If you Group the players, systematically power down one member of the Group and then play for a while. While Mother Nature is never fair, it would be unusual for all of the units to fail at the same time — unless there was some sort of catastrophic event, such as a nearby lightning strike or other power “event”.

Also, systematically pick a different player as you start to build the Group.

Failing memory in Amps had me quite worried, seeing they are Connect Amps, less worried.

An earlier topic here discussed moving Connects and Connect Amps to the S1 controller / firmware as it was easier on the internal memory.

Folks don’t appreciate the huge advances made in memory over the last few years, not just size but durability. Hopefully all Sonos Amps have enough spare memory to deal with failures and enough durability that the failures will be longer in coming.

As already said, you would have to be unlucky for all of them to develop a hardware fault at the same time but… having said that, the older ZonePlayer/Connect generation of units had power supplies that were not especially reliable - I probably rebuilt half a dozen of them when I owned ZonePlayers.

Failing memory sounds possible, but you wouldn’t expect to see anything other than a small number of failures. Typically they conform to the well-known electronic component failure pattern, where you get most failures early in life and then they will go on for many years without issues. Generally the worst component by a long, long way is the electrolytic capacitor - which explains why the majority of failures are in the power supplies.

As suggested, you may find that taking just one or two out of the system solves the problem - though of course it’s time consuming to diagnose a problem in that way… Another option would be to try reducing the reliance of the system on wifi, by hardwiring more of them directly via Ethernet.

The Sonos memory issues are not in RAM, which does follow the curve you mentioned but in the re-programmable memory used for the operating system and persistent data. While it follows the same failure curve initially it also has a different failure mode when cells can no longer accept another cycle of new data and have to be discarded from the available pool.

Excellent discussions of the details in the old topics.