It's might hurt bitbucket more.
I'm in the "lovely" position of using github (for opensource), gitlab (on-prem hosted), bitbucket (because we were using this before we were acquired, paid-for-private stuffs); oh, and some of the stuff on github is mirrored into gitlab.com.
We were all about (pre-acquisition) "use my own infrastructure" for doing the CI/CD; whether that's circleci/appveyor/travis-ci/ my very own jenkins or whatever. In fact I have circle-ci/appveyor/github-actions/travis-ci setup for most of our opensource projects to at least run tests: mainly because we can, but also because using multiple pipelines mean that when we have intermittent network-y style failures we can eyeball the PR and still approve it. If 3 out of 4 work, then it's probably an isolated problem that will eventually go-away... if none of them work, then we know it's a regression..
The org itself is in the process of standardising on gitlab on-prem; which means that we were going to move bitbucket into gitlab on prem; but now, we probably won't bother, we'll just move all the existing BB repos into github as private repos, since the opensource stuff will have to live somewhere...
Atlassian can't ignore github or gitlab, so jira/trello will always almost certainly have connectors for them; bitbucket may end up becoming unloved.