Also by the time this Mac Pro goes vintage the number of new SSD NAND modules probably won't be that big.
Don't disagree but it's not about size of the SSD!
I could have phrased that differently but stripping out the prepositional phrase.
"Also by the time this Mac Pro goes vintage the number .... probably won't be that big. "
It isn't about the capacity size of the SSD modules. It is about the number of SSD modules. These are 'spare parts'. By the end of the service lifetime the number of new spare parts will have sunk to a relatively small number of do accurate forecasting.
Apple recently added an "special exception" policy to their Vintage/Obsolete coverage where they'll do repairs after the windows if the parts just happen to be available. ( e.g., the deployed component service failure rate was slower than expected. Or some special corner case popped up and they have to restock and bump the supply along the way.). However, Apple's general strategy is to not hold extensive inventories and to run critically low on parts at the end of the service lifetime they have explicitly announced they are covering.
Remember these systems cannot boot without a paired T2/Apple SSD module in them (even not from other media: the MP7,1 loads its boot code from the Apple SSD, and only from there - it cannot boot from any medium if the SSD isn't in it and working).
Which has exceedingly little to do with "upgrades" of the drives. This will also lead to more repairs of "worn out" drives during the Apple covered service lifetime. Which means the inventory of spare parts that Apple sets aside will even
more likely be small by the end of the active service lifetime. if there is a higher need for spare parts, then you probably should prematurely deplete that supply chasing capacity upgrades that could be readily provisioned on another drive.
Should Apple keep an extra "Plan B" emergency reserve of 256GB modules on hand in case their projections are way off ? Probably. Somewhere before they shutdown production orders for spare parts they add some more 256GB modules so if they can coast past their operational lifetime targets with some options to "degrade" folks to keep the boot process still working. That is substantively different than what most folks are pointing at though ( capacity upgrades later). [ The more common representative "downgrade' case so far is folks who "ordered wrong" and want to backpedal. ]
If the T2's SSD drops into a "no more writes" mode it can still read the boot firmware. The state that folks are extremely agitated about really hasn't been demonstrated to be a common occurrence state at all. What would need to happen is that the SSD controller drives the NANDs off a cliff in some way that somehow destroys the firmware at the precisely the same time.
Apple should provide better "drive health" diagnostics for the T2's SSD. Again though, better, accurate proactive diagnostics should lead to higher rates of repair of the modules (because it is an Apple part service solution). So once more the low term levels of inventory at the end will lkely be smaller as the draw down rate will get incrementally higher.
Better drive health would allow users to pick the time when the stop flogging the T2's SSD like an old mule and shift day-to-day usage to another drive. The folks pushing 8-12 years out will need to pick a point in the future. But a >= 1TB drive that is only used for maintenance ( e.g. , a 120GB partition with a minimal macOS instance and the rest of the rest of the drive unformatted and used. ) should in a very high number of cases last for years. There is gobs of spare space fro the SSD controller to put a extremely small amount of write changes into. And read-only from a NAND is extremely unlikely going to cause any failures at all.
So, if you look at today's MP5.1 population: how many of those vintage machines still have their original Apple supplied bootdisk in them (even if they don't boot from it), how many still have that disk in a working condition ?
I know HDD and SSD's don't have the same lifecycle, but it's not because SSD don't have moving parts that they don't wear out, quite the contrary.
The 5,1 isn't vintage; it is obsolete status. If these 5,1 drive failures largely happened inside the pre-vintage/obsolete window then they would be replaced by new drives later in their overall system's lifetime. An older system on a newer drive is more probably going to last well past the vintage/obsolete date than the ones that didn't have a failure. This "huge" failure rate inside the coverage window you are pointing at only creates
more systems at the end that will coast past the vintage/obsolete date; not fewer.
The problematical ones would be the original ones that had zero failures over the 8-9 year span.
Also past the Applecare coverage window most of these drive failures would not be reported to Apple. So their knowledge of the failure rate of their drives over the extended period would be incomplete. ( they could close the holes there with some configuration surveys of older systems down to the drive top metadata level. Once again though if all of that replace is driven into Apple's spare parts inventory it will need to be larger and not depleted during the pre vintage/obsolete status. ( so selling off stuff sooner that may need later isn't a good idea when the production of new components is limited (or at turned off stage.). )
IMHO, Apple probably doesn't have a solid idea of how many of these they will need in several years. They have an opportunity to accumulate better knowledge over the next 2-5 years but for not they shouldn't be overly confident they know what is going to happen with replacement rates over the whole service lifetime for this system. And folks outside of Apple probably are even less aware ( don't have access to iMac Pro replaement rate data that has already come in and nor the new Mac Pro data or most of the rest of the Mac t2 based soldered line up that has had post-termination diagnostics applied. )
This will affect 2nd hand value of the MP7,1 to those that fear wear of SSDs and a risk that it can't be fix rendering the entire motherboard/SDD combo completely useless once the system is declared vintage by Apple.
The 2nd hand value after Apple has put it into obsolete status? That is already going to be pretty bad.
"fear of wear" would be counterbalanced by better diagnostic reports (more information to make a rational decision on as opposed to a fearful one based on little (not enough) information.
Finally, "wild, wild west" pairing is probably gong to lead to less information. No more. There is a pretty good chance the way all data on the NAND chips is encrypted that all of the wear level metadata on the drive is destroyed in a T2 controller - NAND module pairing. That has no impact when the NAND module is brand new. ( there may be some factory "does it work" test wear but that is negligible. ). Opening it up to pairing by anybody will open the door to folks who will pair up an older worn drive from a scrapped Mac Pro to a system and then selling as a "only used by a little old lady from Pasadena to drive to church on Sundays" drive. Pragmatically using the car analogy the odometer will be rolled back to zero. Brand new modules and 'almost worn out' modules both presenting an 'odometer' reading of zero is actually overall less information for used buyers than more. That would contribute to creating more 'fear of wear' fear more so than if know that the module odometer reading is accurate or not.
These are not self contained SSD drives. They are just subcomponents of a complete SSD. If took a normal SSD drive and soldered off a NAND chip and put another on the integrity of that drive would be likely be compromised because the metadata about the drive is stored in the NAND chips also.
