Discussion about this post

User's avatar
Nan Deng's avatar

The 8% overlap between ChatGPT and Claude subscribers is the number I keep coming back to — it suggests these are becoming habit-level installs rather than interchangeable utilities, which makes the gross-additions-vs-churn split matter more than the traffic rank. Since the spending data is a panel rather than a census, I'd love to see the power-law claim cross-checked against something less selection-biased, like the state-level subscriber saturation you already track.

Avni Shah's avatar

Good read. Useful information.

5 more comments...

Ready for more?