Clojure being a Lisp probably exceeded and overachieved its goals when it comes to popularity. Sadly, not a single Lisp probably will ever become a mainstream.
And for the same reasons why Vim and Emacs would never become more popular than IntelliJ and Sublime. People tend to choose something easy to start with, out of the box thing, not something that has a small, simple core yet extremely extensible and rich ecosystem. Most of the innovation in Clojure-space happening outside of Cognitect, driven mostly by individuals.
Look at Clojurescript. It hasn't stalled, au contraire - with every release it's getting better. Pick any large Clojurescript project - PrecursorApp or Circle CI, look under the hood. Compare Re-frame and Fulcro projects with other popular Javascript frameworks. Check out Re-Natal, Lumo and Planck.
Right now Clojurescript probably the most pragmatic choice if you want to build web-apps using a functional language. GHCJS and Purescript (that appeared few years before Clojurescript) - still hasn't quite reached popularity. Elm - which is 5 years old (same as Clojurescript) - still feels a bit experimental. ScalaJS - I don't even want to talk about Scala. Fable/FunScript/WebSharper? Microsoft has failed to convince even its own herd to start using F#. Largely - C# still in use. And those guys prefer Typescript - not really functional. Only ReasonML someday soon may become true competitor to Clojurescript simply because: a) Facebook behind it; b) ain't a Lisp.