On August 28, 2026, OpenAI published its decision on Cursor after SpaceX acquired the coding tool. The company says it notified SpaceX that it intends to wind down the contract that provides OpenAI models to Cursor, with a proposed shutoff date of November 12, 2026. It says it is giving the maximum notice in that contract so developers can keep access as long as the paper allows. That is a proposed shutoff. It is not a claim that OpenAI models in Cursor are already off.
What OpenAI actually told SpaceX
OpenAI’s wording is specific. It says it cannot be confident that SpaceX will use the technology within OpenAI’s terms of service, based on its experience with Elon Musk’s companies violating contracts. It says that after Musk acquired Twitter — now part of SpaceX — the company broke the terms of OpenAI’s contract, “alongside many others.” It also says that under oath earlier this year, Musk admitted that xAI, now also part of SpaceX, had violated OpenAI’s terms of service, terms which OpenAI says are similar to xAI’s own. Those are OpenAI’s claims in the August 28 post, not findings of this desk.
“Today, we notified SpaceX that we intend to wind down our contract providing OpenAI models to Cursor, with a proposed shutoff date of November 12, 2026.”OpenAI, Our decision on Cursor following its acquisition by SpaceX, August 28, 2026
OpenAI says custom contracts are how it usually works with a large partner like SpaceX, to keep compliance and safety at scale. It says the Cursor agreement gives OpenAI a limited window to cancel after a change of control. It also cites accountability for an upcoming model it names as Astra. Given all of that, it says it will hold the cancellation to the latest date it can while not providing future models to Cursor.
November 12 is proposed, not dark
Do not run the shop as if GPT access in Cursor is already gone. The date in the post is a proposed shutoff. OpenAI says it has worked with Cursor for nearly four years and that the people most affected are developers who rely on OpenAI models there. It says it is ready to go above and beyond to support them in the transition. This briefing does not treat that as a completed cutover.
CNBC reported on August 29 that Cursor CEO Michael Truell said OpenAI models serve about 5% of Cursor user traffic, and that Cursor is speaking with OpenAI. That is Cursor’s line, as quoted by CNBC. It is not an audited share from this desk. CNBC also quoted Anthropic co-founder Tom Brown saying Cursor has been a trusted partner since Sonnet 3.5, that Anthropic will continue to increase compute to support Claude models in Cursor, and that Anthropic is excited for what comes next with them at SpaceX. That is the public report this desk is using for the Anthropic compute line. This briefing does not attach a tweet URL.
What Alberta operators should do before November 12
If a Sherwood Park or Edmonton shop still routes production work through OpenAI models inside Cursor, treat November 12 as a failover deadline, not a surprise. Stand up a second model path now: another IDE, another vendor, or a private stack that does not depend on that one contract. Keep client files, tenders, and credentials on a private path. Do not wait until the proposed shutoff to discover which jobs were still on OpenAI-in-Cursor. BYOK is local Chat and Agent only. Tab, Auto, Cloud, Background Agents, CLI, and API stay on the OpenAI–Cursor contract. Pasting an API key is not the Alberta failover. Failover is pick a default non-OpenAI model in Cursor now; do not hang CI or background agents on OpenAI-in-Cursor.
A second path is AI consulting work, not a prompt pasted into a public thread. If the job has to stay on the desk, it still belongs on a private or on-device route. If the job can move, pick the replacement while OpenAI says access is still up. This desk is not pricing the SpaceX deal. OpenAI’s post does not state a purchase price, so this briefing does not invent one.
The guardrail
Change-of-control language is a contract fact, not a vibe. One vendor inside one IDE is a single point of failure. For AI automation Alberta shops, the useful move is a named backup before November 12, with a private AI security rule that client work does not follow the IDE if the IDE’s model path changes. Opcelerate Neural can map the failover if that is the next step.