Cursor enters the AI agent race: System that solves weeks-long bugs in seconds is on the way

Analyst insight

The models Cursor uses are expensive and subject to rapid changes, so the company wants to develop its own models. This will result in an AI code editor that is less dependent on external sources, produces cheaper and faster solutions. AI is costly, and no one can offer it unlimitedly because API costs skyrocket with heavy usage, causing the company to incur losses. Furthermore, Cursor's team-level product goal anticipates that tasks such as team-based workflows (pull requests, code reviews, etc.) will now be handled by Cursor. This strategy leads to a “product ecosystem” race rather than a model race with competitors. Cursor now wants to win the race to “automate the software production process with AI” rather than the race to “write code with AI.” We believe this is the right move because software is not just about writing code; it consists of intensive processes such as debugging, testing, reviewing, refactoring, and deploying. While competitors remain editors, Cursor is developing a fully functional product.