," Petrie said. "And organizations need all the help they can get because it's complicated and difficult to weave AI into proprietary business processes."
While the partnership will likely aid users as they develop new AI tools, it was those users who provided Databricks with the impetus for collaborating with OpenAI and adding native availability of OpenAI's models, according to Hanlin Tang, chief technology officer for neural networks at Databricks.
"We've seen overwhelming demand from customers … to natively use OpenAI models on the Databricks Data Intelligence Platform and our flagship AI product Agent Bricks," he said.
Meanwhile, native availability of OpenAI models could help Databricks' partnership with OpenAI stand out from OpenAI's deals with Snowflake and other Databricks competitors, according to Catanzano.
Native availability adds deeper integration than simply making a model available in a development environment. Although many of Databricks' peers also provide direct access to OpenAI's models, the way each vendor makes models natively available can differ. Databricks has the opportunity to embed OpenAI's models into features, such as its Unity Catalog for governance, and make the models easy to integrate with customers' proprietary data.
"The integrations and embeddings will be the differentiator," Catanzano said. "What Databricks does to make it easier to use its data and governance capabilities with OpenAI to support customer use cases is the biggest value. The same is true for Snowflake and others."
Ultimately, vendors such as Databricks and Snowflake are racing to create the perfect data platform, he continued.
"This is part of the race that all of the clouds and lakehouses are in," Catanzano said. "If you can get the first AI use cases, you have that customer for the long run. Each is adding more capabilities, making acquisitions and more so that they can have it all for you."
Looking ahead
As Databricks plots out the final months of 2025, its focus remains on simplifying AI development, according to Tang.
"We want to make it easier for all our customers to build AI apps and agents that accurately reason on their enterprise data," he said. "Upcoming innovations will help companies achieve high accuracy while maintaining flexibility and governance."
Catanzano, meanwhile, predicted that Databricks will continue its aggressive acquisition strategy.
While internal product development has been one part building an AI development environment, acquisitions such as buying MosaicML for $1.3 billion in July 2023, Tabular for $1 billion in June 2024 and Neon for $1 billion in May all added capabilities to simplify building AI tools.
"I think [Databricks] will continue to acquire more solutions," Catanzano said. "OpenAI is one piece of building enterprise AI solutions with enterprise data. I think they will acquire a vector database company, and an AI development platform at some point to get closer to being a full platform."
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.
