I will take the best estimate I can find and try to have the numbers represent the facts at the end of the year.
Similar to Sam's original comment, a decade ago we saw 10-15% of adults on antidepressants.
Even in 2023 we're looking at an average of 17% usage (by gender: 24% of women)
I would expect usage for GLP-1s to surge early, see some churn (injections, side effects) and then continue to grow slowly until something better comes along.
So: how likely is it that the something better is an easier to take GLP-1? How likely is it that it's something else entirely (eg something easy to discover with alphafold n)? If it's something else entirely, it would probably stop growth in usage of GLP-1s, but not cut them altogether unless there's a very bad long term side effect of GLP-1s (something I don't expect to see, ie <5% chance). Those tails seem equally likely, we're currently at 6% or something. It's notable that up to 12% have tried (that means half of all users stopped)
Is anyone trading this actually on GLP-1 drugs, or talking with someone who is?
My current thinking is that GLP-1 drugs will be more popular than statins are, and statins are used by about 15% of US adults