OpenAI Pulls Its AI Models From SpaceX-Owned Cursor

OpenAI said Friday night that it will terminate its contract to provide its AI models to coding assistant Cursor following Cursor’s $60 billion acquisition by Elon Musk’s SpaceX. The move escalates the high-profile feud between OpenAI CEO Sam Altman and Musk: OpenAI’s post explicitly cite a lack of trust stemming from alleged past contract violations by Musk-owned entities like X and xAI. (Musk previously said xAI used OpenAI models to help train xAI models.)

“I couldn’t care less,” Musk said in a reply on X.

The decision could shield OpenAI’s next flagship model, Astra, from being used by Musk’s team to develop new models. For his part, Cursor co-founder Michael Truell said in a post on X that OpenAI represents just 5% of Cursor’s traffic and lamented that Cursor had “trusted their platform to be neutral….” (Cursor is generating around $4 billion in annualized revenue overall.)

Anthropic didn’t miss a chance to dunk on its rival, with co-founder Tom Brown saying on X that his company would continue providing Claude models to Cursor. What he didn’t say is that Cursor customers’ use of Anthropic models drives a meaningful amount of revenue to Anthropic, and Anthropic recently became a large renter of servers from SpaceX data centers.

Some critics pointed out the hypocrisy of Brown’s tone, given that last year Anthropic rushed to cut off AI coding assistant Windsurf from having direct access to Claude models because of reports that OpenAI was buying the startup behind the assistant. (Cognition, another AI coding assistant startup, ended up buying Windsurf instead.)

Existing OpenAI models will be permanently pulled from the Cursor platform on Nov. 12. The move is somewhat ironic that OpenAI led a seed investment round in Cursor’s parent company three years ago