I'm trying to think of real world applications for the semantic web, in 2014, which do not require a browser. Struggling.
Well, you could directly query a SPARQL database, but thats the one scenario where a server still makes sense, to provide some decoupling from the database model and the client.
"Use the right tool for the job". :-) There's no particular reason a "modern web-app" has to run completely in the browser... it still makes perfect sense to put heavyweight process intensive stuff, persistence, business logic, etc. on the server-side. If I'm building an app that uses a lot of "semantic stuff", most of that stuff is, indeed, happening on the server side.
So your answer sounds more like an excuse than a solution to me, sorry.