SpaceX to buy Cursor for $60B

The clearest answer people landed on is that Cursor is being bought for three things at once. First, it has real enterprise adoption and a broad developer footprint in a market where most coding harnesses are interchangeable on the surface. Second, it has a stream of valuable training signals from real coding sessions, accepted and rejected edits, prompts, plans, and workflows. Third, it gives xAI a distribution channel into coding, where Grok has lagged Anthropic and OpenAI. A lot of commenters treated the IDE itself as almost incidental. The asset is the data and customer relationship sitting between developers and frontier models. That said, there was no consensus that Cursor is worth anything close to $60 billion as a standalone business. Many saw the price as a function of SpaceX's inflated stock rather than Cursor's moat. Several people pointed out that Cursor's revenue is tied to reselling expensive model access, often with thin or negative margins, and that the coding harness market is already crowded with Claude Code, Codex, Zed, OpenCode, Pi, Cline, and others. The practical read was that Cursor found an exit before the model vendors fully squeezed the middleman. On the product side, the comments were much more mixed than the headline suggests. A sizable group said they had already moved from Cursor to Claude Code or Codex, often keeping their existing editor and using the model directly. Their reasons were cost, speed, stability, and the feeling that once top models got good enough, the harness mattered less. Another large group pushed back hard on the idea that Cursor is obsolete. They said Cursor still wins on plan mode, model switching, cloud agents, bug review, autocomplete, UI inspection, remote workflows, and enterprise administration. Composer 2.5 in particular got real praise as a fast and cheap coding model, even from people who would still reach for Opus or GPT-5.5 on harder tasks. A deeper thread ran underneath the tool comparisons. People using these systems successfully are not just pressing a button. They are supplying context, shaping plans, encoding local rules in AGENTS.md or similar files, and knowing when to keep the model on a short leash. Several comments argued that coding agents are exposing a real skill gap in engineering. Strong developers can model a system clearly, explain hidden business logic, and anticipate what the model does not know. Weak developers let the agent drift. That is why some see "prompt engineering" as fading and others think it is simply becoming normal engineering communication. The overall mood was cynical about valuation, skeptical of Musk, and divided on Cursor itself. Plenty of people said they would cancel on principle or because they do not trust xAI with private code and usage data. But there was also a practical undercurrent that this acquisition makes strategic sense inside the current AI race. If you believe the near-term contest is won by whoever combines compute, distribution, and proprietary training signals fastest, then this is not about buying a text editor. It is about buying a lane into software development before coding becomes just another model battleground.