Though I think they have a lot of stuff they already needed to solve. Trying to fix the # protocol itself and replace the # server with something more efficient, while porting features over from # (both the server and the client ends) is going to increase the strain on their development resources.

The recent migration from # to # (on Android) is already dissatisfying, as the new client does not have all the capabilities of the old one and also introduced new bugs.

As a regular user of Matrix (who once considered hosting a homeserver), I see this mostly like Mozilla's Eudora mail client announcement several years back: "We already lack the needed resources for this project, so we're going to take on additional responsibilities".