OpenAI’s multi-year partnership with Snowflake helps the generative AI model maker put its technology in front of a different category of users served by the data platform provider and show that the AI model race has narrowed.
OpenAI and cloud data vendor Snowflake on Feb. 2 said they had signed a $200 million agreement to deliver AI model capabilities to Snowflake’s enterprise customers. The deal makes OpenAI models natively available to Snowflake’s global customers in Snowflake Cortex AI, a managed service that integrates large language model and machine learning technology into the Snowflake platform. Snowflake will use OpenAI’s ChatGPT Enterprise internally for its employees.
The partnership shows how OpenAI, seen as behind in the race to deliver AI products and services to businesses, continues to target enterprise customers indirectly by forging relationships with enterprise platform providers. Last month, the model maker revealed a partnership with ServiceNow.
Models and Access
The deal is a testament to consolidation in the AI model market, with numerous players starting to winnow to three leaders: Google, Anthropic, and OpenAI. The commoditization of enterprise-ready models indicates that most enterprise platform providers looking to appeal to their customers will seek relationships with these vendors.
“Companies like Snowflake, they’re all trying to build these partnerships as a way to potentially differentiate in the market,” said Arun Chandrasekaran, an analyst at Gartner. He added that a year ago, there was a general consensus that the more models, the better. SaaS and platform providers did not see the need to partner specifically with model makers because the model market was still open, but that is changing, leading vendors to forge deals with key model players and to integrate the models into their technology.
“It’s a way of signaling to the fact that we have what you want, and more importantly, we have tightly integrated that into our product, into our platform and into your workflows,” Chandrasekaran said.
For OpenAI, its partnerships with Snowflake and ServiceNow also help it to reemphasize to the AI market that its models are still important and as useful and dependable as they were in the beginning, he continued.
“OpenAI is trying to create a wide channel ecosystem by partnering with a lot of these … providers as a way to convey that there is still a lot of premium left in the model and also as a way to get access to enterprise customers,” Chandrasekaran said.
With the Snowflake partnership, OpenAI gets access to new enterprise customers and new roles within the enterprise since Snowflake’s primary focus has been on data engineers, data scientists, and business analysts. In contrast, OpenAI has, up to now, mostly focused on developers, he added.
Making Money
However, the biggest question for OpenAI is how Snowflake will compensate the AI vendor for using its models, said Mark Beccue, an analyst at Omdia, a division of Informa TechTarget.
“Is it trading compute, which would be some kind of circular thing? Is it licensing? Is it revenue share?” Beccue continued. “Is the difference that they’re getting paid?”
While Snowflake and OpenAI have revealed the value of the agreement, it is unclear whether that is in revenue for OpenAI. Moreover, for a company preparing to go public, such as OpenAI, deals like this help, but they do not resolve OpenAI’s profitability problem, Beccue added.
The model maker projects 2029 to be the year it breaks even, with what it expects as $125 billion in revenue and achieves cash flow.
“Only 22 companies on the planet currently have revenue of $125 billion or more,” Beccue said.
To achieve its goal, OpenAI is shifting away from its no-ad strategy to introduce ads on the free tier of ChatGPT, which might be a way to generate cashflow, Beccue continued.
“Ad-supported consumer is the only viable path, and that’s shaky,” he said.
