* Slow start up times (a second or half a second) for anything that includes the python standard library.
* Avoiding the standard library means writing javascript helper functions so you end up needing to know and write javascript anyway. In some ways Brython feels like "Python for javascript programmers"
* Integration with browser for error messages isn't as convenient as for javascript where I can "click into" the js file in dev tools.
* Pierre is brilliant but it is clear that it is a labor of love. He doesn't want corporate sponsorship or anything that would make it a job for him. He works tirelessly on Brython, the man is a machine, but it really is just him having fun. What happens when its not fun any more? There is no obvious succession. The lack of corporate sponsorship/marketing means that the community is small, there's no "virtuous cycle" of adoption.
I hope the above doesn't mis-represent Pierre, this is my perception only.
Lastly, in recent years many people came to python for the scientific/ML libraries which don't work with Brython. To these people, a python without numpy is no python at all?