1.6 billion tokens in Cursor: what I actually spent them on

I used 1.6 billion tokens in Cursor in a few months. Most went on asking why things broke and rethinking decisions, not on generating code.

Abdullah Abdul Samad, . A 1 minute read. First posted on LinkedIn in January 2026.

I did not realise how much I was using AI until I saw the number: 1.6 billion tokens. That is what I had spent in Cursor over a few months.

At first the number felt overwhelming. Looking back over how I had used it explained it.

Where the tokens went

Most of them were not spent on “generate code for me”. They went on:

  • Asking why something broke.
  • Rethinking architecture decisions.
  • Debugging things I did not fully understand yet.
  • Rewriting code until it felt right.
  • Learning faster than I ever had before.

What changed

Somewhere along the way my workflow changed. I stopped getting stuck for hours. I stopped hesitating to start something new. I began thinking more like an engineer and less like someone fighting syntax.

Cursor did not make me a better developer overnight. It removed friction from learning and building, and that effect compounds quickly.

The takeaway

AI works best when you treat it as a thinking partner, not a shortcut.