Tom Brown, co-founder of Anthropic, has announced that the company will continue to supply its Claude models to Cursor while adding more compute behind them. He made the statement within 24 hours of OpenAI informing SpaceX (NASDAQ: SPCX) of its decision to cut Cursor off, in response to Elon Musk's $60 billion purchase of Cursor's parent company.

Two AI labs, two different responses

OpenAI announced plans to stop supplying its models to Cursor on August 28, aiming to end the contractual relationship by November 12, 2026. The company says it is acting reluctantly, but Musk's purchase of Cursor's parent left it no choice.

The Altman-led firm claimed Musk-owned companies have a habit of violating contracts, citing Twitter's violations of its terms, which Musk himself admitted to under oath.

Anthropic did the exact opposite. "Cursor has been a trusted partner of Anthropic since Sonnet 3.5," Brown said. He added that Anthropic would "continue to increase compute to support Claude models in Cursor."

The split reflects how much leverage now sits with the labs that own frontier models. AI coding assistants like Cursor do not build their own foundation models; they resell access to those of OpenAI, Anthropic, Google, and others. That makes their product roadmaps dependent on the licensing decisions of a handful of suppliers — decisions, as this episode shows, that can shift with corporate rivalries as much as with commercial terms.

The $60 billion purchase that kicked things off

SpaceX purchased Anysphere in an all-stock transaction in June, closing the deal on August 14, according to an 8-K filing with the SEC. The filing values Cursor at $60 billion, with a total of 389.3 million shares of SpaceX Class A stock issued. The deal ranks among the largest startup acquisitions to date.

The change in ownership gave OpenAI a legal pathway out of its contract with Cursor. The agreement contained a clause permitting cancellation after a change in ownership, within a specific time frame. OpenAI chose to time the cancellation on the latest date possible under the agreement. OpenAI also based its decision on Astra, an upcoming model that it says raises the bar on who can access its systems.

Musk shrugs it off, and Cursor downplays the loss

Musk gave a blunt response: "I couldn't care less." He went on to call OpenAI's Sam Altman and president Greg Brockman untrustworthy, reiterating his claim that they "stole an open source nonprofit."

The episode is the latest in the long-running feud between Musk and OpenAI. Earlier this year, Musk sued OpenAI for $150 billion, but a federal jury dismissed the case after finding it exceeded the statute of limitations.

Cursor is not losing sleep over OpenAI's actions. Co-founder Michael Truell, now a SpaceX executive, said in an X post that OpenAI's models account for just 5% of Cursor's traffic, while the two sides remain in talks to find a resolution.

Nonetheless, OpenAI's departure may weaken Cursor's pitch to developers as a one-stop shop offering access to any leading AI model. A key question going forward is whether other model providers follow OpenAI's lead or Anthropic's, and whether the SpaceX-owned Cursor retains neutral-gateway status or becomes increasingly tied to models from within Musk's orbit.

Anthropic has set the standard before

Anthropic's decision to stay with Cursor is notable given its track record. In January, Anthropic cut off xAI after its staff used Claude internally via Cursor. The company also cut off OpenAI from its systems just before the launch of GPT-5, saying OpenAI was "a direct violation of our terms of service." This time around, Anthropic is staying put.

The divergent moves by two leading AI labs have split the market into camps, as access to top AI models has become an arms race. For developers, the result is fewer OpenAI options in one widely used tool, Cursor, and the realization that access to any AI model on demand can no longer be taken for granted.

Source: Cryptopolitan