Nvidia brought in 30,000 of its engineers to work on the AI assistant Cursor, and the code volume tripled.

NVIDIA triples internal commits thanks to AI integration

*Since all engineers switched to AI‑enabled tools, the number of internal code commits in the company has tripled.*

1. What is used and by whom
- Tool: the integrated development environment *Cursor* from Anysphere

- Number of users: more than 30 000 Nvidia developers

- Scope of application: almost every stage of software creation – from writing and reviewing code to generating comments and quality control

> “Cursor covers nearly the entire development lifecycle. We have created many custom rules that fully automate workflows,” noted Vice President of Development Wei Luio.

2. How AI speeds up work
Task Role Cursor Writing code Automatic generation and verification Debugging Efficient search for rare, persistent bugs and deployment of agents to fix them Project management in Git Extracting context from tickets and documentation; automatic patching with test validation
> “Before Cursor we had other AI tools, but after implementation we noticed a noticeable acceleration in development,” added Luio. He emphasized that the system especially excels at analyzing distributed databases – a task difficult for humans.

3. Impact on staff
- Interns and newcomers quickly master Cursor.

- Experienced developers are freed from routine tasks and focus on more creative work, thereby narrowing the gap between idea and implementation.

> Generative AI is intended specifically for repetitive actions, not to replace human ingenuity.

4. Code quality
- Error: the number of bugs remained at the same level despite a threefold increase in code volume and overall productivity.

- Why it matters: critically important components – graphics processor drivers – depend on code quality that is now partially generated by AI.

In short: Nvidia successfully integrated Cursor into its development ecosystem, leading to significant productivity gains without compromising quality.