Scramjets require the engine to already be going pretty fast, so they can't be a first stage engine. Then, with a rocket, most of the speed to get into orbit is built up after getting out to essentially vacuum, so, again, a scramjet is not that useful.

At an abstract level, it's the same issue as with aerospikes, the idea has too many caveats, too much expense and not enough benefit to justify the relatively small improvements in efficiency.