I was actually thinking of making a topic on this very thing a few days ago.
The audience for single-player linear games on average (not just the break-out hits) is about the same as ever, they grow at a slow rate as new people coming into games mainly go into multiplayer titles like Fortnight, PUBG, League, Overwatch, etc. Single-Player grows a lot slower, but the problem is the cost of making these games to an industry standard has escalated in price insanely in the last 10 years. Remedy and a few other studios commented on this previously, making a single-player AAA game now is 2-3x more expensive than it was 10 years ago, and the actual audience and performance for these games haven't increased by a whole lot to make up for the inflated cost of these games, and the actual price these are sold at has remained the same.
Then add to this that most people don't actually complete single player games, the number of people who complete these games are sadly low, like most people who see an ending of a single player game are only about 10-20% of the people who play the game usually, and not just bad single-player games, even the really good ones.
I think her two suggestions are quite good, making shorter single-player games at a cheaper pricepoint is one thing I think these will head towards, both as players right now aren't finishing single-player games and their bloated cost to make, so focusing on shorter experiences frankly is what will probably occur in the near future as the market has spoken, and this is apparently the direction it wants to move in if you aren't some big open world/multiplayer game. Some individual people may not like this at all, but frankly if the previous size of single-player games are getting too expensive to be able to hold its own, and most people aren't actually finishing these games, all signs point towards shorter experiences.
And I do think a game subscription service is on the horizon as well to support these shorter games as well, I think one will happen before the other, but I get the feeling such a service is almost a certainty at this point somewhere in the future. I don't think we're QUITE there yet, some other technologies need to advance for this to be a viable thing, but I am almost certain it'll occur somewhere in the future.