I think it's a mistake to assume the database will always be the size it currently is. Once the public fingerprint submission process is in place I plan on making various tools so that users can easily submit their fingerprints from their own music libraries, allowing us to crowd-source the problem of improving the database.

In addition to this EchoNest have a database of 30 million fingerprints using their old proprietary ENMFP algorithm, it's not unreasonable to assume that at some point in the future they might run these same tracks through the EchoPrint algorithm as well. Alongside that MusicBrainz is also getting involved in trying to improve the EchoPrint system and have their own EchoPrint test server setup (http://echoprint.musicbrainz.org/).

So while the database is too small at the moment, it will grow in time.