OpenAI told Cursor on Friday it will stop supplying its models to the AI coding tool, effective November 12, now that SpaceX owns the company. [1] "We're ending our partnership with Cursor following its acquisition by SpaceX," OpenAI wrote in a post announcing the decision, adding that it "cannot be confident" SpaceX will honor the contract's terms given Elon Musk's history of legal disputes with OpenAI. [1][2]

This is the second consecutive edition in which a Musk-adjacent compute move has forced a reaction from a rival lab — this paper covered Thursday how Musk's SpaceX-Nvidia orbital data center plan reshaped the AI infrastructure conversation, and Friday's Cursor cutoff extends the same pattern in a different direction: a Musk acquisition triggering a competitor's retreat rather than an expansion.

Cursor's access to OpenAI's GPT models will end under OpenAI's proposal on November 12, giving developers roughly ten weeks to migrate. [1] OpenAI's statement to Cursor's users struck an apologetic note even as it announced the cutoff: "We know that the people most affected by this decision are the developers who rely on OpenAI models in Cursor. We care." [2] SpaceX completed its roughly $60 billion acquisition of Cursor on August 14, closing a deal years in the making and years documented in this paper's own coverage of Cursor's valuation climb from a $50 billion coding-seat rush to a SpaceX-controlled asset. [1]

Anthropic wasted no hours capitalizing. Within the same day, the company said it would increase Claude's compute allocation inside Cursor to absorb whatever demand OpenAI's exit leaves behind, framing itself as a stable partner despite competing directly with both OpenAI and, now, SpaceX-owned Cursor's underlying business. [3] "Cursor has been a trusted partner of Anthropic since Sonnet 3.5," the company said, pledging to "continue to increase compute to support Claude" inside the product. [3] The offer costs Anthropic almost nothing — it already competes for the same developer base — and lets it look accommodating exactly where OpenAI chose to look defensive.

Reuters' coverage treats the decision as straightforward IP-risk management: a company protecting proprietary model weights and usage data from a competitor that now, through Cursor, has commercial access to some of OpenAI's technology stack, delivered in the context of an escalating feud with Musk that predates the acquisition by years. [1] That framing is defensible on its own terms — OpenAI has genuine reason to worry about a Musk-controlled company's incentives around its models. But it also elides the commercial cost: OpenAI is voluntarily surrendering a share of the coding-assistant market it fought hard to win, on a ten-week runway, to a rival that will happily take the business without having to compete for it on product terms.

The Musk-Altman rivalry supplies an obvious motive, and neither side has bothered to hide it. OpenAI's blog post cited "a history of breaking contracts" without naming specific instances; Musk has sued OpenAI multiple times over its transition away from nonprofit status, and Cursor's new ownership gives him a fresh venue to extend that fight by proxy. What is missing from the coverage on both sides is any acknowledgment that this is a fight OpenAI cannot win commercially. Cutting off model access does not weaken SpaceX's balance sheet or dent Musk's ambitions; it hands developer relationships and revenue to Anthropic while OpenAI's own justification remains a prediction about future bad faith rather than a documented breach.

Cursor itself has said little publicly beyond confirming it received the notice, leaving its roughly ten weeks of runway to look for alternatives — a list that now includes Anthropic's expanded offer and, presumably, whatever coding-focused models xAI or SpaceX itself might eventually supply. For a product built on giving developers a choice of underlying models, losing one of the two largest labs as a supplier is a real degradation, dressed up in corporate statements from both companies as everyone behaving reasonably.

The practical math makes OpenAI's position harder to defend the longer the ten-week window runs. Cursor's product has never been model-exclusive; developers routinely toggle between GPT, Claude and other underlying models inside the same editor depending on the task, which means the cutoff does not remove Cursor from the market or meaningfully damage SpaceX — it simply removes one menu option that developers were already free to swap away from at will. Anthropic's expanded compute commitment costs it nothing it was not already spending to compete for the same coding-assistant users, and every Cursor developer who defaults to Claude during the transition window is a data point OpenAI handed to a rival for free, in service of a grievance about contract terms that has not yet produced a lawsuit, a breach finding, or anything beyond a company's own stated prediction about what Musk might do next.

-- THEO KAPLAN, San Francisco