Author(s): Kashif Mehmood

Originally published on Towards AI.

OpenAI’s models carry 5% of Cursor’s traffic. Anthropic’s Claude runs on 220,000 of SpaceX’s GPUs. Read those two numbers before you read either company’s statement.

On Friday 28 August, OpenAI published a short post explaining that it would stop supplying models to Cursor, the code editor SpaceX finished buying two weeks earlier, with a proposed shutoff date of 12 November. The reason, in OpenAI’s words: “we cannot be confident that SpaceX will use our technology within our terms of service, based on our experience with Elon Musk’s companies violating contracts.”

The article argues that the public feud narrative misses the real driver: each model provider enforces a “change of control” / non-compete clause in direct proportion to how much it needs (or doesn’t need) that customer’s business. It walks through what OpenAI cited—Twitter-data access cut off by Musk and xAI’s distillation admitted under oath—then contrasts Anthropic’s behavior, noting it previously blocked xAI via Cursor and later chose to keep supplying Claude while adding compute capacity tied to SpaceX’s Colossus 1 deal. From these cases (plus earlier similar cutoffs like Anthropic vs. Windsurf), the author develops an “exposure test” (traffic share served versus dependency on the customer’s owner) and concludes that the stated reasons are real, but the stakes are what determine when a vendor chooses to walk away—often at the moment an acquirer can be blamed.

Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI