This article from the Washington Post provides a somewhat depressing but mostly accurate look at the state of the original cast album today: recording costs have grown while the recording market has shrunk, and there aren't even many good recording studios left in New York. The result is that whereas big labels used to routinely record Broadway and off-Broadway shows, today it's a niche market (in some ways, even more of a niche market than classical, which has international appeal) and there's no guarantee that any but the big hit shows will get recorded at all.
The article does make things sound a little worse than they are. Hit shows and even shows that aren't hits continue to get recorded; on niche labels, yes, but the recordings exist. Also, there have always been some shows that didn't get recorded. In the days when cast albums were big sellers, shows would get a recording contract before they opened (a record company would invest in the show in exchange for the privilege; Columbia put up a lot of money for My Fair Lady, and RCA did the same to get Fiddler On the Roof). But if the show closed very early, there would be no album. (An exception was Columbia deciding to record Anyone Can Whistle even though the show had already closed; too bad they didn't do the same the following year for a much better show and score, Drat! The Cat!) Today, what was considered a long run in those days is not enough to make back any of the show's money. So the fact that Cry Baby didn't get recorded isn't that strange; its 68-performance run today is the equivalent of a 1960s show that closed after a week.
Still, the obvious point is that there will be fewer shows that get recorded, and shows that are not genuine hits will have a lot of trouble finding the money for a recording. Which is a problem. American musicals didn't have original cast recordings until Oklahoma! (though the Brits had been doing such albums for years), but back then, they didn't need them; musicals were only expected to run the season -- 300 performances was considered a long run -- and what helped it do so was getting lots of pop recordings of the show's hit songs. But as time went on, producers wanted and even needed their shows to last a long time beyond their initial productions; some shows, like Sweeney Todd, lost money at first and finally paid back their investors based on all the post-Broadway productions, amateur rights, etc. A recording may not make much money for the show, but it serves as the show's calling card.
As former RCA head of theatre music Bill Rosenfield notes at the end of the article, he recorded Jason Robert Brown's score fopr an unsuccessful Off-Broadway show. It has since had many subsequent productions -- not because people saw it, but because they heard the recording. A show with no recording will not get revived, because only a few people are familiar with the score; a show with a recording will come to the attention of those who might want to produce it. One of the driving forces behind the 1994 revival of Show Boat was the then-recent release of EMI's big studio recording of the complete score, which called people's attention to the show and to the fact that it was a much richer score (with many variants and opportunities for interpolation of cut numbers) than anyone would have known from the hit tunes alone. And She Loves Me, a truly great musical that ran only 300 performances and lost money, has become a favorite in amateur productions because the recording exists to tell people how wonderful the score is, and leads them to investigate the idea of producing the whole show.
So eventually it seems like cast albums might work the way recordings do for most classical artists these days: not as a source of revenue, but a promotional tool. In the case of musicals, what's really being promoted is the idea that this show is worth producing.