My problem w/ pip is that it is unsound. It's particularly problematic that it works for simple projects and not hard to live with for complex projects with a single dev, but if you have a few freshers working on a complex project you will always be helping them unbreak their builds and sometimes their whole Python installation.
The Python community is chasing new shinies that are 98% correct (e.g. Poetry... which you'd think wouldn't be so damn slow) instead of insisting on 100% which is usually cheaper and faster than the interminable process of pushing a bubble around under the rug.
I had my own problems w/ conda back in the day but got back into it recently because today you can be up and running w/ transformer models on the GPU in conda in minutes, which is about 3000x faster than it was back in the day. I sketched out a design for a correct package manager for Python that only worked on wheels. Packages based on eggs would have to be built into wheels... It turns out that you pretty much have to do the same thing if you need to install pip packages into a conda installation and have it work 100% of the time.