If we accept as probable that we won't see L4 until after Tesla launches its own SoC/board presumably as part of the platform upon which such niftiness as L4 can exist, and if we accept that such a board is at least a year away, I wonder how much of the insurance discussion concerning FSD is even relevant. Neverminding that to some extent, 5G (ideally to then be supplanted by Starlink 10 years hence) would probably be needed to manage the additional data requirements...
As it stands today, only one company (VW Group - see Audi) has even L3 AP in a form that a customer will be able to actually buy and use.
Soooooo... considering that, along with any responsible person's caveat concerning Elon time and tweet parsing, does anyone credibly think that we'll see *any* FSD features this year? I don't doubt that we'll see the first EAP features (thereby elevating the E in EAP from its current decidedly silent status), but c'mon, folks. v9.0 on August 31st will beget v9.1 before the holidays, and then something interesting in late Jan/early Feb after it's been sanity-checked after the usual year-end madnezz.
In fairness, if you just read Elon's tweets for what he didn't type as much as for what he did, nowhere in said tweet did he suggest that all of FSD will be at hand this year or even next anyway - and that's an improvement in Elonspeak right there, or at least in our abilities to parse it.
Seems that the insurance industry has plenty of time to reconcile the benefits of driver assist versus the potential costs. Remember, these are the same scrimey bastahds ^H^H^H^H^H fine upstanding members of society that started charging more to insure SUVs because while safer for the occupants of said SUVs, they were causing too much damage to those vehicles impacted during collisions. Any way to make a buck, and that will not change anytime soon.